There is no standardised questionnaire. Instead you build the questionnaire based on what you learn about SAM causality during SQUEAC (e.g. from histories, interviews with program staff, key-informant interviews, &c.). A questionnaire can then be built from standard parts (e.g. WASH questions) but you will usually have to have bespoke questions which you have to make up and test as needed. This is a very common approach in service epidemiology. I suppose that you could have a "usual suspects" questionnaire but this may miss local risk factors. I think the best we could do ius to work on a set of modules that could cover most situations. We would still need to allow for locally appropriate question-sets.

Thanks Mark, that will be quite helpful. What modules exactly do we need to go through? And what of the sample size?

I think that, as a core set of modules: (1) IYCF : I would start with the new streamlined indicator set and (maybe) pick additional individual indicators from the WHO adapted for retrospective collection. (2) WASH : Basic JMP questions for latrines, water, &c. plus USAID KAP type questions about hand-washing. (3) WDDS : Dietary diversity for mothers. (4) Poverty : MPI indicator. This will also include EPI, health, wealth, education dimensions and individual indicators. (5) Household hunger scale (6) Recent morbidity (child) Much of this is / can be made standard. VALID and GAIN have been using this in their S3M surveys and have a good set of indicators. The main thing is to have indicators that can be used with small samples and to allow for additional questions to be added that may have arisen from the SQUEAC investigation. Sample size will depend on wether you look at GAM or SAM. If cases are easy to find (e.g. MAM, low HAZ) then I’d suggest a simple case-control study with (e.g.) 65 cases and 65 controls (more = better). For SAM I’d suggest a matched design with 35 cases (doable with SQUEAC) with two or three (more up to five = better. Not much point in going form > 5 controls per case) age, sex, and neighbourhood matched controls per case. You should treat these sample sizes are guideline minimums. Such a study will not have great power but this may be what you want as it will detect the larger effects only. You can also use SAM cases in treatment (and recovering SAM cases) as cases in the case-control study but probably best to pick up cases in the SQUEAC stage III survey. A MUAC / oedema case-defintion is most sensible and also most practicable for identifying cases and controls in the field. I hope this helps. Let me know if more detail is needed.

Thanks Mark, I'll get back to you when we need further details. Thanks