INTRODUCTION: There are few documented examples of community networks that have used unstructured information to support surveillance during a health emergency. In January 2015, the Ebola Response Consortium rapidly implemented community event-based surveillance for Ebola virus disease at a national scale in Sierra Leone.
METHODS: Community event based surveillance uses community health monitors in each community to provide an early warning system of events that are suggestive of Ebola virus disease transmission. The Ebola Response Consortium, a consortium of 15 nongovernmental organizations, applied a standardized procedure to implement community event-based surveillance across nine of the 14 districts. To evaluate system performance during the first six months of operation (March to August 2015), we conducted a process evaluation. We analyzed the production of alerts, conducted interviews with surveillance stakeholders and performed rapid evaluations of community health monitors to assess their knowledge and reported challenges.
RESULTS: The training and procurement of supplies was expected to begin in January 2015 and attain full scale by March 2015. We found several logistical challenges that delayed full implementation until June 2015 when the epidemic was past its peak. Community health monitors reported 9,131 alerts during this period. On average, 82% of community health monitors reported to their supervisor at least once per week. Most alerts (87%) reported by community health monitors were deaths unrelated to Ebola. During the rapid evaluations, the mean recall by community health monitors was three of the six trigger events. Implementation of the national system achieved scale, but three months later than anticipated.
DISCUSSION: Community event based surveillance generated consistent surveillance information during periods of no- to low-levels of transmission across districts. We interpret this to mean that community health monitors are an effective tool for generating useful, unstructured information at the village level. However, to maximize validity, the triggers require more training, may be too many in number, and need increased relevance to the context of the tail end of the epidemic.