 It was a three-year project, it just ended at the end of 2015 and the end line was just completed last month as well. So this is the learning-to-date as of the end of 2014, which was the end of the second year of a three-year program and we're eagerly awaiting the end line results, I'm sure the Tufts is crunching data as we speak, but we thought the findings were interesting enough to present as they are to date. So just two disclaimers, one is that when about the end line this story could change once we get those results obviously and the second one is that I'm not an agriculturalist, I'm a nutritionist and you won't get, there is a strong agricultural component in this program but you won't get a lot of rich detail unfortunately in this presentation. So I thought I'd start with two key messages at the risk of stealing my own thunder, just putting them up there, because I think they complement some of the agri-diet conclusions which is that wash in fact, our findings from the midline is that wash in fact may be at least as important as agriculture to improve nutrition in this context and it just underscores how important it is to understand your own context where you're working or living and the second headline is that size matters and specifically we've found that village size matters and I'll explain a bit more about that but for us that underscores that despite trying as we're all striving to develop and deliver multi-sectoral programs with different interventions that we can't ignore the coverage that we're achieving or saturating or converging all these different multi-sectoral interventions and making sure that we are in order to have an impact. Okay, so this is just background on our program area, I might just highlight that it's largely agro-pastoralist, the main harvest October, November, it's millet, sorghum, ground nuts, it's French speaking and Arabic speaking, malnutrition is high, gender inequality is widespread, polygamy is practiced not that the two go hand in hand but it definitely affects the social dynamic, migration levels are high, often up to Libya, the health system is very weak and there's a history of displacement due to conflict and drought and that's it, it's just on the border with Sudan. So how our CRAM model which is what we described it as first because it was in fact a bit of a headquarters conceptual model that then quickly was taken up by our concerned Chad team and fully developed but it was based on our experience in many countries realizing that trying to reach these rather elusive development goals in places that are constantly experiencing shocks and stresses such as droughts, floods, earthquakes, particularly in Chad it was drought and occasionally floods just wasn't working. So from this conceptual model which was in kind of technical advisors heads and a little bit on paper it was fully fleshed out in Chad, turned into an integrated multi-sectoral design with this contingency plan so a lot of trying to detect when a particular shock was going to come and be prepared to respond to these localized emergencies and I've put this in blue because I think and not just because I think Irish aid is in the room but because I think it's a message that has to get out there that it was really the opportunity of the multi-annual Irish aid funding we get institutional funding from Irish aid for a five-year time span and this is just the kind of opportunity that you need to develop these kinds of programs especially in difficult contexts such as Chad. So it was a three-year pilot program I put pilot in quotes there only because we've already moved on with different funding for a similar program even though the end line results are in but it was seen as an opportunity for learning and that's why the partnership with the Feinstein International Centre to develop a very strong impact evaluation and operational research was at the core of the program. This is the conceptual model so we aim to improve child nutrition food security and resilience to future shocks and stresses just around the theme of measurement we have very clear metrics around child nutrition and some around food security we're still struggling a bit to figure out how we measure resilience to future shocks and stresses but it's there in principle and we're working on that. So essentially a conceptual model was an integrated set of interventions agriculture wash health system strengthening social and behavior change and gender equality all being appropriate and that's to be delivered in all years and then an early warning system as I said to detect when for example a poor harvest is coming that would try sorry trigger timely response and that would be delivered in a bad year. So that was kind of the theory in Chad these are the interventions that more detailed interventions roughly that were delivered and this was focused on predicting poor harvest so in terms of agriculture it was really lead farmers to per village who were trying to promote particularly conservation agriculture and some climate smart agricultural practices vegetable gardens for women and we have behavior change communication to improve mothers understanding and also build their confidence around caring for their children and seeking care at the health centers very importantly we drilled one borehole or ensured that there was one borehole per village and also supported latrine building and then we provided support to the health system and some other aspects around animal health workers. We did try to promote women's or in the in theory we were going to really make an effort in promoting women's decision-making. I almost put as my third message that promoting women's or gender equality is very hard but it seemed a bit defeatist but we haven't really shifted much in the equality aspect but we're working on it. Okay so here's the impact evaluation design so 35 villages were randomly received uh sorry randomly selected to receive the whole integrated package and 35 villages randomly selected to not receive basically they were the control villages and we had a baseline survey midline survey and line survey and the same households were followed up at the baseline the midline and the end line so there's a lot of tracking going on there a similar design as the agri-diet and that respect qualitative research as well some longitudinal monthly data collection which I'll show you in a minute. So in terms of program impacts there are kind of two pieces of findings the first is around did this integrated model that we designed to work and as compared for those who received it as compared to those in the control group we don't have multiple arms so we can't say for example that it was the addition of WASH the addition of community health workers to say what we just have it worked it didn't work so at midline what we found is a direct result of the program that levels of acute malnutrition among children have reduced albeit in smaller villages not in larger villages and we don't fully understand yet why households are better able to cope with the hunger gap and this was as measured by the coping strategy index but and I'll get into this in the next slide just it was only captured from the monthly data so this idea of seasonality not just in how it affects people but how your monitoring systems you have to include some kind of seasonality when you're dealing when you're trying to see what works in these contexts is is also emerging and more parents brought their sick children for treatment at a health center hospital or mobile clinic in our program villages so we had a community health component in the intervention or program villages and not in the control villages and that seems to have made a difference so just in terms of measure the measures of food security we used sorry there are a few more findings but just while we're on this topic of coping with the hunger gap we used a community or coping strategy index which I won't go into but it's basically 10 questions of how people of how you're coping with food shortages total months of food insecurity and household diet diversity so I'm glad the the theme of seasonality has already been introduced because this is one of the key findings and again it's it's really in terms it's partially in terms of being able to measure impact of a program like this so this is the coping strategy index so basically the higher your score yet the worse off you are so fewer coping strategies down below here and the worse off there the top line are our control villages so this is the longitudinal longitudinal data that was collected from 60 households every month to say did you do any of these 10 behaviors it starts from kind of moderate or mild food coping strategies to really extreme I think the last one is going without food for an entire day so the control villages as you can see overall had higher they were doing worse they were employing more coping strategies that our program villages were lower and kind of able to cope and the interesting thing there then is just that the harvest which happens around October November they kind of converge again so in our in our kind of annual surveys or our baseline midline online we wouldn't have picked that up at all if it wasn't for this sub-sample so just back to the last I think for findings headline findings more households this is around wash so more households reported using an improved source for drinking water which was a borehole which isn't surprising because as I said we established one in each village but just to note that increase was greater in smaller villages again I'll come back to that more households knew the critical times to wash hands and there were further signs of improved hygiene were seen as well in terms of the portion of households who had a hand washing station or could actually display proper hand washing technique was higher in the control villages but only in the smaller control villages are sorry in the intervention villages but only in the smaller intervention villages so basically we're seeing an impact on hygiene in our program villages and especially in the smaller villages and finally there were no improvement seen in female decision making the household which is a shame but to be perfectly honest we didn't actually we weren't actually able to implement a lot of gender activities and that has been addressed we have a new gender strategy going forward so the final thing is less about impact of our program but looking at the wider data set so looking at the the program and the control villages and the households within them together as as a whole set he keep all nutrition and children under five was found to be significantly linked to the frequency with which households reported washing their water containers and the presence of lives it was also linked to the presence of livestock in the village so if livestock in particular in that village if there was a greater concentration of livestock and if they were also reported to be using the same you know in a borehole you know using the same water point rather than a separate waterhole for animals then the children in those villages were more likely to be acutely malnourished and probably the most important thing is that no such link was found between the keep monetization and any of the three food security measures that I mentioned before so it's really just and it could be specific to this context and I think as Ed said there's it's complicated there's probably a lot more going on but for us this is very very interesting it's just we don't have there's a paper being produced it's been produced by Tufts we're just waiting for it to be published so that we can share it more widely and then again the end line findings will hopefully either strengthen or potentially refute that that mid line finding but basically watch the space there's more coming um do I have time yeah okay okay so this is uh so the last I was meant to about an early warning system I think in fairness what we have so far something that Tufts has developed it's an early warning um modeling so it's it's able to predict this is I just took this picture off the website because we I don't think we I don't know if we have a map this is I think North America somewhere but it's from the same website that has the data sets that Tufts has used to develop this model so they basically use rainfall data um from satellites from satellite source showing the total amount and the distribution of rainfall and it's I think it's down to a 10 kilometer square kilometer area they compared that with historical rainfall data or they they compared that so they took that data set it's a massive thing I think NASA and the Japanese um outfit organizes it Paul knows more um and they uh they took the the last 10 years of that data and they compared it against what the Ministry of Agriculture and had as the last 10 years of millet and sorghum harvest and it was very strongly correlated so and they also compared it against they they took a sample of people and asked older people and asked them to remember the last 10 years and rate the last 10 years as very bad bad good very good um and that also correlated quite well with the harvest data and and the rainfall data so it seems to be a good a good fit a good um proxy the rainfall data for food security and the harvest um it appears to be quite predictive uh probably more work needs to be done I mean it's similar to fuse net but it's really down to a very localized level and strictly on rainfall and it gives you a better it gives it has given us for the past two years roughly two month advance um prediction of what the harvest is going to be like which did allow us I think in 2013 to actually scale up a seed distribution um a bit in advance so more works to be done to build the whole early warning system so as I said there's a lot going on at the regional level the national level that we need to look at um we need to look at the whole picture and kind of figure out how this fits in and make sure we're not contradicting or going against so quick implications for programs so the multi-sectoral program seems to be working to prevent acute malnutrition but as I said we need to ensure sufficient coverage and this is this small village versus large village so I think it's probably obvious that if you dig a borehole in a village with 500 people and one with 150 people that the borehole in the village with 150 people should have it will probably have greater impact in terms of they'll have more water they'll have more regular access to it they don't have to wait in line potentially um and the same would go for interactions with community health workers if you have I don't know I don't have the figures in front of me but if you have the same number chasing 150 villages versus 500 there's probably a lot more that can be done at the same time there could be something else intrinsically about the social network in a small village maybe it's more coherent maybe there's a tipping point at which people the whole village just says right we're gonna start breastfeeding or we're gonna start washing our hands or so we don't really fully understand that um and just this hygiene a lot along the water chain I think I didn't mention that we did a lot of testing along the water chain to confirm some of this and hopefully we'll have as I said more on this and the livestock link with acute malnutrition so these are just more general ones strong impact evaluation very helpful and useful the importance of the longitudinal data alongside that and the qualitative data to explain the why and the how so there's a strong qualitative element that um has actually just come through in the last day or so from Tufts so we should have a lot to say hopefully in about two to three months and the last one sorry it's getting knocked off there but it's just that this partnership between NGOs and academic institution like Tufts can really work and we can really advance learning and leaps and bounds but they do require work they do require clear budgets contracts expectations regular concentration and flexibility and we used a small steering committee internally that that seemed to work last slide is just end line results as I said to be analyzed and shared and and the continuation so cram has finished and the randomized control trial has finished a randomized control design has finished but we've already that's overlapped with new funding from DFID under the building resilience and adaptation to climate extremes in Chad and Sudan where we're basically going to expand this cram approach it's a three-year programming program will continue with Tufts and the and we'll also be working with the world agroforestry center incorporate more try to adapt the model to pastures particularly working with the pastures focused NGO in Sudan and more to come