Deficiencies in transportation and communication, low frontline staff morale, and mistrust among communities, among other operational challenges, greatly limited Ebola case investigation in Sierra Leone. Recommendations for future outbreaks: (1) timely compensation for frontline staff, (2) context-appropriate transportation and communication resources, (3) systematic data collection, storage, and retrieval systems, (4) sound linkages between frontline staff and communities, (5) daily meetings between frontline staff and epidemiologists, (6) clear and appropriate operational chain of command, and (7) political and funding support to operational agencies.

ObjectiveTo assess whether the change in death swabbing policy in SierraLeon has begun to affect community death reporting, we analyzedtrends in death reporting before and after the policy change.IntroductionStemming from the 2014-6 Ebola virus disease (EVD) outbreak,community event based surveillance (CEBS) was implemented inSierra Leone using community health workers to generate alerts fortrigger events suggestive of EVD transmission. Through September30, 2015 (last month of active EVD transmission), the majority (86%)of alerts reflected community deaths; this was beneficial as Ebola-related deaths were detected with delay during the epidemic’s peak.The Government had implemented a policy of mandatory swabbingand testing of all dead bodies. The policy changed on June 30, 2016wherein only swabbing of deaths deemed to be high-risk for EVD isrequired. To assess whether this policy change has begun to affectcommunity death reporting, we analyzed trends in death reportingbefore and after the policy change.MethodsThis analysis was conducted using data from nine districts duringperiod 1 (January-June 2016) and period 2 (July 2016). Weeklychanges in the reporting of death alerts during the two periods wereassessed. An interrupted time series analysis (ITS) with a segmentedlinear regression was also used to assess the immediate impact of thepolicy change.ResultsDuring period 1, monthly changes in death alerts across districtswere variable (-8% to 16%). Comparison of the weekly averagebetween periods 1 and 2 showed a 33% reduction in death alerts.During period 1 (before the policy change), there was an overallsignificant increase of 3.2 death alerts per week (p=0.00) andno immediate impact or changes in the trend afterwards. At thedistrict level, on average 354 death alerts were generated weekly inJune, compared to 237 in July (33% reduction); Moyamba districtexperienced the largest drop in death alerts from 46 to 16 (65%).ConclusionsCommunity death reporting provides early warning of EVDtransmission by rapidly capturing death alerts where vital registrationis not fully functional. Although we have one month of data post-policy change, this preliminary analysis suggests that the changein swabbing policy may have halted an observed increase in deathreporting. Further community mobilization efforts and training arewarranted to prevent a drop in death reporting.

Port Loko district in Sierra Leone had over 1,400 confirmed Ebola cases since the start of the outbreak. Stronger surveillance systems were critical for the early detection of potential EVD cases, thus containing further spread of the epidemic. Community structures such as clinics, village/section chiefs, community Ebola responders, religious institutions and community social mobilization teams were critical to developing robust integrated surveillance systems that could report significant EVD events to the District Ebola Response Center. This ensured that all significant events were investigated. Continuous engagement of community and section structures proved critical in stopping the spread of EVD in Port Loko

and ≥60) and district. Results: Of the 33,957 CEBS alerts that were generated during the 16-month period, 72% were death alerts and 28% were sick alerts. Of these alerts, 48% were male and 52% female; percentage of death alerts and sick alerts did not vary by sex. Alerts by ages group showed comparable trends between male and female alerts, with the largest percentage of alerts being from deaths in people over 60 (22% of male alerts and 20% of female alerts). Out of nine districts, a majority of districts generated a greater percentage of female alerts compared with male alerts within the 10-19 and 20-29 age groups, while seven districts had a higher percentage of male alerts compared with female alerts among people over 60 years of age. Conclusion: Understanding community-based surveillance data is crucial for ensuring proper disease surveillance, targeting appropriate responses during outbreaks, and to improve community surveillance. The consistency across districts in sex-and agespecific demographics among both live and death alerts generated by CEBS suggests that the reporting of trigger events was not influenced by sex or age.