 I thank you very much for having me here to present on these evaluations on behalf of everybody who took part in them. So I'm actually going to talk about two different programmes in two very different settings, so there's a lot to get through but I'm happy to clarify at the end. We're talking about the integration of non-communicable disease care in the Democratic Republic of Congo and in Swaziland, the MSF experience. So as I said, these are two very different settings. roedd ymateb o'r defnyddio gyda'r coff diabyt i gyfrifiad ar y cwmgylch yma, yn Nôrth Aquif, yn y Ddemolwgr Republicau Congol, ac ymddych â'r gyfrifiad yma yn y ddeifud wedi eu gyfrifiad ymddych, yn y ddodio gyfrifiad, i ddodio gyfrifiad, yn y Swasiland. Rydych chi'n unig o gyfrifiad mewn cyfrifiad ym unig o gyfrifiad, ac mae'n yn qudweithio i ddigon i gyfrifiadau gyfrifiad. So MSF decided to evaluate both programmes in collaboration with the London School of Hygiene. And the main objectives of these evaluations were to try to understand the programme effectiveness, to look at some of the factors influencing treatment outcomes. And in the DRC this involved the effect of conflict and heightened insecurity on treatment outcomes. And we also looked at programme costs for each programme from the provider perspective. So just to underline the differences in the two settings, on the left you see a map of the DRC and in the circle is the province of North Kivu, so it's an unstable chronic conflict setting with repeated displacement of the population. And on the right you'll see the map of Swaziland, which you've seen already today. And this study was based in Matsafa and it's a, as we said earlier, it's a high HIV prevalence, high TB prevalence resource constrained setting. Also the programmes were quite different in each place. In the DRC it was nurse-led outpatient care in a hospital setting. Nurses provided most of the care with doctors reviewing at enrolment and on a six-monthly basis. And nutrition support, psychosocial support were integrated into the programme. Whereas in Swaziland this was broader NCD care delivered within a primary care level outpatient and HIV service. Nurses reviewed the stable patients, doctors reviewed those that were unstable and patients on enrolment. And in contrast there was no specific adherence or psychosocial support integrated into the programme. So to describe the methods from Weso, there was a retrospective analysis of the routine cohort data done on all enrolment patients from the beginning of the service in January 2014 until February 2017. And we also analysed the relationship of clinical outcomes with different time periods within the programme which were defined by programmatic changes or contextual changes that were related to heightened insecurity. And as I said we did a costing analysis from the provider perspective for 2014 and 2015. And then in Swaziland again we did a retrospective analysis of the routine cohort data for a one-year period from July 2016 to June 2017. In addition routine patient exit surveys were done. The team did some regression modelling looking at predictors of outcomes, clinical outcomes. And we also did this costing analysis from the provider perspective looking at annual total and the unit costs. So from Weso back to the DRC, to pull out from this slide we had 243 diabetic patients enrolled. You'll notice in bold that almost a third of the adult cohort was underweight. Also the alcohol, so self-reported recent alcohol intake at enrolment was relatively high at 22.8% compared to what we know about self-reported alcohol levels in the region. The fact that many of the patients were underweight might explain what's happening in the pie chart on the right where almost a quarter, the portion in grey, were not classified into type 1 or type 2 diabetic patients by the staff. And this could be because there is a reported phenotype of diabetes related to malnutrition in Sub-Saharan Africa where patients have a non-cutotic hyperglycemia, so very high glucose levels. They have high insulin requirements and they're often young and malnourished or underweight. So it doesn't fit neatly into type 1 or type 2 diagnosis. Now this is a very complex looking graph, but I'll explain what was done. So we divided the time period into six different sections which were related to training that happened in the program. Suspension of service when there was heightened insecurity. The next suspend to in green means that the drug supply actually ran out and then there was a gradual return in service. So what the graph represents is actually the delay in attendance at appointment. So on the Y axis you have the number of days after the appointment time that the person actually arrived for their visit. And on the X axis you have each month of the program. So each bar represents the average for the month. And the low bars between the red and the green lines represent the fact that there was a short delay. People were not late for their appointment. Whereas during the return period, the high bars after the dark blue line, they represent the fact that there was a long delay. So people were arriving up to a month late for their appointment. So clearly they were unable to reach the clinic during the heightened insecurity. And that's the circle. So again, this is a complex graph, but I'll try to pick out the most important points. Again, we're looking at these same program periods. And what we're looking at here is the clinical outcomes. So blood pressure and glycemic control and how they related to these different periods. And we have on the Y axis the proportion of visits that were considered controlled. And on the X axis, again, it's the month of each program. And we stratified the patients by whether or not they were on insulin. So the key messages really are that control was fairly consistent. There was just a modest deterioration and disease control during the period of interrupted service. And that mainly happened after the drug supply was exhausted. And the lowest line at the bottom in black, that represents the blood sugar control in diabetics taking insulin. So overall they were less well controlled and they had a fairly stark dip during the period of interrupted service. Okay, so now over to Swaziland for some of the results there. So this was a much bigger cohort. We had 895 patients enrolled in a one year period. 17.4% of them were known to be HIV positive. And it's notable that 46.5%. So a large portion were actually obese, which is quite a contrast to Moeso. It's a broader NCD program. So people were treated for hypertension, diabetes and chronic respiratory disease with hypertension. The main diagnosis at enrollment. And it's notable also that 28.5% of the cohort had two or more of the target NCDs, not including HIV. So we looked at some indicators of clinical quality and some clinical outcomes. So among the hypertensive patients almost all had had a blood pressure check performed at their last visit. And of those almost two thirds were at target. And then we looked at type 2 diabetics and again most of them had had their blood sugar checked at their last visit. And about two thirds were at target. And those are actually very good results in terms of disease control compared to the data that's available for the region on NCDs. So lots in this slide, but basically a convenient sample of patients were surveyed as they exited the pharmacy. 85 patients took part. They were strong at being able to name their NCD diagnosis and name a selected drug and its indication. But they were less good at being able to explain the detail of their diagnosis and particularly to link lifestyle advice and behaviour change with disease prevention and disease control. And also the team looked at some of the predictors of achieving control amongst the population. So in hypertensive patient there was a weak association between being obese and being HIV positive with having uncontrolled blood pressure. And in the type 2 diabetic patients we didn't identify particular predictors in this cohort. So as I mentioned we did costing analyses for each programme from the provider perspective and they were incremental costing analyses. So from WESO in 2015 annual costs were 32,000 international dollars per patient per year this was 224 international dollars. In WESO for 2016 the per patient per year costs were almost double at 441. But to put this in context there's some data from chronic HIV programmes at primary care level from PEPFAR and others that shows that the cost per patient per year of delivering this care is between 208 and 642 international dollars. And it's higher in places like South Africa. Some lessons learned from the implementation of these two different programmes. So in WESO it was nurse led diabetic care and it was well integrated into an outpatient hospital setting. In Matsawa task shifting to nurses or task sharing to nurses did not actually occur as was intended. And we felt this was due to the complexity of the care, the lack of experience generally in providing this type of care and regulatory barriers to nurses initiating. And changing medications. So it is, there are of course limitations to both studies and to both evaluations. They were small clinic based cohorts with no control groups and it's retrospective data with inevitable gaps. Our conclusion is fairly long but it is that NCD care can be integrated into MSF supported national health services. Can achieve acceptable intermediate clinical outcomes and this is at a cost that's similar to the delivery of HIV care at primary level. In terms of next steps we did think that NCD care could be simplified for stable patients. That lessons learned from HIV care could be integrated into NCD care for example community adherence groups and lay counselling and support. And also that contingency planning is required to minimise treatment interruption for NCD care for patients in unstable settings. And again lessons have been learned from HIV in this regard. So I'd like to thank everybody involved in the evaluations especially the staff and the patients of both services in the DRC and in Swaziland. And thank you all very much for your attention.