Background:
Twitter has emerged as a critical source of free and openly available information during emergency response operations, providing an unmatched level of on-the-ground situational awareness in real-time. Responders and survivors turn to Twitter to share information and resources within communities, conduct rumor control, and provide a “boots on the ground” understanding of the disaster. However, the ability to tune out background “noise” is essential to effectively utilizing Twitter to identify important and useful information during an emergency response.

Methods:
This article highlights a two-prong strategy in which the use of a Twitter list paired with subject specific Boolean searches provided increased situational awareness and early event detection during the United States Department of Health and Human Services (HHS), Office of the Assistant Secretary for Preparedness and Response (ASPR) response to Superstorm Sandy in 2012. To maximize the amount of relevant information that was retrieved, the Twitter list and Boolean searches were dynamic and responsive to real-time developments, evolving health threats, and the informational needs of decision-makers.

Conclusion:
The use of a Twitter list combined with Boolean searches led to enhanced situational awareness throughout the HHS response. The incorporation of a dynamic search strategy over the course of the HHS Sandy response, allowed for the ability to account for over-tweeted information, changes in event related conversation, and decreases in the return of relevant information.

Introduction: Data from an electronic medical record (EMR) system can provide valuable insight regarding health consequences in the aftermath of a disaster. In January of 2010, the U.S. Department of Health and Human Services (HHS) deployed medical personnel to Haiti in response to a crippling earthquake. An EMR system was used to record patient encounters in real-time and to provide data for decision support during response activities.

Problem: During the Haiti response, HHS monitored the EMR system by recoding diagnoses into seven broad categories. At the conclusion of the response, it was evident that a new diagnosis categorization process was needed to provide a better description of the patient encounters that were seen in the field. After examining the EMRs, researchers determined nearly half of the medical records were missing diagnosis data. The objective of this study was to develop and test a new method of categorization for patient encounters to provide more detailed data for decision making.

Methods: A single researcher verified or assigned a new diagnosis for 8,787 EMRs created during the Haiti response. This created a new variable, the Operational Code, which was based on available diagnosis data and chief complaint. Retrospectively, diagnoses recorded in the field and Operational Codes were categorized into eighteen categories based on the ICD-9-CM diagnostic system.

Results: Creating an Operational Code variable led to a more robust data set and a clearer depiction emerged of the clinical presentations seen at six HHS clinics set up in the aftermath of Haiti’s earthquake. The number of records with an associated ICD-9 code increased 106% from 4,261 to 8,787. The most frequent Operational Code categories during the response were: General Symptoms, Signs, and Ill-Defined Conditions (34.2%), Injury and Poisoning (18.9%), Other (14.7%), Respiratory (4.8%), and Musculoskeletal and Connective Tissue (4.8%).

Conclusion: The Operational Code methodology provided more detailed data about patient encounters. This methodology could be used in future deployments to improve situational awareness and decision-making capabilities during emergency response operations.

Background: Hurricane Isaac made landfall in southeastern Louisiana in late August 2012, resulting in extensive storm surge and inland flooding. As the lead federal agency responsible for medical and public health response and recovery coordination, the Department of Health and Human Services (HHS) must have situational awareness to prepare for and address state and local requests for assistance following hurricanes. Both traditional and non-traditional data have been used to improve situational awareness in fields like disease surveillance and seismology. This study investigated whether non-traditional data (i.e., tweets and news reports) fill a void in traditional data reporting during hurricane response, as well as whether non-traditional data improve the timeliness for reporting identified HHS Essential Elements of Information (EEI).

Methods: HHS EEIs provided the information collection guidance, and when the information indicated there was a potential public health threat, an event was identified and categorized within the larger scope of overall Hurricane Issac situational awareness. Tweets, news reports, press releases, and federal situation reports during Hurricane Isaac response were analyzed for information about EEIs. Data that pertained to the same EEI were linked together and given a unique event identification number to enable more detailed analysis of source content. Reports of sixteen unique events were examined for types of data sources reporting on the event and timeliness of the reports.

Results: Of these sixteen unique events identified, six were reported by only a single data source, four were reported by two data sources, four were reported by three data sources, and two were reported by four or more data sources. For five of the events where news tweets were one of multiple sources of information about an event, the tweet occurred prior to the news report, press release, local government\emergency management tweet, and federal situation report. In all circumstances where citizens were reporting along with other sources, the citizen tweet was the earliest notification of the event.

Conclusion: Critical information is being shared by citizens, news organizations, and local government representatives. To have situational awareness for providing timely, life-saving public health and medical response following a hurricane, this study shows that non-traditional data sources should augment traditional data sources and can fill some of the gaps in traditional reporting. During a hurricane response where early event detection can save lives and reduce morbidity, tweets can provide a source of information for early warning. In times of limited budgets, investing technical and personnel resources to efficiently and effectively gather, curate, and analyze non-traditional data for improved situational awareness can yield a high return on investment.