When we move from looking a SAM coverage to looking at MAM we actually make things easier because it becomes easier to find a sample. Imagine we have: SAM : 1% MAM : 8% and an average village size of 500 and a proportion 6-59 months of 20%. We would, on average find: 500 * 0.20 * 0.01 = 1 case of SAM per village, and: 500 * 0.20 * 0.08 = 8 cases of MAM per village. The main difference when moving from SAM to MAM is that we use door-to-door sampling with a brief verbal household census to find cases. This is more work than with SAM but we sample in far fewer villages. If we take a sample of 8 villages we get a sample of something like n = 64 cases. This is quite large for both SQUEAC and SLEAC. Given a moderately strong prior we would have a 95% CI of ± 10% or better on a final coverage estimate. If coverage were (e.g.) < 30% (i.e. when we need most barriers data) then we would have >= 0.7 * 64 = 44 uncovered cases to give us a ranked list of barriers to confirm / rank SQUEAC stage 1 findings. We can also use a quick and dirty indirect estimate of coverage from SMART surveys, distribution lists, maps of distribution points, &c. to give us a prior. We might (e.g.) have: Population : 100,000 p(6-59 months) : 20% MAM : 8% (95% CI = 5% - 11%) SFP beneficiaries : 1208 We have expected cases of: N : 100000 * 0.2 * 0.08 * 2.6 = 4160 LCI : 100000 * 0.2 * 0.05 * 2.6 = 2600 UCI : 100000 * 0.2 * 0.11 * 2.6 = 5720 An indirect coverage estimate is: P : 1208 / 4160 = 29% LCI : 1208 / 5720 = 21% UCI : 1208 / 2600 = 46% We could then use a Beta(13,30) prior which would add: 13 + 30 - 2 = 31 to the effective sample size for the coverage estimate. SQUEAC and SLEAC have been used most to investigate TFP coverage but there are no real problems (apart from political will) using these methods to investigate SFP coverage.

I have actually used SQUEAC in assessing the coverage of MAM in Kisumu and Nairobi, Kenya and the process was actually helpful in determining the barriers to the tSFP program. In fact, to get the MAM cases was much easier through the door-to-door screening since the prevalence of MAM in the two areas was high compared to that of SAM (This is always mostly the case).

Hi Samuel and Mark Thanks for your replies. Samuel would you be willing to share the reports of your SQUEAC surveys? Did you face any other limitations practically with using the SQUEAC methodology for SFPs and if so how did you get round them? Many thanks,