SAM is a set of conditions. We might call these "severe wasting" and "kwashiorkor". Sometimes they occur together in the same child in a condition that used to be known as "marasmic kwashiorkor". Kwashiorkor shows strong spatial heterogeneity (a fancy term for patchiness) at all scales. Recent work (in press - I will post a link here when it is published) shows kwashiorkor and marasmic kwashiorkor accounting for about 33% of SAM in the DRC. This is an average figure from a large number of surveys (264 surveys, 227,390 children) so 50% in a survey or higher would not be unexpected particularly given the patchiness of prevalence.

Many thanks for this. Some surveys in DRC have only measured SAM prevalence by W/H z-score, with no assessment for oedema. In order to estimate annual SAM burden (including all its conditions), would it be correct to try to ‘adjust’ severe wasting prevalence to account for oedema cases? If so, what adjustment would you recommend?
Thanks as well for mentioning the recent work that has been conducted. I will keep a look out for the link once you have posted it.

Surveys like DHS and MICS often do not collect the oedema sign intentionally (i.e. there is no space on data collection forms for it). SMART surveys do collect the oedema sign. I am surprised that in a country like DRC in which kwashiorkor is known to be relatively common that the oedema sign is not routinely collected.

Calculating the annual SAM burden from prevalence is not straightforward. This has only become a matter of interest since the adoption of the CTC/CMAM delivery model which allows for greatly increased coverage compared to the centre-based inpatient delivery model. A number of methods have been tried by a number of different teams. A team based in Harvard are currently assessing the different methods. Having said this ... estimating annual burden from prevalence does need accurate and precise prevalence estimates. There are a number of issues here with survey methodologies but these are not relevant to you question.

It is desirable and possible to correct prevalence estimates for the failure to collect and analyse data on the oedema sign. The information that I gave in my previous post is not very useful here as they were based on the MUAC < 115 case-definition and it is possible to have:

You will have prevalences for the first two case definitions from the surveys you mention but will not have prevalence for the oedema only case-definition.

I have looked at the database of surveys that I have to hand and in the 266 surveys from DRC (n = 226,767 children) I found:

The correction factor would be:

Just checking ...

Note that this is for the WHZ case-definition and ignores MUAC. Unfortunately DHS and MICS (and some SMART surveys) do not collect MUAC makin these surveys next to useless for assessing burden for CMAM programs.

I hope this helps.

Many thanks - very helpful