Dear José, programme boundaries are NGOs issues and not population issues; the population probably attends the most convenient health centre based on accessibility (distance), services offered and, most probably, quality of services. Is there any evidence that barriers may be different from the 5 communes you support and the commune supported by the other NGO? For example, supplies shortage may have a big impact on coverage, regular or disrupted supplies will impact on coverage (in your area or in the other area). Also don’t forget that the main objective of SQUEAC is identifying barriers (and boosters) in order to take programmatic decisions. Will you be able to take decision (advice) for the other NGOs? In my opinion, unless agreed with the other NGO, I would cover only your communes of intervention, however I would pay attention to where the children admitted will come from (your communes or the other commune). Amongst others, this can have an impact on defaulter’s rates (very difficult to do the follow up of children living in area not covered by the programme). Briefly, pay lot of attention at the data analysis; try to “isolate” children coming from the commune not covered by your NGOs and see if these children impact on your coverage. Discussions should obviously follow with the other NGOs. Good luck

There is, I think, no correct answer to the original question. It depends on what you want to know ... coverage in your five communes or coverage in all communes? You may want to consult the other NGO and the responsible governments departments before doing M&E work in their area.

Dear Mark Myatt One thing I would like to ask that if Jose is intrested to see the coverage of the services, his/her organization is providing then the boundaries of the Communes must be specified because his/her intrested to see their own coverage of the services and if Jose just wants to see the coverage of services providing for whole population then there is no need for Communes boundaries specification.

I think there are a few issues here. We need to know what is meant by "commune boundary". These may have a real meaning in the sense that the commune has a strong influence on the program and program coverage by (e.g.) proving training, logistics, or integration with other commune-led services. In this case it would make sense to look at coverage in each commune perhaps using SQUEAC small-area surveys or separate SLEAC surveys. If "commune boundary" has no real meaning with regard to CMAM programming and CMAM program coverage then we might not bother with looking at coverage by commune. Note that we can look at the "whole population" whilst looking at subpopulations divided by administrative boundaries (see the Sierra Leone national SLEAC survey in FEX42 for an example). We also have to take politics into account. It will be much easier to undertake surveys in your own area(s) rather than in others' areas. I would be very wary of doing uninvited M&E in an area without consulting those responsible for the program in those areas. I have worked in some countries where this could cause very big problems for programming. I see tools like SQUEAC as audit tools that inform program reform. I know that I have a good chance of reforming the program that I am responsible for. I may not have a good chance of reforming a program that I have nothing to do with. I think it makes little sense to do SQUEAC in programs that I cannot hope to reform. The old prayer asking for "courage to change what can be changed, serenity to accept what cannot be changed, and wisdom to be able to distinguish between the two" applies here. What is needed is an informed and wise decision. I'm not sure that we have the information here for that.

Hi, I think Mark point very well what we need to remind when we do a SQUEAC and translated indeed most of what we look for, it is has to say more than a figure but indeed the reasons for which this figure can be more or less raised under the circumstances. On the other hand I think that it will be important for José Luis in his analysis to target well the children who result from the other “communes” and to make it a separate and distinctive point. Mark I still did not have time to read the FX 42, you puts back us has the SLEAC to identified the coverage of small area if I understand well. In the case of our investigation were we found only few malnourish child (28) could it be more justified to use the combination of both methodology’s for a better understanding of the coverage in every small “commune” which is was not possible with only SQUEAC? Thank you Chantal

SLEAC is designed to classify coverage in a "service delivery unit" (SDU). This could be a clinic catchment area or a commune or a district. In the Sierra Leone case we classified health districts since from the viewpoint of the national MoH the SDU was the health district. If I were doing this is a district then the SDU might be the locality or ward or clinic catchment area depending on the structure of the program. SLEAC can work in low prevalence settings. The normal SLEAC sample size is n = 40. This provides accurate and reliable classification when you have large numbers (e.g. > 1000) of SAM cases in the survey area. If there are fewer SAM cases then you do not need such a large sample. If (e.g.) you have good reason to think that there are only about 100 SAM cases in a survey area then you can make accurate and reliable classifications with a sample size of n = 28 cases although you may need to work very hard to find those 28 cases. There are sample size tables in the SLEAC section of the SQUEAC / SLEAC handbook. Another approach is to use the simplified LQAS method in SQUEAC. This will only works well with very small sample sizes if you are testing a prior hypothesis about coverage in a small area (this is the methods main use in SQUEAC). If you need a detailed map of coverage then you could use the new S3M (Simple Spatial Survey Method) which is currently being developed by UNICEF, VALID International Ltd., and Brixton Health in Niger and Ethiopia. Results of pilot surveys will be available soon. Work is ongoing to use this method to map prevalence of GAM and SAM, IYCF, WASH, and other indicators.

Thanks a lot for all the interesting advice. I'd like to give you a little bit more of information about the area. First: both NGOs have a good relationship. The NGO that works in only one "commune" has facilitated us access to their data and I will visit their centres this week. Also I believe they are interested in the results and will use them to improve their program. Second: Community boundaries (1 Commune is around 30.000 inhabitants here) are defined according to administrative divisions in the Haiti State. We think that people do not respect this "boundaries" in terms of access to health centres and many people goes to the centre that suits them best. Main reasons seems to be distance but also knowledge about centres that have been recently open. Third. Both programs face similar problems though management of the health centres are different. Some are more integrated in the government health centres than others. But there seems not to be big differences in terms of barriers: there is no supplies shortage in any of them. Fourth: the NGO working in the one commune has very few admissions since prevalence of SAM in Haiti is generally low. I have the idea of making the survey in 6 communes (5+1) but I still have my doubts. Both NGOs have agreed that a general coverage figure will suit them but I am sure that at the end they will ask to know if there are differences in coverage rates between both programs. In order to do that, must I conduct two phase 3 surveys? thanks again and congratulations for this very interesting forum

Thanks for this information. Some comments ... First : I think it a very good thing that the two NGOs can work together like this. You have to be aware that they might not like you so much if you find low coverage. Second : Very likely. I have seen cases in Niger (CSAS) and Zambia (SQUEAC) of beneficiaries not respecting national borders. Third : OK. I think that you may need to do a little more work to make NGO-specific results. I have noted a tendency for rejection of low results by NGOs ("shooting the messenger") and it will be easy to reject disaggregated results with "Our coverage is good. It is mixing it with your coverage that brings the overall result down". I don't mean to pick on NGOs. I have seen this behaviour in MoHs and UNOs too. Forth : Very few admissions can mean several things ... low prevalence is one but flight to other centres and low coverage are others that should be considered. An attempt at an answer ... I do not think that you will need to do two separate phase 3 surveys. You might consider the "risk mapping" approach in phases 1 and 2 ... as you are collecting and reviewing the data required to identify areas likely to have high / low coverage, you should mark the areas of interest on a map. You are interested in sorting communities into two distinct groups: (1) Communities / areas where the coverage is likely to be high. (2) Communities / areas where coverage is likely to be low. This process is known as "risk mapping". You might call its "coverage mapping". The term "risk" is used because it implies that low coverage puts children at risk. You might find it useful to mark clinic catchment areas on the map together with a summary of data from interviews and group discussions with clinic staff, clinic workload returns, time-to-travel data, MUAC at admission data, carer interviews, &c. You might also find it useful to use transparent overlays that show (e.g.) the probable distribution of identified barriers. The final output of the assessment will include a risk map. Working with risk maps at an early stage in the assessment allows you to create material useful for inclusion in reports as you work. This is much the same as we do in SQUEAC phase 1 already but we may place a little more emphasis on producing a detailed risk map. At the end of phase 1 of a SQUEAC assessment you should have collected a large amount of data relating to coverage. Before proceeding further you should probably document the evidence and assumptions used to make the risk classifications (i.e. low coverage, high coverage). We would now use small-area surveys to confirm the risk mapping. This will include confirming your high and low coverage areas. In your case you would want to place one or more surveys in the second NGO area which mirror those done in the first NGO area. This would yield a map of coverage based on confirmed hypotheses. That might be good enough. It is important to note that if you have a great deal of patchiness (heterogeneity) in coverage then an overall estimate is nit very meaningful. I hope this helps. BTW : The forum is what we all make of it.

Thanks Mark, I will follow this approach. If I have understood well I can use several hypotheses to select areas of low and high coverage. Specifically I think about 1) In rural areas distance seems a major problem, I expect low coverage in some mountainous localities on the far north 2) In cities there seems to be other problems than distance. Neighbourhood with impoverish migrant/displaced population that moves a lot and also were screening activities do not happen seem to have low coverage rate. I will test theses hypothesis with small are surveys, and mirror these surveys in both NGO's areas. Then I will move on a global phase 3. Does this seem correct to you? thanks a lot

These seem reasonable to me. let us all know how you get on with this.

Phase 2 confirmed our hypothesis about low and high coverage areas. Unfortunately in the second NGO's area only 2 SAM cases were found and therefore I can not show evidence of differences between one NGO's coverage and the other's. I have moved to phase 3 since I am running out of time. I suppose I can not use phase 3 results to show difference among the two NGO's coverage rates? Anyway I think a global figure about coverage will satisfy both NGOs, Thanks again

Phase 3 is usually a wide area survey. You could do two surveys but that would probably be a waste of resources. Do you have sufficient information to say something about the probable coverage in the second NGO's area?

yes, I do not have enough resources for two phase 3 surveys. I will stay with one coverage rate for the whole area. To say something about the coverage in the second NGO I can only use data from phase 1, which is not bad, but will stay as hypothesis.