Disasters – PLOS Currents Disasters http://currents.plos.org/disasters Thu, 08 Nov 2018 15:48:50 +0000 en-US hourly 1 https://wordpress.org/?v=4.5.3 Exploratory Qualitative Study of Fire Preparedness Among High-rise Building Residents http://currents.plos.org/disasters/article/exploratory-qualitative-study-of-fire-preparedness-among-high-rise-building-residents/ http://currents.plos.org/disasters/article/exploratory-qualitative-study-of-fire-preparedness-among-high-rise-building-residents/#respond Fri, 31 Aug 2018 09:45:17 +0000 http://currents.plos.org/disasters/?post_type=article&p=39763 Introduction: Fire hazards are an extreme risk to occupants of high-rise buildings. Little attention has been paid to emergency and evacuation preparedness among people living in high-rise buildings. This paper reports on emergency fire preparedness among residents of a high-rise building that has experienced multiple fires in the past.

Methods: An exploratory qualitative pilot study was conducted using key informant interviews. Six residents participated. Themes on preparedness for fires and emergency evacuation were extracted.

Results: Findings indicated varying levels of preparedness for fires and emergency evacuation among residents. Factors influencing residents’ emergency preparedness included fire risk perception, owner or renter status, and building-level emergency preparedness. Fire alarms were considered to be an ineffective evacuation cue. Severe cues such as seeing fire or smoke were more likely to prompt evacuation. Participants provided a series of suggestions to keep high-rise residents safe during fire emergencies.

Discussion: The study revealed fire preparedness knowledge, decision-making processes, and actual behaviors of residential high-rise occupants who experienced a fire emergency in their building. Main findings of the study are discussed in two themes: influences on fire emergency and evacuation preparedness, and evacuation decision-making and response to fire. Results from this pilot study will be used as the basis for a follow up study involving residents from multiple high-rise buildings.

Keywords: disaster, emergency preparedness, evacuation, fire, hazard, high-rise building, pilot study, qualitative research


Exploratory qualitative study of fire preparedness among high-rise building residents

Currently, over half of the world’s population lives in urban areas, a proportion expected to rise to two-thirds by 2050.1 This global trend portends an increasing number of urban disasters.2 Sustained growth in urban centers is leading to higher concentrations of high-rise buildings, and an amplified risk to people living and working in these structures in the event of an emergency.

Fire hazards are an extreme risk to occupants of high-rise buildings. Between 2009 and 2013, U.S. fire departments responded to an estimated 14,500 structure fires in high-rise buildings per year. The majority of these incidences (62%) occurred in apartments and other multi-family housing. During this period, there was an annual average of 40 civilian deaths, 520 injuries, and $154 million in direct fire damage.3 Specific populations are much more vulnerable to such emergencies, namely the elderly, children, and people with disabilities or other health conditions that impair mobility. Deaths among older adults (> 65 years old) and very young children (< 5 years old) in residential fires are disproportionately higher to their proportion of the U.S. population.4 Persons with mobility impairments are also at a higher risk for injury and death during residential fires.5

Engaging in emergency preparedness is an effective way to reduce the risk of death or injury during crises.6 Factors influencing household and community emergency preparedness have been the focus of many research studies.7, 8 Little attention has been paid to specific factors influencing emergency preparedness for fires in high-rise buildings. Most of what is known on this topic has arisen from research involving office or commercial buildings. Less is known about fire preparedness among people living in residential high-rises. An exploratory qualitative pilot study was conducted to examine fire and evacuation preparedness among residents of a high-rise building that has experienced multiple fire emergencies in the past. Findings from this study are shared in this article, including specific factors influencing high-rise building occupant fire and evacuation preparedness, cues for evacuation, and suggestions to keep residents safe during actual fire emergencies.

All-hazards preparedness in high-rise buildings

High-rise buildings are defined by the National Fire Protection Association3 as buildings greater than 23 m (75 feet) in height from the ground level to the highest floor. The main uses for this type of building are residential (apartments, dormitories or other multi-family housing), office buildings, hotels, and facilities that care for the sick.3 Practicing all-hazards emergency preparedness can improve the likelihood that building occupants remain safe during disasters such as fires and other crises. An all-hazards approach to emergency preparedness “addresses a full range of threats and hazards, including natural, human-caused, and technology-caused.”9 According to the Federal Emergency Management Agency (FEMA)10, basic preparedness strategies common for all types of disasters include getting information about hazards and emergencies affecting families in a given location, developing an emergency plan, assembling a disaster supply kit, learning where to seek shelter from all types of disaster, identifying warning systems and evacuation routes, and practicing and maintaining an emergency plan.

High-rise building residents encounter environmental elements that make planning for all-hazards emergency preparedness unique. Services that people in high-rise buildings depend upon, such as water, electrical, transportation and communication systems, can be disrupted if a disaster damages or destroys critical infrastructure. High-rise building residents must also be able to evacuate quickly in emergency situations. Residents may have to travel long distances to get from the area of residence to safety, especially if they reside at higher elevations. The route to safety may be crowded with others trying to evacuate at the same time. Evacuation routes may be more challenging to navigate for persons with disabilities or other health conditions that impair mobility. During emergencies, evacuation routes may become too dangerous, so residents must also know how to safely shelter in place until first responders arrive.

Commercial high-rise building emergency preparedness

Studies on the evacuation of the World Trade Center (WTC) buildings following the September 11, 2001 terrorist attacks have provided significant understanding of factors influencing preparedness for emergency and evacuation of commercial high-rise buildings under extreme conditions.11, 1213, 14 Research on incidences of high-rise building fires and terrorist attacks has also been informative in describing building occupants’ preparedness for emergencies and evacuation behavior.15, 16, 17 One study developed the Emergency Preparedness Knowledge and Experience Scale, to measure building familiarity and knowledge among WTC evacuees.18 Evacuees scored low on the scale, signifying low levels of knowledge regarding the location of building fire stairs, how to enter the stairwell, or where the stairs lead. Occupants were also unfamiliar with building evacuation plans, were not confident evacuating the building without instruction, or lacked knowledge regarding evacuation routes. While most had experience participating in emergency drills, participants reported not having experience actually entering stairwells or exiting the building during evacuation drills, or were not aware that stairwell doors were locked from the inside, preventing reentry from the stairwells on certain floors. Proulx and Reid17 found similarly low levels of knowledge regarding emergency procedures among evacuees of a high-rise office building fire. Other studies have shown higher levels of emergency preparedness among commercial high-rise building occupants.16, 17, 19 These studies reported that most participants knew the location of exit doors and/or had knowledge of building emergency evacuation plans. Participants also reported high levels of participation in some form of emergency preparedness training, such as receiving information on emergency procedures and participating in evacuation drills.

Residential high-rise building emergency preparedness

In a study comparing occupants of residential and commercial high-rise buildings, Zmud19 found that residential high-rise occupants had a lower degree of emergency preparedness knowledge than people in commercial high-rise buildings. In comparison to commercial high-rise building occupants, residential building occupants were less aware of building evacuation procedures and less likely to participate in fire drills. They also lacked awareness of occupant emergency training, or building fire safety features, such as fire pull stations, fire exits, or the building public address (PA) system. In another study that surveyed residential high-rise dwellers in Egypt, participants reported high levels of knowledge regarding their buildings’ emergency plans and locations of emergency exits. However, most participants in the study hadn’t received fire safety training and were not sure what to do in fire situations.20


Study design and sample

An exploratory qualitative pilot study was conducted to improve understanding of the knowledge and attitudes regarding fire and evacuation preparedness among people living in residential high-rise buildings. The study was approved by the University of Hawaii Human Studies Program as exempt from federal regulation pertaining to the protection of human research participants.

All study participants were residents of a high-rise building that had experienced multiple fires in the past. Three fires had occurred in the building within a span of seven years. Two of the building’s previous fires, one on the 8th floor and another on the 23rd floor, had been confined to the rooms of origin. The most recent fire in the building had occurred seven weeks prior to the start of the study. This fire originated on the 26th floor, and spread to multiple units on three floors of the building.

Six participants were included in this study. Study participants were identified using a snowball sampling technique, and invited to participate in-person, or were contacted by phone or email. Inclusion criteria included English-speaking adults age 18 years or older who were able to provide informed consent and had experience living in a residential high-rise building. Each person was provided information about the study’s purpose, methods, human subjects protections in place, and eligibility criteria prior to participation. Oral consent to join the study was obtained from each participant.

Data collection and analysis

A semi-structured interview guide was developed using information on emergency and evacuation preparedness in high-rise buildings. Interviews lasted an average of 60 minutes. All of the interviews were conducted in-person from September to October 2017. During each session, one researcher would ask questions, another would take handwritten notes. Handwritten notes were summarized and read back to participants intermittently to check for accuracy. Each session opened with an uninterrupted narrative where participants shared their experiences during the building’s most recent fire. The semi-structured interview guide was used to hone discussion on specific aspects of interest that had not been previously mentioned.

A short demographic information sheet was used to collect information about each participant’s age, gender, number of years living in high-rise buildings, floor of residency during the most recent fire, and owner/renter status. Unique code numbers were assigned to each participant. No personal identifiers of any kind were noted. All study materials were stored in password-protected files on an encrypted computer only available to the research team.

Qualitative thematic analysis21 was used to identify themes that emerged from the data related to participant preparedness for, and actual experience during building emergencies. Handwritten notes were reread several times and initially coded. Codes were sorted and combined to generate overarching themes and subthemes. Themes were reviewed and refined to assure they accurately reflected the data set as a whole. Strategies for establishing the trustworthiness of the study were used to maintain methodological rigor.22 Handwritten notes were read back to participants as a form of member checking. Two researchers reviewed and coded the handwritten notes independently at first, then compared codes with each other. Differences in codes were discussed and mutually resolved. Researchers kept the raw data and analysis to ensure an audit trail.


Participant demographic information

Demographic data collected from participants of the study are summarized in Table 1. Of the six participants, two were male and four were female. The age of participants ranged from 24 to 77 years old. Three participants were owners and three were renters. Three participants lived between the 8th and 19th floors; the other three lived between the 20th and 35th floors. The number of years that participants lived in their current building ranged from 1 to 35 years. The total number of years that participants lived in any high-rise building, including their current building, ranged from 4 to 49 years.

Table 1

Characteristics of study participants (N=6)

Characteristic n (%)
Male 2 (33)
Female 4 (67)
20-39 1 (17)
40-59 2 (33)
60-80 3 (50)
Owner 3 (50)
Renter 3 (50)
0-7 0 (0)
8-19 3 (50)
20-35 3 (50)
Years lived in current high-rise building
0-5 2 (33)
6-10 2 (33)
11-15 1 (17)
>15 1 (17)
Total years lived in high-rise buildings
0-5 1 (17)
6-10 1 (17)
11-20 2 (33)
>20 2 (33)

Qualitative results

Qualitative findings have been organized into four sections: 1) pre-incident, 2) peri-incident, 3) post-incident phases of the fire, and 4) a summary of participants’ suggestions for improving residential high-rise building preparedness for fire emergencies.


Preparedness for fire emergencies. Participants described varying levels of household fire preparedness. Some participants had multiple fire extinguishers and smoke alarms in their apartment unit. Others had smoke alarms but no fire extinguisher, or neither of these items. All participants knew the location of the emergency exit staircases and had experience entering the stairwells in the past. Some also had actual experience evacuating the building during two fire emergencies in the past. One participant who had experienced three fires in the building maintained a high level of evacuation preparedness, stating “I keep my purse in the foyer, always by the door… ” One participant believed a generational difference in “fire literacy” exists, with the current generation lacking a basic understanding of fire safety, stating, “Earlier generations knew how to manage fire…everyone had a fire plan.”

Fire risk perception. Risk perception of fire among participants was generally low despite most having previous experience with fire in the building. One participant, who had experienced one fire in the building, described never having thought about fire hazards, stating, “Before the [most recent] fire, we didn’t have a plan for fire… never considered fire in the building, sprinklers, or proximity to the stairs.” Among participants who were worried about fire, the risk of fire did not outweigh the benefit of building amenities, as one participant stated, “When we bought the unit, I wanted fire sprinklers, but [my spouse] wanted guest parking.” There was also a perception that fires are inevitable, as one person stated, “One fire every 10 years is expected in a community of this [building’s] size.”

Owner/renter perspectives on fire safety responsibility. A difference between owners and renters’ perception of responsibility for fire safety emerged from the data. Owners felt that renters expected apartment unit owners to be responsible for fire safety and equipment. An owner stated, “Owners put [fire extinguishers] in units they own. Renters abdicate [responsibility] to owners.” Conversely, renters described feeling responsible for the residence but felt excluded from building decisions regarding fire safety and preparedness. One renter stated, “Renters have no voice.” One participant didn’t see any difference between renters and owners, stating, “All are equally fire illiterate.”

Ineffectiveness of fire alarms as an evacuation cue. Participants provided multiple reasons why they believed the building’s fire alarm was not an effective mechanism for initiating evacuation. Past experiences with false alarms resulted in many residents not initiating evacuation immediately after hearing the fire alarm. One participant said, “I heard the alarm before, but did not react…it would go on for a long time.” Another said, “At times of false alarms, security would announce it was a false alarm with a bullhorn.” Participants also described the difficulty in hearing alarms, or not being able to distinguish the alarm from other sounds. One person said, “I called security to check… usually, the building alarm was faint… it was like, ‘Oh, is that the fire alarm?’” Another person said, “I wouldn’t hear the alarm because of multiple [emergency vehicle] sirens from the street.”

Building-level fire preparedness. Participants described an overall low level of building fire safety culture. One participant stated, “Never had a drill… no fire safety pamphlet.” Participants were all aware that the building did not have fire sprinklers. There was uncertainty regarding whether or not the building had a fire safety plan for persons with disabilities, or whether the building possessed a list of people who would need assistance evacuating. One participant described observing confusion about how to evacuate residents with mobility impairments during each fire that had occurred in the building. The participant described seeing people with assistive devices (wheelchairs/walkers) moving towards the stairwells but not knowing what to do when they got there, remarking, “All [fires] were the same… people trying to go down the stairs with walkers… a family with a female in a wheelchair, blocking the stairwell exit… People do not know how to exit with a wheelchair.”

Preparedness for other types of disasters. Varying levels of household preparedness for other types of disasters were also noted. One participant stated, “We’re better with hurricane preparedness.” Emergency stockpiles of supplies varied, from having a full week of food/water and supplies (flashlight, batteries), to having only limited food and water, to having no emergency supplies at all. The occupation of one participant’s spouse was very influential to family emergency preparedness. This individual’s household was well-stocked with emergency supplies and had a well-developed family reunification plan that included more than one meeting place. Another individual described being prepared for riot and civil insurrection and had listed weapons and a military-grade gas mask among the items included in an emergency supply kit.

Reasons cited for low levels of household preparedness included a lack of storage space in the apartment unit and plans for sheltering at a relative’s home (rather than in their own apartment) in the event of a natural disaster. In describing personal preparedness for disaster, one participant stated, “Not well. If anything [we] keep cases of water in the closet for hurricanes, random canned goods, none in storage… [It’s] hard in an apartment building, there’s limited storage.” Another participant described similar preparedness for disasters, stating, “Nothing… Plan to go to mom’s [house] during hurricanes.”


Evacuation cues. Two types of cues for evacuation emerged from the data: speculative cues and severe cues. Speculative cues were those that prompted residents to seek more information but did not immediately prompt evacuation. Such cues included hearing the building fire alarm, emergency vehicle sirens, and seeing emergency vehicles (fire engines/helicopters) and equipment. One person who had experienced three fires in the building said, “I saw a helicopter… thought it was unusual.” Severe cues included seeing or smelling smoke, seeing fire/flames, hearing people screaming, or hearing commands to evacuate. One person said, “I heard the alarm go off… Looked out the window and could see fire, heard screaming… We all started heading towards the door.” Another person remarked, “I heard the alarm, waited three minutes, opened the door and smelled smoke… My neighbor was running, knocking on doors. Lights were out in the hallway past the elevator.

Evacuation process. Initiation of evacuation was delayed by the gathering of supplies prior to moving toward the stairs. One person said, “I went into ‘auto-mode,’ gathered computers, purse… went to the stairs.” Movement down the stairwells was described as very orderly. During the evacuation process, people provided assistance to others who needed help down the stairs, such as older residents or persons with disabilities. One person said, “I saw an elderly female in a bathrobe, frenzied, dazed… went into ‘take-care mode’ to care for her… helped her contact her husband.” Another person said, “Whenever I have to down the stairs [during a fire], I go slower… help others.”

Reliability of fire warning systems. Despite the building having a smartphone app capable of sending push-notifications to residents, no notifications were received during the fire incident. Participants described waiting for messages on their phone about what was happening, though none came. One person said, “I did not receive notification… called [spouse], he didn’t get a notification on the app…”


Danger, stress, and uncertainty. Participants described strong feelings of danger, stress, and uncertainty during the immediate period following evacuation. One person said, “While on the ground, things were falling off the building… glass, debris, sides of the building.” Another person said, “People stood in the grassy area looking at the burning building. Not sure… was this safe? Smoke, soot, fire, glass, metal… Was there asbestos in the smoke? Was lead being melted in burning paint?” One person noted residents seemed to be in shock, remarking, “Many people were in a ‘zombie state.’” During the event, participants felt stress being separated from their family members. One person said, “I went into ‘stress/panic mode.’ I called friends… got daughter reunited with the family, then felt better.

Feelings of stress and uncertainty continued during the weeks and months following the fire, driven by concerns about the costs of the damage, conflicting messages regarding hazard mitigation, and worry about health and the safety of their environment. Participants described fear of exposure to hazardous materials and byproducts of the fire, such as asbestos, mold and lead paint. One participant shared, “We were not sure the unit was safe… worried about asbestos coming from vent… we decided to cover the vents in the apartment.” One participant stated, “I feel like I’m experiencing PTSD.” The participant suspected a young child in the family might also be similarly affected, remarking, “My son said, ‘I don’t want you to die in a fire.’” Another participant whose apartment was not damaged in the most recent fire described his feelings as, “Survivor guilt…why was I spared?

Recovery activities. Immediately following the event, people took action toward their own recovery. Pictures of damaged property were taken for insurance purposes and then discarded. One person said, “We were not sure what to keep/discard. We took video of water damage. Saw black soot on shower bathroom walls. Not sure to keep or wash down.” Participants also shared their children’s reactions to the incident. One participant said, “After, in a hotel my son said, ‘I feel safe now.’” Another participant realized her young child needed to continue playing despite disaster recovery activities, remarking, “I put blue tape [on the carpet] to tell my son don’t come inside [damaged areas]. I saw my son playing with a small item as a toy.”

Social support. Residents, family members, and community organizations provided support to people affected by the fire. Friends and family provided displaced residents with temporary housing. Community agencies provided residents with shelter and supplies. One person said, “The next morning [after the fire], food, donations, clothes, supplies were donated… it was uplifting.” Survivors of the event were giving and receiving support from each other. One person said, “Those in need were helping others. The building, as a community, all came together.”

Impact on future fire and emergency preparedness. Impact on preparedness for emergencies among participants was mixed. Most stated that they had made no changes to their personal/family fire or emergency preparedness since the fire. One person remarked, “After the fire, I made a list… but it’s still just chaos.” Some participants described intentions to increase their family’s emergency preparedness supply stockpile, updating their family emergency plans for future disasters, or seeking out training on fire safety. One participant said, “[Before the fire], we had one week worth of water and food supply… Will increase to 14 days. [My spouse is] now talking about where to meet when tsunami hit [sic].”

Suggestions to prepare high-rise residents for fire emergencies

Participants provided suggestions regarding how to better prepare residents of high-rise buildings for fire emergencies. Suggestions were organized into three levels: individual/family, building management/condo association, and systems (community/public policy) levels.

Individual/family level. Participants’ suggestions for individuals/families pertained to evacuation, basic fire safety, and health impact of hazardous material exposure (Table 2).

Table 2: Individual/ family level suggestions for residential high-rise fire preparedness
Category Suggestions
Know how to evacuate • Develop household evacuation plan. Include the route to safety/what items to bring • Place copy of the plan on the entryway door • Close widows prior to evacuating • Knock on neighbors’ doors while evacuating • Consider using hanging ladders to evacuate lower floor balconies
Practice basic fire safety • Install/maintain apartment smoke alarms • Practice proper fire extinguisher use • Unplug appliances when not in use • Close windows/doors when leaving apartment
Know health impact of fire events • Learn signs/symptoms of exposure to hazardous materials

Know how to evacuate. Residents should have a plan to evacuate the building that all members of a household are aware of. The plan should be written and placed on the back of the door. The plan should include what route to take to safety, and what items to bring when evacuating, such as a flashlight, water, phone and charger, and any important medications. Prior to evacuating, residents should quickly close their windows to protect their unit from smoke or water damage. While evacuating, people should knock on their neighbors’ doors to warn them to get out. Residents of lower floors should consider using hanging ladders to evacuate from their balconies.

Practice basic fire safety. All occupants should have smoke alarms and at least one fire extinguisher in their residence. They should replace smoke alarm batteries annually. Proper use of the fire extinguisher should be practiced. Appliances should be unplugged when not in use. Windows and doors should be kept closed when leaving the apartment.

Know the impact of the event on health. People should be aware of and be able to recognize adverse health effects resulting from exposure to hazardous materials in a building affected by a fire. These include exposure to smoke, mold, asbestos, and other hazardous byproducts of a fire.

Building management/condo association level. Suggestions for the building include improving building fire safety systems, emergency communications, fire drills and training, emergency planning, and social support systems (Table 3).

Table 3: Building management/condo association level suggestions for residential high-rise fire preparedness
Category Suggestions
Enhance building fire safety systems • Place fire extinguishers in every kitchen • Install fire sprinklers, automatically-closing doors, illuminate pathways to exits, and audible fire alarms • Install pumps to fill standpipes prior to the fire department arriving
Robust emergency communication system • Develop an emergency communications plan that designates how warning messages are sent, and frequency/timing of messages • Messages should provide clear instructions on what to do/whether to evacuate • Establish multiple channels for communicating emergencies • Ensure communication system redundancy
Building drills/training • Hold annual fire drills • Provide voluntary fire safety training to residents • Schedule fire alarm testing so that residents can check whether alarms are audible • Develop building self-sufficiency so that residents can extinguish fires prior to the fire department arriving
Develop plans to keep all residents safe • Distribute pamphlets describing building fire safety and evacuation procedures • Designate fire captains on each floor who know which residents require evacuation assistance • Mark doors to signify residents needing evacuation assistance • Allow renter representation on the building’s governing board • Establish emergency vehicle access for all sides of the building without immediate roadway access
Establish a building social support network • Hold social events to foster a sense of community within the building • Develop a method for following up with residents after an emergency

Improve building fire safety systems. Participants recommended enhancing building fire safety by installing fire sprinklers, doors that close automatically in the event of a fire, and an emergency lighting system to illuminate pathways to the exits. Fire alarms should be audible and should automatically alert building management/security personnel. Fire extinguishers should be placed in the kitchen of every apartment in the building. The building should also consider installing pumps to fill standpipes prior to the arrival of the fire department.

Robust emergency communication system. Multiple methods for communicating emergencies to residents should be in place. Suggestions included installation of a PA system, a reliable mobile alert system, and a TV broadcast warning system. Warning systems should alert residents of emergency situations, and instruct them whether evacuation is necessary. The building should develop an emergency communications plan that designates which building personnel is responsible for sending out warning messages, how messages can be sent, what information the messages should contain, and the frequency and timing of such messaging. The plan should include redundancies in case portions of the building (or management staff) are not accessible or available.

Building drills and training. Fire drills should be held once or twice a year. Building management should partner with the fire department to provide residents with voluntary training on proper use of fire safety equipment. Residents should develop self-sufficiency for fighting fires prior to the arrival of firefighters. Fire alarm testing should be scheduled so that residents can check if alarms are audible from all rooms of the apartment.

Develop plans to keep all residents safe. A pamphlet describing fire safety and evacuation procedures should be made available to residents. Special considerations should be arranged for persons with disabilities such as designating a fire captain on each floor who knows which residents require assistance evacuating. Doors could also be marked to signify the residence of someone who needs assistance. Renters should have representation on the building’s governing board to provide diverse input. Emergency vehicle access should be established for all sides of the building, including sides of the building that do not have immediate roadway access.

Establish building social support network. Building management should organize social events to foster a sense of community within the building in periods prior to and following an emergency. A method should be in place for following up with persons suffering from post-traumatic stress disorder (PTSD). Networks for post-disaster support should also be established to provide affected families with resources such as temporary housing, psychological counseling, and interpretation services.

System (community/public policy) level. Wider changes to policy and education regarding fire safety were also suggested (Table 4).

Table 4: System (community/public policy) level suggestions for residential high-rise fire preparedness
Category Suggestions
Policy suggestions • Mandate annual fire drills • Conduct regular fire safety inspections • Enforce fire safety rules
Education • Improve community fire safety education • Provide school children with fire safety education, including how to use fire extinguishers.
Social support network • Develop community networks to support families post-disaster

Laws should be created to mandate fire drills at least twice a year in residential high-rise buildings. Fire safety inspections should be conducted regularly, and safety rules should be strongly enforced. Finally, more should be done to better educate the population about basic fire safety. Specifically, elementary school children should receive fire safety education and be taught how to properly use fire extinguishers.


Findings from this exploratory pilot study offer valuable insight into emergency fire and evacuation preparedness among residents of a high-rise building that experienced multiple fires. The main findings of the study are discussed in two themes: influences on fire and evacuation preparedness, and evacuation decision-making and response to fire.

Influences on fire and evacuation preparedness

Risk awareness, critical awareness of disasters, and hazard anxiety are required to motivate preparedness for disasters.6 Overall, all participants had some knowledge about emergency preparedness, and half had actually taken some action to prepare for disasters. Some participants described being more prepared for hurricanes than a building fire. Participants seemed aware of the risk for fire in the building, yet there still existed a low level of anxiety regarding fire hazards among some participants. Not all participants were aware that the building lacked a fire sprinkler system prior to the most recent fire. The building was built before fire sprinkler systems were mandated by local building codes. Further inquiry into beliefs about emergency fire preparedness and hazard risk among high-rise building residents is recommended to better understand these findings.

All participants described limited support from the building management or owners’ association for communal fire preparedness efforts. Participants of this study, some who had lived in the building for many years, were not aware of any fire drills, or other types of fire preparedness activities ever having been conducted in their building. This lack of emphasis on fire preparedness at the building level may have influenced household fire preparedness among study participants. This is analogous to findings from studies done in commercial high-rise buildings. For example, Gershon et al.18 found that WTC evacuees who reported lower organizational support for emergency preparedness and response capabilities on their floors were less knowledgeable about building fire safety features and evacuation procedures, and took longer to evacuate the building during the September 11, 2001, WTC terrorist attacks. The influence of residential building association emergency preparedness safety culture on household preparedness is not well documented, and further research on this topic is recommended.

Differences in fire preparedness among renters and owners were noted in terms of perceived responsibility for preparedness and actual fire preparedness behaviors. Previous studies have described multiple factors affecting renters’ and owners’ preparedness for natural and technological disasters, including access to resources, exposure to hazard information, differences in motivation, and perceived responsibility for preparedness.2324 Findings from this study suggest similar factors may play a role in fire preparedness in residential high-rise buildings. The three participants who were owners tended to be more prepared than the renters for fires and other disasters. One participant who was a renter stated that the inability to participate in condominium board meetings precluded this individual’s ability to provide input into building safety matters. This resulted in a feeling of exclusion regarding important building fire safety decisions. Previous studies have reported that persons who believed that they had less choice in preparedness matters were less likely to prepare for natural disasters.24 Findings from this study suggest that similar factors may also influence personal fire preparedness of high-rise building renters.

Evacuation decision-making and response to fire

Previous studies on commercial high-rise building evacuation have described factors influencing evacuation decision-making among occupants. These studies have shown that environmental cues, if perceived as indicating the existence of a threat, can interrupt an individual’s normal activities, and influence protective action. Individuals will either seek additional information, engage in actions to protect people or property, perform actions to reduce psychological stresses, or resume normal activities. In order for an individual to perceive the existence of a threat, the individual must first receive cue(s), pay attention to the cue(s), and comprehend the cue(s).25, 26

In this study, participants described their inability to hear the fire alarm clearly or discern it from other noises. Furthermore, participants had experienced multiple false alarms in the past. The sounding of the fire alarm was not perceived as a cue that a true fire threat existed. Rather, upon hearing the fire alarm, participants sought more information (listening or waiting for more cues) or did nothing (continuing what they were doing). This behavior is a common reaction to building fires. Building occupants often ignore fire alarms because they fail to hear the signal, fail to recognize the signal as a fire alarm, or have lost confidence in the system due to nuisance or false alarms.27

Upon seeking and receiving more extreme cues, such as seeing smoke or flames, or hearing screaming, participants were more likely to perceive the presence of a real fire threat and began to take protective action. Participants engaged in a variety of activities to protect people or property, such as gathering possessions, assembling family members and neighbors, and moving towards the stairs. These actions are consistent with what is known regarding human behavior during fires. Occupants are more likely to interpret a situation as a fire when they are presented with a higher number of cues, a consistent set of cues, and unambiguous cues.28

Fire drills and fire safety education can help occupants better judge a fire situation, understand the importance of quick evacuation, and become more familiarized with exit routes.29 Findings from this study indicate a need (and desire) among high-rise residents for fire safety education and training. Participants in this study described a period of 3 or more minutes of seeking further cues after hearing the fire alarm sound. Such delays in initiating evacuation increase the risk of injury or death in fire situations. Providing training and education on fire and evacuation preparedness should improve high-rise residents’ ability to quickly recognize emergency fire situations, decrease the amount of time it takes to initiate and complete evacuation, and potentially save lives.

Limitations and conclusion

This study involved a small number of participants from one residential high-rise building. Transferability of the findings to other residential high-rise buildings may be limited. Furthermore, findings may not pertain to emergency preparedness for other types of high-rise buildings, such as commercial office buildings or hotels. Additionally, one member of the research team was a resident of the building from which participants were recruited. Some participants knew the researcher prior to the study, which may have introduced some bias into the study’s findings. Conversely, as a result of knowing the researcher, participants may have been more open in sharing their experiences. Despite these limitations, this study contributes new knowledge on the subject of fire and evacuation preparedness among people living in high-rise buildings and highlights multiple areas for further research. Findings from this study will be used as the basis for a future study involving more participants from multiple residential high-rise buildings. It is hoped that further study on this topic will contribute to policies and programs that improve residential high-rise building fire safety.

Data Availability

Data available upon request to the University of Hawaii Human Studies Program.

Contact information: University of Hawaii Human Studies Program, 1960 East-West Road, Biomedical Sciences Building B104, Honolulu, HI 96822 ( email: uhirb@hawaii.edu, phone: +1(808) 956-5007 )

Publicly posting the data might reveal the identity or location of participants. This is because the transcripts of the qualitative key informant interview data collected during interviews contain multiple specific references to the building name/location, names of surrounding community features (parks, schools) and even the floors of the building that participants live(d) on in the building. Therefore, it is not appropriate to publicly post qualitative key informant interview transcript data. Request for this data can be done through a request from the University of Hawaii Human Studies Program.

Competing Interests

The authors have declared that no competing interests exist.

Corresponding Author

The corresponding author is GARY GLAUBERMAN (glauberm@hawaii.edu)

http://currents.plos.org/disasters/article/exploratory-qualitative-study-of-fire-preparedness-among-high-rise-building-residents/feed/ 0
A Typology of Secondary Stressors Among Refugees of Conflict in the Middle East: The Case of Syrian Refugees in Jordan http://currents.plos.org/disasters/article/a-typology-of-secondary-stressors-among-refugees-of-conflict-in-the-middle-east-the-case-of-syrian-refugees-in-jordan/ http://currents.plos.org/disasters/article/a-typology-of-secondary-stressors-among-refugees-of-conflict-in-the-middle-east-the-case-of-syrian-refugees-in-jordan/#respond Thu, 10 May 2018 15:00:37 +0000 http://currents.plos.org/disasters/?post_type=article&p=37839 Introduction: As the years of displacement accumulate, the burden of secondary stressors (i.e., stressors not directly related to war) increase on the shoulders of millions of refugees, who do not have the option of either returning home due to war or having a sustainable livelihood in the host countries. This paper aims to shed light on the overlooked importance of secondary stressors among refugees of conflict in developing countries; it will do this by highlighting the experience of Syrian refugees in Jordan, and developing a typology of these stressors.

Methods: We approached this issue using two levels of exploration. In study 1, we used participant observation and 15 in-depth interviews in Irbid, Jordan. Data were analysed qualitatively using thematic analysis to explore the different types of stressors. In study 2, a questionnaire survey among Syrian refugees in Jordan (n = 305) was used to collect data about a wide range of stressors. Responses were subjected to factor analysis to examine the extent to which the stressors could be organized into different factors.

Results: The thematic analysis suggested three different types of secondary stressors: financial (money related), environmental (exile structures and feelings created by it), and social (directly related to social relations). The factor analysis of the survey data produced a similar typology, where secondary stressors were found to be grouped into four main factors (financial, services, safety, and relations with out-groups). The final result is a typology of 33 secondary stressors organised in three main themes.   

Discussion: Syrian refugees in Jordan suffer the most from financial stressors, due to loss of income and high living expenses. Environmental stressors arise from exile and are either circumstantial (e.g., services and legal requirements) or created by this environment (e.g., instability and lack of familiarity). Social stressors were observed among a considerable section of refugees, varying from stressors due to being targeted as a refugee by the locals (e.g., discrimination) to more traumatic stressors that came from both locals and other refugees (e.g., assault). The typology of secondary stressors suggested by the present analysis needs to be investigated in a larger sample of refugees of conflict in other countries in the Middle East, in order to determine its generality. We suggest that it is a basis for a framework for practitioners and academics working with refugees in the region.      

Key words: Armed conflict, Syrian, Jordan, refugees, forced displacement, exile, daily stressors, secondary stressors, trauma, typology.



The aim of this paper is to identify the stressors arising from the exile environment among refugees of conflict in the Middle East and to explore the relationship between different stressors, in order to provide the foundation for classification and measurement of such stressors that could be helpful to both researchers and practitioners working with refugees in the region. To achieve this goal, we first conducted a qualitative study (ethnography) that included observations and 15 in-depth interviews with Syrian refugees and local relief workers in Jordan to identify the main themes of such stressors. The second study was a survey (n = 305) among the same population to get a sense of the most prominent stressors and to validate the thematic structure of study 1, in addition to identifying factors of related stressors that can be used as a first step towards building a scale for assessing secondary stressors in the region. Before describing the research we provide the background and explain the need for our analysis.

The current refugee crises

The United Nations High Commissioner for Refugees (UNHCR) global trends report for 20151 shows that the current forced displacement crisis is the largest in modern history, exceeding the situation at the end of the Second World War. More than 65 million people have been forced to leave their homes, and 86% of them are hosted in developing countries1. These people are not expected to return home soon, as the current forced displacement is characterised by a prolonged duration, where refugees spend an average of 26 years in host countries1.

Jordan is an example of the refugees’ impact on countries in areas of armed conflict like the Middle East. Jordan’s current population is 7.5 million, including 656,170 Syrian refugees who have arrived since 20122, and 2,117,361 Palestinian3 refugees who were forced to leave their country by the Israeli occupation, more than 50 years ago4. However, as with many countries in the region, Jordan has not signed the 1951 Refugee Convention5.

The international response to the forced displacement crisis falls short in meeting the persistent challenges facing the majority of refugees, where funding for the relief responses in developing countries is far from meeting the needs in the areas where demand is high. For example, the humanitarian funding appeal in Jordan in 2016 covered only 22% of projected costs.6 This shortfall leaves refugees with little support in facing stressors associated with prolonged displacement, which increases the negative impact of such stressors on their well-being.

War trauma and distress, in context

When thinking about refugees of war, it is easy to recall many mainstream media images of the trauma of war forcing refugees to flee their homes. However, the untold story of exile is different as “affected people often identify as their greatest sources of distress, not the memories of past violence and horrific events but the problems of everyday living” (7 p. 8). The World Health Organisation and International Medical Corps8 undertook an assessment of the mental health and psychosocial needs of Syrian refugees (n = 7964) in different regions of Jordan. They found that the most common mental health concern reported by the refugees was “Worry and concern over the situation, relatives and the future”, followed by “Fear from environmental threats”. When asked about the help they needed, the top priority was “To improve services and living conditions”, while only 13% expressed a need for “Counselling or psychological support”.

Recent studies have found a relatively high prevalence of post-traumatic stress disorder (PTSD) among Syrian refugees in neighbouring countries.9 However, for a more contextualised understanding of the status of trauma among Syrian refugees, we should take into consideration that there are factors other than war exposure that contributes to PTSD. Rather, some of these arise from the exile environment. Al-Smadi et al.’s study10 shows low scores for PTSD among the majority of Syrian refugees in Jordan, and that the higher scores are associated with factors not directly related to war – chronic disease diagnosed in exile. Basheti, Qunaibi, and Malas11 also found an association between psychological distress and living conditions; refugees in tents had double the rates of refugees in caravans. The previous literature is best summarised by Miller and Rasmussen12 who argued that “daily stressors” have a direct effect on mental health among refugees of conflict, in addition to increasing the negative effect of trauma on mental health. This conclusion also supported by later studies (e.g., 13,14,15).

Examples of stressors

Stressors arising from the exile environment cover a wide range of challenges among refugees from different regions, although some of these stressors are sensitive to context. The literature on refugees of conflicts in developing countries provides us with examples of such stressors: among refugees in Afghanistan, these include overcrowded housing, poverty, unemployment, the security situation, violence in the home, poor health and pollution16; El-Shaarawi17 reports inaccessibility to (work, education, and health) services, social isolation and separation from family and friends, decline in living standards, in addition to a status of uncertainty about the future among Iraqi refugees in Egypt. Even using unfamiliar transportation was found to be stressful, as in the case of Darfur refugees in Sudan.18

Secondary or daily stressors

Some studies among refugees of conflict use the term “daily stressors” in referring to stressors arising from the exile environment (e.g., 12,16), which we suggest is problematic for two reasons. First, some of these stressors are not daily (e.g., physical attack). Second, the term is sometimes used in the literature in contrast with traumatic stressors although some of the “daily stressors” can be traumatic (e.g., domestic violence). Therefore, we recommend using the term “secondary stressors”, as found in the disasters literature, which describes this group of stressors as “continuing or chronic problems that … are not directly related to or inherent in the event” (19, p. 3). For this study, we define secondary stressors as stressors not directly related to war (arising from exile environment), and which are socially mediated in nature (operating through the societal response – or lack of societal response – to the events).

Goals of the paper

Based on the literature showing that secondary stressors among refugees of conflict are important but receiving less attention, and the fact that the few examples of research that focus on them were conducted outside the Middle East when the Syrian crises is a main contributor of the current global displacement crisis, we decided to focus on Syrian refugees in Jordan (which is a major refugee host country in that region) to explore the forms of secondary stressors among them, in two stages. In study 1, using an ethnographic framework we conducted in-depth interviews to identify types of secondary stressors and common themes between them. In study 2, we undertook a survey (n = 305) in order to examine the relation between stressors (factor analysis), in addition to assessing the most common stressors among the refugees.

In addition to advocating using the term “secondary stressors” to describe the stressors facing refugees of conflict in exile, the typology we developed can be helpful to both practitioners and academics working with the refugees of conflict in developing countries, by identifying the specific domains that the refugees are in need of support, and the relation between stressors.

Study 1

The aim of study 1 was to identify some of the range of stressors that the Syrian refugees in Jordan encounter, some of which could be specific and sensitive to the setting. To achieve that, we conducted an 8-month ethnography among a community of Syrian refugees, where we collected contextual data from trust-based in-depth interviews and participant observation by the first author.


Participants. The interviews (n = 15) all took place in the northern region of Jordan, by the Syrian borders. Most of the participants were from the nearby Syrian region of Dara’a, which is on the southern Syrian borders with Jordan, and were recruited through the first author who served as a volunteer in a neighbourhood heavily occupied by Syrian refugees. We interviewed 13 refugees who were residents of the neighbourhood, in addition to two local relief workers who worked with Syrian refugees for several years. Only three interviews were with women, due to the local culture restricting private interaction between females and males (the first author being male).

Procedure. The first author collected data in the form of in-depth interviews and participant observations, during his 8-month period as a volunteer in a school located in a neighbourhood predominantly inhabited by Syrian refugees. The first author made it clear (whenever meeting someone new, in addition to later reminders) to the school management, other volunteers and the community members that the main purpose of being there was to collect data in the forms of observations and interviews.

The interviews were undertaken after establishing trust-based relations and then snowballing recruitment. Interviews lasted between 40 minutes to one hour and were conducted in Arabic and audio recorded using the first author’s phone. The researcher (who is a native Arabic speaker) transcribed the interviews and then translated them to English. The researcher also kept a regular diary (21,062 words) that included information obtained through casual chats and observations he made in the neighbourhood, in Syrian family homes, and while accompanying the refugees through activities such as shopping using food coupons and getting necessary documents.

Analysis. Due to the exploratory nature of this study, which aimed to lay out a foundation for a typology of secondary stressors, we adopted a grounded approach (bottom-up) to analyse the data. We used Big Q thematic analysis, which is flexible and involves an organic coding process.20 We started coding the interview data from the lowest level possible and then identified relations between codes themes and merged or split some of them where needed. The extracts and observations we present in this study are chosen because they are representative of statements found in the dataset.

Ethical review. A written consent was obtained from the interview participants. This study (ER/KHTA20/3) have been approved by the Sciences & Technology Cross-Schools Research Ethics Committee at University of Sussex (crecscitec@sussex.ac.uk).


We identified three main groups of secondary stressors (financial, environmental and social) among the participants, which they suggested were common among Syrian refugees in Jordan. Consistent with the literature21, the interviews suggested that these participants struggled more with the secondary stressors than the memory of war or its direct effect on their life (e.g., deaths of relatives in Syria).

Financial Stressors

This group of stressors is a result of the forced displacement of refugees from subsidised Syria to the relatively more expensive Jordan, losing their previous sources of income, and facing challenges finding jobs in exile due to restrictions on refugees’ ability to work. We identified poverty as the main financial stressor that most of the participants faced, in addition to three other stressors (residence, education and health) that arose from the financial strain that the refugees faced.

Poverty. Syrian refugees left their homes to travel to Jordan in order to escape from a life-threatening military operation and deteriorating basic services, with the hope that their forced displacement would be for only a few months:

We came [to Jordan] on a temporary base. We couldn’t wait to get back. I even did not say goodbye to my family. We were thinking 2-3 months maximum.

Participant 7
With the priority to save life and as a temporary solution, many of Syrian refugees walked away from their jobs, farms and shops across the borders with Jordan with little savings:

Even if you have some savings it will be gone, especially when you are not allowed to work [pause] The financial ability of people diminished as the crisis prolonged.

Participant 9

In addition, the Syrian refugees faced a serious financial challenge as they moved from less expensive ‘socialist’ Syria (where basic services were subsidised by the government) to more expensive free market Jordan, and had to face additional inflation caused by hundreds of thousands of refugees who moved out from self-owned houses in Syria to rent limited housing units in Jordan for high prices:

We did not have to worry about rent, electricity and water bills. [pause] We had free education and health services [pause] Here everything needs money.

Participant 2

Refugees have little financial support in facing such challenges, as they depend on the UNHCR which provides food coupons (around $21 per person monthly) for registered refugees, in addition to some irregular aid from charities:

In the beginning, the aid from UN was available [pause] it was fine, but now it’s less. Instead of 24JODs($33)/person, now they give 10-15JODs($14-$21) or even cancelled it for some people. I still receive the [food] coupons 15JODs($21), which are only enough to provide us for 20 days. The rest of the month we have to take loans from friends or find any work to survive. Each month, we receive a text message to inform us that the coupon debit card is charged, then we go to the authorized outlets to buy grocery. We are only allowed to buy food [using the coupons].

Participant 2

A smaller group of refugees receive a monthly allowance after being evaluated and approved by home visits undertken by a team from UNHCR, although such payments are not reliable. The refugees call it “the retina” because the payment is through ATMs with a retina scanner:

We receive 100JODs($141) to pay toward the rent, that’s all [pause] the Retina [monthly salary][pause]. They came to evaluate our home, twice in the same month [pause] two persons each time. They evaluated and approved us, but then it was taken away. So we went back struggling with the rent. This is an on-going suffering. The monthly salary was cut off for 4 months and then returned, but they have cut it again now.

Participant 8

During the first author’s visits, he noticed the simple furniture in the crowded homes of Syrian refugees, and torn clothes on some of the children. In addition, there were examples of how living in poverty can increase expenses, like in the case of one Syrian family caught in a situation where they were using most of their income to pay rent for a furnished apartment, which made them unable to save money and buy furniture and move to a less expensive unfurnished apartment (observation made on March 21st, 2016).

Residence. Due to the influx of hundreds of thousands of Syrian refugees, there was a scarcity of housing units that led to an increase in rents. Many refugee families had to share housing with each other, especially upon arriving. A refugee described her experience of staying in her husband’s brother’s two-bedroom apartment, when they arrived from Syria:

When we first came to live with my older brother in law, he was living with his wife and kids. So, we lived with other two families in one apartment, which was very hard. It was only two-bedroom apartment with a small living room, one toilet and a small kitchen [pause] The older brother had a wife and two kids, we had three kids and the youngest brother had a pregnant wife. [pause] 11 persons.

Participant 7

Sharing small apartments is a practice that Syrian families sometimes continued to do not just on a temporary basis upon the arrival of a new family, but also as a permanent solution to reduce the cost of rent and utilities. The first author visited many Syrian families’ homes and in one case, there was an extended family of ten people who had been living for years in a three bedroom apartment, including their married son and divorced daughter with their own children (observation made on February 7, 2016). Due to the high rents, refugees move to a cheaper apartment whenever possible (especially with little furniture), even if the place is not well maintained or is affected by pollution (e.g., burning garbage and flooded sewers) (multiple observations).

Education. Basic education (up to 12th grade) is available for Syrian refugees in Jordan free of charge, but challenges facing Syrian students in the Jordanian government schools pushed two participants to consider private school as an option:

Education is expensive in Jordan because government schools do not take care of refugee kids. They go to school in the afternoon and get little attention [pause] I pay 45JODs($63) a month for my son’s private schools. It is a burden, but I’m handling it.

Participant 3

Also, there was the stress of sending children to college; this was an aspect that they did not have to worry abut in Syria, where free college was offered to all citizens:

My older son is 19 years old and finished 10th grade in Syria, but now he cannot go to school because he’s working full time. He didn’t finish high school because I don’t have the money to send him to college.

Participant 2

Health. Nine interviewees said that they were satisfied with the health services available to them, especially after they registered with UNHCR. However, three indicated that that health services available to them were basic and even the medical missions and subsidised private health services could not cover some of their serious health needs:

I have serious health issues that the [public] health service did not treat. I need serious medical care, which is expensive.

Participant 6

One participant complained about discrimination in providing health care for Syrian refugees in specific health centres, which he believed was sub-standard:

When I get sick I don’t go to the doctor. I have no medical insurance. There is a health centre in [location] for refugees. Once I took my friend there and it was very bad. They treated us like cockroaches, not patients [pause] openly, everybody there did. Even if I was dying, I won’t go there. Sometimes I have health problems and go to a docto, who is the mother of a Jordanian friend.

Participant 4

Environmental Stressors

This group of stressors have in common that they arise either from the circumstances of the exile environment (official documentation issues) or a general feeling created by this new environment (instability and lack of familiarity).

Official stressors. Some of the Syrian refugees arrived in Jordan through the official borders gates with Syria (closed since April 201522,23), while others arrived through the border fence and then on to refugee camps. Refugees had the option of leaving the camp if they secured a Jordanian guarantor. However, some refugees had their own ways of leaving the camp, as reported by one of the participants:

My family entered Jordan using passports, but that was dangerous for me. So, I entered through the border fence with the help of Free Syrian Army. Jordanians [Jordanian Army] took us to Alzaatri refugee camp, where I left in the same day with smugglers for 70JODs($98). Three months later I came back to the camp with a guarantor [to issue official documents]. I did not know him [the guarantor], there were people would arrange it for 50JODs($70)

Participant 10

UNHCR registration was important for many refugees because it was a requirement to receive aid and services provided by the government or the NGOs. At the peak of the influx of refugees, the wait for a UNHCR registration interview appointment was months:

The situation went from bad to worse. We were not receiving food coupons, because we had to wait for a UNHCR [registration]appointment due to huge numbers [of refugees]. It was easier for people who came earlier. We arrived in August but had to wait until December to register with UNHCR.

Participant 7

It is worth mentioning that UNHCR registration should be renewed annually:

The UNHCR registration has to be renewed every year, while the Jordanian ID [for refugees] does not expire.

Participant 2

Many services require a valid UNHCR registration and one example is a health centre that had signs explaining that blood tests will not be done without a valid (up-to-date) UNHCR registration (observation made on 9th of February 2016).

The UNHCR subsidizes the cost of refugees’ health care, which could be the reason for asking for a valid UNHCR registration as a requirement to receive health care services:

… let’s take health services as an example, Syrian [health care expenses] are paid by the UNHCR while the Jordanian [expenses] are paid by the ministry of health

Participant 15 (a local relief worker)

However, being registered with the UNHCR is not the end of the struggle with documentation issues, as the refugees are required to acquire other legal documents. This realisation came to the first author after a short period in Jordan when he was walking with a Syrian student when he asked the first author to avoid passing by a policeman. The first author was puzzled, as he knew that all students in the school were registered with UNHCR, but the other children told him that the student did not have “the ID” (observation made on December 16, 2015). A few months later, the boy’s family was deported.

The Jordanian government issued a “security identity card” for Syrian refugees and required them to carry it and show it at official departments and at security checkpoints. The Jordanian security identity card has been updated with a new magnetic one, which contains the refugee’s retina scan information. A refugee explains the process and the requirements:

They [Jordanian Government] required us to have the white identity card at first. I gave them my [Syrian] family card and they granted me the white card. Then they asked us to get the magnetic identity card, which requires having the white card. It was a matter of time due to pressure. We also had to do a blood test, which I paid 30JODs($42) for, in the [name] hospital last August. I’ve heard that the price was reduced to 5JODs($7). Only adults have to do the test. I took the result to the nearest police station and got the new magnetic security card. They also asked for UNHCR registration and a certified rent lease contract.

Participant 7

Some refugees did not get the new identity card because of the high cost of the blood test, which used to cost 30JOD($42) per person before it was subsidised to 5JODs($7). However, not holding the new Jordanian identity card can lead to serious consequences, as a participant explains:

In a few months, they would not register my kids in the school, for the next year. You need it to goto health centres and hospitals. If I wanted to travel from Irbid to Amman [the capital] I need it, otherwise I would get into trouble [pause] I would be legally questioned by the police patrol [road block] […] They might just give you a warning, depend on the police officer [pause] they might throw you in al Azraq camp [more of a detention camp] or let you go. Especially if you talked back, they could throw you back to Syria.

Participant 2

Syrian refugees in Jordan are only allowed to work if they have a working permit.24 They find it hard to get a permit, and face serious consequences if they have worked without one:

They [refugees] don’t have working permit. They cannot have a permit. Anytime they could be caught and deported. Syrian refugees don’t get a permit. We tried and it didn’t work.

Participant 5

A local relief worker noticed an additional reason for Syrians not to get a working permit:

The Syrian refugees try to stay away from getting a working permit because it restricts them to do just one specific job (e.g. builder or sweets maker), while the refugees try to work in more than one job to compensate for the low wage. One day he works in a farm, next day he will work as a builder.

Participant 15

However, Syrian refugees have ways around restrictions on work, as the first author noticed that some refugees chose to work night shifts, outside the working hours of Ministry of Labour inspectors. Other refugees decided to send their children to work (they would not get deported if caught) for shops and restaurants for very low pay (less than $3/day) (observation made on May 15th, 2016 ).

New Environment

Many of the Syrian refugees in Jordan come from the southern region and settle in the northern region of Jordan, due to the proximity (20 kilometres) and similarities in culture. However, Syrian refugees in many cases found themselves in a new environment where they felt disoriented, especially those who came from rural areas and settled in a city:

We did not know the laws, refugee regulations. If I had a sick child, I did not where to go. I have diabetes and high blood pressure and did not know where to get medicine.

Participant 10

This new environment creates a lack of familiarity that could have a deep impact on a newly arrived refugee, who feels a deep sense of estrangement and powerlessness:

I was walking back home when I got lost. It was hard to go from one area to another, without knowing where you should go! You take a taxi that asks for 5JODs($7), but you don’t know if that’s too much or not. You don’t know the value of the 5JODs, or how long the distance is!

Participant 12

Instability. The participants expressed a deep feeling of instability and loss of aspiration of the future. This is mainly because they consider their stay in Jordan as being temporary, especially when they do not have the option of permanent residence. They felt deep instability that pushed some of them to consider relocating to a third country or even returning to Syria:

The biggest struggle for a refugee is the temporary situation. Where are we going? then what? the instability!

Participant 7

Another effect of the exile environment on Syrian refugees was the lost sense of normal life, and one clear example of this lost normality can be found at bedtime. As most refugees are not allowed to work and their children go to schools after the Jordanian students finish school, they seem to adopt a nocturnal lifestyle where the whole family stays up late (multiple observations).

Social Stressors

Although many of the environmental and financial stressors are socially mediated, we labelled this last group of stressors as ‘social stressors’, because social relations or forms of social interaction (or lack of interaction) are the main source of these stresses.

Separation from relatives. The participants belonged to a collectivist culture, so it was not a surprise that they showed great concern for their relatives beyond the borders, in Syria and other countries:

There was suffering [pause] to stay away from my family. [pause] I have a brother in Syria and another brother in Libya. Like the destiny of the Syrian family, we are scattered all over the world.

Participant 3

Relations with host-community

Across ten of the interviews, refugees mentioned the word “Jordanians” nine times in a positive context, nine times in a negative context and three times in a neutral context, which may indicate that the relation between Syrian refugees and their Jordanian hosts is a love-hate relationship, were refugees have a mix of negative and positive experiences with locals and local authorities (e.g., Participant 4, who was treated badly at the health centre, but have very supportive Jordanian employer). The general perception of the relationship seemed fine, but there were still everyday indications that Syrians were no longer welcome. One refugee attributed it to the commercial aspect of the crisis:

At first, people were very welcoming to Syrians and treated them very well, but with the increase of influx, things turned to be materialistic. At the beginning, Syrian families were offered housing for free, then turned to rent only. My first landlord even used to bring me breakfast and pick some fruits for me.

Participant 9

On level of personal relations, the picture is more positive in that we found reference to many strongly supportive relations:

It was struggling until we managed to know good people. Every community has good and bad people [pause] My Jordanian co-workers are a good example. They visit me and I visit them, which made me no longer feel as a stranger. I have excellent relationships with them, which reduced the struggles I have.

Participant 3

the owner pays all [working permit] fees [pause] He [Jordanian employer] told me not to worry about it, even if it’s 600JDs ($846) [a year], he would pay it [pause] Even if it goes to a 1000JDs ($1410), he would still pays it. This is hard to find anywhere els.

Participant 4
Prejudice. Some participants reported verbal and physical attacks, and while they stressed that these were rare,they felt they were targeted because of their refugee status:

We’ve been treated somewhat bad. Jordanians do not accept Syrians. They don’t treat Syrians well.

Participant 8

However, it seems that some refugees blamed other refugees for such treatment:

They think Syrian women are cheap. Some Syrian women caused this reputation, but it annoys me when they generalize it.

Participant 7

Discrimination. Some interviewees said that they felt welcomed by Jordanians, who treated them well on both personal and official levels. However, they also reported some aspects of discrimination:

At the airport, everybody enters with dignity, except Syrians. They treat him like if he has a bomb in his bag.

Participant 1

Some Jordanians apparently held a belief that refugees should have lower status, and they treated them accordingly. The first author witnessed this first-hand while sharing a taxi with a Jordanian, who said during the ride said that Jordan gave a lot to the Syrian refugees who are ungrateful and have a strong voice. He used a traditional proverb “a guest with a sword” to illustrate that the refugees are not at the same level as Jordanians and should behave accordingly (observation made on February 19th, 2016).

Exploitation. Some interviewees reported exploitation practices that targeted refugees either by taking advantage of them based on their identity or needs as refugees (e.g., work and housing):

When I want to rent a place I suffer [pause] either they raise the rent or stuff like that. Even a taxi driver when he hears my Syrian dialect asks for more. [pause] Bullying. This does not happens to Jordanians [pause] The salary [for Syrians] is too low compared to other workers, even to the minimum wage set by the government -which was 180JODs($253)/month-. They offered me 150JODs($211) for working 10 hours. The 180 minimum is for 6 hours! Take it or leave it.

Participant 4

There was a restaurant in the neighbourhood where a Syrian worker resigned but did not get paid, and the first author witnessed when a group of Jordanian young men threatening the restaurant’s manager to pay the former worker his unpaid salaries. Once they left, the manager called the restaurant’s owner and told him that he will “get rid of the worker and send him back to Syria” (observation made on May 16th, 2016).


Syrian refugee interviewees reported the issues that they face in exile and that challenge them the most. Occasionally, some refugees reported primary stressors caused by war (like loss of relatives or property back home), but the main challenges were those that they have to face in exile (secondary stressors).

We found three main sources for stressors among the refugees in exile; first, financial stressors where the refugees lost income and found themselves in a relatively expensive host country, which led to poverty and other struggles caused by that poverty (crowded housing, education, and health stressors); second, environmental stressors created by the circumstances of exile (e.g., documentation issues) and the feeling created by this environment (unfamiliarity and instability); third, social stressors where social interactions are the main source of stressors like prejudice, discrimination, and exploitation.

Secondary stressors can be intense and could lead to traumatic experiences, which led some refugees to consider returning to war-ravaged Syria, as in the case of Participant 5 who was struggling for years to provide for his family without any improvement in living conditions, and Participant 14 who was a single mother with three disabled children, with no stable income.

Establishing personal relations with community members, and providing contributions as a teacher and a part-time relief worker helped the first author establish trust for uncensored and safe disclosure during the interviews and casual chats with refugees and volunteers. The immersion of the researcher in the community for a relatively long time allowed him to make sufficient observations to avoid a superficial and restricted perspective of the community that a passing observer could be exposed to. One example of such a superficial and partial image of the community was experienced by the first author once he arrived in the city. He encountered Syrian boys begging on the main street, while the neighbourhood behind that same street (200m away) included Syrian families, who we found out later had to be convinced to accept aid..

However, our approach had some drawbacks in regard to face-to-face communication, which limited interaction with females due to cultural restrictions with the male first author and also made it harder to discuss sensitive issues like sexual harassment. (One female participant mentioned harassment but did not use the word “sexual” directly and a male participant reported that it happened without discussing specific cases.)

Study 2

In order to advance our exploration of secondary stressors among Syrian refugees in Jordan, we conducted a questionnaire survey of secondary stressors among a convenience sample of Syrian refugees in Jordan. This study first aimed to get a sense of how common the stressors are that we identified in the previous study; its second objective was to test and validate the themes resulting from qualitative analysis; its third intention was to detect any other stressors that could have been missed by the ethnographic case study (e.g., rare, unobservable or not accessible face-to-face).


Participants. The participants in this study were 305 Syrian adults (18 years and above) who left Syria because of the war and were now living in Jordan. The majority of the participants were male (64.3%).

Building the questionnaire items. The survey items were created using multiple sources, mainly guided by the qualitative data from study 1, and then consulting local experts (two academics and one relief worker) who worked with Syrian refugees in Jordan. Finally, we drew upon other sources relevant to the refugees’ situation16 or others from different context.19 The survey included 51 questions (see Table 1), each asking how frequently the respondent was exposed to the secondary stressors (with a 5-point Likert scale, from ‘never’ to ‘always’) for all the items, except for seven items (18,38,40,42,44,46 and 48) which expressed the degree to which they agreed with a statement (on a 5-point Likert scale, from ‘totally agree’ to ‘totally disagree’). At the end, the questionnaire included nine demographic questions. The survey questions were translated to Arabic by the first author (who is a native Arabic speaker) and then back translated and checked against the English questions by a research assistant (who was English-Arabic bilingual).

Recruitment and data collection. In September 2016, we used an online survey service (Qualtrics) to construct the survey. We recruited participants using Facebook, based on our knowledge that Syrian refugees are widely using the site to stay connected to their family and friends who were separated due to war, in addition to sharing information about services in exile. Refugees access Facebook using phones, so we designed the online survey to be mobile-friendly. The first author used the social relationships that he built during the ethnography in Jordan with refugees and community organisations to spread the word about the survey, in addition to sending invitations through public Facebook groups dedicated to Syrians in Jordan. We sent link invitations to people with public profiles with the following criteria: adult, Syrian, and living in Jordan.

Analysis. We started with basic descriptives of the sample characteristics, followed by the percentages that show how common the different stressors are among the participants. Finally, we conducted a principal factor analysis (with Varimax rotation) to identify possible groups of stressors that are related to each other and independent from other stressors.

Ethical review. A page of participant information and a consent was obtained from the online survey participants by answering a clear question confirming their consent to participate. This study (ER/KHTA20/4) have been approved by the Sciences & Technology Cross-Schools Research Ethics Committee at University of Sussex (crecscitec@sussex.ac.uk).


The sample (n = 305) was fairly young (more than 60% < 30 years old) and almost half of them (47.7%) were single, which could be due to the recruitment method used, whereby Facebook users among the refugee community tend to be younger. Furthermore, it is worth mentioning that the majority of participants (84.8%) spent three years or more in Jordan (due to the closure of the border), which we expected to establish among them a good awareness of secondary stressors of prolonged displacement.


In Table 1, we find a percentage analysis that highlight the most common stressors that sample of Syrian refugees suffered from in high intensity (“most of the time” and “all of the time”). On the top of the list we find the financial stressors that the majority of refugees suffer from (e.g., 73.9% lost usual income), while the structural environmental stressors appears to affect a large percentage (quarter to third) of the participants. The stressors regarding the relation between The Syrian refugees and the locals were reasonably low (<15%).

Although the previous analysis gives, us a snapshot of the most common stressors among the refugees, it might fail to detect the importance of some highly sensitive social and political stressors (e.g., sexual assault and working without a permit), where refugees might be hesitant to report encountering such stressors. Also, this approach cannot detect some serious (high intensity) stressors (e.g., domestic violence) that only occurs to some refugees, but nevertheless should be highlighted in order to be addressed. To highlight such underlying stressors and explore the relations between the general stressors we conducted an exploratory factor analysis (EFA).

Factor Analysis. A principal factor analysis was conducted on the secondary stressors items with Varimax rotation. The Kaiser-Meyer-Olkin (KMO) measure was 0.73; all KMO values for individual items were greater than .64, which is above the acceptable limit of .5 which indicates an acceptable sample size.25

An initial analysis was run to obtain eigenvalues for each factor in the data; 15 items fall in four factors had eigenvalues greater than 1 and in combination explained 60% of the variance. After rotation, the three factors retained corresponded to four dimensions of secondary stressors. Reliability scores for most factors were acceptable: secondary stressors α = .67; financial stressors α = .86; stressors related to services α = .72; safety stressors α = .72; and relations with locals stressors α = .5



As expected, financial stressors were the most common severe stressors among the participants, due to the lack of financial support and restrictions on employment of refugees and expensive life in Jordan. The factor analysis identified four coherent factors. The low reliability of the factor for relations with the locals (Jordanians) factor could be due to the sensitivity of the questions, and thus the responses of some participants. This suggestion is based on observations by the first author during interviews and casual conversations, where most of the Syrian refugees were very careful not to talk in a negative way about Jordanians, to avoid being seen as criticising their hosts. The relation between the Syrians and Jordanian is relatively good, but during casual talks with refugees who were close to the researcher, some of them reported negative encounters with Jordanians that left an impact on them. It is possible that some of the refugees who had negative experiences with Jordanians did not report it in the questionnaires due to two possible factors: first, because of cultural traditions that consider it highly inappropriate to criticise your host; and second, due to the political status quo where refugees suffer from vulnerability in relation to locals and the government (e.g. stressors 20, 27, 29, 35).

General Discussion

We found that the qualitative analysis produced three main themes or types of stressors, which were financial, environmental and social, while the quantitative analysis produced four themes or types of stressors, which were financial, services, relation with locals and safety. We identified that the quantitative themes broadly map into the three main types of secondary stressors that resulted from the qualitative analysis. The financial stressors theme was matched in both analyses, services can perhaps be considered as a sub-type of a broader environmental category, and relations with locals and violence can be considered sub-types of the main social stressors category.

Based on the two studies described above, we therefore propose a typology of 33 secondary stressors (Table 3) that is organized into three main themes (financial, environmental and social stressors). The selection of these reflects, first, the qualitative analysis that generated the three main themes; second, the factor analysis that generated four factors, which map onto the three qualitative themes; and third the percentage of each stressors among the sample. The scale of fifteen secondary stressors items (Table 2) is therefore situated within the typology, with its four sub-scales (financial, services, relations with locals ,and safety stressors).

Financial stressors consistently appeared as a top priority in both the qualitative and quantitative analysis, while there were less frequent – but not less serious – issues that seek our attention (e.g. 5.8% reported physical sexual harassment by another refugee). We had indirect reports (not using the word “sexual” and second-hand experiences) of sexual harassment from two participants during the interviews, but, combined with the fact that 17 participants in the survey reported being sexually harassed, we can conclude that these were not isolated incidents among the refugee community.


In the typology, we divided the stressors into main themes and then to smaller categories in an attempt to explore the relation between different stressors, but in reality, such division is not pure as overlaps are expected and the nature of a specific stressor can arguably fall under more than one possible category. Our main reason for creating the typology was to identify the main sources of secondary stressors. As we argued in the introduction, by definition secondary stressors arise from the exile environment and thus are environmental in nature, and socially mediated which emphasises the social nature of such stressors; and we acknowledge financial ability to buffer many stressors in situations like forced displacement. However, we focused on the main source making the stressor most salient in the refugee’s experience, which is either the structure which forces conditions collectively on refugees (e.g., documentation restrictions/environmental stressors), or social interactions with individuals that spark stress (e.g., exploitation), or the lack of financial ability that determines accessibility to services (e.g., advanced healthcare). Thus, we decided to consider residence, education, and health expenses stressors as financial stressors because these are the “stressors of the poor”. We considered worrying about relatives in Syria as a secondary stressor instead of treating it as a primary stressor, directly caused by war, because we believe that the main stress source in this case is the separation itself, not to the dangerous situation that the relatives are experiencing, which we think is also relevant. What gives us confidence that the mere separation from family is the main source of stress is that the participants were found to worry about relatives in Syria (25.4%) to a similar degree that they worry about relatives in other countries (24.1%).

Financial stressors

This type of stressor is the most important, as it was clearly found in both qualitative and quantitative analysis, and the top five stressors common among the participants were all financial (see Table 1). This came as no surprise due to the fact that most of the refugees lost their income and homes and depleted their savings during many years in exile, and due tothe high living expenses in Jordan compared to Syria or other neighbouring countries. Two specific financial stressors (poor housing condition and low quality of education) did not group well with the other stressors (maybe because these two result from them), although a considerable number of the survey participants (30% and 17.7%) reported suffering from these stressors.

Environmental stressors

This category included stressors that are circumstantial either in the structure of the exile environment (e.g., services and legal requirements) or relate to a general feeling created by this environment (e.g., instability and lack of familiarity). Most Syrian refugees – especially those who were displaced early in the crises – left their homes as a temporary solution and were not prepared (financially or mentally) for many years in exile with an ongoing civil war without an end in the horizon. This situation is complicated by the fact that Jordan is not a member of the 1951 Refugees Convention, which ensures basic services to refugees (e.g., housing, work, and education)5. The Syrian refugees could face these harsh conditions indefinitely as Jordan is very clear in its resistance to refugees being integrated (there are still Palestinian refugee camps in Jordan since 1967). Syrian refugees know that they cannot stay in Jordan or go back home because of war, and relocation to a Western country is not an option for the majority of the refugees because these countries only take a small percentage of refugees1 and additionally many refugees are not interested in living in a Western country which takes them far from their relatives and traditional culture.

Social stressors

This group of stressors directly arises from social relations, either with locals or other refugees, or the lack of social interaction in the case of separation from relatives. The stressors arising from social interactions in exile were found to be grouped into two types: stressors related to relations with locals and safety stressors.

In the interviews, most of the refugees said that they have good relations with Jordanian people and that they appreciate the efforts and services offered to them in Jordan. However, some refugees reported occasional issues in their relationships with locals (prejudice and exploitation) and more common issues in regard to systematic discrimination against them (e.g., work restrictions, lower quality education and multiple official document requirements). This analysis was supported by the quantitative analysis, which showed that discrimination against Syrian refugee appears to be more common (25% of the participants) compared to experiencing prejudice (less than 15%).

The social violence (safety) group of stressors (physical and verbal sexual assaults) is important, as it represents how the stressors of the exile environment can be traumatic, and at the same time does not have to be common or “daily”. This supports our argument for using the term “secondary stressors” to describe these stressors arising from the exile environment instead of “daily stressors” that is widely used in contrast to traumatic stressors.


Both studies reported here were conducted using a convenience sample of adult Syrian refugees in Jordan, and thus we should not assume it is representative of the general population of Syrian refugees in Jordan. Assumptions based on our findings should be examined among larger and more diverse samples of refugees in the Middle East, which will also help to refine the data collection and measurement tools used in both studies. By definition, secondary stressors arise from the exile environment, and thus we predict different secondary stressors among Syrian refugees in other countries, due to the difference in the environment (e.g., the language barrier to Syrian refugees in Turkey). Therefore, we need to extend our examination to include more hosting countries in the region as they have different situations (e.g., financial, security situations and demographics) that could shape the exile into an environment different than what we examined in Jordan.

We recommend more exploration among more refugees in more locations in order to produce more comprehensive and more reliable typology of secondary stressors among refugees of conflict in the Middle East and developing countries in general. Although we laid a foundation to build a scale for secondary stressors in the region, it still needs to be tested it in a wider sample and the items refined.


As shown in the introduction, the relief literature does recognise the importance of stressors arising from exile, but it lacks an organised conceptual framework to classify these stressors or to measure them. The only effort that we are aware of is a typology of secondary stressors in disasters,19 which was not designed with armed conflicts in mind. We recommend using the term secondary stressors instead of daily stressors, as it includes stressors that are not daily in nature, in addition to traumatic stressors that arise from exile. Our data showed that refugees did occasionally report primary stressors caused by war such as the loss of relatives or property back home, but they clearly emphasised that the main challenges were secondary stressors. Some of the secondary stressors reported by the participant were quite serious and could be considered traumatic (e.g., physical and sexual attacks) which again suggests the flawed use of the term daily stressors, especially when used in contrast with traumatic experiences.

We found that Syrian refugees in Jordan suffer the most from financial stressors, due to the loss of income and high living expenses, and deficiency of the services available to them. Social stressors were found among a considerable number of refugees, where they varied from a group of stressors due to being targeted as a refugee by the locals/government (e.g., discrimination) to more traumatic stressors that came from both locals and other refugees (e.g., assault).

We hope that this typology will offer help to both interventions and future research to identify and classify secondary stressors among refugees of conflicts.


The survey data are available on Figshare at DOI: 10.25377/sussex.6169316. The interview transcripts contain sensitive data and cannot be anonymized, therefore they are available upon request. These restrictions are imposed by the Sciences & Technology Cross-Schools Research Ethics Committee at University of Sussex. Interested researchers can request access to the data from the Sciences & Technology Cross-Schools Research Ethics Committee at University of Sussex at crecscitec@sussex.ac.uk.


The authors have declared that no competing interests exist.


Khalifah Alfadhli who may be contacted at alfadhli@ksu.edu.sa.

http://currents.plos.org/disasters/article/a-typology-of-secondary-stressors-among-refugees-of-conflict-in-the-middle-east-the-case-of-syrian-refugees-in-jordan/feed/ 0
Do Natural Disasters Affect Voting Behavior? Evidence from Croatian Floods http://currents.plos.org/disasters/article/do-natural-disasters-affect-voting-behavior-evidence-from-croatian-floods/ http://currents.plos.org/disasters/article/do-natural-disasters-affect-voting-behavior-evidence-from-croatian-floods/#respond Fri, 06 Apr 2018 11:00:45 +0000 http://currents.plos.org/disasters/?post_type=article&p=37169 Introduction: Studies show that natural disasters influence voters’ perception of incumbent politicians. To investigate whether voters are prone to punish politicians for events that are out of their control, this study was conducted in the previously unstudied context of Croatia, and by considering some of the methodological issues of previous studies. Method: Matching method technique was used, which ensures that affected and non-affected areas are matched on several control variables. The cases of natural disaster in the present study were floods that affected Croatia in 2014 and 2015. Results: Main results showed that, prior to matching, floods had an impact on voting behaviour in the 2014 and 2015 elections. Voters from flooded areas decreased their support for the incumbent government and president in the elections following the floods. However, once we accounted for differences in control variables between flooded and non-flooded areas, the flood effect disappeared. Furthermore, results showed that neither the presence nor the amount of the government’s relief spending had an impact on voting behaviour. Discussion: Presented results imply that floods did not have an impact on the election outcome. Results are interpreted in light of the retrospective voter model.



Some of the ongoing questions regarding voting behaviour are related to voters’ competencies. Can citizens correctly perceive information or are they prone to biases 1,2, are they informed enough3,4,5 , are they motivated to participate in politics6 , etc.? These questions relate to the idea, or the norm, of a rational citizenry. They are of great importance because assumptions and findings on voters’ rationality are used to justify different political institutions and systems7. If voters are rational, their broad political participation and inclusion should be warranted; if they are not, their role in the political system should be minimised.

(Ir)rational voters

One of the main models of rational voting is the retrospective voter. This model is based on the idea that citizens use voting to punish or reward incumbents 8. In that sense, elections are used to control the democratic processes and hold politicians accountable for their actions 9. Politicians, in turn, are incentivised to keep their promises and to perform well during their mandate. The evaluation, or ‘performing well’, is usually conceptualised via economic indicators, and connected with the concept of economic voting10,11,12. Studies mostly use either macro-level indicators (sociotropic voting), such as inflation, GDP, economic growth, etc. or micro-level indicators, such as individual changes in economic welfare (egotropic, pocketbook voting)10,13.

If the economy, either on a national or individual level, is doing well, voters will reward the incumbent by voting for her/him in the upcoming election. While some researchers claim that the retrospective economic voting is a robust phenomenon 11,14, other authors point out two caveats.First, it seems that economic voting depends on contextual factors, such as the number of political alternatives 10, media attention 15, or incumbents’ ideology 16. Second, and more important for the idea of retrospective voting, are findings that show that citizens do not know enough about politics, are prone to biases, and hold the government accountable for things which are out of its control 17.

The first set of objections to retrospective voting does not necessarily endanger the concept of voters’ rationality. Citizens can still be prospectively rational, i.e. vote for the party or candidate whose positions on political issues are closest to their positions, known in literature as issue voting 18,19,20 .

On the contrary, the other set seriously undermines the concept of voters’ rationality. For example, how can citizens evaluate government success if most of them do not know what parties form the government 12? Furthermore, it is not just that citizens do not know enough about politics; studies have shown that citizens base their evaluations of candidates on politically irrelevant characteristics, like the candidate’s physical appearance or vocal characteristics 21,22,23,24,25,26 . It is assumed that these physical characteristics are related to candidates’ leadership skills. More masculine facial and vocal features signal dominant behaviour and good leadership qualities, and those findings hold in cross-cultural contexts and in different political systems. Besides candidates’ traits, some external factors, which are beyond the control of politicians, could affect voters’ decisions. One example of an external factor are natural disasters.

Natural disasters and voting behavior

It has been shown that natural disasters, as well as some other negative events, can influence voters’ perception of incumbent politicians and the government. Perhaps the most striking results are reported by Achen and Bartels 27,28. They showed that shark attacks in New Jersey in 1916 decreased President Wilson’s vote share in the election held the same year. Authors also examined the effects of droughts and floods on voting behaviour in the 1896-2000 period and reported that floods had a negative effect on electoral support for the incumbent president’s party. Moreover, they point out that these irrationalities were the result of egotropic retrospective voting, by which citizens could not correctly attribute the changes in their economic status. The main assumption is that citizens engaged in blind retrospection, which means that when voters “are in pain they are likely to kick the government… In most cases, incumbents will pay at the polls for bad times, whether or not objective observers can find a rational basis for blame” 28 (p.118). This has implications for accountability in democracy, and Achen and Bartels 29 (p. 23) conclude that it “significantly degrades the efficacy of elections as mechanisms for inducing incumbent leaders to pursue their citizens’ subjective well-being”.

Since then, several studies have focused on the effects of irrelevant random events on voting behaviour. For example, it has been found that a win by the local college football team before Election Day led to an increase in the vote share of the incumbent party in presidential, senatorial, and gubernatorial elections in the team’s home county 30. Moreover, it has been shown that voters punished the incumbent party in presidential elections for damage caused by tornadoes 31. Furthermore, Gasper and Revees 32 reported that voters “punish” both presidents and governors for weather events. In a similar vein, using data on Indian elections, it has been reported that rainfall, just one standard deviation from the optimal level, reduced the vote share for incumbent coalitions by 3.25 percentage points 33. In addition, different types of natural disasters, like earthquakes, landslides, volcanic eruptions, and floods increase the likelihood of a political crisis through anti-government demonstrations, riots, guerrilla warfare, and intra-state conflict 34. Chang and Berdiev 35, using data from 156 countries in a period of 35 years, found that natural disasters, like floods, extreme temperatures, earthquakes, and windstorms, increased the probability that an incumbent government is replaced. Natural disasters have an impact on turnout as well. Using data on Hurricane Katrina, Sinclair, Hall and Alvarez 36 showed that flooding decreased participation in the following election. However, authors also showed that voters who experienced more than 6 ft of flooding were more likely to vote, compared to those who experienced less flooding. Furthermore, after the 2010-2011 floods in Pakistan, turnout was higher among citizens from impacted areas 37(but see 38).

As pointed out above, there is an extensive body of research that shows that citizens punish incumbents at elections following a natural disaster, which might seem irrational. The reason for this could be that natural disasters often cause resource scarcity, which leads to unequal distribution of resources, and may lead to low satisfaction with the present government, although it did not cause the scarcity 35. At first glance, it seems that these findings put democratic accountability in a tough position. However, Healy and Malhotra 17 point out that voters can and should hold government accountable for failure in relief after disasters. From their perspective, it could be even rational for voters to respond negatively to disasters because any negative event could be a signal of the incumbent’s competencies. Since it is costly to distinguish if that was the case, it is logical to punish the government, as a sort of “just-in-case” scenario. Moreover, the government could be responsible with respect to the way it deals with the disaster. For example, Arceneaux and Stein 39 investigated the effects of massive flooding after Tropical Storm Allison in Houston, which happened a few months before the mayoral election in 2001. They have found that, even though the government could not be blamed for natural disasters, voters evaluated government performance in terms of how it handled the disaster (see also Lay 40). If voters believed that the government could have done more to prevent the level of damage, they were willing to attribute blame and punish incumbents accordingly. Similar results were found in a study of effects of tornado damage on voting behavior 31. The incumbent government lost voters’ support when a formal declaration of natural disaster was absent. This finding suggests that voters punish the incumbent party if they perceive lack of competence in dealing with natural events. However, voters also reward incumbents. Healy and Malhotra 41 showed that relief spending had a positive impact on the vote, and Velez and Martin42 found that Hurricane Sandy had a positive impact on re-electing Barack Obama. Overall, it seems that voters’ immediate response to natural disasters is to punish incumbents, but if they perform well, i.e. give out disaster declarations or approve relief spending, they can counteract the negative tendency 32.

In order to investigate whether voters are indeed prone to punish politicians for events that are out of their control, the present study focuses on a previous unstudied context – the case of 2014’s and 2015’s floods in Croatia. Furthermore, we take into account some of the methodological issues of previous studies. The starting point was a recent study conducted by Fowler and Hall 43. They criticised Achen and Bartel’s 27 original study for the way certain counties were included/excluded, pointing out that 1916 shark attacks were not in any way a peculiar case of shark attacks. Fowler and Hall 43 collected data on every fatal shark attack in U.S. history and county-level returns from every presidential election in the period between 1872 and 2012 and found no effect of shark attacks on the president’s vote share. Based on their finding, the authors concluded that Achen and Bartel’s 27 results were most likely a false positive (for a similar re-evaluation of football game effects on voting behaviour see Fowler & Montagnes 44). Keeping these contradictory findings in mind, we conceptualised natural disasters as a form of natural experiment with an experimental and control group. We were also aware that disasters do not create equivalent groups, which means that we cannot just assume that our research has the form of a natural experiment, we need to ensure that this indeed is the case 38. One way to achieve this is by using a matching method technique. This technique ensures that all areas affected by natural events are matched to non-affected areas on several control variables which are important for explaining the vote share (e.g. lagged vote share, unemployment rate, mean income, population size, etc.). In this way, “starting” differences between affected and non-affected counties are accounted for, which leads to more reliable results regarding the impact of natural disasters on voting behaviour (for an example see 42).

Floods in Croatia

During 2014, a total of 148 Croatian municipalities experienced floods, with two major events being floods in February and May. During the February floods, the Sisak and Karlovac areas were affected by high water levels in the rivers Sava, Kupa, and Odra. For ten days, in certain areas, the Sava had water levels over 7.7 metres, the Kupa had a maximum water level of 8.27 metres, and water levels at the Odra mouth were 8.9 metres, which was a historical maximum 45. Due to above average rainfall in April 2014, the Sava with its tributaries experienced extreme water levels during the May floods. The most affected were Vukovar-Srijem and Brod-Posavina Counties, for which 214 kilometres of Sava were defended against overflowing. For all eight measuring stations in that area, water levels were higher than the historical maximum (sometimes over a metre higher). At certain places the embankments did not hold, causing floods in municipalities (such as Rajevo Selo, Gunja, and Račinovci), and evacuations of residents 46. TThe government response was swift with almost four thousand firefighters and soldiers deployed. Among other things, over 1.5 million bags of sand and over 300 hundred machines were stationed in the affected regions 47. To help the affected areas, the public raised over 50 million HRK (≈7,400,000 USD)48. The governmental officials, including the Prime Minister, Minister of Internal Affairs, Minister of Agriculture, and the President of Croatia, were present in the affected areas. However, there was public outrage at the way the floods were handled, mainly because the embankments did not hold the water 49.

Compared to the floods in 2014, the number of floods was much lower in 2015, with 34 municipalities being affected. The floods in May and October were less severe than those from the previous year, with the Samobor and Karlovac areas being hit. Still, at certain areas, the embankments did not hold, and the water level of the Kupa was higher than 8 metres. Due to better weather conditions, and more work being done for the reconstruction and improvement of the embankments, 2015 was characterised as being a “normal” year for floods 50.

In Croatia, activities regarding floods are bound by the legal framework on natural disasters. The Croatian Act on Protection against Natural Disasters 70 states that the mayor of a county oversees the declaring of a natural disaster (art. 6). After a natural disaster is declared, a county’s Committee for Natural Disasters, following the prescribed Methodology for making threat assessments 71, assesses the damage for all municipalities within the county (art. 38, par. 2). The assessments are reviewed by the National Committee for Natural Disasters, which proposes the amount of relief spending (art. 38, par. 1). The final approval of relief spending is made by the government (art. 33, par. 3).

The present study

The goal of this study was twofold. First, we wanted to investigate whether floods had an impact on the following elections, i.e. did Croatian citizens in flooded areas punish incumbents for the floods. Second, we wanted to examine whether relief spending had an impact on the voting behaviour of citizens affected by the floods. In other words, did citizens in flooded areas reward or punish incumbents based on the presence of financial help? To answer these research questions, a natural experiment was conducted, with independent variables being presence of flood and presence of relief spending, and dependent variables being change in incumbent’s vote share and turnout from previous to the current election.

Data from four elections was used: the 2009 and 2014 presidential elections, and the 2011 and 2015 parliamentary elections. Both presidential and parliamentary elections featured an incumbent running for re-election. For each incumbent, we used his (or his party’s) electoral success from the previous election (i.e. success in the 2014 presidential election was compared with success in the 2009 presidential election, and for the 2015 parliamentary election, we used data on electoral success in the 2011 parliamentary election). This way we could take into account the overall change in support within each election cycle. The first round of the presidential election was held on 28th December 2014, so we were able to investigate more immediate effects of the 2014 floods on voting behaviour. To investigate more long-term effects of floods, we used the parliamentary election which was held on 8th November 2015. This election was also used to see if relief spending had an impact on voting behaviour.


Research design

In the present study, floods were considered as a natural experiment, where some municipalities formed a treatment group (flooded areas) and others a control group (non-flooded areas). This should not imply that causal mechanisms can be easily discussed because one cannot be a priori certain that the only difference between these two groups is exposure to floods. Lay 40 (p. 649) points this out by stating that “disasters do not affect all people equally”. Additionally, this method falls within the scope of observational design in which “treatment is not randomly assigned but rather passively observed” 51. This means that the differences in dependent variables can be a result of bias, and not the treatment itself. Since floods cannot be controlled as an experimental variable, and random assignment of municipalities to either group is not possible, we are forced to deal with this issue post hoc. While this can be done via regression models, there are several issues with that type of analysis that can result in remaining biases that confound the causal findings (more details in 52,53). To address this issue, researchers have focused on the use of propensity scores, which are a “probability…that a subject will be treated based on an observed group of covariates.”52 (p. 954; emphasis added). ). Propensity scores are used in a wide array of research areas 54 and have recently been used in assessing the impact of natural disasters on voting 42,55. By using propensity scores, we are able to calculate a unified probability of being affected by floods for each municipality, based on several variables that could impact the dependent variable (voting decision).

Next, based on the propensity scores, it was possible to create a subsample of municipalities that were not affected by floods and match them to those municipalities that were affected. In the present study, the nearest neighbour matching algorithm was employed 54 . Matches were formed based on the nearest propensity score. This method was adequate for our sample because the subsample of control (non-flooded) municipalities was large enough (Npresidential election = 404, Nparliamentary election = 402)56. To form propensity scores, several variables related to voting behaviour were used: unemployment rates, income per capita, municipal income per capita, level of education, lagged incumbent vote share, and lagged turnout.

Data collection

Three sets of data were collected: a) data on election results, b) data on flooded areas, and c) information used to form propensity scores.

a) Data on election results. Relevant information about the presidential and parliamentary elections was obtained from the State Electoral Commission of the Republic of Croatia (http://www.izbori.hr/). For each election, data was obtained in sheet form for each electoral district and was later combined into one sheet with data for each municipality. Municipality level data on the number of voters, total votes, and votes for the current incumbent was collected for elections before the floods (2009 and 2011) and elections following the floods (2014 and 2015).

b) Data on flooded areas. To identify flooded municipalities, we contacted each county (total of 21 counties, including the capital Zagreb as its own county), and received a formal assessment for which of the municipalities a natural disaster, specifically a flood, was proclaimed in 2014 and 2015. In addition, information on damage appraisal was obtained. To obtain data on relief spending, governmental decisions, which are available online (http://www.mfin.hr/hr/odluke-vlade-rh-o-dodjeli-sredstava-pomoci), were collected for all flood related relief spending in the period 2014-2015 for each municipality.

c) Information used to form propensity scores. In order to form propensity scores, to produce samples similar in their demographics, data on unemployment rates, income per capita, municipal income per capita, and level of education were collected from the Ministry of Regional Development and European Union funds (https://razvoj.gov.hr/o-ministarstvu/regionalnirazvoj/indeks-razvijenosti/112). The Ministry periodically measures these variables to form a development index (a composite measure of socioeconomic variables). This data was downloaded from the Ministry’s website for each municipality in Croatia. We used data from 2013, which was the last data point available.


Data was analysed with R 3.2.4., using MatchIt 57, dplyr 58, lm.beta 59, and car 60 packages. Analyses were done separately for presidential and parliamentary elections. Independent variables were the presence of flood and presence of relief spending, and dependent variables were change in incumbent’s vote share and turnout in the current election compared to previous.

For each of the elections, we reported results obtained before and after matching the affected and non-affected areas in key variables: unemployment rate, income per capita, municipal income per capita, level of education, lagged incumbent vote share, and lagged turnout. In addition, we investigated the impact of relief spending on the 2015 parliamentary election.

Flood impact on 2014 presidential election

To investigate the flood effect on the 2014 presidential election held in December, data on floods that took place in 2014 during February and May was used. Main results regarding flood effect on voting behaviour in the 2014 presidential election are reported in Table 1. Before matching flooded and non-flooded areas, results showed a significant difference between flooded and non-flooded municipalities in the decrease in support for the incumbent president. Flooded areas showed a higher drop in incumbent vote share in the election following the floods. At first glance, it might seem that voters in flooded areas punished the incumbent. However, there were also significant differences in all socioeconomic variables between flooded and non-flooded areas. Unflooded areas had lower unemployment, higher individual and municipal income per capita, and higher education level. This implied that the matching of affected and non-affected municipalities was warranted. To match the subsamples, MatchIt R package was used and the nearest neighbour method was applied. This resulted in a new sub-sample of non-flooded municipalities (the rest were excluded from the analysis).

After matching, there was no significant difference in any of the socioeconomic factors. Analysis using matched samples also revealed that there was no difference in turnout, nor in the incumbent vote. This means that, once we accounted for relevant variables, the difference between flooded and non-flooded areas in vote change disappeared. In other words, the impact of floods on the 2014 presidential election was absent.

Table 1. Flood effects on 2014 presidential election for matched and unmatched samples

Unmatched data Matched data
Non-flooded Flooded Non-flooded Flooded
Unemployment 17.34 (8.78) 23.46*** (9.1) 22.98 (9.67) 23.46 (9.1)
Income 22471.44 (6119.15) 19237.45*** (5288.13) 19299.04 (5620.64) 19237.45 (5288.13)
Municipal income 2220.85 (1830.85) 1340.43*** (1033.65) 1196.84 (824.99) 1340.43 (1033.65)
Education 70.56 (10.91) 63.23*** (9.74) 63.63 (10.53) 63.23 (9.74)
Lagged turnout (2009) 49.05 (8.94) 49.29 (7.51) 48.69 (10.03) 49.29 (7.51)
Lagged vote (2009) 60.64 (17.11) 60.64 (17.11) 61.39 (16.44) 60.09 (11.37)
Turnout change (2014 – 2009) 8.20 (6.64) 7.89 (5.4) 8.70 (6.38) 7.89 (5.40)
Vote change (2014 – 2009) -13.12 (7.56) -14.82* (6.62) -13.37 (8.64) -14.82 (6.62)
N 404 148 148 148

Sample means with standard deviation in the parenthesis for: Unemployment – average percentage of unemployed individuals 2010-2012, Income – average yearly income per capita 2010-2012 (in HRK), Municipal income – average yearly municipal income per capita 2010-2012 (in HRK), Education – percentage of population aged 16-65 with high school of higher 2011, Lagged vote – incumbent vote share before floods, Lagged turnout – turnout in election before floods

Ψ p<0.1; * p<0.05; ** p<0.01; *** <0.001

Flood impact on 2015 parliamentary election

For the 2015 parliamentary election held in November, we used data on floods that took place in 2014 during February and May, and/or 2015 during May and October, which was in concordance with the time-frame used in previous research 36,41,61. Results showing the impact of the 2014-2015 floods on the 2015 parliamentary election can be seen in Table 2.

Table 2. Flood effects on 2015 parliamentary election for matched and unmatched samples

Unmatched data Matched data
Non-flooded Flooded Non-flooded Flooded
Unemployment 17.22 (8.69) 23.59*** (9.1) 23.03 (8.91) 23.59 (9.1)
Income 22505.18 (6155.99) 19270.14*** (5247.34) 19327.11 (5363.62) 19270.14 (5247.34)
Municipal income 2236.21 (1831.04) 1316.95*** (1011.99) 1135.46 (636.05) 1316.95Ψ (1011.99)
Education 70.57 (10.94) 63.44*** (9.68) 64.09 (10.13) 63.44 (9.68)
Lagged turnout (2011) 59.04 (7.23) 59.76 (6.19) 59.36 (7.46) 59.76 (6.19)
Lagged vote (2011) 38.04 (15.29) 35.15* (10.87) 34.68 (15.60) 35.15 (10.87)
Turnout change (2015-2011) -0.12 (6.30) -1.53** (4.54) -0.31 (6.69) -1.53Ψ (4.54)
Vote change (2015-2011) -7.13 (11.85) -5.34* (6.49) -5.22 (8.39) -5.34 (6.49)
N 402 154 154 154

Sample means with standard deviation in the parenthesis for: Unemployment – average percentage of unemployed individuals 2010-2012, Income – average yearly income per capita 2010-2012 (in HRK), Municipal income – average yearly municipal income per capita 2010-2012 (in HRK), Education – percentage of population aged 16-65 with high school of higher 2011, Lagged vote – incumbent vote share before floods, Lagged turnout – turnout in election before floods

Ψ p<0.1; * p<0.05; ** p<0.01; *** <0.001

Similar to the case of the presidential election, there was a significant difference in vote change before matching, but in the opposite direction – non-flooded areas showed a higher drop in incumbent vote share in the election following the floods. Also, the fall in turnout was more severe for the flooded areas. In addition, non-flooded areas had higher lagged turnout, lower unemployment rates, higher personal and municipal income, higher education levels, and higher lagged incumbent vote. Therefore, we repeated the matching procedure, which showed that all differences between the two subsamples disappeared after matching. Subsamples did not differ in any of the socioeconomic variables, lagged incumbent vote, nor vote change; the difference in turnout was marginally insignificant. To conclude, results indicated that floods did not appear to influence voting behaviour in the 2015 parliamentary election.

Relief spending impact on 2015 parliamentary election

Relief spending for floods in 2014 and 2015 was given during 2015, after the presidential and before the parliamentary election. Thus, data on government relief was used to investigate whether it had an impact on voting in the 2015 parliamentary election. Relief spending was approved for 103 out of 154 municipalities, and the amount of relief spending had a median of 3.1% of the amount of estimated flood damage. First, we investigated whether relief approval (approved vs. not approved) had an impact on vote change between the two elections by conducting a regression analysis with relief spending and other political and socio-demographic variables as predictors. The results are summarised in Table 3.

Table 3. Predicting the parliamentary election vote change (2015-2011) based on the approval of relief spending, political and socio-demographic variables

B β
Intercept -7.951 (5.156) /
Lagged vote (2011) -0.209 (0.057) -0.348***
Relief approved 0.176 (1.409) 0.013
Income -0.0002 (0.0001) -0.128
Municipal income -0.0009 (0.0005) -0.136
Education 0.221 (0.079) 0.329**
Unenmployment 0.001 (0.079) 0.002
R2 0.183 /
N 154 /

Standard errors are in parenthesis; Lagged vote – incumbent vote share before floods, Relief approved – 0 (not approved) and 1 (approved), Income – average yearly income per capita 2010-2012 (in HRK), Municipal income – average yearly municipal income per capita 2010-2012 (in HRK), Education – percentage of population aged 16-65 with high school of higher 2011, Unemployment – average percentage of unemployed individuals 2010-2012

** p<0.01; *** <0.001

Regression model was statistically significant (F(6, 147)=5.494; p<0.001). Specifically, we could predict the change in vote for the incumbent (2015-2011) for the flooded areas based on the lagged incumbent vote (2011) and the educational structure of the local population. Keeping other variables constant, one percent vote increase for the incumbent in 2011 resulted in 0.21 percent fall in the 2015-2011 vote change; one percent increase in share of people with high school or higher education resulted in 0.22 percent increase in 2015-2011 vote change. However, relief approval did not predict vote change. Furthermore, we conducted regression analysis in order to predict vote change with the percentage of approved relief by including only those municipalities that received flood relief. Besides the percentage of approved relief, predictors were political and socio-demographic variables. The model was not significant (F(6, 96)=2.079; p=0.061). Therefore, it could be concluded that the presence of relief spending did not have an impact on vote change.


The backdrop of studies which focus on the impact of politically irrelevant influences on citizens’ behaviour is the concept of the (retrospectively) rational voter, according to which voters use elections as a means of punishing or rewarding incumbents. If they do this rationally, then their role in the political system is justified; i.e. the representative model of democracy is justified 62. However, if citizens are prone to making mistakes, like irrationally blaming the government for things which are not under its control, citizens’ role in the political system is not justified. Thus, we turn to other democratic models, such as elite, or minimalist democracy 63. A newer take on this discussion flourished with Achen and Bartel’s study 27,28 in which they showed that citizens punished the American incumbent president for shark attacks. Several other studies showed that citizens’ political behaviour is influenced by weather events, such as natural disasters 31,32,33,35 . They paint a picture of irrational citizens and point to the inability of elections to be adequate mechanisms of democratic accountability.

Following these studies, we attempted to replicate their findings for the case of floods in Croatia in 2014-2015. The goal of this research was to investigate whether floods, a form of natural disaster, had an impact on the following elections. Moreover, we wanted to check whether the government’s response to the floods, in the form of relief spending, had an impact.

It is important to note the methodological differences between present and several previous studies. In most of the previous studies, authors conducted regression analysis with the dependent variable usually being the change in incumbent vote share, the independent variable being the presence of a natural disaster, and several control variables, such as county size or lagged vote share 27,32,33,35,64. However, some authors have argued that this approach is not adequate because of the assumption that natural disasters function as random, exogenous shocks to the political system 42. Bodet, Thomas and Tessier 38 (p. 92) state clearly that “assuming the natural disaster acts as an as-if random exogenous shock, and treating it as a natural experiment, can lead to incorrect conclusions…studies that wish to treat natural disasters as experiments should justify this analytical choice by demonstrating the disaster produces equivalent experimental and control groups. Failing to do so would, in this case, lead us to inappropriately reject the null hypothesis”. Thus, it is possible to pinpoint the influence of natural disasters on voting behaviour only if flooded and non-flooded areas are comparable on variables that could have influenced the incumbent vote share. Moving from conceptual level toward the statistical one, the same problem arises when regression models are used in observational studies; i.e. “when there are large differences in distribution of covariates between treatment groups, adjusting for these differences with conventional multivariable techniques may not adequately balance the groups, and the remaining bias may limit valid causal inference” 52 (p. 954). One way to deal with this issue is to use matching method technique, which has been used in several previous studies 38,42,55 .

In the present study, when we compared the flooded and non-flooded areas, it seemed that the people from flooded areas punished the incumbents in both elections. However, once we accounted for differences in flooded and non-flooded areas in several sociodemographic and political variables, the effect disappeared. In other words, by using appropriate methodology (matching method technique), we found no effect of floods either on incumbent vote share or in the turnout for the following elections.

The conclusion following these results can fall into one of three categories. First, there is a possibility that floods usually have an impact on voting behaviour, but the current research has methodological issues, or the floods were not disastrous enough to elicit a response among the population of flooded areas. Considering the methodological issues, we feel that we have followed the above-mentioned suggestions on both the conceptual and statistical level. Also, we have obtained data from official sources considering vote shares, flood presence, and damage, as well as municipal level data on income, education, and unemployment. Considering that the floods studied in this paper were the biggest in (recorded) Croatian history, that they resulted in the destruction of several villages, and displacement of more than 13,000 people (and two dead)65 , we can conclude that they were disastrous enough to elicit a response. Even though it’s not unusual for studies to focus on a single case of a natural disaster (e.g. 36,39,40,42,61,38 ), the certainty of our conclusion would be improved by using a wider timeframe (e.g. 32,35,41).

Following the second argument from this category, that the effect of natural disasters was not found because the disaster was not big enough, was deemed an “act of God”, or the government was not blamed, etc. 27 , we think that this idea opens the possibility that we deem all null-results as being the result of a “bad” natural disaster, or to put it in other words – if the disaster would have been more extreme, we would have found an impact (for a similar discussion see 43 ).

A second possible conclusion drawn from the present results is related to the idea that natural disasters usually have an impact on political behaviour, but the negative response was counteracted by other variables, mainly concerning the government’s response to the disaster. Several studies have shown that natural disasters have a negative impact on the incumbent’s vote share in the following election 31,32,33,39 . Other studies have found that this negative impact can be lessened or counteracted by announcing a formal disaster declaration 31 , relief spending 32 , or by a general positive evaluation of the government’s performance following a disaster 39,40.

Moving back to our results; when we tried to predict voting behaviour within flooded areas, we found that the presence of relief spending had no effect on incumbent vote share. We could conclude that the relief was too low, or given too late, to counteract the impact of floods on the vote, but then we would expect to find the difference between flooded and non-flooded areas in vote change. There are several other possibilities. First, it is possible that some other aspect of voting behaviour was responsible for these results, such as the (dis)satisfaction with the way the government handled the floods. These individual-level variables were not available for this research, but even if they were, we would still then expect to see the overall effect of floods on incumbent vote share. Second, it is possible that we stumbled upon a floor effect. The incumbent party in the period 2011-2015 was the Social Democratic Party of Croatia (SDP) with its coalition partners. In 2011 SDP won 52.98% of votes, but only 37.09% in 2015. Thus, it is possible that the overall fall in support masked the effects of the floods. However, when we look at the presidential election, the incumbent in the period 2009-2014 was Ivo Josipović, who won 60.26% in 2009 and 49.26% in 2014. This difference in vote change is smaller for this election, and the overall incumbent vote share in 2014 is relatively high compared to the parliamentary election. Still, the flood effect on voting behaviour was absent in both elections. However, it’s important to note that the number of flooded areas was too small to use a matching method technique, so we used a regression analysis in which we tried to predict the incumbent vote change based on relief spending, several socio-demographic and political control variables. A suggestion for future studies would be to find cases of natural disasters within which there are enough cases of flooded areas that received and those that did not receive relief spending.

Finally, it is possible that floods indeed do not have an impact on the voting behaviour of citizens, and it seems our results point to this conclusion. However, we must attend to several issues. First, there are more studies published that found an effect of floods than vice versa 37,43,67 . This could be the case of publication bias 68 , or, as Fowler and Hall 43 pointed out and showed in their analysis, published articles could be based on false positive results. If that was the case, then one could expect that using more adequate methodology and statistical analysis would yield clearer results. Unfortunately, studies which have used matched methods have found negative flood effects on the incumbent vote 55 , positive effects 42 , and no effects 38 , as in the present study. Second, even though the goal of the study was to replicate previous findings in the previously unstudied context of floods in Croatia, we used two cases of natural disasters within a single country, which constrains the generalisation of our findings.

Thus, it seems that the verdict on the impact of natural disasters is not clear cut, and should not be taken as such. Even if we look at research done with adequate methodology, we find results going in all directions, with our results showing that floods did not have an impact on the incumbent’s vote share or voter turnout in two elections. Further research is needed, as well as opening up to publishing null results. First, we would suggest matched method as the methodology of choice for this type of research. Second, since results seem to vary in direction, it is probable that some moderator or mediator variables should be considered. For example, one could use type of natural disaster (which could be differently portrayed as “an act of God” versus “an act of Man” 69 , the incumbent’s approval in the period before the disaster, different legislature on dealing with and helping after natural disasters, public support for the affected areas, etc. Third, it could be useful to use more individual-level data, as in some previous studies. For example, survey studies can assess the way voters perceive the amount of destruction, as well as the way the incumbent government dealt with the disaster 39,40 . Studies can also combine individual level data with macro variables 36. Finally, we suggest using multilevel analysis, which can be used to combine data on political knowledge, political cynicism, ideology, etc. with county-level data to get a clearer idea of the impact of natural disasters on political behaviour.


The goal of the present study was to study the impact of natural disasters on voting behaviour for the context of the 2014 and 2015 floods in Croatia. Results showed that, once socio-demographic and political differences between the flooded and non-flooded areas were controlled, floods did not have an impact on voting behaviour in the 2014 presidential or in the 2015 parliamentary election in Croatia. Furthermore, results indicated that neither presence of relief spending nor its amount had an impact on voting behaviour. However, keeping in mind the constraints of this study, further research is needed to get a clearer picture of the way natural disasters are related to political behaviour.

Competing interests

The authors have declared that no competing interests exist.

Corresponding author

Kosta Bovan: kosta.bovan@fpzg.hr

Data Availability

All supplemental materials used for this article (data and R scripts) can be download via https://figshare.com/articles/Supplemental_materials_-_Do_natural_disasters_affect_voting_behavior_Evidence_from_Croatian_floods_/5821674

http://currents.plos.org/disasters/article/do-natural-disasters-affect-voting-behavior-evidence-from-croatian-floods/feed/ 0
Comparison of Jurisdictional Seismic Resilience Planning Initiatives http://currents.plos.org/disasters/article/comparison-of-jurisdictional-seismic-resilience-planning-initiatives/ http://currents.plos.org/disasters/article/comparison-of-jurisdictional-seismic-resilience-planning-initiatives/#respond Thu, 29 Mar 2018 14:00:24 +0000 http://currents.plos.org/disasters/?post_type=article&p=38045 Introduction: Three jurisdictional earthquake resilience planning initiatives respectively conducted in California, Washington State, and Oregon are compared: SPUR Resilient City (SPUR), Resilient Washington State (RWS), and Oregon Resilience Plan (ORP). This paper presents an exploratory analysis that reveals divergent and convergent themes across the resilience planning initiatives, with the goal of informing similar initiatives in the future.

Methods: Data for this exploratory study comes primarily from the reports produced by the initiatives, but also initiative presentations, limited correspondence with initiative organizers for clarifications, and personal experience. Extensive computer-assisted text analysis was done to analyze, synthesize, and visualize the content of the SPUR, RWS, and ORP reports Results: The SPUR initiative was the inspiration for both RWS and ORP. As such, an evolution of ideas is evident from the first initiative (SPUR) to the most recent (ORP). While the SPUR initiative was a model for the RWS and ORP initiatives, the process and outcomes of the latter two initiatives were more similar than to the original SPUR initiative. Both the RWS and ORP initiatives were significantly smaller in scope. These two initiatives also made creation of recovery-based performance measurement frameworks--timetables of expected and desired recovery estimates--even more central to the process of identifying seismic resilience recommendations.

Discussion: The SPUR, RWS, and ORP initiatives have had demonstrated impact on jurisdictional pre-event planning, mitigation, and preparedness efforts. However, the impact of the specific innovations developed by the three earthquake resilience planning initiatives is not clear because of the limited degree that the resilience definition and performance measurement framework for each initiative were explicitly integrated to produce the respective recommendations. For example, Washington State's Seismic Safety Committee made recommendations similar to their RWS recommendations as part of past initiatives that did not use a resilience lens or a recovery-based performance measurement framework.

Conclusion: More systematic research into the innovative elements of the SPUR, RWS, and ORP initiatives, such as development of the recovery-based performance measurement frameworks, is warranted given the initiatives’ popularity and influence on the National Institute of Standards and Technology’s Community Resilience Planning Guide.



The purpose of this review is to analyze and synthesize three jurisdictional earthquake resilience planning and policy initiatives conducted in California, Washington State, and Oregon. The study provides evidence-based information for use in evaluating the analyzed initiatives, as well as related current and future initiatives. An objective of the study is to reveal divergent and convergent themes across the initiatives and their respective documents, as well as to help identify possibly overlooked or underemphasized themes. The intention is to understand the unique contributions from the approaches taken by the three seismic resilience planning initiatives, as well as the resulting documents and outcomes. Data for this study comes primarily from the reports of the initiative, but also initiative presentations, personal experience, limited interviews of initiative participants for clarifications, and feedback from participants. In addition, extensive computer-assisted text analysis was done to summarize and compare initiative reports. Ultimately, it is hoped that this study engenders an informed discussion and systematic research on seismic resilience planning initiatives.

SPUR (San Francisco Planning + Urban Research Association) conducted the first of the seismic resilience planning initiatives. It was called the Resilient City initiative and it focused on the City and County of San Francisco 1. The SPUR initiative produced nine reports, published between 2008 and 2013 2,3,4,5,6,7,8,9,10. SPUR Resilient City inspired two state-level earthquake resilience initiatives conducted by the State of Washington Seismic Safety Committee (WASSC) and the Oregon Seismic Safety Policy Advisory Commission (OSSPAC). The Resilient Washington State (RWS) final report and the Oregon Resilience Plan (ORP) were published in 2012 and 2013, respectively 11,12. The author is a member of the WASSC and played a significant role in the design and facilitation of RWS.

This article begins with a brief overview of the SPUR, RWS, and ORP initiative processes to put the initiative reports into context. The third section attempts to differentiate and synthesize the content across the SPUR, RWS, and ORP reports. The fourth section attempts to reconcile a core feature across the initiatives. The SPUR, RWS, and ORP documents all include timetables — referred to in the initiatives as performance measurement frameworks — that present estimates of expected and desired target recovery times for different assumed sectors of resilience, such as housing, utilities, and economic development. A sample of estimates from each initiative are converted and reformatted to facilitate comparison. The fifth section of the report provides an overview of the respective recommendations that are presented in the SPUR, RWS, and ORP documents. Subsequently, the sixth section describes the relative progress towards implementation of the respective recommendations. The paper concludes with a critique of insights revealed from the study in context with potential future earthquake resilience planning initiatives.

Seismic Resilience Planning Processes

This section provides an overview of the motivation and approach taken for each of three earthquake resilience initiatives. Of course, the story of each initiative is much longer and nuanced than can be summarized in a single article. The initiation of each effort was distinctly sequential: SPUR Resilient City began first, then RWS, and ORP third. This sequence allowed WASSC and OSSPAC to review and incorporate ideas from what had been done to that point by the other initiatives. It is possible then to see what ideas were carried forward from each initiative, what ideas were not, and what new ideas were required to achieve objectives unique to a particular initiative. The approaches are described in sequence below. Throughout the article, an attempt is made to describe features and insights related to each initiative and the respective reports in sequential order. This facilitates seeing the evolution of ideas for doing earthquake resilience planning.

Before describing the process and products of each of the initiatives, it is insightful to present the working definitions of resilience used for the respective initiatives. In each case a contextual definition of resilience was developed for the purpose of guiding the different multi-stakeholder jurisdictional-level initiatives. For each initiative, developing a unique definition was seen as a way to build a shared understanding among participants and stakeholders regarding the goal of each process and inform the jurisdictionally-specific recommendations for increasing seismic resilience. These definitions are juxtaposed for comparison in Table 1.

Table 1

Definitions of resilience developed for the SPUR, RWS, and ORP initiatives, respectively.

SPUR: “[A resilient San Francisco has the ability] to remain safe and usable after a major earthquake A resilient city is able to contain the effects of earthquakes when they occur, carry out recovery activities in ways that minimize social disruption, and rebuild following earthquakes in ways that mitigate the effects of future earthquakes.”7
RWS: “[A resilient Washington State] is one that maintains services and livelihoods after an earthquake. In the event that services and livelihoods are disrupted, recovery occurs rapidly, with minimal social disruption, and results in a new and better condition.”11
ORP: “[In a resilient Oregon State] citizens will not only be protected from life-threatening physical harm, but because of risk reduction measures and pre-disaster planning, communities will recover more quickly and with less continuing vulnerability following a Cascadia subduction zone earthquake and tsunami.”12

The RWS definition takes some inspiration from SPUR’s definition. Both focus on minimizing social disruption during the recovery process, for example. Definitions for both SPUR and ORP focus on maintaining safety in the face of earthquakes, while the RWS presents a higher standard of maintaining livelihoods of residents. SPUR and ORP state goals of improving mitigation or reducing vulnerability during recovery; the RWS recovery goal focuses on more than just seismic safety.

SPUR Resilient City

SPUR brands the Resilient City initiative as “creating a new framework for disaster planning” 1—a claim that manifested by the initiative’s influence on subsequent earthquake resilience planning initiatives. The initiative was a volunteer effort, but some funding for the initiative was provided by Degenkolb Engineers, U.S. Geological Survey, and U.S. Department of Homeland Security. The intended audience of the initiative is not explicitly stated in the initiative reports, but most recommendations were aimed generally or specifically at the executive and legislative branches of the City and County of San Francisco.

The initiative was initially organized by SPUR’s Disaster Planning Policy Committee to evaluate the seismic performance needs of buildings and infrastructure in San Francisco at a city scale based on the initiative’s definition of resilience. The original goals of the initiative as stated in its first report are the following 7:

  1. Define the concept of resilience in the context of disaster planning
  2. Establish performance goals for the expected earthquake that supports [the] definition of resilience
  3. Define transparent performance measures that help … reach [the] performance goals
  4. Suggest next steps for San Francisco’s new buildings, existing buildings and lifelines [transportation and utility infrastructure]

The SPUR Disaster Planning Policy Committee first created the Before the Disaster Force of about sixteen volunteers to work on four reports of the Resilient City initiative. Arguably, the work of this task force and their four reports are the biggest innovations from the initiative and what SPUR Resilient City is known for most. The four reports largely focus on defining earthquake engineering performance goals for buildings and infrastructure, as well as developing a means to measure progress towards these goals 5,6,7,8. The task force met once a month for over two years. The recommendations in the four reports are meant to be implemented over a 50-year timeframe. The time horizon was chosen to make costs for any new structural design and retrofit requirements or related policies politically daunting and financially onerous.

The Before the Disaster Task Force considered the performance of buildings and infrastructure for an expected M7.2 San Andreas earthquake through the lens of post-disaster recovery. While the M7.2 scenario was emphasized in defining San Francisco’s resilience and determining engineering performance goals, routine and extreme earthquake scenarios were also defined to provide context for the initiative. The task force worked to understand gaps between expected and desired recovery outcomes to facilitate working backwards to develop their policy and planning recommendations. To do this the task force innovated a new performance measurement framework in the form of a recovery timetable, which is presented in the task force’s first report 7.

The recovery timetable explicitly outlines how quickly certain aspects of San Francisco should recover from the expected earthquake to determine what levels of performance—relative to current performance levels—for new buildings, existing buildings, and lifelines are necessary for increasing San Francisco’s earthquake resilience. Specifically, the timetable depicts estimated anticipated and target recovery times for four categories of facilities and systems with respect to the expected earthquake scenario: 1) critical response facilities and support systems, 2) emergency housing and support systems, 3) housing and neighborhood infrastructure, and 4) community recovery. The recovery time estimates and targets for each category were determined by the task force using expert judgement, as well as outside consultation. The timetable was intended by the Before the Disaster Task Force as a tool for three sub-committees to develop reports and recommendations about existing buildings 8, new buildings 5, and lifelines 6, respectively.

The Resilience City initiative was significantly expanded beyond the work of the Before the Disaster Task Force to include other issues of earthquake resilience. Additional task forces were formed to make recommendations for better emergency response capacity, as well as increased capacity for rebuilding and recovering after a disaster. The various initiative task forces did not necessarily work closely together, but were aware of each other’s work.

A Shelter-in-Place Task Force was created to leverage the Before the Disaster Task Force recommendations and make their own recommendations to improve the likelihood that San Francisco residents can stay in their homes after a future earthquake while their structures are repaired 7. A Disaster Preparedness Task Force assessed and made recommendations to improve San Francisco’s culture of disaster preparedness 9 and its emergency response capacity 10. A report was created by a separate task force that outlines how San Francisco can rebuild its transportation infrastructure more quickly and effectively after a future earthquake 4. A Land Use Planning and Rebuilding Task Force published an extensive report laying out recommendations to increase the likelihood for successful post-earthquake rebuilding and recovery 2.

Resilient Washington State

The origination of Resilient Washington State initiative was directly tied to the outcome of the SPUR Resilient City initiative. A member of WASSC presented the SPUR performance measurement framework timetable to the committee, suggesting that WASCC adapt the city-scale initiative for Washington State. The WASCC member was aware of SPUR’s initiative because of their affiliation with Degenkolb Engineers. (The Before the Disaster Task Force chair was also affiliated with Degenkolb Engineers.)

The RWS initiative sought to develop a set of recommendations for improving the state’s earthquake resilience within a 50-year timeframe—the same timeframe used for the SPUR Resilient City initiative. The initiative’s primary audience was the governor, executive cabinet, and state legislature. The audience of RWS was more explicitly stated than for SPUR Resilient City and of course from a different jurisdictional level. Six volunteers from the WASSC created the RWS subcommittee. It was tasked with adapting the Resilient City initiative for the state-level, conducting the process, making recommendations, and drafting the initiative reports. The RWS subcommittee laid out four goals in their first report 13 (pg. 1), taking inspiration from the work of SPUR’s Before the Disaster Task Force. The RWS goals hew closely to the development of performance measurement framework timetables and do not reflect goals of the other SPUR Resilient City task forces:

  1. Evaluate the current condition of infrastructure in the state relative to earthquake resilience
  2. Develop targets for the desired level of performance
  3. Develop target timeframes for the restoration of services
  4. Prepare recommendations for statewide action to achieve desired targets

The RWS subcommittee meant infrastructure to refer to more than lifelines and, in fact, is less specific than the SPUR Before the Disaster Task Force’s scope of new buildings, existing buildings, and lifelines. Unlike SPUR, RWS did not emphasize a single expected earthquake because of the state scale—multiple earthquake scenarios are expected for the various seismic sources in the state. The RWS subcommittee based the initiative on multiple earthquake scenarios for different sources in the state, collectively referring to the scenarios as earthquakes of statewide significance. The scenarios provided a basis for discussing the state’s earthquake resilience, but were not individually considered explicitly in the analyses and recommendations of the initiative.

A kickoff workshop was held in September 2010 that involved approximately 50 volunteer participants from all levels of government, the private sector, and academia, representing disciplines such as civil engineering, emergency management, and human services. Following the kickoff workshop, four workgroups were organized around four statewide sectors: 1) critical services, 2) transportation, 3) utilities, and 4) housing and economic development. Taking inspiration from the development of the Resilient City performance measurement framework, the RWS sector work groups were asked to evaluate the current recovery capacity of their sector, develop recovery targets for state-level earthquake resilience, and prepare recommendations to achieve desired recovery targets. The four workgroups were given autonomy to achieve their objectives how they saw fit over the spring and summer of 2011. The work groups accomplished their work through ad hoc combinations of in-person meetings, teleconferences, online work (e.g., web surveys), and individual contributions. A second workshop was held in December 2011 to present and get feedback on the sector workgroup outcomes.

The RWS subcommittee synthesized the recommendations produced by the workgroups, which were reviewed by participants of the second workshop, to produce a final unranked list of ten detailed recommendations. Each of the ten recommendations included specific action items, as well as references to state agencies that could potentially lead implementation of the action items. The WASCC released the final report with these recommendations and the finalized performance measurement framework to the public 11. The report was presented to the state’s Emergency Management Council, which advises the governor and the director of the state’s military department, in 2012.

Oregon Resilience Plan

The Oregon Seismic Safety Policy Advisory Commission learned of the SPUR Resilient City and Resilient Washington State initiatives prior to the completion of either initiative. There were professional connections between the SPUR Disaster Planning Policy Committee, WASCC, and OSSPAC at the time that facilitated this awareness. More so, there was a direct connection between three members of the three jurisdiction’s committees because of their affiliation with Degenkolb Engineers. This connection heightened OSSPAC’s awareness of the first two initiatives and facilitated subsequent knowledge sharing.

OSSPAC created a volunteer steering committee to shepherd the ORP initiative. Similar to RWS, the ORP initiative’s primary audience was the governor and state legislature. The stated goals of the ORP initiative are similar to the SPUR Before the Disaster Task Force and the RWS initiative. Like the RWS initiative the ORP goals focus largely on development of performance measurement framework timetables; conversely, like SPUR Resilient City emphasize a single scenario earthquake 12 (pg. xiv):

  1. Determine the likely impacts of a magnitude 9.0 Cascadia earthquake and tsunami on [different] sector[s], and estimate the time required to restore functions in [each] sector if the earthquake were to strike under present conditions
  2. Define acceptable timeframes to restore functions after a future Cascadia earthquake to fulfill expected resilient performance
  3. Recommend changes in practice and policies that, if implemented during the next 50 years will allow Oregon to reach the desired resilience targets

The goals of the ORP initiative appear to be the broadest in scope compared to the SPUR’s Before the Disaster Task Force and the RWS initiative. Thus, the ORP’s scope is closer, but lesser, in scope to the entire nine-report scope of the SPUR Resilient City initiative than the original RWS initiative scope. To match this broader scope, eight task groups were established for developing the ORP. One task group was responsible for developing the earthquake and tsunami scenario. The other seven task groups conducted analysis and developed recommendations related to seven sectors: 1) business and workforce continuity, 2) coastal communities, 3) critical and essential buildings, 4) transportation, 5) energy, 6) information and communications, and 7) waste and wastewater. In sum, approximately 170 volunteers participated in the ORP task groups. Compared to the other initiatives, the ORP initiative emphasized lifeline sectors the most.

A kickoff workshop was held on January 26, 2012 so that the eight task groups could meet and form their work plans and approaches to collaboration. Task groups met at least once a month over roughly a year and were given freedom to collaborate as they saw fit. In July 2012, there was a check-in workshop to allow task groups to share information and consult with each other. Another larger meeting was held in October 2012 to focus on developing recommendations. The final 300-page ORP report was published in February 2013 and delivered by OSSPAC to the 77th Oregon Legislative Assembly on February 28, 2013. The initiative continued with the establishment of the governor’s Resilience Plan Implementation Task Force, mandated by Senate Bill 33. The governor’s task force developed a two-page report, published in 2014, that presented next steps for implementing the recommendations in the ORP.

Seismic Resilience Plan Contents

Considering the collective length of the SPUR, RWS, and ORP reports, it is difficult to manually compare the published results of the three initiatives. To facilitate more efficient comparison, two text analysis methods were employed to summarize the report contents: collocations and dispersion diagramming. These methods were implemented using the Python library called Natural Language Toolkit NLTKA. (A large variety of text analysis software and programming libraries exist; NLTK was chosen because of the author’s familiarity with the Python programming language, as well as the library’s popularity.) Collocations are meaningful phrases automatically extracted from a document having a specified length that appear prominently throughout the document’s text. Dispersion diagrams graphically locates each instance of a single word along the length of a corpus (set of documents).

Several sets of words were selected for conducting dispersion analysis of the reports published by each resilience planning initiative. These words were selected to nominally represent 1) phases of disaster management, 2) level of government, 3) critical services and socio-economic sectors, 4) buildings and housing, and 5) lifeline infrastructure. In cases where obvious synonyms exist, each synonym was analyzed to determine which word occurred more frequently. If there was significant difference between the initiatives, the choice of word was made based on the frequency of occurrence in the SPUR reports because it is the standard for the other initiatives. In cases where multiple reports were published (SPUR and RWS initiatives), the reports were concatenated in order of publication date before being analyzed.

Table 2 and Figure 1 summarize the content of all nine SPUR reports (shown in sequential order). The collocations communicate the place-based nature of SPUR Resilient City, as well as its intended focus on a local jurisdictional scale—very little focus on the state jurisdictional level. The collocations and dispersion diagram clearly show that the SPUR initiative addresses all phases of disaster management, which is not surprising given the initiative’s nine-report scope. Interestingly, it appears that the concept of resilience is not carried through the entire set of reports. Response and preparedness are not concurrent with the use of the term resilience. While recovery co-occurs with the use of resilience in the initial reports, it is prominently independent of the term resilience in the last two reports 2,4. Topically, the SPUR reports dedicate a lot of space to various discussions about the performance of city’s building and housing stock. While buildings are discussed throughout the reports, discussion of lifeline infrastructure is somewhat contained in the initial reports, with transportation issues having continued prominence in subsequent reports. Surprisingly little reference is given to aspects of critical services or socio-economic sectors. However, based on the collocations, the SPUR reports appear to be the only documents that make prominent reference to historical preservation, faith-based organizations, affordable housing, low-income populations, and land use planning.

Table 2

SPUR. Prominent two-word phrases (collocations) in SPUR’s Resilient City report series.

local governments; near term; long term; emergency management; planning departments; emergency response; disaster response; disaster recovery; red cross; responsible parties; east bay; bay bridge; golden gate; soft story; unreinforced masonry; wood frame; building code; historic preservation; safe enough; enough stay; shelter place; seismic performance; task force; faith based; transportation infrastructure; land use; hazard mitigation; low income; first responders; affordable housing


Fig. 1: –

SPUR. Occurrence of selected words across all SPUR reports. Tick marks indicate one occurrence at the location in a particular report. Reports are shown in topical order according to SPUR 2,3,4, 5,6,7,8,9,10.

The RWS reports are summarized by Table 3 and Figure 2. Based on the corresponding collocations, the RWS reports clearly focus most on the state jurisdictional scale and little reference is made to the city or local level—opposite the focus of the SPUR Resilient City reports. More specifically, the RWS reports make prominent reference to liquefaction hazard, law enforcement, backup generators, mutual aid crews, and economic development-relatively unique compared to the other initiatives. The collocations for the RWS reports show prominent references to phrases related to time. These relate to descriptions of the various RWS performance measurement timetables and their development. The RWS reports are much more succinct than the publications from the other initiatives; the reports do not include much background information or analysis and reflect the smaller stated scope of the initiative relative to the other two initiatives. As a result, the timetable references are more prominent in the RWS report in comparison to the other initiatives. The post-event phases of disaster management (response and recovery) are referenced more often than the pre-event phases of preparedness and mitigation. The phases of recovery and response are emphasized relatively equally throughout the reports. It appears that topical focus on critical services and socio-economic issues shifts after the first report. The topics of facilities and transportation are referenced most frequently throughout the reports. Closer inspection of the RWS reports reveals that the term facilities is used to refer to a wide range of critical built assets (e.g., commercial facilities, industrial facilities, daycare facilities, school facilities, medical facilities, etc.).

Table 3

RWS. Prominent two-word phrases (collocations) in the RWS report.

resilient state; short term; long term; current capacity; response recovery; timeframe current; target timeframe; anticipated timeframe; time required; timeframe response; 24 hours; week month; months year; law enforcement; mutual aid; emergency response; repair crews; critical services; cascadia subduction; cascadia zone; pump stations; following earthquake; backup power; life safety; minor


Fig. 2: –

RWS. Occurrence of selected words across all RWS reports. Tick marks indicate one occurrence at the location in the report. Reports are shown in order of publication date. 11,13,15

Table 4 and Figure 3 serve to summarize the Oregon Resilience Plan, including the two-page follow-up report by the Resilience Plan Implementation Task Force 12,14. Similar to RWS and opposite of SPUR Resilient City, ORP is most focused on state-level government, with prominent place-based references. Compared to the SPUR and RWS initiative, there is more discussion of connection between state and local government. The term resilience is used most concurrently with references to the recovery phase. Within the ORP, the post-disaster phases of response and recovery are emphasized over the pre-event phases. Reflecting the initiative’s broader goals, the latter topic is more prominent in the ORP than in the RWS reports, but less so than for SPUR Resilient City. Topically, discussion of critical services and socio-economic sectors is lighter than the focus on buildings, housing, and lifeline infrastructure. It’s worth noting, however, that the ORP frequently uses the term “critical buildings” to refer to critical services. Discussion of the built environment in the ORP exceeds that found in the RWS reports. In comparison to the SPUR reports, the ORP has roughly equivalent space given to discussion of buildings. But significantly more space is given o components of lifeline infrastructure, particularly energy infrastructure, reflecting the more numerous task groups, sectors, and participants of the ORP. The collocations from the ORP show uniquely prominent references to the tsunami hazard on Oregon’s coast, which is reflective of the initiative’s scenario earthquake and statewide scale.

Table 4

ORP. Prominent two-word phrases (collocations) in the ORP.

coastal communities; life safety; emergency response; long term; liquid fuel; natural gas; cei [critical energy infrastructure] hub; water wastewater; pump stations; treatment plants; information communications; life safety; public transit; police stations; state highway; raw sewage; unreinforced masonry; non-ductile concrete; subduction zone; zone event; ground shaking; tsunami inundation; inundation zone; lateral spreading; columbia river; klamath falls; willamette valley


Fig. 3: –

ORP. Occurrence of selected words within the Oregon Resilience Plan. 12,14 Single tick mark indicates one occurrence relative to the location in the plan.

Seismic Resilience Performance Measurement Frameworks

A central feature of the SPUR, RWS, and ORP initiatives are the respective performance measurements frameworks that take the form of timetables showing expected and target recovery times for key sectors. The role of the timetables was much more central for both the RWS and ORP initiatives than for SPUR Resilient City, which only utilized the timetables in two of nine reports. Both the RWS and ORP initiatives were explicitly organized around the construction of multiple sector-specific recovery timetables, while only the work of the SPUR Before the Disaster Task Force was organized this way. The scope of the RWS and ORP performance measurement frameworks and the effort to develop them exceed that for SPUR’s framework. Whereas the SPUR performance measurement framework is included in two tables across one and a half pages, the RWS framework spans four timetables filling three pages, while the ORP framework requires ten timetables that fill 16 pages of the plan. One could see a city-scale performance measurement framework being more detailed because of greater feasibility to do refined analysis. Or a state-level framework might be larger because of the greater geographic area. For these three initiatives, the difference is less likely an issue of the difference in jurisdictional scale and more a matter of emphasis between the three initiatives (e.g., as stated in their goals above).

It is informative to compare the timetables created by SPUR, RWS, and ORP given their persuasiveness and prominent role in the latter initiatives. A key question is how the expected and target recovery time estimates compare across initiatives. Unfortunately, comparison of the published timetables is difficult.

The initiatives organized and approached development of the timetables in a significantly different manner. There are not many instances where the sector components and facility types are identified in the same way to facilitate one-to-one comparison. SPUR made estimates for specific or named on-the-ground systems in only a couple of instances. RWS provides estimates for several specifically identified components of the state’s transportation network. The ORP presents a high level of detail both geographically and with respect to specific transportation system components. The approach that each initiative took to determine current and target recovery estimates varied considerably. In general, each subsequent initiative used more time, analysis, and people in determining the estimates. In other words, the process was least emphasized in the SPUR initiative and most so in the ORP initiative. For example, some ORP task groups utilized loss estimates produced by FEMA’s Hazus software to inform their assessment-the only initiative to publish that quantitative analysis was employed in determining current recovery time estimates. The ORP is also the only report to reference specific past earthquake disasters as a baseline for estimates. The approach taken by RWS split the middle between the SPUR and ORP approaches. Considerable time and perspectives were used, but with a relatively ad hoc or experiential approach to estimation similar to SPUR.

Another reason it is difficult to compare the recovery timetables across the initiatives is the lack of consistent temporal units and resolution. SPUR made distinct but nominal estimates in units of hours, days, and months. RWS made estimates using inconsistent time intervals using units of hours, days, weeks, months, and years. The format and content of the ORP timetables are a synthesis of the SPUR and RWS approaches. The units and resolution are typically similar to the SPUR timetables. But the approach taken for each ORP timetable is different for each sector (task group).

To facilitate comparison for this exploratory study, a subset of sector components was extracted from each plan and the corresponding estimates were converted into common units of days. For RWS and ORP, a significant degree of simplification was required in most instances to facilitate the comparison. The comparisons are grouped by three themes: critical services (Figure 4), buildings, housing and economic development (Figure 5), and lifeline infrastructure (Figure 6). No effort was made to normalize for geographic scale; it was assumed that each initiative focused on the entire area within their respective jurisdictional boundary.

Overall, estimates for both current and target recovery times vary widely across the initiatives, except for a few instances. For some components, estimates differ by more than 1000 days; in one instance a single RWS recovery estimate varies by 275 days. Critical services recovery estimates are most straightforward to compare across initiative—at least as interpreted for this study. There is little consistency between estimates for current critical services recovery times across the initiatives, but there is relative agreement for target recovery times. For the buildings, housing, and economic development sector, there is very little convergence across the three initiatives. However, it is more difficult than for critical services to create straightforward comparisons between sector components. Even though the SPUR initiative put a lot of effort into advocating and planning for sheltering in place, their performance framework presents the longest acceptable target recovery time for housing across the three initiatives. For the RWS initiative, it is apparently acceptable for housing recovery to take longer for multi-family dwellings than for single-family dwellings. Of the three sector themes in this section, recovery estimates for lifeline infrastructure are the most varied. A major reason for this is the large difference in how detailed each initiative was in representing the different lifeline sectors. SPUR’s estimate for current utilities recovery time is significantly different, in most cases, from the estimates generated for RWS and the ORP, which have more instances of similarity. The ORP and RWS initiatives included many participants in general and utility provider representatives specifically, while the SPUR initiative had more limited participation generally and specifically. With respect to estimated target recovery times, SPUR provides only one estimate for all lifeline components: 30 days. Whereas, target times for RWS and the ORP vary considerably with respect to each lifeline component (within and between initiatives).

Comparison of estimates for current recovery capacity (Current) and target recovery (Target) timeframes related to critical services. Time units have been converted to days. Common shading indicates category similarity across reports.

Fig. 4: –

Comparison of estimates for current recovery capacity (Current) and target recovery (Target) timeframes related to critical services. Time units have been converted to days. Common shading indicates category similarity across reports.

Comparison of estimates for current recovery capacity (Current) and target recovery (Target) timeframes related to housing and economic development. Time units have been converted to days. Common shading indicates category similarity across reports.

Fig. 5: –

Comparison of estimates for current recovery capacity (Current) and target recovery (Target) timeframes related to housing and economic development. Time units have been converted to days. Common shading indicates category similarity across reports.

Comparison of estimates for current recovery capacity (Current) and target recovery (Target) timeframes related to lifeline infrastructure. Time units have been converted to days. Common shading indicates category similarity across reports.

Fig. 6: –

Comparison of estimates for current recovery capacity (Current) and target recovery (Target) timeframes related to lifeline infrastructure. Time units have been converted to days. Common shading indicates category similarity across reports.

Seismic Resilience Plan Recommendations

Ultimately the purpose of the SPUR Resilient City, Resilient Washington State, and Oregon Resilience Plan initiatives is to affect change to build earthquake resilience. The documents published by each initiative lay out an extensive list of recommendations to catalyze this change. Although the initiatives share similar roots, the audiences, jurisdictional-levels, goals, approaches, and participants of the three initiatives result in different specific sets of recommendations. To facilitate some level of cross-comparison, the recommendations for all three initiatives were clustered into five loose themes. Because of the large number of recommendations and the overlapping nature of them, other themes can certainly be extracted. The five themes are 1) critical services, 2) buildings and housing, 3) lifeline infrastructure, 4) business and economic development, and 5) planning and preparedness. Of the five themes, critical services and business and economic development have the fewest recommendations. For the most part, economic issues were only considered in final Resilient City report on recovery 2. The ORP does not give much attention to economic issues either, as well as critical services. In contract, planning and preparedness is a peripheral focus of the RWS reports, reflecting the fact that RWS had the narrowest scoped goals of the three initiatives.

SPUR Resilient City

SPUR produced at least 272 recommendations. There are at least 23 recommendations related to critical services. At least 42 recommendations are about buildings and housing. For lifelines, roughly 91 recommendations are given, including a recommendation to form a lifelines council. 111 recommendations were identified related to planning and preparedness. In categorizing the recommendations, at least five are clearly associated with the theme of business and economic development, although the recommendations were not made specifically for that context. Table 5 lists several prominent two-word phrases (collocations) used in the recommendations extracted from the SPUR Resilient City reports. The collocations give a broad sense of what the SPUR recommendations emphasize.

Table 5

SPUR. Prominent two-word phrases (collocations) across all SPUR recommendations.

occupancy resumption; golden gate; mutual aid; building resumption; building occupancy; retrofit replacement; mutual agreements; aid agreements; soft story; bay area; park ride; reserve fleet; bay bridge; resumption plans; non-ductile concrete; concrete structures; occupancy plans; shelter place; reliably functional; restricted vehicle; facilities safe; low income; post disaster; historic preservation; safe functional; safe reliably; ride locations; ferry landings; replacement redundancy; financial incentives; spontaneous volunteers; buses required; city owned; park locations; parking lots; credits deductions

Of all the initiatives, SPUR puts the greatest emphasis and effort on pre-event recovery planning. SPUR makes a blanket recommendation to develop strategies for post-earthquake economic recovery as part of ongoing economic development efforts and advocates outreach to small businesses to inform them of building occupancy resumption programs. The SPUR initiative focused heavily on recommendations for pre- and post-event resilience of the transportation system, with an entire report dedicated to the topic 4. Considering the minor focus on utilities in their performance measurement framework, it is not surprising that less focus is given to making specific recommendations for other lifeline components compared to RWS and the ORP. The Resilient City report lays out many recommendations to improve seismic safety of existing and new buildings, including residential dwellings, such as mandatory retrofits of existing soft-story buildings and a program to mitigate critical existing non-ductile concrete buildings. Great emphasis is placed on enabling as many residents as possible to use their home as post-earthquake shelters. In general, the recommendations in the SPUR reports are more targeted than the RWS and ORP recommendations. The initiative focuses on a well-defined local jurisdiction, while the other initiatives focus on more heterogeneous jurisdictions that in some cases have limited authority over their scope of recommendations.

Resilient Washington State

The RWS initiative produced at least 153 recommendations. This is a considerable number given the smaller scope of the initiative compared SPUR Resilient City, but also likely reflects the larger number of RWS participants and perhaps the larger jurisdictional scale. It is often cited that the RWS initiative resulted in only ten recommendations. This refers to the ten categories of recommendations given in the final report’s executive summary. Each of the ten categories has additional recommendations (or action items). In addition, the second RWS report includes a significantly longer list of recommendations put forward by RWS workgroups and workshop participants 15. The RWS subcommittee analyzed the long list of recommendations (with the help of Western Washington University students) to create the consolidated priority recommendations included in the final report. Of the over 150 RWS recommendations, at least 41 deal explicitly with critical services; 33 with buildings and housing, 58 with lifelines infrastructure; 13 about business and economic development, and 8 for the theme of planning and preparedness.

Table 6 lists collocations from the recommendations extracted from the RWS reports to give an overall sense of the RWS recommendations. As already observed, fewer recommendations regarding general planning and preparedness efforts are given in the RWS reports in comparison to the other initiatives. For this theme, developing more and better information (e.g., higher resolution liquefaction hazard maps) connects most of the recommendations. The RWS reports provide the highest number of recommendations specific to business and economic development. RWS made multiple recommendations in relation to improving the resilience of buildings and housing, though often at a more basic level than SPUR (e.g., creating a statewide inventory of vulnerable buildings) because of the different jurisdictional roles between cities and states. Like SPUR, RWS highlights the issue of sheltering in place. RWS initiative recommendations related to critical services focus primarily on the assessment and mitigation of hospitals, as well as schools. Lifeline infrastructure is a major focus of the RWS recommendations. Like SPUR, RWS made a recommendation for a lifelines council, but at the state level. Greater emphasis on transportation is made in the RWS recommendations than for utilities lifelines. This is similar to the SPUR initiative, but the RWS recommendations are more numerous for each sector. Recommendations for electricity are most frequent in the RWS reports among utility-specific recommendations.

Table 6

RWS. Prominent two-word phrases (collocations) across all RWS recommendations.

mutual aid; following earthquake; local jurisdictions; operations plans; continuity operations; supply chains; continuity plans; robust continuity; robust operations; aid agreements; mutual agreements; rules regulations; school districts; service providers; backup generators; transit agencies; enact requires; rerouting traffic; post disaster; actual stock; detailed actual; stock types; inventory actual; detailed inventory; enact legislation; legislation requires; self sufficient; interagency wsdot; vertical evacuation; agreements local; develop robust; restoration service; home owners; interagency agreements; retrofits across; facilitate rerouting; facilitate traffic; requires districts; state federal; requires school; building sold; owner building; assess risk; able withstand

Oregon Resilience Plan

The ORP document includes at least 103 recommendations. This is the case even though the ORP document is similar in length to all nine of the SPUR reports combined and was created by the most participants focused on the most sectors. At least 19 recommendations can be related to the theme of critical services. 18 recommendations relate to the theme of buildings and housing. 42 or more were associated with lifelines. Business and economic development was not explicitly assessed, but at least 6 recommendations are associated with the theme for this study. Lastly, 18 or more recommendations were given that deal with planning and preparedness. Table 7 lists identified collocations in the recommendations that were extracted from the ORP document.

Table 7

ORP. Prominent two-word phrases (collocations) across all ORP recommendations.

cascadia subduction; subduction zone; earthquakes tsunamis; rating system; wastewater agencies; seismic vulnerability; vulnerability assessments; communities develop; comprehensive plans; coastal communities; compliance applicable; number days; following cascadia; citizen expect; days citizen; expect wait; homes multi-family; interoperability among; sharing among; sharing interoperability; statutory prescriptive; within local; recovery efforts; retrofit existing; sector companies; retirement full; accelerate retirement; accelerate full; full upgrade; retirement upgrade; gap analysis; help coordinated; periodic updates; permitting processes; position help; prescriptive routine; prescriptive waiver; quality DEQ; environmental DEQ; regulatory applicable; regulatory compliance; routine permitting; transit rail

Like SPUR, planning and preparedness was an explicit part of the ORP initiative (not so for RWS). Related recommendations are mostly related to coastal/tsunami issues, as well as the need to promote public self-sufficiency for a longer time than has previously been communicated. Recommendations are not specifically listed for all components of critical and essential buildings that were assessed during the initiative, though particular attention is given to critical services within the tsunami zone. ORP recommendations focus significantly on business continuity, again with an emphasis on coastal communities. Similar to both RWS and SPUR Resilient City, the ORP advocates inventorying and evaluation of critical buildings. In addition, recommendations are made for a building vulnerability rating system and incentives for owners to upgrade existing vulnerable buildings. Like RWS and SPUR, transportation is a major area of concern within the ORP recommendations, particularly related to maintaining immediate post-earthquake access. A number of recommendations are given specific to utility lifelines, with particular emphasis given to wastewater systems compared to the other two initiatives. A unique recommendation across the initiatives was to create a state-level resilience officer position.

Seismic Resilience Plan Implementation Progress

Between the three seismic resilience initiatives—SPUR Resilient City, Resilient Washington State, and the Oregon Resilience Plan—approximately 500 recommendations have been put forward that might increase city- and state-level resilience to earthquakes. Implementation (and planning for implementation) of these recommendations is at different stages for each initiative and always will be. Regardless, only limited progress should be expected at this point given the multiplicity of recommendations and the five-decades long implementation horizon for each initiative. It is premature to systematically evaluate the individual effectiveness of the SPUR, RWS, ORP initiatives. The three initiatives have certainly had and likely will have demonstrably positive impacts in their respective jurisdictions, even if just raising public and political awareness.

It is difficult to determine the specific impacts of each initiative within their respective jurisdictions and any progress described below is now out of date. But at the time of writing, the RWS initiative has made least amount of progress and impact. Considering the higher stature of seismic resilience issues in San Francisco compared to Oregon, it is easy to argue that the ORP initiative has achieved as much or more relative progress and impact as SPUR Resilient City. The different implementation progress between the three initiatives reflects the relative amount of policy activity and coverage between the three states 16. Washington State has lower state-level policy activity and coverage related to earthquakes than Oregon, even though Oregon has significantly lower seismic risk. Whereas, Oregon has the most seismic policy activity and coverage in the United States other than California. For example, as of the time writing the Washington State Seismic Safety Committee, which is not a mandated body, has met only once in the past three years. This reduces the capacity for RWS organizers to advocate for RWS recommendations. In contrast, the Oregon Seismic Safety Policy Advisory Commission is a legally mandated body that meets on a bi-monthly basis, have a state legislator on the body, and publish meeting minutes. It is conceivable that implementation progress for each seismic resilience plan is more reflective of each jurisdiction’s political activity and culture on the topic of seismic resilience in general than because of the specific process or outcomes of the resilience planning initiatives. Of course, the jurisdictions of the initiatives are different scales. It is also possible that it is politically and practically easier to make progress at the local level, which favors success for the SPUR initiative.

Unfortunately, all three initiatives have relatively little content focused on the recovery goals stated in their respective definitions of resilience for their jurisdiction. The initiatives put a great deal of emphasis on the goal of increasing the speed of recovery. Of course, recovery time is the core of their respective performance measurement frameworks. However, moderate to no attention is given within the initiative reports on how to minimize social disruption (SPUR and RWS), reduce vulnerabilities (ORP), mitigate effects of future earthquakes (SPUR), or facilitate new and better conditions (RWS) beyond simply building anew. The performance measurement frameworks of all three initiatives make no consideration of these goals. The SPUR Resilient City initiative did publish an entire report focused on making recovery recommendations 2. However, the RWS and ORP initiatives use recovery to inform and prioritize recommendations for mitigation and preparedness (e.g., drills and continuity planning), rather than how to plan for improved post-earthquake recovery and building back better.

SPUR Resilient City

A significant measure of progress for the SPUR Resilient City initiative is just the number of task forces, corresponding reports, covered topics, and years of sustained volunteer work—well beyond the states of the RWS and ORP initiatives after similar durations. Additionally, the initiative has already had impact within the City and County of San Francisco and, of course, its inspiration for other resilience building initiatives. SPUR Resilient City has likely hastened progress on San Francisco’s Community Action Plan for Seismic Safety (CAPSS). CAPSS lays out 50 tasks to be taken in 30 years to improve seismic safety. In April 2013, San Francisco passed an ordinance requiring the evaluation and retrofit some types of soft-story buildings. In the years since the last Resilient City report was published, San Francisco has allocated significant funds for improving seismic safety of city facilities. In February 2013, Pacific Gas and Electric announced allocation of $1.2 billion over five years to improve utilities infrastructure, with a focus on earthquake resilience. The City of San Francisco Earthquake Safety and Emergency Response Bond (Proposition A) was easily passed by a public vote in 2014. The San Francisco Lifelines Council was formed in 2009 and published an interdependencies study in 2014. In April 2016, San Francisco published their resilience strategy as part of their Rockefeller Foundation 100 Resilient Cities grant. The strategy involved SPUR Resilient City organizers and explicitly references the initiative as informing aspects of the strategy 17.

Resilient Washington State

The final report of the Resilient Washington State initiative was released in November of 2012. In conjunction with the report’s completion, the RWS subcommittee briefed the state’s governor, adjutant general, and emergency management director on the RWS findings and recommendations. The report received limited print media attention in the months following completion. The RWS initiative of course informed the approach taken by OSSPAC to develop the Oregon Resilience Plan. The most obvious influence is on the development of the recovery-focused performance measurement frameworks. The emphasis and detail of the frameworks in both the RWS and ORP approaches exceed what occurred in the completion of SPUR’s initiative. Since the RWS final report was published, significant progress has been made on Washington State’s Project Safe Haven-a program to promote vertical evacuation for areas on the coast in tsunami inundation zones—a recommendation presented in the RWS reports. In April 2013, a school bond was passed in Westport, WA to fund reconstruction of Ocosta Elementary School. $2 million of the bond was allocated to construct the first vertical evacuation structure in the state, which is now complete. In October 2015, FEMA and the Earthquake Engineering Research Institute provided funding to support seismic evaluations of three school districts in Thurston County, WA. The intent of the program is to provide a model that can be replicated statewide to achieve recommended school seismic safety. The RWS initiative met its goal to provide a model to local government when King County, WA kicked off the Resilient King County initiative in September 2013. In 2016, the RWS initiative has gained much more exposure from local news media because of heightened interest in seismic safety issues. To date, the most recent progress is the formation of the governor’s Resilient Washington subcabinet, which was tasked to identify recommendations to move forward on and determine associated implementation strategies. The subcabinent published a report in September 2017 to recategorized the RWS recommendations and ranked them based on cost to implement 18. No specific recommendation from the RWS initiative has been intentionally implemented to date.

Oregon Resilience Plan

The Oregon Resilience Plan received significant media attention during its development and after the publication of the plan. The initial media attention exceeded that given to the RWS initiative and arguably what was given to SPUR’s initiative within a similar timeframe. More political awareness has been achieved for the ORP initiative than for RWS. The ORP was formally given to the Oregon’s 77th Legislative Assembly in February 2013. Following, the ORP steering committee presented their results at multiple legislative hearings to brief officials on what is needed to improve Oregon’s resilience. For example, in May 2013 the ORP findings were presented to the Oregon House Committee on Transportation and Economic Development. Subsequently, Senate Bill 33 was passed in requiring the creation of the Oregon Resilience Task Force. This task force was mandated to develop a report summarizing how to implement the ORP recommendations. The report was submitted in September 2014 14. A $680 million capital bond was approved in Beaverton, OR to fund the construction of seven schools to performance standards that ensure full functionality immediately after a major earthquake. Beaverton school district representatives cited the ORP in the news media as the major reason for this standard. Most recently, a high priority ORP recommendation was fulfilled with the appointment of a permanent full-time resilience officer in the governor’s office. Among other responsibilities, the resilience officer will champion the implementation of the other ORP recommendations. Also in June 2017, the governor signed a bill that requires OSSPAC to prepare policy recommendations to increase earthquake insurance coverage, as well as for post-event mass care and displacement strategies.

NIST Community Resilience Planning Guide

More broadly, the SPUR, RWS, and ORP initiatives have helped transform thinking about how to approach increasing jurisdictional seismic resilience at the local and state levels. One clear outcome of this is the development of the National Institute of Standards and Technology’s community resilience planning guide 19, which is intended as a model for other jurisdictions to design and conduct similar seismic resilience planning initiatives. The NIST guide was inspired and patterned after SPUR Before the Disaster Task Force reports 5,6,7,8 and the ORP. (The guide cites RWS once to note that the ORP built on the initiative, but otherwise RWS is not referenced.) The chair of the SPUR Before the Disaster Task Force and one of the ORP steering committee members were selected as NIST fellows to assist in developing NIST’s community resilience planning guide. The San Francisco, Washington, and Oregon initiatives have grown a professional network for discussing earthquake community resilience planning. The initiatives and the eventual use of the NIST guide by other jurisdictions have and will create collegial inter-jurisdictional competition that should lead to further innovation. To hasten the positive influence of the seismic resilience planning initiatives and the related NIST Community Resilience Planning Guide, it is important in conclusion to take a critical look at aspects of the efforts.


The performance measurement frameworks (recovery timetables) developed as part of the SPUR, RWS, and ORP initiatives are one of the most notable and innovative features the work and so is the primary focus of conclusion. This innovation is sure to live on and evolve because it is the core feature of the NIST guide. SPUR’s performance measurement framework timetable undoubtedly the primary persuasive force behind initiation of the RWS and ORP efforts. SPUR’s framework was easy to communicate and compelling to WASSC and OSSPAC. Interestingly, the SPUR performance measurement framework played a prominent role in inspiring the other initiatives and the NIST guide, but the framework was not referenced in seven of the nine Resilient City reports—only two of the reports from the Before the Disaster Task Force who developed the framework. The SPUR framework was designed with structural mitigation and engineering performance but the framework’s unique recovery-lens can also inform recommendations for preparedness, response, and rebuilding as evidenced by the ORP. The performance measurement framework was the central organizing concept of the RWS and ORP initiatives to the exclusion of other aspects of the Resilient City initiative.

Given the role of the performance measurement frameworks, it is surprising that the timetables and recovery estimates (current or target times) are almost never referenced in the respective SPUR, RWS, and ORP reports after first being presented within each document. For example, in the collective recommendations there does not seem to be references to identified recovery gaps as justification or prioritization for a recommendation. This is most surprising for the RWS and ORP initiatives, which were organized entirely around development of the timetables. Future seismic resilience planning initiatives should more explicitly reference their performance measurement frameworks in their recommendations and other outcomes. This will hopefully be facilitated by the NIST guide, which provides significant more detail than the original reports on how to develop and use a resilience planning performance measurement framework.

The two SPUR reports that include their performance measurement framework do not describe the methodology used to develop it, such as how to estimate the recovery times presented in their recovery timetables 7,8. The number of participants and the jurisdiction itself are smaller than the case for RWS and ORP; the smaller scope perhaps made it less critical to predesign and describe an explicit collaborative process. The RWS subcommittee had the benefit of seeing SPUR’s innovation and in turn saw the value of formalizing the development process to facilitate the much larger number of participants necessary to involve at a state-level. The ORP report does not provide much detail on the process for developing their framework, but OSSPAC had the RWS process to build upon and chose to have the eight ORP task groups develop their own approaches. The development process outlined in the NIST Community Resilience Planning Guide appears much more effortful and resource intensive than the SPUR, RWS, and ORP approaches. Again, NIST had the benefit of detailed knowledge of the three initiatives and logically sought to expand and improve upon their approaches. The most significant expansion is inclusion of steps for identifying social dimension and explicitly linking them to individual sectors of the built environment. The inclusion of social dimensions goes well beyond the original scope of SPUR’s Before the Disaster Task Force of existing building, new building, and lifeline performance, but follows the broadening trend set by RWS and ORP. Consideration of social dimensions certainly complicates the development of a resilience planning performance measurement framework and it is yet to be seen how well the NIST guide lays out the process to do it.

A justification for putting increasing effort into developing performance measurement framework timetables is better understanding of a jurisdiction’s recovery gaps and improved ability to identify recommendations that would not be obvious if the recovery timetables were not developed. As described above, the estimated current and target recovery times vary considerably between the initiatives. This exploratory study does not provide evidence that the ORP estimates, which were produced with the most time and effort, are more realistic or meaningful than the others. It is difficult to conclude whether development of the timetables resulted in the three initiatives making more numerous, unexpected, or creative recommendations than they would have otherwise. A more formal study is needed to better understand this. The ORP initiative spent the most effort to develop the most recovery timetables and it produced about 103 recommendations, which covered the broadest scope of topics. The RWS initiative allocated fewer workgroups, participants, and timetables in development of their performance measurement framework and came up with at least 153 recommendations, which covered a smaller scope of topics than the ORP. The SPUR Before the Disaster Task Force utilized the least amount participants and time to develop the fewest timetables, but produced approximately 79 moderately-scoped recommendations of the 272 broadly-scoped recommendations from the full initiative. It’s currently difficult to argue that one initiative’s recommendations are more unusual or creative than another’s, while all the initiatives have recommendations that refer work that was already in progress. It is not clear then that more time and resources spent on creating increasingly detailed timetables translates to more informed or systematic metrics and recommendations. This is an important consideration for jurisdictions that use the NIST guide to design their own resilience planning initiative. What is clear for future jurisdictions is that a broadly scoped initiative—promoted in the NIST guide—results in more comprehensive recommendations, even if only a moderate effort is spent on developing their performance measurement framework.

The innovation of how to develop the performance measurement recovery timetables can largely be attributed to the RWS and ORP initiatives based of the relatively limited original explanation given by SPUR. Development of the recovery timetables is now a central part of NIST’s Community Resilience Planning Guide. The NIST guide does not compare, critique, or synthesize the respective approaches taken by the RWS and ORP initiatives to justify or inform the new approach presented in the guide. The approaches to recovery timetable development for each of the initiatives, as well as presented in the NIST guide, are widely different. No discussion is given in the SPUR, RWS, ORP, and NIST documents on how to qualitatively or quantitatively measure progress of whether implemented recommendations are closing gaps between estimated current and target recovery times. In other words, no metrics are given for how to measure performance relative to the respective performance measurement frameworks. Thus, systematic research is needed to understand how the outcomes would differ based on the approach taken and, in turn, how best to go about developing such a performance measurement framework (e.g., more or less time; fewer or more participants; qualitative or quantitative).

Corresponding Author

Scott B. Miles, milessb@uw.edu, http://resilscience.com/

Competing Interests

The author is a member of Washington State’s Seismic Safety Committee and in that role helped design and facilitate the Resilient Washington State initiative.

Data Availability

The tables and figures were generated from text analysis and visualization of text extracted from all of the reports published by the three earthquake resilience planning initiatives. The citations of each report are included in the references section and should be cited if the extracted text is used for future research. No primary data was generated for this study. The extracted text data, as well as the Python code for analyzing and visualizing it, can be downloaded at the following permanent URL: https://osf.io/2vjrb/

The data and code can be cited separately with the following citation:

Miles, S. B. (2018). Data and code for “Comparison of Jurisdictional Seismic Resilience Planning Initiatives.” http://doi.org/10.17605/OSF.IO/2VJRB

http://currents.plos.org/disasters/article/comparison-of-jurisdictional-seismic-resilience-planning-initiatives/feed/ 0
Community’s Emergency Preparedness for Flood Hazards in Dire-dawa Town, Ethiopia: A Qualitative Study http://currents.plos.org/disasters/article/communitys-emergency-preparedness-for-flood-hazards-in-dire-dawa-town-ethiopia-a-qualitative-study/ http://currents.plos.org/disasters/article/communitys-emergency-preparedness-for-flood-hazards-in-dire-dawa-town-ethiopia-a-qualitative-study/#respond Wed, 21 Feb 2018 16:00:04 +0000 http://currents.plos.org/disasters/?post_type=article&p=37095 Background: Emergency preparedness at all levels (individuals and communities) is the corner stone of effective response to the increasing trends of global disasters due to man-made and natural hazards. It is determined by different factors, including (among others) past direct and indirect exposures to hazards. This study was carried out in Dire Dawa town, Ethiopia, which in the past experienced frequent flooding events, yet dearth of information exists about preparedness in the area.  The aim of the study was to assess the levels of emergency preparedness for flood hazards at households and communities levels.

Methods: The study was conducted in a qualitative approach and was conducted in Dire Dawa town, which has been divided into nine administrative-units called Kebeles. Two focus group discussions were held in two of these units (Kebele-05 and 06), each focus group comprising twelve people (all above 18 years of age), and in total 24 people (13 females and 11 males) took part in the study. Open ended questions were used that could guide the discussions, and the discussions were audio-taped and transcribed. The results were translated from local language to English and qualitatively presented. 

Results: The findings of focus group discussions showed that the local government in collaboration with the federal government built the flood protection dams in areas where flood hazards have been thought to be repeatedly wreaking havoc, specifically after the flood disaster of the year 2006. In addition, in Kebele-05, where one Non-Governmental Organization (NGO) was operating on flood hazards prevention and mitigation program, some non-structural emergency preparedness measures were undertaken by the communities. These non-structural measures (the major ones) entailed: establishment of committees recruited from residents and training them to raise awareness among communities on emergency preparedness; some residents made changes to their own houses (retrofitted) and put sandbags around their houses to temporarily protect the flooding; establishment of communication channels between communities to alarm each other in the event of flood disaster; and reforestation of the already deforested mountainous areas surrounding the town. However, concerns were raised by study participants about strengths of the constructed flood protection dams. Furthermore, the non-structural emergency preparedness measures identified by this study were not comprehensive; for example, residents were not trained in first aid, first aid kits were not provided, there was no linkage being established between communities and health facilities so as to provide emergency medical care to victims in the event of flood disaster. 

Discussion: The findings of this study concur with some of the previous quantitative studies’ results in that the past direct and indirect disaster experiences invoke preparedness intention and actual preparedness for flood hazards at individuals, communities and organizations levels. The only one quantitative and behavioral based study conducted thus far in Dire Dawa town reported the strong association of past flood disaster experience with household emergency preparedness. Among the residents there was a tendency to rely on the dams to be constructed with “good quality” and “higher strength” than making preparedness efforts on their own at their households. Structural measures such as building of dams, dikes, levees, and channel improvements could be means of mitigation measures; however, solely relying on these measures could have far reaching consequences.

Conclusions: To mitigate flood hazards, dams were built and in addition, in Kebele-05 where an NGO was operating, some non-structural emergency preparedness measures were undertaken. In the course of construction of flood protection dams, ensuring communities involvement is needed; and at the same time undertaking comprehensive non-structural emergency preparedness measures in all Kebeles is highly recommended.

Key words: Emergency, Preparedness, Flood, Dire Dawa, Ethiopia.



The global increasing trends of natural disasters makes preparedness for natural hazards a corner stone of effective response to disaster events (should they happen), and preparedness saves the loss of human lives and properties 1,2. Depending on the hazard type, preparedness (informed by scientific evidence) at community, individual and organization levels is recommended 3. Therefore, to help individuals and communities get prepared for hazards, provision of information or education is needed along with the necessary skills and resources. However, giving information to communities doesn’t necessarily lead to disaster preparedness 4 and studies have found that levels of preparedness at households and communities are low despite provision of information to communities and individuals and acknowledgement of risks by the vulnerable residents 5,6,7,8.

Hence, comprehending as to what motivates individuals to get prepared for natural hazards and what determines the preparedness behaviors (particularly those at risk of natural hazards) is quite a complex issue. Since preparedness has to do with human behaviors, a number of studies have been conducted applying behavioral theories to disaster and emergency health preparedness so as to better understand factors associated with preparedness behaviors 9. These theories mainly include the Health Belief Model (HBM), the Extended Parallel Process Model (EPPM), the Theory of Planned Behavior (TPB) and the Social Cognitive Theories 9. The constructs of these and other theories have been mainly made up of cognitive and affect (‘positive or negative evaluations of object, behavior or idea’) factors at individuals level, and social factors at communities level, which all in turn are responsible for risk perception primarily emanating from past hazard exposure experiences 5,10,11,12. Despite the complexities, other studies too reaffirmed the importance of risk perception, prior experience, and better information about the protective mechanisms on preparedness behaviors 13,14. Therefore, prior direct or indirect experiences about hazards are very important for risk perception, which ultimately determines the disaster preparedness behavior status. Studies from developing countries, where shortage of information exists, on how the prior experience may influence the preparedness behaviors of individuals and communities could help to enrich the existing literature.

Ethiopia has been experiencing a number of natural disasters and associated consequences, chiefly due to drought and flood hazards. Flood hazards rank second (next to drought) in Ethiopia and they claimed a significant number of human lives and properties in the past, for which preparedness is highly needed. The frequency of flood disasters and their impacts are increasing over time, the most catastrophic flood disasters across the country being documented in the year 2006-killing an estimated of over 600 people and affecting more than 500,000 people 15. Injuries, disruption of pure water supply and sewage disposal systems (that potentially predispose the population to diarrheal diseases) and malaria epidemic outbreaks are the major consequences of flood disasters reported in Ethiopia 16.

Dire Dawa, a town with a current population of 293,000 (projected/estimated by Ethiopian Central Statistics Agency for the year 2017) 17 and located at 515 KM from the capital city to the east, is one of the areas prone to flood disasters in Ethiopia. The town has experienced flood disasters repeatedly, and the flood disaster of 2006 was the most devastating one in terms of claiming human lives and economic losses. It caused 256 deaths, resulted in 244 people missing, and displaced 10,000 people 18. Due to the fact that paucity of information exists in the country on preparedness for natural hazards in general and on flood hazards in particular, this study was conducted in the Dire Dawa town. Thus, the main aims of this study were to assess the levels emergency preparedness for flood hazards at the households/individuals and communities using exploratory method, and to assess how emergency preparedness for flood hazards has been influenced by the past disaster experiences and provision of information/education (if any) by external organizations (beyond the domain of local communities).


Study setting and data collection

The data for this study was collected on June 29 and 30, 2015, from Dire Dawa town (Ethiopia), which is located in eastern part of the country. The town has nine administrative units called Kebeles and each Kebele has its own distinct population and administrative office. Two focus group discussions were held in two of the Kebeles (Kebele 05 and 06), whereby in the past they frequently experienced flood disasters and have been more prone to flood hazards. Each focus group discussion comprised of twelve people (all above 18 years of age). From Kebele-05, six males and six females participated, while from Kebele-06, five males and seven females took part in the study. Participants were recruited and included into this qualitative study mainly based on the knowledge of socio-demographic, familiarity with the study area, knowledge of tradition and culture of the town, and regularity of participating in community based activities including disaster risk reduction initiatives. Study participants were not categorized according to their ethnic groups and religious denominations as selection criteria, however, the participants were from the dominant ethnic groups and religious denominations of both Kebeles.

Open ended questions were prepared that could guide the two focus group discussions. These questions were about whether the town was vulnerable to flood hazards, experiences of flood related losses in the past, whether the risk of flood hazards could be averted through preparedness actions at households and communities levels, if there were any prevention and mitigation measures undertaken by government or non-governmental organizations, and if the residents were feeling safe as a result of emergency preparedness (prevention and mitigation) actions undertaken at households and communities levels (please refer to annex for the full open ended questions). The focus group discussions were audio-taped and transcribed; the results were translated from local language to English and qualitatively presented.

Ethical consideration

Ethical approval was obtained from the Ethiopian Ministry of Science and Technology, which is the responsible body in the country for ensuring the ethical standards of studies, and permission to conduct the study was obtained from Dire Dawa Administrative Council Health Bureau and from each Kebele Administrative Office. Information was provided to participants about the purpose of the study and participants were informed that taking part in the study was solely based on their willingness. The consent sought was verbal, because some of the study participants couldn’t read and write. The ethics committee from Ethiopian Ministry of Science and Technology approved this verbal consent procedure. The study participants verbally consented to their opinions including direct quotations published without revealing their personal identities. After all study participants agreed to take part in the study and verbally consented for the discussions to be audio-taped, the focus group discussions were held and audio-taped. The verbal consent procedure was documented by the author.


Given the repeated flood disasters that happened in Dire Dawa town in the past, all the participants of the focus group discussions agreed that Dire Dawa town was vulnerable to flood hazards, though some of them argued that its exposure level to flood hazards lessened in the recent times owing to the construction of flood protection dams. Furthermore, the participants stated that major flood disasters events were witnessed in the year 1981, 1997, and 2006. From these events, the year of 2006 flood disaster was reported to be the most devastating one in the history of Dire Dawa town-in terms of claiming human lives and loss of properties. A number of reasons were cited by participants that could increase the vulnerability of the town to flood hazards, and the main ones were: topography of the town (relative to its surrounding districts, Dire Dawa town is situated in lower altitude and surrounded by mountains), deforestation in mountainous areas surrounding the town, and construction of houses in the vicinity of ditches and rivers (that cross the city during rainy season), contributing to the narrowing of the flood-ditches and rivers’ ways, often causing flooding due to the overflow of water into the nearby households. One of the participants from Kebele-05 put the vulnerability of Dire Dawa town to flooding in this way: “it is possible to categorize into three groups about the vulnerability of Dire Dawa town: firstly, because of deforestation, there is nothing that guards against the rivers that flow from the upper stream to the town and they [rivers] simply overflow into the town; secondly, previously the mountains were filled with forests, now all those forests have been cleared and when it rains, it directly comes down to the town; and third, there are houses built on the ways of floods (ditches) that caused the narrowing of the ditches, and floods that supposed to go through those ditches overflow into the nearby houses”.

According to the views of the participants, after the major flood disaster event of 2006, some actions had been taken to reduce the vulnerability of the town to flood hazards. These actions were: construction of flood protection dams (in both Kebeles), organizing communities for planting various trees in the mountainous areas surrounding the town and in previously deforested places (Kebele-05), increasing the awareness level of communities on flood protection measures and on communications between communities of different villages (Kebele-05). In addition, by virtue of experiencing past flood disasters, the residents of the town had made some changes to their own respective houses and the surroundings (Kebele-05). These changes entailed mainly putting sandbags around their houses to protect the flooding, and construction around their homes to elevate the floors higher above the normal surface in which the floods run, which were ultimately meant to make the houses a little bit “flood-resistant”. However, until the question was posed to them, whether they had done something to their homes and the surroundings on their own initiative or not, they were repeatedly raising about construction of dams to protect and mitigate the flood disasters, which shows the residents were heavily putting their hopes on constructed or yet to be constructed flood protection dams. This pattern was similar in both group discussions (Kebele-05 and 06).

The flood hazards prevention and mitigation actions were carried out by the initiatives of the government, a non-governmental organization operating in the town and by residents of the town. The government built dams after the major flood disaster of 2006, and the participants said that construction of dams had somewhat reduced risks associated with flood disasters. For example a participant of focus group discussion from Kebele-06 being heard of saying, “since dams are being constructed in challenging and difficult places where floods are thought be causing problems, I don’t think the flooding in the town would create the likes of problems it caused in the past”. And the other participant from the same Kebele added, “it is clear to everyone that the past disasters left behind terrible memories, after those periods, both government and the community have been making efforts [to prevent and mitigate flood hazards]; however, it is impossible to say ‘reliable’.’’ Therefore, participants of focus group discussions from both Kebeles repeatedly raised their concerns surrounding the strengths of these dams as some of these dams had already been demolished by relatively what participants called, “smaller floods”.

According to the focus group discussion held in Kebele-05, there was an indigenous non-governmental organization called “Jerusalem Children and Community Development Organization (JeCCDO)”, at the time of the data collection this organization was registered as “an Ethiopian Residents’ Charity Organization”, rendering services to communities on flood disaster prevention and mitigation activities. On accounts of participants, the services that JeCCDO was offering include: organizing the communities and establishing committees recruited from four villages of Kebele 02 and 05, building the capacity of these committees through training that could enable them to carry out awareness raising activities on flood disaster prevention and mitigation. As a result, trees were planted in areas previously deforested, communication channels were created between communities of the upper and lower streams so that residents of the upper stream could inform the residents in the lower stream by mobile phones when flooding was about to come. However, concerns were raised again on the means of communication, for instance, one woman from Kebele-05 said, “this means of communication is not that much satisfactory and sustainable, because during flooding, during rainy times, there are times when the telephone/the mobile itself not functioning”. In kebele-05’s group discussion, study participants also raised the presence of ‘bell’ in the Kebele-05 Administrative Office to ring it in the event of flooding as a warning for evacuation, and the presence of gun firing into the air by police, this too, aimed to alarm the residents. On the other hand, in Kebele-06, participants said that most of the residents in that area were too poor to make changes to their homes on their own and community based initiatives of flood hazards prevention and mitigation efforts were not mentioned. In addition, participants said that apart from the government there was no other non-governmental organization operating in Kebele-06 on flood hazards prevention and mitigation program.

Participants were asked to what extent they and their families were relying on the preventive and mitigation measures being undertaken at households and communities levels thus far (at the time of data collection). And their responses were that they couldn’t rely on those measures-“we cannot say that we are threat free from the flood disaster, because every time situations are changing”, said one man from Kebele-05; the other person from the same Kebele added, “dams have been constructed, but it is impossible to rely on these dams, because just recently last time it was only the ‘smaller floods’ that managed to demolish and take away those dams”; a women from the same Kebele further added, “the dams were constructed long before the establishment of committees (the committees were established by JeCCDO in order to work on prevention and mitigation activities in Kebele 02 and 05) and they (dams) do not have quality, and it would have been good if committees were consulted and involved in the construction process”; and another man from Kebele-05 voiced his worries in saying,” we have repeatedly expressed our concerns to the Kebele administration and Dire Dawa town Administrative Council about the threat of flood disasters”. In general, participants from both focus groups stressed that the constructed flood protection dams had no quality and strengths, and these dams were not something to be reckoned with. And they recommended that the residents of the Dire Dawa town should be involved and allowed to be part of the construction process of the dams so as to ensure the strengths of these dams and to avoid the catastrophic situation that may come in the near future due to flood disasters. Moreover, like any other cities (towns) in Ethiopia, there are people residing on streets, and in the past, streets dwellers were also the victims of flood disasters in Dire Dawa town, and participants (particularly from Kebele-06) recommended that shelters should be provided to those people. In addition, the group discussion’s participants of Kebele-06 acknowledged that there were no communications between the communities and they pointed out the necessity for the establishment of committees in their Kebele that would convene meetings on a regular basis to discuss and share information among the residents on how to prevent and mitigate the food hazards.


This study used a qualitative approach (focus group discussions) to investigate the levels of preparedness at individuals and communities in the town at risk of flood hazards. Furthermore, it was aimed at assessing if the previous flood disasters might have influenced the residents, the local and federal governments, and other organizations in taking some proactive emergency preparedness actions in preventing and mitigating future flood disasters in the area. The findings of focus group discussions show that the local government (Dire Dawa Administrative Council) in collaboration with the federal government built the flood protection dams in areas where flood hazards have been thought to be repeatedly wreaking havoc, specifically after the flood disaster of the year 2006. Even though the findings were controversial among studies, the findings of this study concur with some of the previous quantitative studies’ results in that the past direct and indirect disaster experiences invoke preparedness intention and actual preparedness for flood hazards at individuals, communities and organizations levels 5,19,20,21. The only one quantitative and behavioral based study conducted thus far in Dire Dawa town reported the strong association of past flood disaster experience with household emergency preparedness 21.

However, according to the views and observations of study participants, the strengths of the built flood-protection dams have been under question whether they could really protect the residents from heavy flood hazards or not as some of the dams had already collapsed due to relatively “smaller floods”. Hence, the results of the focus group discussions from Kebele-05 and 06 demonstrate that the residents of the Dire Dawa town did not have confidence in the constructed dams and the participants from both groups raised their concerns about the strengths of already built dams, and calling and recommending for the dams to be constructed with “high quality and strength”. However, the participants didn’t raise any other alternative means of saving their own lives and properties until the question was posed to them whether they had done anything at individuals’ level on their own or by the initiative of other external organizations. This means the residents were putting much of their hopes primarily on the dams to be constructed with “good quality” and “higher strength” than making preparedness efforts on their own at their households. Structural measures such as building of dams, dikes, levees, channel improvements could be means of mitigation measures; however, solely relying on these measures could have far reaching consequences if alternative preparedness of non-structural emergency measures are not put in place as these structural mitigation measures may fail when the hydrological load exceeds their capacities 22,23,24,25. And hence, from the view point of cost implication to put in place the structural measures to mitigate flood hazards might not be affordable always and due to the fact that there might be unforeseen consequences of relying on these measures (even if afforded), emergency preparedness involving non-structural measures will pay off in the long run.

Nevertheless in kebele-05, where there was a community based non-governmental organization working on flood hazards prevention and mitigation program at the time of data collection, there were communications between the communities through the established committees about prevention and mitigation, preparedness at the households and communities levels existed, and the communities were better informed about flood hazards prevention and warning systems. As this study employed the qualitative approach, study participants were not requested to rate their responses of preparedness measures they undertook as often being done by relatively quantitative studies. Even though the study participants of kebele-05 didn’t characterize the level of non-structural preparedness measures in terms of scale or number, a local NGO operating in the Kebele-05 has stated the extent of some of these measures in its annual report of the year 2015. According to this report, for example, “37 community leaders of disaster prone areas networked to run Community Based Early Warning system”, 2822 people (1,095 females) received “awareness raising information” on disaster risk reduction through different forms of activities, “2 community owned and managed nursery sites [were] established and strengthened to support afforestation programs”, and “20 hectares of land rehabilitated through soil and water conservation practices” 26. As mentioned in the result part, this local NGO’s name was “Jerusalem Children and Community Development Organization (JeCCDO)”; at the time of the data collection this organization was registered as “an Ethiopian Residents’ Charity Organization” providing services to communities on disaster risk reduction and others 26. On the contrary, in Kebele-06, such community based preparedness initiatives were not mentioned by participants and it looks they are lacking. This shows even though the communities experienced the flood disasters in past repeatedly, provision of information is needed on better emergency preparedness, particularly not only about the risk of flood hazards but also about the ramifications of flood disasters and the cost effective methods of mitigation measures at households levels 25.

In general, this exploratory study found out that apart from the preparedness efforts put in place by the government, non-governmental organization, and the communities that were identified by these focus group discussions, there were no other essential emergency preparedness activities. For example, there were no linkages established between the health facilities and the communities to provide emergency medical care in the event of flood disaster, training were not provided to communities’ members on first aid and first aid kits were not given to them, there were no routes of evacuation plans and temporary shelters. Hence, more work remains to be done to engage communities in structural and non-structural flood hazards mitigation measures (early warning, emergency and precautionary measures among others).

Limitations of the approach

The most commonly recommended number of participants for a single focus group discussion is from 8 to 12 people, and in this study 12 individuals took part in each discussion, and even some qualitative research text books put the maximum figure as 8 individuals 27. Hence, the sample size of this study per focus group discussion was adequate. From nine existing kebeles in the town, two kebeles were selected purposely based on their exposure status to flood hazards, which could be one of the limitations of this study. Another limitation of this study could be only focus group discussions were employed instead of mixing up with other qualitative study types such as “an individual based in-depth interview”, which could have enriched the information on disaster preparedness status. Future qualitative studies to be conducted in the town could benefit if they address these limitations.


In the history of Dire Dawa town, the flood disaster of 2006 was the turning point in prompting some emergency preparedness actions. The local government in collaboration with the federal government initiated the construction of flood protection dams in some flood prone areas of the town. However, according to this study’s findings, some of the already constructed dams had already collapsed and the residents had no confidence in those dams. In Kebele-05 where an NGO existed to operate on flood hazards prevention and mitigation, some emergency preparedness actions were undertaken by the communities. These major non-structural measures undertaken include: putting sandbags around households for temporary mitigation, making changes around homes, planting of trees in the mountainous areas surrounding the town, establishment of communication channels between different communities to notify each other during the event of flood disaster, and the existence of flood disaster “means of warning” at the Kebele level. Involvement of communities in the course of construction of flood protection dams to ensure dams’ strengths, and provision of comprehensive information on non-structural measures of emergency preparedness (including linking the emergency preparedness with the health facilities for emergency medical care during disaster event and having routes of evacuation plans) are recommended, particularly in Kebeles other than Kebele-05, where non-governmental organizations are inactive on flood prevention and mitigation program.


The original audio-taped interviews were transcribed and translated from the local language to English, it can be accessed at: 10.6084/m9.figshare.5853582, https://figshare.com/articles/Transcriptions_pdf/5853582.


The author has declared that no conflicts of interest exist.


Luche Tadesse Ejeta, BS, MPH, PhD, e-mail: tadesse.ejeta@gmail.com


Appendix I

Questions for Focus Group Discussions

No Question
1 In your opinion, do you think that Dire Dawa town is prone to flood disaster? If yes, why?
2 When did the major flood disaster happen in Dire Dawa town? Why you think that was the major flood disaster in the history of Dire Dawa town?
3 Do you think that the residents of Dire Dawa town are at risk of flood disaster that might happen during rainy season in the near future? If you think that the residents are at risk, why?
4 Do you think the flood disaster could be prevented or mitigated through some actions by human being in your area?
5 To avert the flood disaster risk, are there any specific things you have done around the home to minimize the effects of flooding? (Why did you/why didn’t you?)Where did you get this information from?Why did you think this information was useful?
6 Have you discussed flood issues with others? (Why did you/why didn’t you?)What/who influenced you to discuss these issues?Did this help in clarifying issues?Would it influence you to become more involved in the wider community?
7 Have you had any contact with the Dire Dawa council or the local office (kebele) and any health facilities (hospitals or health centers) regarding floods?Have they contacted you?How satisfactory was this contact?
8 Is there any action taken by the government (could be local or federal) to mitigate the flood disaster in your area? If yes, what are those actions?
9 Is there any action taken by the non-governmental organizations (could be civil society organizations, local-NGOs, and International NGOs) to mitigate the flood disaster in your area? If yes, what are those actions?
10 How much do you rely on the preventive and mitigation measures being undertaken by you personally at household level, government, and non-governmental organizations? Do you feel that you and your families and the community at large are safe as result of these actions (individually, government and non-governmental organizations)?

http://currents.plos.org/disasters/article/communitys-emergency-preparedness-for-flood-hazards-in-dire-dawa-town-ethiopia-a-qualitative-study/feed/ 0
Listening to the Voices of the People: Community’s Assessment of Disaster Responder Agency Performance During Disaster Situations in Rural Northern Ghana http://currents.plos.org/disasters/article/listening-to-the-voices-of-the-people-communitys-assessment-of-disaster-responder-agency-performance-during-disaster-situations-in-rural-northern-ghana/ http://currents.plos.org/disasters/article/listening-to-the-voices-of-the-people-communitys-assessment-of-disaster-responder-agency-performance-during-disaster-situations-in-rural-northern-ghana/#respond Wed, 20 Sep 2017 15:30:51 +0000 http://currents.plos.org/disasters/?post_type=article&p=31803 Introduction: In Northern Ghana, a combination of torrential rains coupled with the spilling of the Bagre dam in neighboring Burkina Faso in the past few years has resulted in perennial flooding of communities. This has often led to the National Disaster Management Organization (NADMAO) the main disaster responder agency in Ghana, being called upon to act. However affected communities have never had the opportunity to evaluate the activities of the agency. The aim of this study is therefore to assess the performance of the main responder agency by affected community members to improve on future disaster management.

Methods: A mixed qualitative design employing a modified form of the community score card methodology and focus group discussions was conducted in the 4 most affected communities during the last floods of 2012 in the Kasena-Nankana West district of the Upper East Region of Northern Ghana. Community members comprising of chiefs, elders, assembly members, women groups, physically challenged persons, farmers, traders and youth groups formed a group in each of the four communities. Generation and scoring of evaluative indicators was subsequently performed by each group through the facilitation of trained research assistants. Four Focus group discussions (FGDs) were also conducted with the group members in each community to get an in-depth understanding of how the responder agency performed in handling disasters.

Results: A total of four community score cards and four focus group discussions were conducted involving 48 community representatives. All four communities identified NADMO as the main responder agency during the last disaster. Indicators such as education/awareness, selection process of beneficiaries, networking/collaboration, timing, quantity of relief items, appropriateness, mode of distribution of relief items, investigation and overall performance of NADMO were generated and scored. The timing of response, quantity and appropriateness of relief items were evaluated as being poor whereas the overall performance of the responder agency was above average.

Conclusion: NADMO was identified as the main responder agency during the last disasters with community members identifying education/awareness, selection process of beneficiaries, networking/collaboration, timing of response, quantity of relief items, appropriateness of relief items, mode of distribution of relief items, investigation and overall performance as the main evaluative indicators. The overall performance of NADMO was rated to be satisfactory.

Key words: Kasena-Nankana West district, NADMO, community score card, Rural Northern Ghana



Over the past several decades, both naturally occurring and man-made disasters have increased in frequency and number worldwide with Sub-Saharan Africa considered to be the most vulnerable to climate variability in particular.1,2 The growing frequency and severity of extreme events such as droughts and floods along with shifting rainfall patterns, threaten to overwhelm the natural resilience of African communities risking livelihoods and food security. Widespread poverty, fragile ecosystems, weak institutions, and uncoordinated disaster response among responding agencies exacerbate Africa’s vulnerability to climate change.3,4

In Northern Ghana, a combination of torrential rains coupled with the spilling of the Bagre dam in neighboring Burkina Faso in 2007, 2010 and 2012 spelled doom for the area. According to the UN Office for the Coordination of Humanitarian Affairs, the 2007 floods alone killed 22 people, destroyed or completely damaged 11,239 homes, washed away 12,220 hectares of farmland and affected 275,000 people.5

As the reported number of people affected by disasters has risen over the past decades, so has the expectations placed on responding agencies by donors, the public and affected populations also increased.6 In contrast to financial accountability mechanisms which are now comparatively well developed within donor and implementing agencies, the mechanisms for ensuring accountability to the population being served are poorly developed. Apart from the fact that there is the need to improve accountability of relief interventions through the development of appropriate criteria and methods to measure their successes, there is perhaps more importantly, the need for the creation of a management culture which exacts the high standards of accountability from donor and implementing agencies to recipient communities.7

The Kasena Nankana west district was one of the hardest hit in all the previous disastrous events resulting in widespread humanitarian effort by government; non-governmental organizations (NGOs) such as the International Federation of the Red Cross and the Inter-NGO Consortium; and individuals. All the humanitarian aid was channeled through the National Disaster Management Organization (NADMO) the main disaster responder agency in Ghana established by act 517 of 1996 to manage disasters and similar emergencies in the country.8 However there have been a number of allegations such as over concentration of relief items to areas that were not hardest hit, unfair distribution of relief items and poor coordination of the disasters among others labeled against NADMO by affected community members.

There is currently scarcity of knowledge concerning the performance of the main disaster responder agency by community members in Northern Ghana especially after the disastrous floods of 2007, 2010 and 2012. We therefore conducted this study to assess the performance of the main responder agency at the grass root level in Northern Ghana and to set the platform for an interactive feedback between responder agency and community members for future disaster management.


Study area and Design

This study was conducted from May to June 2014 in the four most affected communities (Navio, Kajelo Nyangenia and Nakong) of the last floods of 2012 in the Kasena-Nankana West district of the Upper East Region of Northern Ghana. The Kassena-Nankana West District is one of the thirteen districts in the Upper East Region of Ghana. Its population according to the 2010 Population and Housing Census is 70,667 representing 6.8 percent of the region’s total population. Males constitute 50.8 percent and females represent 49.2 percent. Seventy nine percent of the population is rural. The population is engaged in various economic and social activities including; subsistence farming, business/trade, rearing domestic animals among others.9

The map of the Kassena-Nankana West District showing the study communities is shown in Figure 1.

Map of Kassena-Nankana West District showing study communities

Fig. 1: Map of Kassena-Nankana West District showing study communities

Due to the growing interest and importance in the use of qualitative, especially participatory techniques in evaluating relief programs7, a qualitative design employing a modified form of the community score card methodology and focus group discussions were used in this study.

The Community Score Card methodology

The community score card (CSC) process is a community-based monitoring tool that is a hybrid of the techniques of social audits and citizen report cards. The CSC is an instrument to exact social and public accountability and responsiveness from service providers. By linking service providers to the community, citizens are empowered to provide immediate feedback to service providers. The CSC process uses the “community” as its unit of analysis, and is focused on monitoring at the local/facility level. It can therefore facilitate the monitoring and performance evaluation of services, projects and even government administrative units by the community themselves. Since it is a grassroots process, it is also more likely to be of use in a rural setting. CSC process involves basically four components: input tracking scorecard, community generated performance scorecard, self-evaluation scorecard by service providers and an interface meeting between users and providers to provide respective feedback and generate a mutually agreed reform agenda.10,11,12

For the purpose of this study, we employed only the component of community generated performance scorecard as this involves the generation of evaluative indicators by community members themselves.

Selection of study participants

At the community level, members comprising of chiefs, elders, assembly members, women groups, physically challenged persons, farmers, traders and youth groups formed a group in each of the four communities. This target population was those actually chosen by community members to represent them in the communities’ dealings with the responder agency during the disasters. Each group was considered to be heterogeneous since the members were from different backgrounds, ages and genders. All participants were however aged 18 years and above.

Data collection process and analysis

Two experienced research assistants were trained on the community score card approach, the study protocol and procedures to assist the investigators in data collection.

In all, four community visits where made to each community by the research team. The first visit was to explain the purpose of the study to the chiefs, elders and opinion leaders of the communities. During the second community visit, a community meeting was held where the purpose of the study and scorecard methodology (with much emphasis on indicator generation) was explained to community members after which each community chose its own group members. On the third visit, the field research team facilitated an indicator-generation process where groups generated and defined a set of indicators. The definitions of indicators were given entirely by group members themselves after which group representatives from each group met to consolidate and harmonize these definitions under the guidance of the research team. In the final community visit, group members met to score each indicator where a circle was drawn on the ground and divided into different sections by one of the participants in each group. Items like sticks, sandals and stones were used to denote an indicator in each of the divided sections of the circle. Participants cast a small stone each in the divisions of the circle (figure 2) to represent their individual grading/score of an indicator as arrived at by consensus. The groups further settled on a collective scoring measure or rating of 1-100 to score some indicators. These scoring processes in the form of a scoring matrix were then summarized into a score card and presented in the form of tables.

Focus group discussions (FGDs) were also conducted with the group members in each community during the third community visit to get an in-depth understanding of how the responder agency performed in handling disasters. FGDs were recorded with an audio tape and later transcribed verbatim from the local language (kassem) into English in a text form for analysis. The investigators read through all the transcripts exhaustively to check for inconsistencies after which coding and categorization was done manually using the conventional analysis approach as outlined by Hsiu-Fang and Sarah.13 Categories were then grouped under thematic areas which corresponded with community generated indicators by the community score card approach for the write-up.

Ethical considerations

Ethical approval was given by the Navrongo Health Research Centre Institutional Review Board (ID: NHRCIRB 183) and Kasena-Nankana West district assembly and district NADMO office. Permission to carry out this survey at the community level was verbally sought from the chiefs and elders of the respective communities first, after which a verbal informed consent was also gotten from individual members of the participating groups. The consent procedure was reviewed by and approved by the ethics committee.


In total four community score cards and four focus group discussions were conducted involving 48 community representatives. Males accounted for 56% (27) of the population with the remaining 44% (21) being females. The average age of participants was 52 years. An average of 12 members formed a group in each of the communities.

Findings of both the community score card and focus group discussions are presented together.

Community Score Cards

All four communities identified NADMO as the main responder agency during the last and previous disasters. Community generated indicators included: Education/awareness, Selection process of beneficiaries, Networking/Collaboration, Timing of response, Quantity of relief items, Appropriateness of relief items, Mode of distribution of relief items, Investigation and Overall performance of NADMO.

The consolidated and harmonized definitions of the indicators is presented in table 1 whereas the summarized score card (individual and consolidated scoring) for each of the communities are presented in tables 2 to 5.

Education/awareness The extent to which people were educated/made aware on flooding, storms etc in the community using durbars, radio or other means; sensitizing communities and not waiting for disasters before you intervene
Selection process of beneficiaries Degree of satisfaction on who will benefit from relief items and the selection criteria of beneficiaries
Investigation Assessing degree of destruction and its accompanying effects on affected people for deciding on who gets what and by how much
Timing The time between the writing of names of affected persons and when the relief is brought to them
Quantity of relief items The amount of relief given in relation to what the victim actually needed/required
Appropriateness Whether relief given is what the victim needed/ lost due to disaster
Mode of distribution Mode of distribution of relief breeds conflict among those who get and those who do not
Networking Assessing who/which entities collaborated in assessing, deciding and distributed relief items to affected persons
Overall performance NADMO overall performance graded

Education/awareness Media/radio 11 NOT APPLICABLE
Community durbars 0
No education 1
Selection process of beneficiaries Good 0 20
Poor 0
Very poor 10
No investigation 2
Networking/Coordination Involves a committee 0 NOT APPLICABLE
External distributors 2
NADMO officials 8
Party leaders 2
Timing Timely 0 20
Late/average 2
Very late 10
Never came 0
Quantity of relief items Adequate 2 30
Inadequate 2
Woefully inadequate 4
Not at all/Absent 4
Appropriateness Appropriate 0 50
Average 2
Somehow 9
Not at all 1
Overall performance Good 7 60
Average 2
Poor 3

Education/awareness Media/radio 0 NOT APPLICABLE
Community durbars 0
No education 12
Selection process of beneficiaries Excellent 12 100
Good 0
Poor 0
Mode of distribution Satisfactory 0 10
Not satisfactory 12
Timing Timely 0 20
Late/average 0
Very late 7
Never came 5
Quantity of relief items Adequate 0 10
Inadequate 0
Woefully inadequate 12
Not at all/Absent 0
Appropriateness Appropriate 0 10
Average/inappropriate 0
Woefully inappropriate 12
Overall performance Satisfactory 6 50
Somehow satisfactory 6


Responses where based on earlier encounters with NADMO and not the recent 2012 floods

Education/awareness Media/radio 1 NOT APPLICABLE
Community durbars 0
No education 11
Timing Timely/appropriate 0 1
Inappropriate 0
Somehow 0
Not at all 12
Quantity of relief items Adequate 0 1
Inadequate 0
Woefully inadequate 0
Not at all/Absent 12
Overall performance Good 0 50
Average 6
Poor 6

Education/awareness Regular education 1 60
No education 3
Somehow 8
Selection process of beneficiaries Very good 0 40
Good 5
Poor/selective 7
Investigation Proper 0 30
Poor 6
Not at all 6
Timing Timely 0 20
Late/average 1
Very late 11
Quantity of relief items Adequate 0 30
Inadequate 12
Woefully inadequate 0
Appropriateness Appropriate 0 30
Inappropriate 12
Somehow 0
Overall performance Satisfactory 4 40
Unsatisfactory 8

Focus Group Discussions


Two of the communities (Nyangania and Nakong) said they did not receive enough education and sensitization on the flooding situation resulting in community members being hit hard by the floods. This they said also affected their resilience level adversely.

“The local FM station did a good job by constantly announcing and reminding us of when they were going to open the dam from neighboring Burkina Faso. This helped members of the community to prepare adequately and so we were able to adapt to the floods quite well with the help of NADMO” (Group member of Navio community-FGD)

“Our community did not receive information or education from any organization. So when the flood came we were not prepared at all and this led to loss of lives and properties of community members” (Group member of Nakong community-FGD)

Section process of beneficiaries

Members of three communities (Navio, Nakong and Kajelo) identified this as an indicator with only one of these communities (Nakong) rating this indicator high.

“In fact we were really very satisfied with how NADMO selected those who were to benefit from the relief items they brought to us” (Group member of Nakong community-FGD)

“The way NADMO selected those who were to benefit from the relief items was poor. For example some of the affected people who wrote done their names did not receive any item whiles some people who didn’t write their names at all were given relief items” (Group member of Kajelo community-FGD)


This indicator was identified by only one community with members of the group identifying the existence of networking and collaboration between NADMO officials and other stakeholders in the management of the last disasters.

“NADMO officials, leaders of some political parties and other organizations who donated relief items to us worked together in deciding those who got those items” (Group member of Navio community-FGD)

Timing of response

Whereas three of the communities (Navio, Nakong and Kajelo) got some response from the responder agency, the response rather arrived very late with the fourth community (Nyangania) not receiving any response at all during the last floods in 2012.

“Although we eventually got some items such as cups, mattresses, blankets and buckets from NADMO, these items arrived very late. We hope next time there is a disaster such as this, NADMO and any organization which has items to donate, will do that as quick as possible to help the community recover more quickly” (Group member of Kajelo community-FGD)

“For the last floods that we experienced here, NADMO visited us to write the names of those who were affected through the community elders. However they never came back again and so no single person received any relief item in this community” (Group member of Nyangania community-FGD)

Quantity of relief items

Majority of respondents from communities were the responder agency responded opined that the quantity of relief items were generally inadequate.

“The quantity of relief items we received was not enough at all. Imagine after writing down the names of so many affected people only 8 mattresses were distributed to the worst affected areas of Kajelo through the elders of the community” (Group member of Kajelo community-FGD)

Appropriateness of relief items

Those communities that got a response from NADMO rated the appropriateness of this response to be inadequate. All the 3 communities that got some help from NADMO said the relief items they received were exactly what they needed to cope and recover from the floods.

“Though officials of NADMO came to write our names and the items we actually needed, what they eventually brought to us was different and this did not help us that much especially on how we coped and recovered from the floods” (Group member of Navio community-FGD)

Mode of distribution of relief items

Only one (Nakong) community evaluated the responder agency using this as an indicator with group members stating that the mode of distributing relief items was unsatisfactory.

“The unfair manner with which the relief items were distributed to affected members of the community rather brought about some disagreements and conflicts amongst the beneficiary since some people thought they were most affected and therefore deserved more items. Some members of this community have also come to complain to me that the relief items were distributed along partisan lines and not based on merit” (Group member of Nakong community-FGD)


Group members in only one (Kajelo) community identified community needs assessment by the responder agency as an evaluative indicator with 50% of the members saying that this was poorly done and the other 50% saying this was not carried out at all.

“Although NADMO came around to assess the extent of damage by the floods before writing the names of those to benefit from any relief item in my area, this was poorly done as some of the people who were seriously affected didn’t get the right quantity of items to help them deal with the situation” (Group member of Kajelo community-FGD)

“NADMO never did any investigation or assessment of the level of damage by the floods in my locality before writing down the names of those affected. And so some of those who were not seriously affected got the same items as those who were seriously hit by the floods” (Group member of Kajelo community-FGD)

Overall performance

All four communities evaluated NADMO using this as an indicator. Apart from one (Kajelo) community that rated the overall performance of NADMO to be unsatisfactory, the rest of the communities thought that NADMO performed creditably well in its handling of the recent disasters.

“I will say that overall, NADMO performed quite well when you look at how they handled the entire disaster situation” (Group member of Navio community-FGD)

“Even though NADMO did not respond to our needs this time round, judging by what they did during the previous floods I will say that their performance was satisfactory” (Group member of Nyangania community-FGD)

“I wasn’t happy with the performance of NADMO at all. They were not fair with the way they responded to this particular disaster. Those who really needed the items never got them and for those who eventually received their items got them very late. So how can you say that NADMO did well?” (Group member of Navio community-FGD)


The methodology employed in this study can be said to be one of the new emerging participatory techniques used in evaluating relief programs and agencies especially against the background that the CSC is being used for the first time for this purpose.

The Hyogo World Conference on Disaster Reduction in 2005 stressed on the importance of information management and exchange which is meant to provide easy understanding of information on disaster risks and protection options, especially to citizens in high-risk areas, so as to encourage and enable people to take action to reduce risks and build resilience.14 However in this setting, half of the affected communities did not receive sufficient information and education on how to reduce their risk and were hit severely by floods which affected their resilience adversely. Disaster management at the community level involves the need for information flow in a coordinated fashion through both a multi-organizational and multi-level structure, requiring that responder agencies or organizations not only depend on their internal interactions, but on the interactions with other agencies as well.15 NADMO over the years has worked hand in hand with other agencies in achieving one of its objectives which is to set up monitoring and early warning systems to aid the identification of disasters in their formative stages, to disseminate timely information and warning, and hazard/disaster awareness creation.8 However during this disaster, this was not the case as information dissemination and awareness creation was generally found to be unsatisfactory by community members. It is important that the responder agency takes advantage of the privileges at its disposal especially during disaster situations to embark on massive community awareness creation and information dissemination on how communities can mitigate, adapt and cope with the shocks and stressors of disasters.

Selection of beneficiaries during and after disasters can be a very daunting task. In this particular disaster, majority of the community members labeled a number of allegations against the responder agency. Similarly in a qualitative study to evaluate post-flood disaster response strategies in southern Ghana, some male participants in a focus group discussion also accused the same disaster responder agency as being discriminatory and actually decided not to violently contest the seeming discrimination they felt so as to sustain the unity in their community.16

Post-disaster governmental manipulations of response processes has been identified to exacerbate existing inequalities within affected communities which has often tend to provoke angry protest and demonstration against the institutions in charge.17,18 In one of the communities in this study, members of a focus group discussion complained of how the responder agency was being manipulated by the ruling government leading to some disagreements and conflicts amongst community members due to the distribution of relief items along partisan lines.

In non-disaster situations, many of the agencies involved in disaster management operate independently of each other. At the community level in a disaster situation however, this is often not the case as complexity arises from a variety of elements, systems, processes and actors in a complex network of many interactions between these agencies at various levels necessary for achieving mutual adjustment and a collective goal.15,19,20,21 Although a number of agencies were involved in managing the floods in the Kasena Nankana west district of northern Ghana by assisting the main responder agency, majority of the affected communities did not see any tentative collaboration or networking between the various actors. Effective networking and collaboration at the community level in particular can serve as a basis of community members developing confidence in disaster management and also help reduce the number of allegations labeled against the responder agency.

It is a usual practice for NADMO to register victims of disasters before resources are mobilized and sent to the community.8 This process however tends to take a long time resulting in delays in supplying relief items as is common with most disaster management agencies.22,23 According to the United Nations, much of the lives loss during a disaster or hazard event occurs in the first 24-48 hours24 and therefore the delays on the part of NADMO in its response as a result of pre- registration of affected people can usually be catastrophic. During the 2012 flood disaster, all the communities which eventually got assistance from NADMO identified the timing of this assistance especially distribution of relief items to be late after most community members where already badly hit by the floods. This was also the situation in southern Ghana when NADMO delayed in distributing relief resources to registered flood victims in 2010.16

In one of the study communities, the need to conduct a needs assessment during and after disasters was identified as evaluative indicator as community members thought this was crucial especially concerning the distribution of relief items by the responder agency. This point has also been emphasized by Oaks in an Overview of the Damage Assessment Process: as he suggests that, after disasters, the damage assessment process is fundamental to relief and reconstruction as it triggers the beginning of formalized disaster relief and recovery aid.25 Similarly research has also shown that, damage assessment helps amongst other things to identify the immediate needs of disaster victims, whiles enabling first responders and emergency managers to recognize required materials and human resources.26

Majority of group members in the affected communities in the Kasena Nankana west district of northern Ghana rated the overall performance of the responder agency to be above average. Although some group members rated certain specific generated indicators low, they thought that in the whole, the responder agency performed creditably well. This finding only emphasis the fact that community members tend to assess performance when certain specific needs are met and not just an overall scoring scale of evaluative indicators as is usually found in quantitative studies. However contrary to the findings of our study, an earlier qualitative study involving the same responder agency found the overall performance of the agency to be poor after the catastrophic 2010 flooding situation in southern Ghana.16 Similarly in Ekiti State in Nigeria, the performance of the State Emergency Management agency which is the main disaster responder agency was found to be comparatively poor.27

The observed differences in the results across the four communities can be attributed to the fact the these communities are often affected differently by the perennial floods with the response from the responder agency being different depending on the level of preparedness by NADMO at the local level and availability of support from all stakeholders.


Notwithstanding the strong qualitative methodological approach adapted by this study, the study had some limitations. We conducted the study in only four communities and about two years after the disaster occurred and therefore the accuracy and precision of the information provided could have been affected especially that some of the community leaders would have relocated, died or forgotten certain vital events that occurred.

Although the composition of the groups was heterogeneous, group members were only elders and representatives of the communities and therefore their opinions might not necessarily be the views of the ordinary members of the communities who were hit hardest by the floods.


This study used a combined qualitative methodological approach of focus group discussions and the community score card in four communities in Northern Ghana to evaluate the performance of the main disaster responder agency. All four communities identified NADMO as the main responder agency during the last disasters with community members identifying education/awareness, selection process of beneficiaries, networking/collaboration, timing of response, quantity of relief items, appropriateness of relief items, mode of distribution of relief items, investigation and overall performance as the main evaluative indicators. Although the indicators were rated differently by each community; the timing of response, quantity and appropriateness of relief items were generally poor whereas the overall performance of the responder agency was generally satisfactory.

Since these communities are often prone and vulnerable to the perennial floods, we recommend that the main responder agency responds more timely and fast whiles engaging the other stake holders in disaster management to ensure that communities get enough of the most appropriate relief items. We also encourage that after each disaster, all the stake holders in disaster management great that platform were the disaster responder agency will be evaluated by the affected communities as a feedback to improve on future disaster management

Competing Interests

The authors have declared that no competing interests exist.

Financial Disclosure

This study was made possible by the generous support of the Global Disaster Preparedness Center and Response 2 Resilience. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. The views presented in this paper are solely those of the researchers and are completely independent of the funder.

Data Availability

All relevant data are within this manuscript.

Corresponding Author

Stephen Apanga

Email: apangastephen@hotmail.com

University for Development Studies, Department of Community Health and Family Medicine School of Medicine and Health Sciences


A demonstration of community scorecard scoring process

Fig. 2: A demonstration of community scorecard scoring process

http://currents.plos.org/disasters/article/listening-to-the-voices-of-the-people-communitys-assessment-of-disaster-responder-agency-performance-during-disaster-situations-in-rural-northern-ghana/feed/ 0
Interventions for Prevention of Intimate Partner Violence Against Women in Humanitarian Settings: A Protocol for a Systematic Review http://currents.plos.org/disasters/article/interventions-for-prevention-of-intimate-partner-violence-against-women-in-humanitarian-settings-a-protocol-for-a-systematic-review/ http://currents.plos.org/disasters/article/interventions-for-prevention-of-intimate-partner-violence-against-women-in-humanitarian-settings-a-protocol-for-a-systematic-review/#respond Wed, 12 Jul 2017 16:27:50 +0000 http://currents.plos.org/disasters/?post_type=article&p=34766 Introduction: Humanitarian emergencies and the number of people who are adversely affected are increasing. In such emergencies, the vulnerability of women and girls to gender-based violence increases signifi­cantly and they often experience high levels of intimate partner violence (IPV). There are a limited number of interventions to reduce gender-based violence (GBV) and IPV in the contexts of humanitarian emergencies, and there is uncertainty about the effectiveness of these preventive interventions. This is the protocol for a systematic review that will synthesize the evidence on interventions for primary or secondary prevention of IPV in humanitarian settings, and assess the effect of existing types of IPV-related interventions in these settings.

Methods and Design: The PRISMA-P 2015 statement has been used to prepare this report. Studies published from January 2000 to January 2017 will be reviewed with no language limits. Any experimental, quasi-experimental, or controlled trials will be included. A combination of four key concepts, including “IPV” AND “population” AND “humanitarian setting” AND “intervention” will be used in the search and a variety of information sources will be used: (1) bibliographic databases; (2) special databases and grey literature; (3) and the reference lists of eligible studies. Two reviewers will independently screen articles, extract relevant data and assess study quality. Discrepancies will be resolved through consensus. Risk of bias will be assessed using the Cochrane Risk of Bias tool and the quality of evidence will be assessed using the CONSORT checklist. A narrative synthesis will be provided. If a sufficient number of studies are found, their results will be pooled using a random-effects meta-analysis. For dichotomous outcomes, summaries of intervention effects for each study will be provided by calculating risk ratios with 95% confidence interval. Standardized mean differences will be used for continuous outcomes.

Discussion: The review will be useful for IPV management policy and related planning. It will help researchers, policymakers and guideline developers with an interest in reducing violence against women among refugees, internally displaced persons (IDPs), and conflict-affected population.



Almost, 11 million people were registered as refugees or internally displaced persons (IDPs) across the world in January 2012, in an estimate from UNHCR 1. The refugee population reached an unprecedented 19·6 million individuals worldwide in 2015 and the number is steadily increasing 2,3. UNHCR also reported the highest number of forcibly displaced people in recorded history in 2015: approximately 59.5 million 4. Humanitarian emergencies and the number of people adversely affected are increasing 5. In September 2016, at a Summit in the United Nations General Assembly, the “New York Declaration for Refugees and Migrants” was adopted expressing the political will of world leaders to save the lives of refugees and migrants, protect their rights, and share responsibility for large movements on a global scale. One of the main commitments of this document was the prevention and response to sexual and gender-based violence 6. Violence against women in conflict-affected settings and complex emergencies is a serious global public health issue 7,8. During periods of conflict or displacement, the vulnerability of women and girls to sexual and gender-based violence increases signifi­cantly 9. Research has shown that approximately one in five refugees or displaced women experienced sexual violence alone 1. Also, emerging prevalence data show that intimate partner violence (IPV) is relatively common in these settings 10. IPV is the most prevalent form of gender-based violence 11 and is an important global public health problem 12.Women are the most common population suffering from IPV and adverse impacts are widely reported among women who experienced IPV 13,14,15,16,17,1819. The Global Burden of Disease Study and WHO (2013) estimates that over 30% of women age 15 or over have experienced some kinds of IPV in their life, such as physical or sexual IPV 8,20,21. Conflict-affected, refugee and internally displaced women often experience high levels of IPV 7. According to some studies, the prevalence of IPV among conflict-affected communities in Bosnia and Herzegovina, Serbia, East Timor, and the Democratic Republic of Congo (DRC) ranged from 24 to 76% 22. In other studies, IPV prevalence among Palestinian refugees and conflict affected displaced women on the Thai–Burma were between 22 to 42% and nearly 8%, respectively 22.

The increases in the number of humanitarian emergencies and growing number of refugee, IDPs and affected people related to them, makes research in this field are an international policy concern and a research evidence priority 5.

Despite the increasing number of humanitarian emergencies 5, increasing IPV occurrence in these settings 21, and IPV related adverse health outcomes 19,22,23,24,25,26, current knowledge on how to prevent IPV is limited 10. Some of the interventions or strategies that have been implemented are women empowerment, gender equity, group training, advocacy and psychosocial support interventions 27.

There are a limited number of evaluations of interventions aimed at reducing GBV and IPV in the contexts of humanitarian emergencies and refugee contexts 27,28,29. Although there has been an increase in the number of programs in refugee populations to prevent and respond to GBV, particularly sexual violence against women and girls, there remains a general lack of evidence on the effectiveness of these efforts (prevention programs, interventions, and strategies, especially among refugee populations) in preventing diverse forms of GBV, and a lack of evaluation of efforts outside of conflict-related sexual violence 28. There is a need for accessible research that evaluates the efficacy and effectiveness of various GBV prevention and management strategies in refugee and displaced populations 29.

This is the protocol for a systematic review that will synthesize information on interventions for primary or secondary prevention of IPV against women in humanitarian settings. It will also investigate the effect of existing types of interventions on prevention or reduction of physical, sexual or emotional IPV in these settings.

Objectives of study

The objective of the systematic review is to identify existing IPV prevention interventions, strategies and programs among refugee, internally displaced, or conflict-affected female in humanitarian settings; and to assess their effects. It will seek to answer the following questions:

  1. What are the existing types of intervention for primary or secondary prevention of IPV against women in humanitarian settings?
  2. What is the effect of various types of interventions on prevention or reduction of physical, sexual or emotional IPV in humanitarian settings?
  3. How do various types of intervention related to IPV affect health outcomes in humanitarian settings?


The Preferred Reporting Items for Systematic reviews and Meta-Analyses for Protocols 2015 (PRISMA-P 2015) statement has been used for preparing and reporting of this systematic review protocol 30.


This systematic review was registered in the International Prospective Register of Systematic Reviews (PROSPERO) on 18 September 2016 and updated in the registry on 15 December 2016 (CRD42016047497 available at http://www.crd.york.ac.uk/PROSPERO).

Eligibility criteria

Studies will be selected according to the criteria outlined below.

Study designs

Any experimental, quasi-experimental, or controlled trials studies (randomized trials or quasi-randomised trials) will be included.


The population will be restricted to refugee, internally displaced, or conflict-affected women aged 15 years and over who live in the situation of humanitarian settings. According to UNHCR, IDPs are defined as “persons or groups of persons who have been forced or obliged to flee or to leave their homes or places of habitual residence, in particular as a result of, or in order to avoid, the effects of armed conflict, situations of generalized violence, violations of human rights or natural or human-made disasters, and who have not crossed an internationally recognized state border” 31.Also, refugees are defined as people who have been forced to flee their country because of persecution, war, armed conflict or violence 32. In this review, refugees and IDPs will be defined as people who have been displaced or fled within or outside their home countries because of armed conflict and other situations of violence (not related to natural disasters). We define “conflict-affected persons” as persons residing in active or recent conflict zones or in a post-conflict settings 33,34. People Affected by Conflict (PAC) will be considered, including those who are forcibly or not forcibly displaced and Conflict Affected Residence (CARs) who, for any reason, did not flee or leave their homes or places of habitual residence 31.

Studies in which it is not clear if the study population were refugees or migrants will be excluded. Studies that only focused on displaced population associated with natural disasters will be excluded. There will be no limits on study participants in terms of country or ethnicity. All refugees or IDPs who relocated in high-income country settings will be eligible for this review.


IPV will include physical, sexual, or emotional forms of violence, abuse or battering between intimate partners. We define IPV-related interventions as any type of intervention conducted in a humanitarian setting for women or men in order to prevent or reduce violence against a woman partner or its related health outcomes. The studies need to have a clear pre/post intervention evaluative component in order to be able to assess its effectiveness. Also, the interventions should focus on efforts that targeted the primary or secondary prevention of IPV or its related health outcomes, including efforts to prevent initial victimization or re-victimization in defined population and settings. Primary prevention aims to prevent initial IPV and secondary prevention aims to prevent or reduce ongoing IPV or its related health outcomes 35.

Eligible interventions include individual interventions; population-, group- or community-based prevention interventions; single or multi-component interventions and programs and may be delivered at various levels 36,37.Intervention types may include, but not be limited to: educational programs, group training, survivor care, advocacy, psychosocial support, home visitation, batterer intervention, livelihood interventions or strategies, community mobilization, personnel interventions, cash transfers, police advocacy, legal strategies, systems and security interventions, and other forms of IPV-related intervention 27,35,37,38.

Interventions that aim to address and modify health-related outcomes following IPV will also be included.


Any comparison group (such as no intervention, or an alternative intervention) will be eligible.

Types of outcome measures

In this systematic review, the primary outcome related to IPV-intervention will be prevention or reduction in male-to-female physical, sexual or emotional IPV or changes in severity, frequency or types of physical, sexual or emotional IPV against affected women after the intervention in humanitarian settings (e.g. reduced sexual abuse).

Changes in health outcomes which occur following the implementation of the interventions will be regarded as secondary outcomes for assessing the impact of intervention. These include, but are not limited to:

  • changes in physical health outcomes related to IPV, including decreased physical health sequela (such as deaths [decreased mortality related to IPV]), decreased injuries and changes in acute or chronic health disorders [decreased morbidity related to IPV] 13,14,15
  • changes in psychological/mental health outcomes related to IPV (such as improvement in mental well-being, reduced depressive symptoms) 1617;
  • changes in sexual and reproductive health (SRH) outcomes related to IPV (including decreased incident of HIV/STIs and reproductive disorders, unwanted or mistimed pregnancy; pregnancy coercion, pregnancy termination; improvement in contraceptive use, improvement in sexual health behaviors)18,19.

Some other changes following the intervention will also be regarded as secondary outcomes. These include:

reduction in the occurrence of any other type of IPV (e.g. economical violence), changes in behaviors (e.g. improvement in controlling behaviors, improvement in healthcare use, use of IPV and community-based resources/referrals, improvement in health/help seeking behaviors), changes in attitudes or social norms (such as reduced acceptability of spousal violence among men, equitable gender norms), increased financial autonomy and security among women, improved economic well-being, enhanced relationship quality, empowerment of women, and promotion of health and health-related quality of life 21,24,27,35,37.

The debate about appropriate outcomes for IPV intervention studies, means that we will not use outcomes measures to determine study eligibility 35.


A humanitarian setting is one in which an event or series of events has resulted in a critical threat to the health, safety, security or well-being of a community or other large group of people. This can be the result of events such as armed conflicts, natural disasters, epidemics or famine, and often involves population displacement 39. In this review, the situation of humanitarian settings is defined using the criteria specified by the Sphere Standards as a range of situations including conflict and complex political emergencies in all countries (including low- or middle-income countries) 40,41. Eligible studies will be those conducted in humanitarian settings, that is contexts affected by war, terrorist attack, political violence or armed conflict.

Refugee/IDP camps or settlements will be eligible settings. All refugees, IDPs, or conflict-affected people who are living in non-camp settings (e.g urban, pre-urban, or rural area) will be included.

Data and Language:

No language limits will be imposed on the search. Studies published between January 2000 and January 2017 will be sought.

Exclusion criteria:

Studies not primarily about the prevention or reduction of IPV or its health-related outcomes will be excluded from this review; as will studies that do not describe an eligible population or intervention. The exclusion criteria include (1) not about IPV (other GBV topics e.g., female genital mutilation, forced or early marriage, rape); (2) female-to-male IPV; (3) not about refugees, IDPs, or conflict-affected; (4) not describing prevention interventions or strategies; (5) not describing or measuring the outcomes of the intervention; (6) no baseline data for comparison; (7) qualitative studies and descriptive quantitative papers with no specific intervention and no outcomes; (8) letters, editorials, commentaries, reviews; and (9) tertiary prevention.

Information sources

A variety of information sources will be used to identify studies meeting the inclusion criteria: (1) electronic bibliographic databases for published studies, using a comprehensive search; (2) special list databases and grey literature; and (3) the reference lists of studies included in the review.

Electronic bibliographic databases

Published literature in the following databases will be searched: PubMed (MEDLINE), Web of sciences, Scopus, Cochrane Library ((The Cochrane Central Register of Controlled Trials (CENTRAL)), PsycINFO, ProQuest, and google scholar.

The dates of the search will be recorded for each database at the time the search is done.

Gray literature

The following databases and websites will be searched for reports and unpublished papers: WHO Global Health Library, UNHCR database, Reproductive Health Response in Crisis Consortium (RHRC), Inter-Agency Working Group on Reproductive Health in Crisis (IAWG), and the Reproductive Health Access Information and Services in Emergencies (RAISE). The PROSPERO registry of systematic reviews will be searched for registered reviews in this field. To ensure literature saturation, reference lists of included studies will be checked.

Documentation of the information sources, including the name of each source, the date range searched (start and end dates) and the search platform (for electronic database searches) will be recorded.

Key search terms and Search strategy

Medical Subject Headings (MeSH), key terms and text-based search terms will be adapted from prior published literature. A combination of four key concepts (main key terms and their synonyms) will be used: “IPV” AND “population” AND “humanitarian setting” AND “intervention”.

The main search terms will be used in different combinations, using the Boolean operators (“AND” and “OR”) and wildcard variants depending on the database being searched. In searching, the Boolean operator “OR” will be used to link each key concept to their synonyms. Then, all key concepts will be combined using “AND” to identify relevant literature. The searches with these defined key words will be limited to the title, keywords or abstract of the article. Search terms will be combined in the following manner:

(All ‘IPV’ terms connected by the search term OR) AND (all ‘population’ terms connected by the search term OR) AND (all ‘humanitarian setting’ terms connected by the search term OR) AND (all ‘intervention’ terms connected by the search term OR). Table shows the key search terms. The full search strategy used for two main databases will be presented in the final report. The initial proposal for the search strategy for PubMed is in Appendix A. This strategy will be applied to the other electronic databases where relevant, modified as necessary.

Table 1-page-001 cropped

Table 1: Key search terms that will be used and their combination.

Data management

All search results will be downloaded and saved into reference management software (EndNote version X7).

Study selection

The description of study selection in final report will be according to the PRISMA guidelines, including the PRISMA flow diagram 42.

Two reviewers (MD and LA) will read the titles and abstracts of citations retrieved from searching, after exclusion of duplicate records, to remove those that clearly do not meet the inclusion criteria or are outside the scope of the review. The abstract of retrieved papers that are in a language other than English will be translated using Google Translate and then will be judged. The same two reviewers will read the full text of potentially relevant studies and confirm their eligibility. In the event of any uncertainty, a third reviewer (EM) will be consulted. The decision of a third reviewer will determine study eligibility. An electronic file that lists excluded references by primary reason for exclusion will be maintained.

Data extraction

Data will be extracted from included papers by using a standardized pre-piloted data extraction form. This form will be adapted from Cochrane Public Health Group’s “Guide for Developing a Cochrane Protocol” 43. The data extraction form will be piloted by two reviewers (MD and LA) on a set of the studies to assess the feasibility of the use of this tool, check its adequacy and consider if any revisions or changes are needed. All included studies will be reviewed and coded according to their relevance. Data extraction will be conducted for all included articles by one team member (MD) and verified by a second (LA). Disagreements between reviewers will be resolved by discussion.

Data items

Data will be extracted for the following:

Study ID/characteristics (first author name, year of publication, evaluated country), participants/population type (e.g. refugee, IDPs), socio-demographics characteristics of participants (e.g. age, Gender, Place, Race, ethnicity, Religion, Education), study setting/situation (e.g. refugee camp, settlement, urban/pre-urban refugee settings, conflict-affected settings, etc.), specific criteria for inclusion or exclusion, study design (e.g. experimental or quasi-experimental), sample sizes, intervention type and description (intervention details), total number of intervention groups, frequency and duration of interventions (exposure period), the specific goals of the interventions, the duration of follow-up, intervention impact, retention rates, control group types and descriptions, outcome definitions and measurement tools, main findings/results, miscellaneous, and key conclusions including implications for practice and/or research from each study. Potential discrepancies will be resolved by consensus.

Risk of bias/quality assessment

Two review authors (AK and MD) will independently assess the possible risk of bias for each included studies by using the Cochrane Collaboration’s Risk of Bias (RoB) tool according to the following domains:

  • Randomization sequence generation

  • Treatment allocation concealment

  • Blinding

  • Completeness of outcome data

  • Selective outcome reporting

  • Other sources of bias 44

Each included study will be rated as low risk, high risk, or unclear and the rating of each study will be included in an appendix in the final report. If insufficient detail is reported for a study, the risk of bias will judged as ‘unclear’. Any disagreements between the reviewers over the risk of bias will be resolved by discussion and consulting with a third author (AA).

A quality appraisal of the full texts of eligible papers will be done using the CONSORT checklist to assess methodological quality 45. A pilot trial on two included studies will be conducted to assess the feasibility of using this tool in this review. Study quality will be independently appraised by two reviewers (MD and LA). Papers will be ranked as high, medium or low quality rating based on the checklist criteria. If any disagreements arise, a third reviewer will be consulted (AA).

Data synthesis and analysis:

A narrative synthesis of the findings from the included studies will be provided. Key findings related to each type of prevention intervention will be analyzed separately and results will then be combined to provide an overall synthesis of publications related to interventions to prevent IPV in humanitarian settings. The narrative synthesis of the findings will focus on target population characteristics, characteristics of included studies (for example, type of humanitarian emergency and study setting), type and content of interventions, duration of the intervention, effectiveness of interventions, types of outcomes measured and findings.

Quantitative synthesis will be used if the included studies are sufficiently homogeneous and comparable across interventions and outcomes measured. Whether or not a meta-analysis is possible, a summary of findings table will be used to present the findings for each primary outcome.

However, it is anticipated that there will be limited scope for meta-analysis because of the range of different outcomes measured across the small number of existing trials 46. If a sufficient number of articles with the same type of intervention and outcome measure are found, the results of these studies will be pooled using a random effects meta-analysis. For dichotomous outcomes, we will provide summaries of intervention effects for each study by calculating risk ratios (RR) with 95% confidence interval (CI). We will use standardized mean differences (with 95% CI) for continuous outcomes.

Dealing with missing data

When there are missing data, we will attempt to contact the original authors of the study (by using email addresses on the study’s publication) to obtain the relevant missing data.

Analysis of subgroups or subsets

If sufficient data are available, subgroup analyses will be used to explore possible sources of heterogeneity, based on different types of interventions.

Reporting of results

This protocol followed the PRISMA-P Statement 47 . and the final report of the systematic review will be presented according to the PRISMA guidelines. For ensuring transparency, a PRISMA flow chart, 48 a table of included studies and the reasons for exclusion of studies will be clearly documented as Figure 1.

PRISMA 2009 flow diagram

Fig. 1: PRISMA Flow Diagram

Interpretation of findings

The results of the review will be discussed in the context of the quality of evidence, and the limitations and the strengths of findings, with emphasis on general interpretation of the results in the context of other evidence and their implications for IPV management practices (including IPV-related intervention, strategies and programs) and for future research in humanitarian settings.


The number of people affected by humanitarian emergencies is increasing and making it an international policy concern and a research evidence priority 5. Despite the increasing number of refugee and displaced population following humanitarian emergencies 49, and the high level of different types of GBV in these settings 50, ncluding IPV, there is a lack of evidence about IPV-related intervention and their effectiveness in humanitarian settings. Also, notwithstanding growing attention to IPV globally, systematic evaluation of evidence for IPV prevention remains limited 35, particularly after humanitarian emergencies. Therefore, the limited and scattered evidence in the literature supports the need for a systematic review in this regard 51. This review will provide a detailed summary of the existing evidence on the implementation and effectiveness of various used IPV-related interventions to prevent or reduce occurring or reoccurring IPV against refugee, displaced and conflict-affected women. The evidence will be useful for IPV management policy and related planning. The review will also help researchers, policymakers and guideline developers with an interest in reducing violence against women associated with the humanitarian setting.

One of the strengths of this review is the use of a systematic approach and employing reliable and robust tools such as PRISMA-P 2015 47. Although systematic reviews of randomized trials are considered to be the highest level of evidence for assessing the effects of interventions 52, restricting our review in this way would be limited by methodological challenges 52,53. Such trials are difficult to implement in the setting of humanitarian emergencies and we may not be able to find sufficient trials to provide users of this review with a body of useful information 53. To overcome this challenge, we will search a wide range of databases and sources, after piloting and revising the search strategy 54.


As this protocol is for a systematic review, formal ethics committee review is not required.

Review output and dissemination plans

Academic review products will be produced, critically appraising the implementation and effectiveness of interventions targeting prevention or reduction of IPV against women affected by humanitarian emergencies. The systematic review articles will be submitted to academic journals for peer-reviewed publication and the findings will be presented at conferences and seminars.

Authors’ contributions:

MD is the guarantor and conceived the original research idea with guidance from EM. MD developed the first draft of the protocol. AA, EM and AK oversaw the development and revision of the first draft of the protocol and contributed to revisions. MD, LA and AK contributed to the development of the selection criteria, and the strategy for the assessment of risk of bias assessment and data extraction. MD, AK and LA will develop the search strategy. MD and LA will extract data from studies. MD and AK will assess the risk of bias for each included study. AA and AK will provide statistical expertise, perform and interpret the analysis. EM and AK will resolve any disagreements. MD will draft the final report of the systematic review. All authors read and approved this manuscript, and will do the same for the final report of the review.

Data Availability Statement

All relevant data are within the manuscript and the public repository Figshare: http://figshare.com/s/cb2feda97952c89eec8e. For more information, please contact the author Marjan Delkhosh (delkhoshmarjan@gmail.com).

Competing Interest Statement

The authors have declared that no competing interests exist.

Corresponding author

Effat Merghati Khoei, PhD. Iranian National Center of Addiction Studies (INCAS), Iranian Institute for Reduction of High-Risk Behaviors, Tehran University of Medical Sciences, Tehran, Iran.

Brian & Spinal Cord Injury, Neuroscience Institution, Tehran University of Medical Sciences, Tehran, Iran. Email: effat_mer@yahoo.com


Appendix A-page-001 cropped

Appendix A: Sample search strategy in PubMed

Appendix A-page-002 cropped

Appendix A: Sample search strategy in PubMed

Appendix A-page-003 cropped

Appendix A: Sample search strategy in PubMed

http://currents.plos.org/disasters/article/interventions-for-prevention-of-intimate-partner-violence-against-women-in-humanitarian-settings-a-protocol-for-a-systematic-review/feed/ 0
In the Field Feasibility of a Simple Method to Check for Radioactivity in Commodities and in the Environment http://currents.plos.org/disasters/article/in-the-field-feasibility-of-a-simple-method-to-check-for-radioactivity-in-commodities-and-in-the-environment/ http://currents.plos.org/disasters/article/in-the-field-feasibility-of-a-simple-method-to-check-for-radioactivity-in-commodities-and-in-the-environment/#respond Tue, 30 May 2017 08:58:21 +0000 http://currents.plos.org/disasters/?post_type=article&p=31937 Introduction: Some release of radionuclides into the environment can be expected from the growing number of nuclear plants, either in or out of service. The citizen and the big organization could be both interested in simple and innovative methods for checking the radiological safety of their environment and of commodities, starting from foods.

Methods: In this work three methods to detect radioactivity are briefly compared  focusing on the most recent, which converts a smartphone into a radiation counter.

Results: The results of a simple sensitivity test are presented showing the measure of the activity of reference sources put at different distances from each sensor.

Discussion: The three methods are discussed in terms of availability, technology, sensitivity, resolution and usefulness. The reported results can be usefully transferred into a radiological emergency scenario and they also offer some interesting implication for our current everyday life, but show that the hardware of the tested smart-phone can detect only high levels of radioactivity. However the technology could be interesting to build a working detection and measurement chain which could start from a diffused and networked first screening before the final high resolution analysis.



The number of nuclear plants in the world, either in or out of service, is growing and many of them are built with obsolete technology and are becoming older and older 1. Therefore some release of radionuclides into the environment can be expected in the future.

For this reason, the average (informed) citizen and big organizations could both be interested in innovative and simple methods to check the radiological safety of the local environment and of commodities, starting from foods. Traditional methods require special expertise with expensive and heavy devices and for this reason they are not very popular and only specialized organizations can afford them. On the other hand, if detecting radioactivity were as simple and cheap as measuring atmospheric pressure or body temperature, it could raise a great interest especially in our networked society and particularly during a radiological emergency.

In recent years, an astounding availability of portable and wearable 2 sensors has flooded the consumer market, as an effect of the explosion of the smartphone market.

Almost anyone can currently monitor georeferenced environmental parameters and can find out and share their measurements in real time 3.

The parameters that can be more easily measured by a smartphone and the appropriate apps installed and running are: local magnetic field (Intensity and direction), visible light (intensity and composition in terms of three-component colorimetric measurement), sound (level and spectrum), atmospheric pressure, atmospheric humidity, latitude, longitude, speed, acceleration, local gravity. In the same way it is well known that almost everyone can take and share in real time georeferenced photos, videos and sound recordings.

It is less widely known that with the same hardware (a smartphone with a built-in camera) and the appropriate software, anyone can monitor radioactivity and measure its intensity, although with sensitivity limitations 4,5,6,14.

In this work, three methods for radioactivity detection and measurement in commodities and in the environment are briefly compared and discussed. They are different in technology, sensitivity, resolution, and cost.

A method that is very recent relies on the current smartphone technology, with no need for additional hardware, which makes it cheap and widely available. It could be valuable for a wide screening activity and for the production, spreading and sharing of information 7. The aim of this work is to verify its putative usefulness in the field.

The two other methods are well known in the specialized laboratory and depend on specialized instruments. They were adopted here as a reference to evaluate the usefulness of the method based on the smartphone technology.


The Reference Sources

To test and calibrate the detectors adopted in this work, a reference standard, characterized by a weak and safe (NRC/IAEA/EU exempt quantity) but clearly detectable gamma ray emission, was needed. A set of eight factory-calibrated point sources was chosen: the RSS8UN set by Spectrum Techniques, LLC., including Ba-133 (t1⁄2 = 10 year), Cd-109 (t1⁄2 = 462.6 day), Co-57 (t1⁄2 = 271 day), Co-60 (t1⁄2 = 5.27 year), Mn-54 (t1⁄2 = 312 day), Na-22 (t1⁄2 = 2.6 year) and Zn-65 (t1⁄2 = 244 day) at 3.7×104 Bq on May 2015, Cd-109 (t1⁄2 = 463 day) and Na-22 (t1⁄2 = 2.6 year) at 3.7×104 Bq on Apr 2015 and Cs-137 (t1⁄2 = 30 year) at 3.7×103 Bq on Apr 2015.

The Instruments

The first kind of detector adopted was based on smartphone technology.

Three common models were tested. They were equipped with a specialized app: “RadioactivityCounter” 4,8 (http://www.hotray-info.de/). The app is available for Android and iOS systems.

The adopted models were Samsung S4, Samsung S7 and Samsung A3.

The S4 is equipped with a 4.4×3.4=15 mm2 sensor (resolution 4128 x 3096 = 12780288 pixel), the S7 is equipped with a 4.2×3.1=13 mm2 sensor (resolution 4032 x 3024=12192768 pixel), the A3 is equipped with a 3.6×2.7=9.7 mm2 sensor (resolution 3264 x 2448 = 7990272 pixels).

CCD or CMOS chips, used as digital image sensors in surveillance or in smartphone cameras, are sensitive not only to visible light but also to higher energy photons. The software analyzes the signals produced by the front or by the rear camera of the smartphone, which has been previously shielded from visible light by an alluminium foil, subtracts the thermal noise and estimates the gamma-ray exposure of the sensor. Furthermore the background emission (measured before) can be subtracted.

The second kind of detector was the Geiger counter PRD 100 (http://www.prd100.com/), made by ITS srl. This device is equipped with a Geiger-Müller tube of 111 mm length x 11 mm diameter (max. section 1221 mm2), which is small enough to allow full portability but whose section is much larger (82 times) than the section of the largest smartphone sensor (15 mm2).

This detector shows additional interesting features: it is cheap (around 100 €), it is small and easily portable (190 g, 123 mm x 91 mm x 35 mm) it works with rechargeable standard batteries (3 AA), it can work in stand-alone mode and/or connected with a smartphone via a Bluetooth radio interface.

A dedicated app (Marie pro PRD-100) running on the smartphone provides some essential real time radiation counting and data displaying, saving, sharing and geo-referencing.

The third kind of instrument chosen in this work was a 1024 channel NaI(Tl) gamma spectrometer made by Ortec, priced around 18000 €.

The system hardware consists of a thallium-doped sodium iodide detector enclosed in a low-background lead shield (30 mm thick), an analog-to-digital converter (ORTEC DigiBase) integrated in an all-in-one spectrometer, and a laptop PC. The digiBASE supplies the multi-channel analyzer function, the high voltage for the NaI(Tl) detector, and all the signal processing electronics. The internal stabilization electronics and the internal check source (K-40 4500 Bq/kg) allow the system to be used over a wide range of environmental conditions. However it can be hardly defined portable, if the 80 kg lead shield is taken into consideration. The NaI(Tl) crystal is a 76.2 mm height x 76.2 mm diameter (3” x 3”) standard. The digiBase is connected to the control computer via a USB interface, which powers the whole system.

Several proprietary software components control the instrument, from the first setting to the final analysis. Ortec MCB Connections-32 acts as a first-level connection driver for the DigiBase. Maestro-32 MCA Emulation Software provides the second-level control of the DigiBase, the live spectral display and the automatic control of acquisition and analysis. This is achieved via a graphical user-programmable interface or via pre-programmed job streams. The software provides also data and results printing and storage. NuclideNavigator is an interactive gamma-ray reference and library program to view, query, and extract gamma-ray energies and yields, half-lives and parent/daughter relations from databases. It can be used to build application libraries or working libraries. ScintiVision-32 is an integrated multi-channel analyzer (MCA) emulator and gamma-spectrum analysis program. It integrates Maestro-32 functions and manages the collection and analysis of gamma-ray spectra. It includes commands that allow you to edit nuclide libraries and automated command sequences or “job streams.”

The set of technologies described regarding gamma-spectroscopy with sodium iodide scintillator allows the identification and quantitative determination of gamma ray emitting radioisotopes, either natural ones such as K-40, U-238, Th-232 or anthropic ones such as Cs-137 and I-131 9, single or in simple mixture 10.

A fourth method must be cited, though it was not experienced in this work, because it represents the state-of-the art in gamma-ray spectroscopy. It is based on High Purity Germanium (HPGe) detectors and can give excellent gamma signal resolution in the whole spectral window, even for energies as low as 3 keV, where NaI(Tl) detectors cannot usefully work and gives a resolution 16 times better than NaI(Tl). The major drawback of germanium detectors is that they must be cooled to liquid nitrogen temperature to produce spectroscopic data. An HPGe system is more complex to manage and on average is priced five to ten times as much as a NaI(Tl) system.

The radon issue

The Radon is a ubiquitous gas, mainly derived from the natural U-238 decay chain.

Radon isotopes emit alpha particles but some radionuclides from Radon’s progeny (Pb-214, Bi-214) emit gamma rays and can add their signals to the spectrum of a sample. To address this problem an independent Radon sensor was adopted to estimate the radon concentration during the analyses by the NaI(Tl) system.

The sensor adopted is the Rstone by Rsens. It can be connected to a computer by a proprietary USB pen. A proprietary program can read and analyze the data stored in the sensor’s memory. The sensor itself runs on a battery whose charge can guarantee up to two weeks of continuous autonomy and monitoring, and has a display to check the current Radon concentration measured during the last 30′.

The Background management

With the “RadioactivityCounter” app running on a smartphone, the sensor noise and the background emission can be measured and stored in the memory, and can be subtracted from each sample measure. Measures are expressed in CPM and saved as total and by-minute counts, together with the sensor temperature (that can affect the measure).

No background compensation is provided by the PRD-100 Geiger counter and by the corresponding Marie PRO PRD-100 app running on a smartphone connected to the instrument. The counting is only instantaneous.

The NaI(Tl) gamma spectroscopy systems allow full control of the background.

The instrument is protected by a lead shield that effectively prevents the variable environmental background radiation from hitting the sensor. The shield itself is a source of radiation mainly from U-238, and Th-232 decay series; anyway this radiation is constant (except for the environmental radon contribution) and can be measured and subtracted with high reliability and precision, acquiring “blank” spectra periodically. The K-40 internal standard can also be included in the background. Therefore a spectrum of the background and of the internal standard was acquired for 963933 s, checking for its stability (on the Energy axis) and saving the result every 1800 s. The stability was assessed by continuously keeping the laboratory temperature as close as possible to 295K and monitoring the stabilizer of the instrument, locked to the K-40 peak: if some adjustment occurred, then the corresponding 1800 s set of data were discarded. This way, the centroid of the photopeak of the K-40 at 1461 keV was always kept corresponding to the channel 552.20+/- 0.20.

The Calibration of the NaI(Tl) system

The spectrum described above, was used either for the background subtraction or for the first calibration step 11 and all the following spectra were acquired with the constraint of the K-40 centroid corresponding to channel 552.20 +/-0.20.

The energy calibration of the NaI(Tl) system was done in several steps. For a first energy calibration the K-40 peak (4518 Bq) was considered with some clearly recognizable peak of the background: the Tl-208 peak at 2614 keV, near the high energy end of the spectral window , from the Th-232 decay chain, which is widely used as a gamma tracer of natural thorium 12,13 and the Bi-214 peak at 1765 keV which is widely used as a gamma tracer of natural uranium 12,13.

During the second step the energy calibration was refined by considering, from the same background spectrum, the Ac-228 peak at 969 keV, the Tl-208 peak at 511 keV and the Pb-212 at 239 keV from the Th-232 decay chain.

For the final energy calibration steps, three reference sources were chosen and their corresponding spectra were acquired: Ba-133 (53 and 81 keV peaks), Cd-109 (88 keV peak) and Cs-137 (662 keV peak). The spectrum of each source was acquired separately, for 66691 s (Ba-133 and K-40), for 14189 s (Cd-109 and K-40) and for 81453 s (Cs-137 and K-40). The three point sources were acquired without any correction for the geometry, at 75mm from the detector surface, to lower the intensity of their signals and make it possible for the stabilizer to work properly.

The final calibration is reported below, where “Channel” is the integer index of the channel ranging from 0 to 1023:

Energy = -7.7883 +2.568091*Channel +0.000165948*Channel2

The full width at half maximum (FWHM) was calibrated as a linear function of energy:

FWHM = 4.8213 +0.038887*Channel

The Efficiency was factory-calibrated.

The test of smartphone sensitivity

The rear (main) camera of each smartphone was shielded from light using a piece of aluminium foil. The “Radiation Counter” app was loaded and launched and the shielding effectiveness was verified by assessing that the background remained unchanged also putting the shielded lens near a strong light lamp. The activity of the Na-22, Zn-65 and Cs-137 sources were measured putting them at 75, 30, 15, 5, 0 mm from the lens’ surface of the rear camera of each smartphone for more than 1200 s.

The Background was checked by putting the radioactive sources at a distance greater than 4m.

The same protocol was followed with the Geiger counter, referring the distances listed above to the instrument surface that directly covered the Geiger-Müller tube.


In the laboratory where the NaI(Tl) spectrometer works and where the radiation counters were tested, the radon concentration varied between 20 and 60 Bq/m3 during the testing periods and at this level did not require any special attention. The main environmental problem was to keep the temperature stable to have the NaI(Tl) spectrometer stable as well.

Only the strongest radioactive reference source (Na-22) could induce a response that was clearly different from the background in all the counters, smartphones included, as reported in Fig. 1 and in Tab. 1. Figure 1 and Table 1 show graphically (Fig 1) and in tabular format (Tab. 1) the measures obtained from each counter, at different distances from the Na-22 reference source.


Fig. 1: Graph showing the relationship between counts and distance, for each counter, from the Na-22 reference source.

The Graph shows the Counts Per Minute (CPM) of the four tested radiation counters and of the NaI(Tl) spectrometer, put at different distances from Na-22 reference source of gamma rays. The numerical values are reported in Tab. 1.


Table 1: The Table shows the output, expressed in CPM (Counts Per Minute) of the tested radiation counters and of the tested NaI(Tl) spectrometer, put at different distances from three different reference sources of gamma rays. For the NaI(Tl) spectrometer, the counts from the whole spectral windows were considered and the output adjusted for the half-life of the nuclides and for the different acquisition date.

The NaI(Tl) sensor showed a very different sensitivity.

For the sake of comparison, the photo-peak of the reference source of Cs-137 (3700 Bq), put on the surface of the NaI(Tl) sensor (18200 mm2) gave a count of 1.31*104 CPM and was “viewed” by the spectrometer as a 1L sample containing 7262 Bq/kg of Cs-137 (Fig. 2). This kind of instrument easily recognizes contents as low as the Tl-208 of the background, “viewed” as a concentration of 19.4 Bq/kg and whose photo-peak area corresponds to 6.60*10-2 CPM (Fig. 2).

Cs137_NaI(Tl)_SpectrumSensor Surface_04

Fig. 2: Gamma-ray spectrum of the Cs-137 reference source

Output of the NaI(Tl) Gamma-ray spectrometer. The point source of Cs-137 (t1⁄2 = 30 year, 3.7×103 Bq on Apr 2015, photo-peak 662 keV) was placed on the sensor surface. Also the photo-peaks at 2614 keV (Tl-208 from the Lead shield) and at 1461 keV (K-40 internal standard) can be clearly observed.

The same reference source of Cs-137 (Fig 3 and Tab. 1) was not sensed by the S7 and was hardly sensed by the A3 and only within a distance equal or less than 5 mm. Only the S4 performed reliably.


Fig. 3: Graph showing the relationship between counts and distance, for each counter, from the Cs-137 reference source.

The Graph shows the Counts Per Minute (CPM) of the four tested radiation counters and of the NaI(Tl) spectrometer, placed at different distances from Cs-137 reference source of gamma rays. The numerical values are reported in Tab. 1.

The Zn-65 reference source (Fig. 4 and Tab. 1) showed intermediate counts between the strongest (Na-22) and the weakest source (Cs-137).


Fig. 4: Graph showing the relationship between counts and distance, for each counter, from the Zn-65 reference source.

The Graph shows the Counts Per Minute (CPM) of the four tested radiation counters and of the NaI(Tl) spectrometer, placed at different distances from Zn-65 reference source of gamma rays. The numerical values are reported in Tab. 1.

The three smartphones performed very differently. The most recent one (Samsung S7), which is equipped with the most advanced (and small) sensor showed too much background thermal noise and was not able to sense the reference sources (Fig. 5 and Tab. 1) , except for the strongest (Na-22) at the minimum distance. On the other hand, only the S7, running under Android 6 (the A3 and the S4 runs under Android 5) showed no problem with file saving.


Fig. 5: Graph showing the response of the S7 smartphone to the different reference sources at the different distances.

The Graph shows the Counts Per Minute (CPM) of the three tested radiation sources placed at different distances from the S7 surface. The numerical values are reported in Tab. 1. The residual background level is also reported.

Among the tested smartphones the most reliable was the S4 which showed a good sensitivity and whose counts followed the expected trend (Fig. 6 and Tab. 1). The A3 showed less sensitivity and performed worse (Fig. 7 and Tab. 1).


Fig. 6: Graph showing the response of the S4 smartphone to the different reference sources at the different distances.

The Graph shows the Counts Per Minute (CPM) of the three tested radiation sources placed at different distances from the S4 surface. The numerical values are reported in Tab. 1. The residual background level is also reported.


Fig. 7: Graph showing the response of the A3 smartphone to the different reference sources at the different distances.

The Graph shows the Counts Per Minute (CPM) of the three tested radiation sources placed at different distances from the A3 surface. The numerical values are reported in Tab. 1. The residual background level is also reported.

The Geiger counter PRD100 (Fig. 8 and Tab. 1) showed a much higher (one order of magnitude) sensitivity of the best performing smartphone (S4) but the software running on the connected smartphone lacks some useful feature which is present in the “RadiationCounter” app. (that is to say data-saving in CSV standard format, a count every 60 s, temperature and battery status and, for each measuring session, total duration, background, noise related to the border of the sensor). Furthermore the app allows real-time background subtraction, 60 s to 1800 s means and general mean calculation.


Fig. 8: Graph showing the response of the PRD100 Geiger counter to the different reference sources at the different distances.

The Graph shows the Counts Per Minute (CPM) of the three tested radiation sources placed at different distances from the PRD100 surface. The numerical values are reported in Tab. 1. The residual background level is also reported.

The NaI(Tl) spectrometer showed (as expected) the best sensitivity (Fig. 9 and Tab. 1) and the highest reliability with the best background management.


Fig. 9: Graph showing the response of the NaI(Tl) spectrometer to the different reference sources at the different distances.

The Graph shows the Counts Per Minute (CPM) of the three tested radiation sources put at different distances from the surface of the sensor of the NaI(Tl) gamma spectrometer. The total counts from the whole spectral window are considered to allow a proper comparison with the other devices. The numerical values are reported in Tab. 1. The background is not reported here because it depends on the K-40 internal standard and on the lead shield, not on the external environment and it can be evaluated observing the spectrum reported in Fig. 2.


The current smartphone technology could open up to a massive radiation monitoring, with some limitations due to low sensitivity, where the number of data could compensate for the possible lack of precision and puts in the hands of the average citizen a direct knowledge that previously only specialized entities could access.

Furthermore, the fact that anyone can check for the safety of the nearby environment leads to a new kind of bottom-up control of the information released by interested stake-holders and by the public authorities, allowing more transparent decision making.

According to the United Nations Scientific Committee on the Effects of Atomic Radiation (UNSCEAR) the worldwide average natural dose to humans is about 2.4 mSv/y (millisievert per year) 15.

On the other hand, the International Commission on Radiological Protection (ICRP) recommends effective dose limits to reduce the risk of stochastic effects to tolerable levels 16 for members of the general public in planned exposure situations and proposes as first reference level 1 mSv/y. The Commission considers also existing exposure situations that are defined as “those that already exist when a decision on control has to be taken” and explains that “there are many types of existing exposure situations that may cause exposures high enough to warrant radiological protective actions, or at least their consideration”. The Commission proposes the reference interval between 1 and 20 mSv/y for these conditions that can be due to NORM (Naturally Occurring Radioactive Material), natural background radiation and radioactive residues within the human habitat. The Commission defines also the levels between 20 mSv/y to 100 mSv/y as “reference levels for the highest planned residual doses in emergency situations”. Therefore according to UNSCEAR and ICRP, an annual dose ten times the “worldwide average natural dose to humans” is already an emergency.

According to our data, at least one of the tested smartphone (S4) is able to reliably distinguish between a residual background of 0.2 CPM and a tenfold greater gamma emission (Tab. 1 and Fig. 6). The cited background can be assumed to be near the average value, according to UNSCEAR’s definition and according to the information available on the geographic area where the tests were done: the alluvial plain of Florence-Prato-Pistoia 17,18. Therefore, in such conditions, the device could be used to detect an environmental radiation level that could signal a radiological emergency, according to ICRP’s definition 16. On the other hand, no real-time detection can be expected in low exposure conditions because to achieve the reported results, several minutes were needed for each measure and all the measured counts, except the S4 measure of the Na-22 put directly on the smartphone surface (99.6 CPM) are less than one count per second. If we consider that one count per second is equal to 60 CPM that in turn, in this case, corresponds to 300 times the residual background, then it can be concluded that only high levels of gamma activity can be detected in real-time by this kind of device and that the low sensitivity of the small sensor of a smartphone can be a serious drawback, that could be compensated only by the adoption of an external sensor. The low sensitivity would not be a problem in a severe radiological emergency, where the usefulness of many working detectors should be unquestionable and where almost everybody could become an active node in an environmental sensory network. In this scenario the value of the information collected can only be imagined.

Of course the huge amount of environmental data that could be produced and shared needs some public revision and lab validation, and here a sensitive, traditional and inexpensive technology like the NaI(Tl) spectrometry can have a new role, before the definite Hi-Res confirmation that only the HPGe technology can give.

Corresponding Author

Stefano Alessandri, Department of Statistics, Computer Science, Applications “Giuseppe Parenti”, University of Florence, Florence, Italy. Email: stefano.alessandri@unifi.it

Data Availability

The data are freely available in figshare repository and can be accessed via the following links:

– Summary of smartphones, geiger counter and NaI(Tl) spectrometer dataset: https://doi.org/10.6084/m9.figshare.4555507.v1

– NaI(Tl) raw dataset: https://doi.org/10.6084/m9.figshare.4644997.v1

Competing Interests

The author has declared that no competing interests exist.

http://currents.plos.org/disasters/article/in-the-field-feasibility-of-a-simple-method-to-check-for-radioactivity-in-commodities-and-in-the-environment/feed/ 0
The Life Process of Children Who Survived the Manjil Earthquake: A Decaying or Renewing Process http://currents.plos.org/disasters/article/the-life-process-of-children-who-survived-the-manjil-earthquake-a-decaying-or-renewing-process/ http://currents.plos.org/disasters/article/the-life-process-of-children-who-survived-the-manjil-earthquake-a-decaying-or-renewing-process/#respond Tue, 04 Apr 2017 15:00:25 +0000 http://currents.plos.org/disasters/?post_type=article&p=30890 Introduction: Among earthquake survivors, children are more vulnerable than other age groups due to their exposure to harrowing scenes of devastation as well as their drastically new living situations that result from an earthquake disaster. The life process of children survivors undergoes many different changes that are affected by a wide range of factors. Understanding the life process of these children may lead to effective outcomes and interventions. In addition, observing children survivors establishes knowledge and understanding of the challenges that correspond with earthquake disasters. Further, observing this group may be further effective in decision-making and establishing types of assistance in similar circumstances.

Objectives: This study was done to explain the life process of children who survived the earthquake of Manjil in northern side of Iran.

Methods: This qualitative study is based on the grounded theory approach. The sampling involved purposive interviews with 12 children who survived the Manjil earthquake and were under 12 years of age at the time of the earthquake. The initial interviews were followed by continuous comparative analysis, and thus the sampling process adopted a theoretical trend. In the end, by the formation of categories and the central variable of the study, interviews were conducted with 16 subjects and sufficient data was provided. Data was collected through face-to-face, in-depth interviews using an interview guide. In order to enrich the categories formed in data analysis, we had also 6 telephone interviews with the same participants in order to complete missed needed information. Data collection began in 2015 and continued up until 2016. Data was analysed using the Strauss-Corbin approach.

Results: The life process of children earthquake survivors consists of ‘unexpected encounter’, ‘transient relief activities’ and ‘long-lasting consequences’. The central variable of this study is ‘the dark shadow of pain and the light shadow of life expectancy’. The life experience of this group of children is immersed in painful memories and varies under different conditions.

Discussion and Conclusion: According to the results of this study, one of the factors affecting the lives of children earthquake survivors which could threaten their health is providing non-specific and transient services. Training relief staff to consider the specific needs of these children at the time of the rescue operation could contribute to improving their health level in various aspects. Considering the effective and comprehensive rehabilitation program in Disaster Management by policymakers can prevent permanent complications caused by earthquakes. Planning and taking action to identify misbehaviours in this group of children as well as raising public awareness, particularly for parents, on how to manage the outcomes of natural disasters are some of the most significant public health priorities. Providing public mental health services for parents and children who survive an earthquake helps to address potential psychological problems in this group of survivors.



The Manjil earthquake1 occurred on June 21, 1990 at 21 GMT near the town of Rudbar, in villages of Gilan Province and the northwest region of Zanjan Province, Tarom Olya located in the north-west of Iran. The Manjil earthquake caused great human and financial losses within a 100-kilometer radius of the epicentre. The earthquake occurred in Rudbar and Manjil was one of the largest and most destructive earthquakes in Iran in recent decades. The earthquake caused nearly 35,000 deaths and 60,000 injuries, and left more than 500,000 people homeless. In addition, 200,000 residential units were demolished, 60,000 of which were completely destroyed. Due to the landslides that occurred during the Manjil earthquake, villages were deeply buried. The initial damage caused by the Manjil earthquake was estimated to amount to more than 800 billion rials, and the disaster caused economic losses equal to 2.5 percent of the gross national product2.


Published statistics of disasters around the world show that in the past two decades, more than 4.3 million people have died, millions of people have been injured and tens of billions of dollars have been spent on financial and life compensations due to natural disasters3. Due to the climatic and geographical situation of Iran, it is one the high-risk countries in terms of natural disasters4. Advances in emergency and injury care systems have caused survivors of natural disasters need wide physical, psychological and social recoveries5.

Among the survivors of the earthquakes, children are more vulnerable than other age groups due to the fact that they are exposed to distressing scenes of devastation. In addition, children survivors are especially vulnerable as their living situations immensely shift following an earthquake disaster. If the special needs of these victims do not receive urgent attention, the evolutionary process of their growth will undoubtedly be interrupted, and they will face serious physical and psychological effects in the near or distant future. Most planning programs emphasise the most immediate needs of disaster victims, including rescue and relief activities and primary care6.

However, it should be noted that in addition to the significant effects and damages associated with natural disasters, the impact of such disasters on victims’ quality of life is much deeper and severe. Therefore, it is crucial to improve the health of victims for prolonged periods after the disaster. Nevertheless, many of the studies conducted in this field have not taken a comprehensive look at this process, and have usually focused on only one aspect such as psychological interventions after disasters, post-disaster stress disorders7,8, physical injuries9, and the role of community participation in physical reconstruction10. Since few studies have specifically explained the lives of children who have survived natural disasters, this study seeks to investigate this particular group. Further, studying children survivors in this context is necessary due to the particular context and culture of Iranian society as it differs from other countries which is based on Islamic beliefs and traditional-local customs. Understanding the life process of these children may lead to effective outcomes and interventions, and explaining the life process of these children according to their experiences could lead to a better understanding of their needs, challenges, issues and problems following their experiences of natural disasters. Furthermore, such a study may reveal the challenges resulting from earthquakes, and may also provide results that are applicable to similar circumstances. Therefore, to address this gap in research, this study aims to explain the life process of children who have survived earthquakes.


This study was conducted using the grounded theory approach, a qualitative research method based on symbolic interactionism. This study examines the social processes in the context of human interactions11.

The sampling involved purposive interviews with 12 children who survived the Manjil earthquake and were under 12 years old at the time of the earthquake. The initial interviews were followed by continuous comparative analysis, and the sampling process therefore took a theoretical trend. Based on the formation of categories and the central variable of the study, interviews were conducted with 16 subjects, and sufficient data for ensuring about the data saturation was provided as a result. Data was collected through face-to-face, in-depth interviews using an interview guide In order to enrich the categories formed in data analysis, we had also 6 telephone interviews with the same participants in order to complete missed needed information. The interviews began with general questions and progressively involved more detailed questions regarding participants’ responsiveness. Questions included, ‘Can you tell me about your experiences after the earthquake?’ and ‘How was your life?’ Interviews were performed in a quiet atmosphere according to participants’ preferences, in locations such as their homes or in a park. According to the tolerance, willingness and environmental factors of participants, 22 interviews were conducted for this study. The duration of face-to-face interviews was 25 to 45 minutes, and telephone interviews lasted 15 to 20 minutes. Data collection started in 2015 and continued up to 2016.

Data Analysis

Data was analysed using the Strauss-Corbin approach11. According to this method, the current study respectively employed concept analysis (immediately after the first interview), context analysis, process analysis, and then combined and integrated the categories and proposed the basic theory (writing analytical stories and reviewing the reminders). The audio-recorded interviews were transcribed, and the typed texts were reviewed several times. Then, the texts were studied line-by-line and word-by-word, and a code was assigned to each key word or sentence. Similar codes were then put together in a category, and the primary classification of 598 codes was obtained. Constant comparison was used to determine the relationship between categories and search the underlying process of data. Then, the researcher integrated the categories using various diagrams to identify the central variable.


The authenticity of the data was examined using the four criteria of Guba and Lincoln12. The data collection took nine months using a combination of face-to-face interviews and telephone interviews. Results were confirmed through peer check and member check. Researchers contributed to the transferability of the data through in-depth descriptions and analysis. The sampling technique with maximum variation (the age at the time of the disaster, gender, losing parents in the disaster, occupation, education level and marital status) was also used, and this improved the suitability or transferability of findings. The researcher precisely recorded and reported on the research process to allow others to investigate the article.

Ethical Consideration

Permission for the study was obtained from the ethics committee of the University Of Medical Sciences Of Babol. Related departments were informed of the study objectives, and their written consent to participate in the study was obtained. The time and place of the interview was determined with the agreement of participants, who were assured of the confidentiality of all personal information and interviews as well as the anonymity of all documents related to the research. The results of the study were made available to the participants upon request. All moral standards relating to the use and publication of texts was observed. Permission to record participants’ voices was obtained. Only one of the researchers listened to the recordings. The recorded interviews did not include any personal identification information. Participants were informed that they could withdraw their information or remove themselves from the study altogether at any time.


In this study, 16 individuals (adults at the time of the study) who were child survivors of the Manjil earthquake were interviewed (Table 1).

M E Table 1

Table 1. Participants’ personal information

Analysis of the data resulted in the obtainment of 598 primary codes. During the process of data analysis, six main categories were extracted.

The life process of children survived earthquakes consists of ‘unexpected encounter’, ‘transient relief activities’ and ‘long-lasting consequences’. The central variable of this study is ‘the dark shadow of pain and the light shadow of life expectancy’. This process is influenced by ‘internal factors’ and ‘modifying factors’ (Table 2).

ME T 2

Table 2. Categories extracted from the data

Unexpected Encounter

Most of these children were sleeping at the time of the earthquake. Therefore, the fact that they did not anticipate the earthquake or its consequences is associated with reactions of panic and feelings of deprivation.

Panic Caused by the Disaster

The first hours after the earthquake were associated with a great deal of fear, apprehension and horror for the majority of these children. These feelings multiplied when they were exposed to the resulting devastation, such as the destruction of shelters and scenes of family members being buried under the rubble, as well as being exposed to wounded and dead victims.

“The ongoing conditions were like the Resurrection. Destroyed and burnt houses, cracked roads, blood, broken bones, dead bodies along the road etc. Everything was a proof of the Resurrection. I was terrified when my uncle was burying his sister, nieces, nephews and other relatives.” (Code 16)


The deep influences of the disaster appeared a few days after the shock and horror of the earthquake. Most children either lost their parents, or their parents were mourning for the loss of their children or other family members. What were previously conceived of as the neighbourhood, kindergarten, school, family and relatives were no longer present. There was no sign of love or loving atmosphere in the family. Most family members were looking for a piece of bread to eat and shelter to live in, and were struggling with many mental pressures. Sometimes the needs of these children were forgotten due to the disastrous situation. Most of these children had no place to sleep and were looking for a clue of memories of good old days in the ruins. Those who had lost their parents and family members wandered around the ruins and cried, and were in search of a safe haven in a world of fear and helplessness. These children were tired, hungry and looking for a way out.

“The earthquake is nothing but the death of love. All my relatives were buried in front of my eyes, the sky of hope was dark and the pain of this separation grew bigger, and as I grew up, it grew up, too.” (Code 14)

Loneliness and the lack of social security in the initial chaos after the earthquake exposed children survivors to mistreatment. This group of survivors trusted people who were available in the tense and cluttered atmosphere, and exposed themselves to mistreatment. On the other hand, these children were neglected or mistreated due to the preoccupation of police, aid workers and family members who were involved in removing corpses and victims from under the rubble. In addition, some children faced threats such as robbery or getting lost in the process of being delivered to the authorities.

“I was crying in a street among the ruins in that confusion, and I saw a man coming towards me. He told me that he came to take me to my mother. As soon as he saw nobody was noticing, he took my bangle by force in such a way that my hand was wounded, he put his hand on my mouth so that I wouldn’t be able to shout and he threatened me.” (Code 13)

Transient Relief Activities

After some time and immediate support efforts, relief activities were provided to enable reconstruction and rehabilitation. However, these steps involved rebuilding houses and buildings, while the important principle of reviving the lives of these children faded out. This category contains the two subcategories of ‘temporary social support’ and ‘ insufficiency of continuity of the care chain’.

Temporary Social Support

What appears in the early hours following a disaster is the influx of government and relief aids to disaster-stricken areas. However, after a short period of time, emotions subside and the amount of aid decreases. This especially diminishes significantly as time goes on. In fact, disaster activities are only vaguely recalled.

“There were lots of people and aids during the first days. But very soon, after one month everything was over. It was [as though] as if the earthquake had involved people only for those days.” (Code 15)

The concept of ‘forgotten’ refers to temporary social support. Although these children require compassionate care, most of them receive humanitarian support for just a few months following a disaster and are gradually deprived of loving attention.

“Children who have survived earthquakes are highly vulnerable and can be emotionally injured by the smallest issues. They need love and a safe shelter more than anything else.” (Code 12)

Insufficiency of Continuity of the Care Chain

A few months after a disaster occurs, continuous care is halted and survivors face serious physical and mental consequences related to the event. The majority of physical and mental health care is limited to emergency and acute care, and rehabilitation services are omitted. The majority of children deal with several complications due to a lack of long-term care services management and planning. ‘Discontinuous follow-ups’ have caused most survivors to suffer from multiple chronic physical and mental problems.

“I was wounded when the earthquake occurred. My leg was stuck under the ruins of the house and was injured. They took me to the hospital and I was taken care of there, but after that they took me home. My parents couldn’t afford to spend money for me. Unfortunately, I limp. If somebody cared about us then and thought [about] what would happen next, I wouldn’t be like this.” (Code 18)

Long-lasting Consequences

Consequences of earthquakes are often delayed and persistent. This category consists of subcategories including ‘persistent anger’, ‘living in the nightmare of the earthquake’, ‘darkness of the sky of hope’ and ‘persistent homelessness’. The consequences of the earthquake proved to be of lower intensity for children who did not lose their parents in the earthquake and could benefit from their physical and mental protection.

Persistent anger

After the earthquake, the games and carelessness of childhood shift toward adult behaviours. ‘The repetitive scenario of the earthquake’ replaces game-playing and living a happy life. Teenagers also experience a shift in perception as they too experience undesirable physical, economic and social conditions. The majority of children survivors cannot enjoy their childhood years as their welfare, physical condition and psychological capacity are lacking in the wake of a disaster. Therefore, this group of survivors refers to their adolescence as ‘unloaded clusters’ The continuous use of wording such as ‘sorrowful dialogues’ or ‘tragedies’ describe the significance of sad memories of the past experienced in their minds and souls. Their spirit is obscured by these dialogues, and these survivors live with the daily repetition of parts of their painful past and memories of the earthquake. Further, these survivors attempt to pay homage to their family members, classmates and friends by keeping their memorabilia. In other words, they consider forgetting the event as a source of guilt and disloyalty.

“At the time I should have been supported by my family, I was taken to an orphanage. As I grew up, I had to leave my friends and move to another place. These changes made me suffer. I didn’t understand how my teenage years passed.” (Code 4)

Living in the Nightmare of the Earthquake

One of the long-lasting and negative consequences for this group is living in the nightmare of the earthquake. The fear of the earthquake’s recurrence along with continuous insecurity and anxiety are always present. Most of these survivors are deprived of a normal life. Participants expressed that they live in fear not just of the earthquake’s recurrence, but also of other disasters occurring. This fear leads them to experience anxiety. Living with prolonged and continuous anxiety leads to conditions such as depression, psychiatric disorders such as drug abuse, and behavioural problems such as delinquency. Living in the nightmare of the earthquake causes ‘excessive sensitivity’ to others, especially family members. Such behaviours of risk aversion cause these people to draw on their memories of the earthquake and, as a result, deter family members from what they perceive to be possible hazards. This problem is associated with fear, and limits the social lives of victims and their family members. In turn, this brings about dysfunction in the relationships between victims and their family members.

“I’ve grown up and aged, but you know, I start shaking even when it’s windy. I go and sit in a corner of the room, hug my baby and get terrified. That fear is still with me.” (Code 11)

The Feeling of Hopelessness

Some of these survivors believe that the feelings of safety and health have disappeared in their lives and have been substituted by fear, uncertainty and loneliness. They reported that after the earthquake, they feel they have lost their direction in life and have lost all hope. They experience a life without financial and emotional support. Their lives have undergone many major changes, and the grief of losing family members and experience of permanent fear have replaced their normal routine of life. Those who have medical problems and severe physical complications caused by the earthquake are exposed to constant medical procedures and treatment recommendations.

“I was young at that time. Our financial situation wasn’t so bad. But suddenly, we lost everything. My father died hopelessly after a few years. I had to work in a factory. It was very hard. I had to work instead of study.” (Code 13)

The ‘guilt’ the respondents feel as the sole survivor, as well as the guilt associated with burying their loved ones, is a significant factor contributing to their feelings of despair. Most of them saw their family members stuck under the rubble and reported that they feel guilty because they could not rescue them.

“I suffered for a long time. I wished it wasn’t me but my parents who were alive. I was little and injured then. I was confused and frightened, and I couldn’t do anything for my family. It hurts me a lot.” (Code 4)

For this group of survivors, disappointment leads to a lack of desire and effort in social activities. Thus, they always feel that they are a ‘social burden’. Lack of adequate physical and academic qualifications for acquiring a job affects this process. Most of these individuals are not married and live with their relatives because they have lost their families. This problem enables the formation of the feeling of being a social burden.

“I went my uncle’s house in another city. That poor man had some children and didn’t have a good financial situation. He was a farmer. I was a burden to their family. I sometimes realised that they couldn’t help doing favours for me, but I also understood that they couldn’t afford to provide for their own children.” (Code 16)

Disappointment causes this group of survivors to refer to their current life situation as ‘bubbles on the water’ (they always think that everything is temporary and is not going to last long like a bubble), and keeps them from envisioning a future for themselves. They feel abandoned because their identity has been buried under the rubble of grief and misery. The missing future is represented in impairment, inability to support family expenses, lack of financial resources for marriage, losing parents and their emotional and financial support. In fact, these individuals believe that they feel they are not in control of their destiny or have failed to fulfill their destinies. The fear of a life without financial resources and family support is always with them.

“A long time ago, I realised that I lost everything in the earthquake. I have neither a good job nor money. I try more and more, but I am always on the first step.” (Code 2)

Persistent Homelessness

A few days after the earthquake, some children who lost their parents and did not have anybody to claim them were taken to children care centres. In their opinion, wandering in these centres and waiting to find a safe haven was the worst experience of their lives. They described the separation from family members, relatives, friends and their hometown in a very bitter manner. Some survivors were given to new families and thus faced many new problems. Orphans who were sent to welfare centres or new families grew up in educational systems that differed from their native culture. Sometimes, this way of life proved contrary to these survivors’ previous beliefs and life experiences, and thus led to behavioural and mood disorders. According to survivors, sympathy was the major factor in their acceptance to these new families or centres. However, over time, many children survivors once again experienced homelessness due to behavioural disorders and adversities.

On the other hand, the lack of necessary physical and mental conditions for adoption led some of these children to stay in care centres for a long time. Further, such conditions at times caused them to wander between different centres as they grew older or their physical or mental complications increased. Children without guardians were taken to welfare centres. There was also a significant difference between genders in child adoption such that most boys were adopted in a short time while girls spent a long time on the waiting list.

“I stayed in the nursery for a long time. Those who were with me were adopted by their relatives or new families. I was sick, I think that’s the reason.” (Code 15)

An important point revealed through data analysis was that some of the parents of children survivors were noted as ‘risky parents’. These parents pose risk factors in the physical, mental, moral and social development of their children due to the loss of their own parents or other children, or due to multiple physical injuries caused by the earthquake. These people cannot provide a safe haven for their children to live. Escape, isolation, and anxiety are just some of the outcomes and emotional experiences of children living with risky parents.

“At first, my father was too nervous because he had lost all his money and two of his brothers in the earthquake. He was usually bad-tempered to me. I remember that he hit me and my mother for small excuses. I grew up in that difficult situation and its result was that I couldn’t study at school. I always started fights at school. Now I am unemployed.” (Code 5)

Internal Factors

The internal factors affecting the life process of children who survived the earthquake are ‘understanding threats’, ‘fatalism’ and ‘dangerous nostalgias’.

Understanding Threats

For children earthquake survivors, facing stressful events, experiencing failures in family and social life, and emotional and educational breakdowns over the course of their lives cause them to feel threatened. The intensity of the feeling of threat shifts according to the context of each individual’s condition. Thus, the intensity and type of stressors influence the intensity of the feeling of being threatened. Feeling threatened affects these survivors’ performance in rebuilding their lives after the earthquake.

“I feel so impatient. Once I was supposed to marry a girl but she left me, I experienced a feeling exactly like which I experienced at the time of earthquake. I reviewed all of my miseries. I felt lonely and threatened very much. I was depressed and devastated for some days.” (Code 15)


Believing in fate and luck are personal factors affecting the life process of these children. According to participants, feeling unlucky suppresses the opportunity to build personal, family and social dimensions of life. It can also cause depression, isolation and social isolation.

“If I was lucky, why should I have been so miserable from childhood? Whatever I do, my bad luck and a part which has been already written in my destiny is with me.” (Code 5)

Dangerous Nostalgias

Internal factors which affect the life process of these survivors include breaks in time and retrieving memories of the earthquake when facing similar scenes or events.

“One day, a motorbike had an accident with a car in the street. The motorbike rider’s leg was stuck under the car’s tire. As I saw this scene, I remembered my mum, her leg was stuck under the rubble of our house. The motorbike rider’s shouts reminded me of stressful scenes and my mum’s shouts. I felt so bad. I cried for many days.” (Code 13)

External Factors

‘Delayed relief activities’ and ‘non-specific care’ are external factors which affect the life process of children who survived the Manjil earthquake.

Delayed Relief Activities

A remarkable finding in the data is that often, treatment measures are delayed due to children’s inability in expressing their immediate needs. This problem is worse in children with injured parents or for those who have lost their parents. Parents help aid groups in considering their children’s needs with the knowledge they have about their psychological characteristics or physical problems. Delayed diagnosis or treatment of different problems of these survivors could have serious and persistent consequences for them.

“I was there when they took a 7-year-old child to the hospital. The child cried and couldn’t answer them when they asked questions. Nurses didn’t even know what medicine the child was allergic to. I remember the doctor gave the child a medicine. The poor child was shocked immediately and died, and it was too painful for me. If they hadn’t given the drug, the child wouldn’t have been dead.” (Code 10)

Non-Specific Care

Health care is not often provided to alleviate the physical and mental needs of children who have survived an earthquake. This reduces the quantity and quality of services. Most attention is focused on mere physical needs, and the mental and emotional needs of these children are neglected. In the area of physical health care, aid workers forget some children, or at times offer them defective or delayed care due to their lack of knowledge of specific care for children.

“They took me to the hospital. I heard two doctors saying that if the aid workers had paid more attention or been trained, they wouldn’t have cut a woman’s leg who was on a bed next to me.” (Code 5)

Modifying Factors

‘Palliative self-care’ and ‘Internal evolution’ are modifying factors in the life process of children who survived the earthquake.

Palliative Self-Care

The lack of long-term social and family services leads to self-action to have a safe and normal social and family life. Children seek different strategies to rebuild various aspects of their lives during different life stages. This category consists of three subcategories including ‘rescue efforts’, ‘religious strategies’ and ‘resilience’.

Rescue Effort

The development of social communications, starting a family, having children and maintaining emotional ties are some of the strategies used by children in rebuilding their lives after earthquake. On the other hand, being optimistic in life, strengthening the ability of forgiveness, being involved in charity works, and public participation have calmed these people. According to participants, empathy is formed in peer groups because of their common pain. By developing social relationships, these people look for those who can be understood easily by them due to the common experience they share. Sharing memories of childhood and using coping strategies proposed in these groups improves their mental relaxation.

“At the moment, I am a stepmother of three orphans. I don’t have money, but this makes me feel good. My sister was six when she died, and I was 7 [at the time of earthquake]. I do it to make her happy.” (Code 6)

Some of the earthquake survivors who are capable of continuing their education attempt to help others by studying. Some even try to make their bereaved or dead parents happy.

“I tried to be a doctor. Any patient who is visited by me to me and doesn’t have money, I don’t charge him. I cooperate with the Red Cross voluntarily. If my family had gotten treatments in time, they would have been alive right now. I became a doctor to help my people if necessary.” (Code 7)

Religious Strategies

One of the strategies for dealing with the persistent crisis of an earthquake is resorting to religious beliefs. Strengthening religious beliefs by reading prayer books and such religious texts as the Quran, participation in religious ceremonies and saying prayers are some of the religious strategies reportedly used by those who survived the earthquake in order to cope with multiple physical, mental and social problems. On the other hand, fatalism and ‘submission to the will of God’ help these children deal with problems after the earthquake.

“As I grew up, I realised that God is the only refuge. Whenever I have problem I say prayers instead of crying and sadness. These beliefs help me out of all pains.” (Code 1)


Resilience is one of the other strategies that children who survived the earthquake use in the process of rebuilding their lives. Resiliency is the result of a shift from being risk-oriented to coping with and modifying stress factors, which leads to the preservation and promotion of health in many aspects. Patience and perseverance, improving self-confidence, seeking familial and social support, and optional forgetfulness are some aspects of resiliency. These earthquake survivors often try to forget what happened in the past through emotional suppression.

“Whenever I give up, I go to my friends and family. My mother understands me very well. She knows how much I was hurt then because I was a kid. I unburden myself [in] this way, and I have withstood many times [through] the help of her.” (Code 8)

Internal Evolution

Internal evolution refers to self-monitoring and positive psychological changes related to the earthquake. Self-monitoring and knowing oneself, others and God are some of the reported positive consequences of the earthquake. These respondents believe that knowing themselves and God gives them the power to deal with the chaos and emotional gaps caused by the earthquake.

“The earthquake was a very bitter memory. But it built me. It made me resistant. It made me appreciate people. It made me more patient. It made me experience things which were productive. When I wonder, I think that it taught me that such moments won’t happen again. I knew my abilities and myself. These made me successful.” (Code 7)


According to the findings of this study, the first step in the lives of children who have survived the Manjil earthquake involves the unexpected encounter with the disaster. King (2006) believed immediately after the incident, the children experienced different reactions of shock and horror. These emotional reactions include a range of physiological, psychological, social and behavioural reactions13. More than other age groups, children are more prone to distress when faced with an earthquake disaster14. The results this study showed that in the early ravages caused by the earthquake, the loss of parents and shelter may have put the children at risk of harm and mistreatment. Therefore proper planning in crisis management, paying attention to the characteristics and needs of vulnerable groups including children, and comprehensive support can reduce the risk of mistreatment15. Based on the results of this study, there is a need for plans to provide long-term services.

It is important to note that threatening factors are persistent due to the long-term and sometimes permanent effects of earthquake disasters, and children are in great need of care at all stages of their lives. Unfortunately, rehabilitation and long-term services are marginal in disaster management planning16. Under these circumstances, there is a significant increase in mortality and a notable decrease in the physical, psychological and social function of these children17. Even if these services exist in society, they do not cover all survivors due to their unavailability and young age. Therefore, providing and sustaining long-term rehabilitation services must be considered as priorities when it comes to children survivors of natural disasters18.

The results this study also indicated that temporary social support and the subsided emotions of people and charities over time cause survivors to be deprived of support services, especially long-term emotional support. According to this group of survivors, after almost 24 years, the disaster has been forgotten and has disappeared from people’s minds. Bolton et al ( 2000) believed this is associated with the lack of understanding by people, especially non-peers19. Concealing the feelings of others and social isolation are some of the strategies that this group of survivors may use to face such reactions20.

Further, this study’s results showed that children whose other relatives are given custody and children who are adopted have the most problems. Since relatives are given priority in adopting a child, in most cases the economic status, emotional situation and social health of these families are not considered before the adoption. The new family members cannot provide a safe haven for these children because they are in crisis and are suffering from many financial, physical and mental problems. These children may experience violence, misbehaviour and secondary displacement in new families21. In such circumstances, these children feel that they are a burden, and experience a lack of opportunity in education, marriage, or suitable occupations22.

Regarding the fact that many of the participants of the present study have lost one or both of their parents in the earthquake, and given that some live with family members in an environment which is not physically, mentally and emotionally stable, most of these survivors face severe emotional issues as their future relationships are threatened. The results of McDermott’s (2005) study showed that distrust of friendly and effective communication with others on one hand and emotional gaps on the other can all cause behavioural and emotional conflicts, and can bring about short-term emotional and social communication issues23.

Failures in communication and constructive attachments cause these child survivors to experience failure in marriage and family life24. Therefore, comprehensive services such as counselling, health care and welfare services are necessary25. This study’s findings also showed that sometimes, parents do not succeed in their parental roles in supporting their children due to the distress they experience as a result of the earthquake. In such cases, these parents in turn can pose a risk to their own children. The results of Gertrud’s (2004) study showed that parents who were themselves severely impacted by disaster reported a reduced ability to assess their children’s reactions, and thereby were unable to provide optimal care2. Interestingly, these parents’ support strategies mirrored the early intervention recommendations put forward in the psychological first-aid guidelines which is a well-accepted and promising practice for helping children after disasters2. In the present study, mental nostalgia is one of the internal factors affecting the lives of those who survived the earthquake as children. These people retrieve memories when they face scenes that they consider as similar to events at the time of the earthquake. Hafstad et al (2004) believed nostalgia of past memories threatens their mental and emotional health during different stages of their lives2.

Believing in bad luck is another internal factor affecting the lives of those who were children earthquake survivors. Similar studies have shown that these beliefs can suppress a person’s effort in coping with stressful atmospheres, and can keep individuals from trying to improve their family and social lives22. Believing in bad luck can also deter this group of people from pursuing their previous dreams and wishes, as well as from social communication26. As the data in this study showed, one of the factors causing threats to the health of these children survivors is non-specific care. It should be noted that ignoring the needs of these survivors, considering the services superficially and providing those services temporarily may threaten their health in various aspects27.

In order to prevent more permanent damages to child survivors, we have to train individuals, aid workers and volunteers to bring up people from under the rubble correctly, identify specific needs of children survivors, and further care provisions28. Participants in this study have tried to resort to strategies such as the development of social communication, spirituality, and psychological strengthening to maintain stability and balance in their individual, family and social lives. In fact, other studies have emphasised religious beliefs as a facilitator of these survivors’ life processes29. Developing social relationships and friendships help improve the quality of survivors’ lives30.


According to the results of the present study, providing non-specific and temporary services is one of the factors affecting the life processes of children who have survived the Manjil earthquake. Training aid workers to consider the specific needs of children at the time of rescue operations could contribute to improving the health of this group in different aspects. Considering effective and comprehensive rehabilitation programs in Disaster Management policymaking can prevent persistent complications caused by earthquakes and other natural disasters. The results of this study also showed that some of the factors threatening the health of children earthquake survivors include homelessness, living in care centres, and living with risky parents. Planning and taking action to identify misbehaviour in this group of children, as well as raising public awareness (particularly for parents) on how to manage a disaster, are public health priorities. Providing public counselling services to children survivors as well as their parents will help to solve the potential psychological problems that threaten the well-being of children survivors.

Further, this study showed that children have tried to use different strategies in the process of their lives to maintain and promote their personal and social life. Strengthening religious beliefs and encouraging children survivors to participate in social groups (especially peer groups) will help them to feel understood, safe and alleviated of many negative feelings resulting from their experience of a natural disaster.

Appendix A: Interview Guide Questions

1- Can you tell me about your experiences after the earthquake?

2- What happened to you at the moment the earthquake happened?

3- Please talk about the harms happened to you.

4- How do you describe the life after the earthquake?

5- What has helped you to deal with the earthquake?

6- What does earthquake mean to you?

7- How was your life?

8- Would you please review a day of your life?

9- Which factors have helped you at the moment of the earthquake or after that?


Among the limitations of this study, restrictions in generalization of the results can be mentioned. As interview method was used for data collection, the veracity of the interviewees might have affected the results. Therefore, conducting more studies focusing on triangulation methods is suggested.

Corresponding Author

Dr Fatemeh Ghaffari. Email: ghafarifateme@yahoo.com

Data Availability

All relevant data are within the paper.

Competing Interests

The authors have declared that no competing interests exist.

http://currents.plos.org/disasters/article/the-life-process-of-children-who-survived-the-manjil-earthquake-a-decaying-or-renewing-process/feed/ 0
Towards Providing Solutions to the Air Quality Crisis in the Mexico City Metropolitan Area: Carbon Sequestration by Succulent Species in Green Roofs http://currents.plos.org/disasters/article/towards-providing-solutions-to-the-air-quality-crisis-in-the-mexico-city-metropolitan-area-carbon-sequestration-by-succulent-species-in-green-roofs/ http://currents.plos.org/disasters/article/towards-providing-solutions-to-the-air-quality-crisis-in-the-mexico-city-metropolitan-area-carbon-sequestration-by-succulent-species-in-green-roofs/#respond Fri, 31 Mar 2017 15:00:32 +0000 http://currents.plos.org/disasters/?post_type=article&p=29803 INTRODUCTION: In the first months of 2016, the Mexico City Metropolitan Area experienced the worst air pollution crisis in the last decade, prompting drastic short-term solutions by the Mexico City Government and neighboring States. In order to help further the search for long-term sustainable solutions, we felt obliged to immediately release the results of our research regarding the monitoring of carbon sequestration by green roofs. Large-scale naturation, such as the implementation of green roofs, provides a way to partially mitigate the increased carbon dioxide output in urban areas.

METHODS: Here, we quantified the carbon sequestration capabilities of two ornamental succulent plant species, Sedum dendroideum and Sedum rubrotinctum, which require low maintenance, and little or no irrigation. To obtain a detailed picture of these plants’ carbon sequestration capabilities, we measured carbon uptake on the Sedum plants by quantifying carbon dioxide exchange and fixation as organic acids, during the day and across the year, on a green roof located in Southern Mexico City.

RESULTS: The species displayed their typical CAM photosynthetic metabolism. Moreover, our quantification allowed us to conservatively estimate that a newly planted green roof of Sedum sequesters approximately 180,000,000 ppm of carbon dioxide per year in a green roof of 100 square meters in the short term.

DISCUSSION: The patterns of CAM and carbon dioxide sequestration were highly robust to the fluctuations of temperature and precipitation between seasons, and therefore we speculate that carbon sequestration would be comparable in any given year of a newly planted green roof. Older green roof would require regular trimming to mantain their carbon sink properties, but their carbon sequestration capabilities remain to be quantified. Nevertheless, we propose that Sedum green roofs can be part of the long-term solutions to mitigate the air pollution crisis in the Mexico City Metropolitan area, and other “megacities” with marked seasonal drought.



INTRODUCCIÓN: En los primeros meses del 2016, la Ciudad de México y su zona metropolitana conurbada ha experimentado una de las peores crisis de contaminación del aire de la última década, urgiendo a las autoridades a tomar medidas drásticas a corto plazo. A fin de proporcionar elementos para investigaciones de sostenibilidad a largo plazo quisimos dar a conocer nuestros resultados de monitoreo de captura de carbono por plantas de azoteas verdes. Estas azoteas son una forma de naturación a gran escala, que constituyen una alternativa para mitigar parcialmente el incremento de dióxido de carbono en las áreas urbanas.

MÉTODOS: Nosotros cuantificamos la capacidad de captura de carbono por dos especies de suculentas ornamentales, Sedum dendroideum y Sedum rubrotinctum, que requieren bajo mantenimiento y nula o poca irrigación. Para tener una idea detallada de la capacidad de captura, medimos el carbono tomado por las plantas durante el día por un año en plantas de Sedum localizadas en una azotea verde en el sur de la Ciudad de México.

RESULTADOS: Estas especies presentan el metabolismo fotosintético CAM. La cuantificación obtenida nos permite estimar que una azotea verde de Sedum de 100 m2 captura aproximadamente 1.8 x 108 ppm de CO2 por año.

DISCUSIÓN: Los patrones de CAM y de captura de CO2 fueron bastante robustos a las fluctuaciones de temperature y precipitación estacionales, por lo cual especulamos que la captura de carbono sería similar en cualquier otro año de una azotea verde de reciente plantación. Azoteas de mayor edad requerirían podas regulares para mantener sus propiedades de captura de carbono, sin embargo, no se sabe qué capacidad pueden tener para capturar carbono. No obstante, proponemos que las azoteas verdes plantadas con Sedum son una parte de la solución a largo plazo para mitigar las crisis de contaminación del aire en la zona metropolitana de la Ciudad de México, y de las megaurbes con una marcada estación de sequía.

Author Contributions

MCO & JRS designed the project; MCO performed experiments; MCO and UR analyzed results; MCO & JRS provided materials and reagents to perform experiments; MCO & UR wrote the manuscript. MCO: orcid.org/0000-0001-6618-4920; UR: orcid.org/0000-0001-5088-2679

Corresponding Authors

MCO (mague.collazo@ciencias.unam.mx) and UR (urosas@ib.unam.mx)


Plant physiology, Crassulacean Acid Metabolism, urban biology, megapolis, environmental crisis, plant urban physiology, Sedum dendroideum, Sedum rubrotinctum, air pollution, bioremediation

Competing Interests statement

The authors have declared that no competing interests exist.

Data Availability statement

Raw data available on doi:10.6084/m9.figshare.3395887


Deforestation, urbanization and the intensive use of fossil fuels are some of the main causes of the increase of carbon dioxide (CO2) since the Industrial Revolution in the nineteenth century, contributing to the current changes in climate 1. These CO2 levels have gone from 280 ppm to 380 ppm, and it is estimated that levels could reach 770 ppm by the end of the twenty-first century 2. This dramatic increase in CO2 levels, particularly in urban regions, has caused a change in the global carbon cycles 2,3. Thus, vegetation plays a crucial role in urban ecosystems, by sequestering carbon excess, storing carbon as biomass, and releasing oxygen and water vapor through evapotranspiration. In recent decades urban ecosystems have drastically expanded and increased in complexity due to mass changes in economic activities, the migration of people from rural environments to cities 4, and widespread use of motorized vehicles. Unplanned growth of urban areas sometimes produces “megacities” where space becomes a luxury, and vegetation is often sacrificed to give way to urban infrastructure.

This is the case of Mexico City and its Metropolitan Area, otherwise known as Zona Metropolitana del Valle de México (ZMVM). This is the fourth most populated metropolis of the world and the largest in Latin America 5,6, with more than 20 million inhabitants according to a 2010 census (INEGI, http://www.inegi.org.mx). According to the Mexican Fifth National Communication to the United Nations on Climate Change, CO2 emissions increased 23.6 % from 1990 to 2010, coming mainly from fossil fuels, land use changes, and lack of silviculture. In the first third of 2016, the ZMVM is experiencing the worst environmental crisis of air quality of the last decade, demonstrating that short-term solutions implemented since the 1980s when the air quality became extremely critical, have been insufficient. In 2009, the Dirección de Reforestación Urbana, Parques y Ciclovías de la Secretaría de Medio Ambiente del Distrito Federal (today Mexico City), pointed out that poor metropolitan growth planning, deficient management of green areas, lack of legislation on urban vegetation, and poor conservation efforts, have all contributed to the deterioration of the green urban landscape, and their associated environmental and social benefits. In addition to better management of public green spaces, it is necessary to urgently reassess the strategies towards sequestering carbon and other air pollutants. To archive this, the mega-polis inhabitants must play an active role in tackling the challenge.

Besides reducing gas emissions from fossil-fuel combustion, in other cities the promotion of urban vegetation has been adopted as a method to eliminate pollutants and purify the air 7,8. Because of the scarcity of space in large cities, a strategy to increase the vegetation cover is through the implementation of what are known as green roofs, eco-roofs, living roofs, or roof gardens . Green roofs are defined as any rooftop covered by layers of roof coating, plant substrate, and vegetation 9. Green roofs have been associated with multiple environmental, ecological, health, and aesthetic benefits 3,9,10,11,12,13,14,15. Despite the relevance of green roofs as a potential solution to improve air quality in “megacities,” there is scarce experimental data in planta on the capabilities of green roofs to uptake CO2. Therefore, quantifying the green roofs’ capacity to sequester and retain CO2, pollutants, and other greenhouse gases is imperative under the current environmental crisis scenario, particularly in the ZMVM.

Currently there is robust evidence supporting the use of plants to capture air pollutants. However, we still need in planta qualifications on the role of green roofs to capture pollutants 16. NOx, SO2, PM10 and O3 are among the pollutants that have been quantified as sequestered by green roof vegetation 7,17,18,19. Experimentally, CO2 green roof-uptake quantification seems to be rather challenging, and mostly estimated using proxies, such as biomass accumulation 13,20, models of carbon-oxygen balance 21,22, macro-estimations of vegetation cover in large cities 7,13,23 or methods more appropriate for urban forests, rather than herbaceous plants 17.

In their studies measuring carbon uptake on green roofs, Li et al (2010) measured the CO2 uptake in a green roof, placing the IRGA within grass-shrubs and comparing it to a location outside the shrub, obtaining an average of 12.9 mg C m-3 per day 8. Ondoño et al (2016) estimated carbon and nitrogen uptake by quantifying the element composition on a green roof of weedy and grassy plants, obtaining 36.52 g C m-2 at the end of 9 months 24. Regarding Sedum green roofs, Getter et al (2009) quantified carbon storage in green roofs of several Sedum species, by measuring accumulated dry matter in the shoot, root, and soil, obtaining 160 g C m-2 during a two year period 13. More recently, Whittinghill et al (2014) measured carbon content and dry weight, and found that Sedum green roofs captured 1940 g C m-2 in a 14-month period and 3910 g C m-2 in a 12-month period 20. Overall, previous studies support the use of green roofs, and particularly, of highly resilient Sedum species to improve air quality conditions, in particular in overcrowded urban areas. However, we still lack data on the carbon uptake by Sedum plants, measuring CO2 exchange.

During the first months of 2016, the ZMVM has experienced one of the worst air quality crises since the 1980s. This has forced the local governments of Mexico City and surrounding States to implement extreme measures, such as emergency “No Driving Today Program”. However, these are often short-term and unsustainable solutions, because they promote for instance, the acquisition of a second car by the inhabitants of the ZMVM. In light of the current air pollution environmental crisis, we felt compelled to release our monitoring of carbon uptake in a green roof in Southern Mexico City, populated with low-maintenance, low irrigation-requirement, crassulacean species of Sedum. Our results show the potential of the Sedum green roofs as a self-sustainable solution. We hope these results help policy makers of the ZMVM and other urban areas with similar environmental conditions to implement informed long-term solutions towards tackling air quality crises.


Green roof location. The green roof was located on the Southwestern area of Mexico City. This green roof was installed in 2009, comprising a total area of 42 square meters, at the Jardín Botánico, Instituto de Biología, Universidad Nacional Autónoma de México, Delegación Coyoacán, within the Reserva Ecológica del Pedregal de San Ángel (REPSA; http://www.repsa.unam.mx) (19° 13′ 45” N and 99° 19′ 37” W), 2330 m.a.s.l. (Figure 1A-C), and subtropical highland climate. The highest temperatures are from March to May (16-19° C), and the lowest are from December to February (12-14° C), with a mean annual temperature of 15.6°C. Precipitation follows a seasonal pattern: the rainy season occurs from June to October, while the dry season is from November to May, with an annual mean precipitation of 833 mm 25.


Figure 1. Green roof location within the Jardín Botánico of the Universidad Nacional Autónoma de México; A. Map of Mexico; B. Zona Metropolitana del Valle de México (ZMVM), according to Instituto Nacional de Estadística y Geografía (INEGI, México), x denotes the location of the Jardín Botánico; C. Google maps view of the Jardín Botánico, arrow (<) denotes the location of the green roof; D. Sedum green roof at the start of the experiment (1-month old plant-cuts); E. Sedum green roof towards the end of the experiment (11-month old plants). *Sedum dendroideum, +Sedum rubrotinctum.

Green roof construction. A completely flat roof was leveled up with concrete and bordered with 10 cm-tall edges (cost $7.5 USD per m-2). The roof was then insulated with a waterproof layer ($0.40 USD per m-2), an insulated asphalt layer ($15 USD per m-2), and a root-proof mesh ($16 USD per m-2). On top of these insulating and protecting materials, a 10 cm deep layer of substrate was uniformly placed, composed of tepojal (to facilitate drainage) and compost (for organic nutrients) in a ratio 1:1 ($11 USD per m-2). It is also estimated that initial plant cuttings cost $12.5 per m-2. All costs are expressed in US dollars as of November 2016. Green roofs require trimming of the excess of plants two years after the initial planting, and every 6 months thereafter. Every trimming removes approximately 30% of the old foliage.

Regarding the soil, we measured the soil apparent and actual densities, pH, electrical conductivity, percentage of moisture and texture 26,27. The bulk density was 1.045g cm-3, organic type, 13.12% humidity, pH 6.05, category moderately acidic, conductance 0.127 S m-1, 27.1 % clay, 12.4% lime, and 72.4% sand, which corresponds to a loam soil.

Biological subjects: We used the crassulacean species Sedum dendroideum Moc. Sessé ex DC. and S. rubrotinctum R.T. Clausen. S. dendroideum is a bushy plant of 50-100 cm in height, and up to 100 cm in diameter. In Mexico, its populations are distributed along the Trans-Mexican Volcanic Belt, and surrounding areas on the states of Hidalgo, Puebla, Estado de México, Oaxaca, and Querétaro 28. S. rubrotinctum has a herbaceous lifestyle, of 15-20 cm in height, heavily branched at the base, originally from the East, South, and Center of Mexico, but thought to be a cultivated hybrid of S. stahlii and S. pachyphylum28. The cuts were obtained from 10 year-old parental plants at another green roof, at the Jardín Botánico de la Universidad Nacional Autónoma de México. S. rubrotinctum’s cuts were 10-15 cm tall, and S. dendroideum‘s were 20-30 cm tall. Cuts were disinfected with Terramycin 1 g L-1 for 24h, and TRICON fungicide. Before planting, all cuts were treated at the base with Radix 10000 ppm to promote root formation. 72 cuts of each Sedum species were planted, every 30 cm in S. dendroideum and 10 cm in S. rubrotinctum, giving a density of 3.4 and 10 plants per square meter, respectively. Planting was done in November 2009, and measurements were recorded from November 2009 to November 2010. At end of the experiment the vegetative growth of both Sedum species gave a 100% cover of the studied green roof area.

CO2 uptake. Quantification of CO2 uptake was performed using an Infrared Gas Analyzer (IRGA) (Qubyt Systems, Canada), which is a device designed to study gas exchange on flat leaves. Thus to quantify the CO2 uptake in succulent plant specimens, we evaluated gas exchange using glass jars. We initially used a glass jar of 418 mL connected with plastic tubes to the IRGA, so we could introduce the entire plant into the jar and seal any openings. As plant size increased, we used a larger glass jar of 1200 mL during the months of April to November 2010. After the month of June, we were only able to measure the part of the plant that fitted into the 1200 mL jar. Measurements were taken monthly on three plants per species randomly chosen at 0:00, 0:40, 1:20; 6:00; 6:40; 7:20; 12:00; 12:40; 13:20; 18:00; 18:40 y 19:20h. The IRGA was calibrated 10 mins prior every measurement, and the quantification of CO2 uptake took 15 mins in each case. The baseline of CO2 in the air was subtracted to the readings, and the differential was considered as the net CO2 uptake by any given plant. In addition we quantified the concentration of organic acids, on the same plants quantified for CO2 uptake, by dissecting the tissues and keeping them chilled until they arrived to the lab. The quantification of organic acids was done according to the titration protocol 29,30, which was done as follow: 1) 2g of tissue were dissected from leaves in each sample, tissues were grinded with 20mL of 50% methanol; 2) the ground tissue was filtered, and 10mL more of distilled water were added; 3) titration was done using a potentiometer, by adding sodium hydroxide 0.01N under constant agitation until a pH of 9 was reached. The added base was used to calculate the acidity per gram of fresh tissue. Raw data available on doi:10.6084/m9.figshare.3395887

Statistical analysis. We used a two-way Analysis of Covariance Variance (ANCOVA) to identify significant differences in CO2 uptake (ppm) due to month and species, and time of the day (time) as a covariate, according to the following model: yppm = Ɓmonth + Ɓspecies + Ɓtime + ɛ. The analysis was done in JMP v12 (JMP, SAS, USA).


Succulent plants, particularly Sedum species, have been extensively used in green roofs because of their resistance to prolonged droughts, high temperatures, and strong winds 10,13,31. Here, we studied the Sedum species Sedum dendroideum Mociño & Sessé ex De Candolle, and S. rubrotinctum R. T. Clausen (Family Crassulaceae). These species have xerophytic morphology and physiology that allows them to cope with extremely dry environments. To assess the carbon capture of Sedum species in a green roof located in Southern Mexico City, we evaluated a plantation of 42 square meters, measuring CO2 exchange using an IRGA system, during the day, over a period of 12 months in 2009-2010. The highest carbon capture in both species was recorded during the night, with a peak at 6:00 (Figure 2). Over the periods of time that were measured, the total CO2 uptake was 2472 ppm (±40.35) from 0:00 to 7:20h, and 152 ppm (± 6.82) from 18:40 to 19:20h in S. dendroideum. Over the same intervals, S. rubrotinctum CO2 uptake was 1823 ppm (±40.70) from 0:00 to 7:20h, and 71 (±4.58) from 18:40 to 19:20h. In both species from 12:00 to 18:00 no CO2 uptake was recorded in most months, but the period of highest CO2 uptake was at 6:00h, with 1602 ppm (±11.05) and 1315 ppm (±14.49) on S. dendroideum and S. rubrotinctum correspondingly. The patterns of CO2 uptake on these Sedum species resemble the Crassulacean Acid Metabolism (CAM) carbon fixation patterns.

Figure 2

Figure 2. CO2 uptake in (A) Sedum dendroideum and (B) Sedum rubrotinctum, during the year, and across the day. Error bars: standard error, n = 3. Raw data available on doi:10.6084/m9.figshare.3395887

In order to validate our observations of CO2 uptake, during the experiment we simultaneously collected tissue from the measured plants to quantify the organic acid concentration using the titration method in the lab. The quantification of organic acids (Figure 3) confirmed our CO2 uptake observation regarding the CAM pattern, also indicating that S. dendroideum and S. rubrotinctum are strict CAM because they uptake most of the CO2, and accumulate organic acid during the night, but not during the day 32.

Figure 3

Figure 3. Quantification of organic acids in (A) Sedum dendroideum and (B) Sedum rubrotinctum, during the year, and across the day. Error bars: standard error, n = 3. Raw data available on doi:10.6084/m9.figshare.3395887

To further investigate the sources of variation on CO2 uptake during our experiment, we performed a 2-way ANCOVA, of carbon uptake (ppm) as a response of months, and species, and time of the day (time) as a covariate, according to the model yppm = Ɓmonth + Ɓspecies + Ɓtime + ɛ. We found no statistical differences between months (F(df 11) = 1.60, p=0.09), or species (F(df 1)= 1.74, p=0.18), but time of the day explained most of the variation in CO2 uptake (F(df 1) = 104.37, p<0.0001). The subtle but not significant differences between Sedum species might be due to their distinctive life forms. S. dendroideum grows erect and bushy, whereas S. rubrontinctum grows herbaceous crawling the floor (Figure 1D-E).

Although it was not statistically significant, February and March were the months with the highest CO2 uptake, coincidentally some of the months with the most chance of high air pollution in the ZMVM. This observation illustrates the physiological plasticity of Sedum species, in which carbon sequestration might be influenced by environmental factors, such as temperature and precipitation (Table 1).

Table 1. Measured net carbon capture of Sedum species in a green roof of Mexico City in a given day per month. Calculated from adding all the 15 mins measured intervals, and averaged from three plants, ± standard error. Tmax: maximum monthly temperature, Tmin: minimum monthly temperature; Tmean: mean monthly temperature; ppt: monthly precipitation. Raw data available on doi:10.6084/m9.figshare.3395887
Month S. dendroideum S. rubrotinctum Tmax Tmin Tmean ppt
November 2009 193 ± 35.76 141 ± 13.89 21.9 8.9 15.4 3.4
December 2009 150 ± 19.67 118 ± 13.71 20.0 7.0 13.0 26.1
January 2010 100 ± 14.89 71 ± 16.29 21.4 7.2 14.0 69.0
February 2010 275 ± 23.05 233 ± 18.09 24.6 9.1 17.2 5.1
March 2010 284 ± 21.13 240 ± 15.12 26.4 10.8 18.5 21.9
April 2010 108 ± 15.33 75 ± 7.03 28.7 13.3 20.7 27.8
May 2010 114 ± 15.24 87 ± 8.69 27.1 14.5 19.7 95.0
June 2010 115 ± 14.58 84 ± 7.05 23.5 14.0 17.9 255.7
August 2010 144 ± 16.09 105 ± 1.06 23.8 14.0 18.8 133.9
September 2010 145 ± 15.90 102 ± 9.69 23.3 13.6 17.7 99.7
October 2010 150 ± 19.05 120 ± 12.21 23.7 9.6 16.2 3.5
November 2010 192 ± 18.39 138 ± 14.43 24.7 7.3 15.2 3.0

The carbon sequestered by Sedum plants can be partitioned into the shoot or root parts of the plant. In order to estimate the biomass partition in the studied species, in a parallel experiment with three plants each, we measured the dry biomass of plants 325 days after planting on a green roof. The largest fraction of the dry biomass was recorded in the shoot (Figure 4). The root:shoot ratio was 0.41 in S. dendroideum, and 0.15 in S. rubrotinctum, which are similar ratios reported for other succulent plants from arid environments 34. This shows that most of the biomass is allocated towards the succulent shoot. However, it is known that roots can produce CO2 because of root respiration and decomposition. The dynamics of carbon sequestration and decomposition in the root and the soil remains to be quantified, however, it is argued that in the short term this ecosystem could be an important sink of carbon 13. Therefore, to mantain a green roof as a carbon sink, we trim about 30% of the old foliage two years after the initial planting, and every six months thereafter. We especulate that trimming practice maintains the plant community as a carbon sink, and reduces carbon emission by decomposition. Nevertheless, carbon sequestration dynamics of older green roofs remain to be quantified.

Figure 4

Figure 4. Dry biomass partition in S. dendroideum and S. rubrotinctum of plants (n=3) measured 325 days after planting on a green roof. The dry biomass of S. dendroideum was 70.3% in the shoot and 29.6% in the root (standard error ±3.1), whereas S. rubrotinctum was 86.8% in the shoot and 13.1% in the root (standard error ±1.1).

It is possible that carbon sequestered by Sedum species is allocated by the plant to the shoot or the root. The shoot adopts a perennial life-style, in which the leaves are succulent, permanently green, and only after several years, they senesce. This is why we believe that shoot senescence would have a small contribution to the re-emission of carbon to the environment. The biomass accumulated in the root, however, has the potential to contribute to the emission of carbon to the atmosphere by decomposition and degradation of the organic matter. Among desert perennial plants, succulents have the lowest root to shoot ratio, ranging from 0.04 to 0.15, which are dramatically smaller than desert shrubs that range from 3.2 to 7.3 33,34. In other words, the root biomass is disproportionately smaller. Moreover, an adaptation of succulent plant roots is to maintain roots, dehydrate them during the drought season and resurrect them during the new rainy season.

Our aim was to estimate the carbon sequestration capacity that green roofs can have. Therefore we used the data that we obtained from measuring individual plants during the day, and across the year to estimate the overall CO2 capture by Sedum plants. To do this we considered the average CO2 uptake from the three measured plants at each time, and month. Assuming that from 3 am to 6 am, Sedum plants have a steady carbon uptake similar to the 6am rate (see Figure 2), averaged the two species and multiplied our estimate of carbon uptake in 15 mins (period of IRGA quantification), by the total number of minutes in 3 hours, and obtained the carbon uptake in a given day per each month. This number was then multiplied by the number of days in any given month. Then we summed the months to estimate the total carbon uptake per plant during a year (July was assumed to have a mean value between June and August, and we averaged the two measurements of November months), giving 524,040 ppm of CO2 per plant per year, or 371.27 g C m-2. Furthermore, assuming that all the plants in the green roof have similar metabolic rates, we can estimate the amount of carbon that was sequestered in the studied green roof, by multiplying the carbon estimate per plant per year, by the total number of plants in our green roof. This gave a total of 75,461,760 ppm of CO2 in the 2009-2010 period of study in a green roof of 42 square meters, with 144 Sedum plants. In other words, a green roof of Sedum in Mexico City of 100 square meters, has the capacity to sequester 179,670,857 ppm of CO2. This estimate is highly conservative given that the number of plants per unit of area increases dramatically over the year as a result of vegetative growth (see Figure 1D-E). Remarkably, our estimation of a Sedum green roof of 100 square meters is equivalent to approximately 46% of the annual CO2 emission of a standard car (up to 7.5 km L-1) 35, driving every day 10km during 365 days, and estimated to produce 388,360,000 ppm CO2.


In this work we evaluated the capacity of two crassulacean Sedum species for capturing CO2, when grown in a green roof, during the day, and across the year. The months of February and March showed the maximum CO2 capture, 75% more than the summer month of June, a response that was possibly influenced by the temperature and precipitation patterns of the ZMVM. Consistent with the CAM, the highest CO2 uptake and accumulation of organic acids occurred early in the morning. We estimated that a green roof of Sedum of 100 square meters is able to sequester approximately 1.8 x 108 ppm of CO2 per year (or 180,000 g C m-2) in the ZMVM. Our quantifications are within the range of previous studies in Sedum species, which range from 160 g m-2 over two years 13, and 1940-3910 g C m-2 over a 12-14 month period 20. It is important to remark that Sedum plants were not irrigated, making these species an ideal candidate, with low irrigation and maintenance requirements, towards promoting naturation efforts in “megacities”.

In our current scenario, carbon concentration in the air is increasing because of the intensive use of fossil fuels. This phenomenon is tightly linked to the growth of human populations and the loss of vegetation in cities, making the design of sustainable urban areas a challenging task. An example of a metropolitan area with poor air quality, linked to decades of deficient urban planning, and short-term solutions to address the air quality issue, is the ZMVM. Particularly, in the early months of 2016, the ZMVM is facing the worst air pollution crisis since the 1980s, which is evidence of the lack of long-term and sustainable environmental policies. Therefore, there is an urgent need to increase the capabilities to fix carbon and other pollutants, in order to improve the air quality in metropolitan areas such as the ZMVM. The widespread installations of green roofs, together with citizen involvement are part of the potential solutions to increase the CO2 sequestration in cities. The green roof system was first developed in Germany in the 1960s 3, and later adopted in several countries such as USA, Canada, and European countries 13,36. Despite their widespread implementation in developed countries and their relevance to tackle air pollution, the understanding of carbon and pollutant dynamics in green roofs is still scarce.

Green roofs are subject to extreme environmental conditions because of the lack of rainfall during certain periods of time, high temperatures, and intense wind. It is because of this that only few species have been used on green roofs, Sedum being the main species worldwide 10,31. As a CAM species, Sedum has fewer capabilities to capture CO2 than C3 or C4 plants; yet, Sedum species are often preferred because of the xerophytic lifeform, and their resilience to extreme conditions.

Carbon sequestration measured in our study is only representative of the environmental conditions of the period 2009-2010. Although we did not measure carbon sequestration in multiple years, we speculate that the pattern should be similar in any given year provided that the variation in temperature and precipitation during 2009-2010 studied period is larger than the corresponding parameters over several years 2005-2015 (CONAGUA, Mexico, 2016; http://www.gob.mx/conagua) (Figure 5). It is remarkable that, despite the extent of variation in temperature and precipitation, the patterns of CAM and carbon sequestration are highly robust during the period of study (Figure 2-3). Therefore, we believe that the estimated parameter of carbon sequestration can be extrapolated to multiple years with similar environmental conditions in the ZMVM. Moreover, our results further support the use of Sedum in green roofs of the ZMVM, as a highly resilient species to temperature and drought.

Figure 5

Figure 5. Historical climate in the Zona Metropolitana del Valle de México (ZMVM) from 2005 to 2015. Data obtained from Servicio Meteorológico Nacional, Comisión Nacional del Agua, México (CONAGUA, 2016).

In our study we estimated the carbon uptake of Sedum species at interval periods during the day and at chosen days of each month, giving a total CO2 uptake quantification time of 122.4 hours. We acknowledge that our study did not monitor the entire 24h, or every single day in any given month; however, we believe that our sampling strategy and the clear recapitulation of the CAM physiology during the day and across the year, makes our data robust enough to make projections of carbon uptake during the 24h, of the 365 days of the year. We consider that such quantification and projection of the CO2 uptake can be highly valuable information for policy makers, especially those who might be currently planning how to address the air pollution crisis of the ZMVM.

As expected, both S. dendroideum and S. rubrotinctum displayed a strict CAM photosynthesis, because they captured CO2 during the night, and early hours of the morning, including the dry (November to May) and wet (June to October) seasons. Remarkably, the plants only received an initial irrigation, and were later only watered by natural rainfall, which emphasizes their sustainability in “megacities” such as the ZMVM with seasonal droughts (Table 1). Moreover, Sedum is a perennial species, reducing the requirement for replanting year after year. Other perennials have been used on green roofs, for instance trees of up to 3m, which can have higher rates of carbon capture. However, trees often require costly maintenance and frequent irrigation. Trees can sequester more carbon at the global level, but as Getter et al (2009) pointed, “due to building weight restrictions and cost, shallow substrate extensive green roofs are more common than deeper intensive roofs” 13. Therefore we strongly support the use of Sedum species, together with other C3 or C4 plants, towards the widespread implementation of green roofs.

In summary, the high survival, fast growth, low or null-maintenance, low or null irrigation requirement, and capacity to capture CO2 make Sedum an ideal species for green roofs, as an alternative strategy for promoting vegetation in megacities with marked seasonal drought. Particularly, Sedum species would be an ideal solution for the ZMVM, which is in urgent need of long-term solutions towards tackling the air quality crisis of the last decades.

http://currents.plos.org/disasters/article/towards-providing-solutions-to-the-air-quality-crisis-in-the-mexico-city-metropolitan-area-carbon-sequestration-by-succulent-species-in-green-roofs/feed/ 0