 Okay, good morning everybody my name is Joe Jaspers. I'm the former surveillance coordinator and like Emily said I'll be presenting with Ruan who's the current epidemiology technical advisor for IRC based in New York And we're gonna be presenting on the evaluation of a national community based surveillance system for Ebola virus disease in Sierra Leone We've divided up the presentation. So as a former field staff I'll be speaking more about the implementation and then Ruan will take over to talking more detail about some of the methodology And the key findings from the evaluation So I'm sure as with Camillini many of you probably had a degree of personal involvement with the outbreak But for those who are unaware to put this in sort of context There were three primary sources for identifying Ebola cases in mid to late 2014 Cases could present to health facilities in which case they'd be screened There was a national hotline set up in Sierra Leone where in community members could report suspected cases and deaths occurring in their villages And then of course for known chains of transmission there was contact tracing The challenge that we encountered was that many cases and many chains of transmission were not known So this is the epicurve for the first 50 weeks of the outbreak and you can see that in beginning in September 2014 There begins a gradual increase or steady increase in the number of cases that were dead when they were detected first This is obviously important from an individual standpoint in that delays in seeking treatment dramatically reduce the chances for survival But also from a public health standpoint in that these cases remained in the communities And were able to continue transmitting the virus onto others Obviously once they died The bodies of of Ebola victims are highly viremic and so that also served as an additional means of further furthering transmission So as in response to these challenges that we designed a program called community event-based surveillance With the objective of limiting geographic spread of the outbreak by improving the sensitivity and the timeliness Of case detection. So this system was essentially sacrificing specificity and Gaining hopefully the speed in detecting cases quicker We did that through training a network of community health monitors at the village level to detect and report on a list of a standardized list of events That were associated with the Ebola transmission. They're on the right here. I won't go into them in detail But just to say that we Thought that training them on an event-based system as opposed to a case-based system would be quicker And would be easier for community health monitors to apply and interpret as opposed to training everybody in the clinical case definition So the design was really exhaustive active surveillance at the village level This is a structure of sort of how the Alerts flow through the system. So beginning on the left, for example, say there was a trigger event that occurred in the community One of the trigger events was a traveler arrives in the village and becomes sick or dies soon after arriving That would hopefully be detected by a community health monitor Who would relay the alert up to a supervisor? The supervisor would then work with the community health officer Which is a ministry of health staff to conduct a preliminary investigation of the alert to determine if it fit The criteria for a suspected case If so, the alert would be forwarded to the district of bowl response center, which was a government structure set up And a formal case investigation team would be dispatched To determine if it was a confirmed case to isolate the case and get it lab testing So the yellow boxes are the components that were added by community event-based surveillance and the blue boxes were Existing ministry of health structures As you can see, our goal was not necessarily to create a standalone surveillance program But simply to improve case finding and integrate our program within the bowl response structures that were already set up So the timeline for implementation In november 2014 we conducted a pilot in around a hundred villages in bow district in the center of the country December 2014 we developed a standard operating procedures with the ministry of health They asked us to scale the program nationally and then from january to april It was scaled up nationally by the bowl response consortium Which was an alliance of ngo's that was formed and led by the irc to enable Essentially for ngo's to implement a bowl response programming in a standardized way at national scale To make sure we were all doing the same thing And three districts were scaled up by ifrc the red cross Using very similar so p's so during that time we trained over 7 000 community health monitors and 150 supervisors Across the nine districts and many of these were pre-existing community health Workers so we're once going to take over and talk a bit more about the evaluation Hi everyone, um, i'll now talk about the epidemiological evaluation So for the methods first we anticipated that a community surveillance program would be more sensitive and and much less specific Therefore, we wanted to investigate whether The system was capable and effective in detecting suspect probable and confirmed cases Um, so this entailed a basic descriptive analysis of alerts across space and time Across nine districts from the start of operations, which was in february 2015 to the end of the study period Which was september 30th We looked at sensitivity and positive predictive value sensitivity of confirm case detection which basically looked at the proportion of confirmed cases detected through Community event based surveillance among all of the confirmed cases detected through the overall surveillance system positive predictive value looked at the proportion of suspect probable and confirmed cases detected Through the alert system among all alerts detected through uh community event based surveillance Finally, uh, given its potential for early warning. We wanted to know whether The community event based surveillance filled a surveillance gap by detecting unknown chains of transmission in a timely manner Um, so we conducted field investigations in the only district still reporting cases at the time, which was cambia To look at timeliness And the source of the surveillance detection during the first six weeks of operation in cambia Over to the results. Um, what you're looking at is a graph of the number of alerts over time Since uh community event based surveillance began its operations on february 27th until the end of the study period in september 30th 2015 the system generated over 12 000 alerts By the end of april eight of the nine districts were online and by the middle of june all nine districts were reporting an average of 79 alerts per day When we looked at the breakdown of alerts, we see that only 5 of alerts total were listed as one of the six pre-defined trigger events The other category that joe mentioned earlier Uh was made up made up 95 of the alerts and 87 of the other category alerts Related to deaths happening in the community that may or may not have been related to uh, ebola However, the death reporting rates by district, which are the white bars you see there were considerably lower than the anticipated crew death rate of 4.66 deaths per 100 000 per day, which is shown by the yellow line So it was not exactly uh mortality surveillance, but it showed a continual flow of death reporting Bear with me. I'm just going to skip over these slides and come back Um, so for sensitivity and positive predictive value, um, the sensitivity was 30 percent The system detected 16 of the 53 confirmed cases recorded in the ministry database in the nine districts Half of those 16 cases were alive when found The positive predictive value was 2.4 percent Which meant 287 alerts produced uh suspect probable or confirmed cases that warranted further investigation In addition though unintended measles clusters were signaled um through the system Um as as I said before cambia was the only district reporting cases during the field investigation period The investigations we did uh aimed to demonstrate how health facility screening community event-based surveillance And contact tracing were involved in the detection Of cases in a timely manner here. We see a community health monitor who's part of the SEBS program Who is also a community health worker recounting the detection of the case in cambia Similarly, here's a surveillance supervisor who's seeking preliminary information on on a case that's just been phoned in And here this is a typical screening post in a health facility that was also involved in detection of one of the cases This chart demonstrates the timeliness of case detection of the six cases from unknown chains of transmission in cambia That was the culmination of these field investigations During the time period six of the 13 confirmed cases were shown to have not appeared on any contact list or been under contact tracing Community event-based surveillance detected four of six of these cases fairly quickly Uh a range of one to three days from data onset of symptoms And the other two cases which were not detected by community of that based surveillance Took five and seven days each for uh detection However, these data are too small to draw a meaningful conclusion on timeliness As with any study, we have limitations first since transmission had declined substantially At the time of evaluation We was not possible to determine how community of that based surveillance performed during high transmission time Second field investigations were carried out on a small subset of the total 53 confirmed cases So it was difficult to say um what the relative contributions of community surveillance and contact tracing were And third as evidenced by the low use of triggers for illness Social and anthropological questions around illness detection are very important and we really focused on the epidemiological questions So to conclude um rumor-based surveillance systems are can be implemented nationally during a humanitarian emergency This was a productive system that produced a lot of alerts And it also helped to confirm that zero transmission was happening in quieter areas where confirmed cases were being detected much less Cebs did generate a lot of false alerts It was reasonably sensitive in that it detected a third of cases And many of the other cases were already under contact tracing It does fill a surveillance gap. It rapidly detected a four out of six cases from unknown chains of transmission So it suggests that it's timely But there was a poor use of triggers as uh described Looking forward We take two important lessons from this experience community health worker systems or other community systems are integrated Already in communities and are key for the rapid rollout of surveillance systems This is important as there's discussions and actions to build longer term detection capacity through community networks As a building block of integrated disease surveillance and response systems in west africa um also less is more Deaths and simple validated events and case definitions are very important to instill from the onset That our study shows that uh reinforcement of training after a rapid implementation process is very important Thank you very much. If you want to learn more, please see these papers. It's one of my favorite pictures from the outbreak And we gratefully acknowledge the cdc ebola response consortium partners serilio and ministry of health and funding from our donors Thank you