What's the spontaneous recovery rate for SAM?

I'll have a go at this ... First ... it all depends on how you define SAM. Spontaneous recovery rates for SAM defined by low MUAC will be different than for low W/H. They will depend upon the case-defining thresholds used. If using W/H, they will also depend upon location, ethnicity, and food-economy zone (amongst other factors). There is plenty of data on mortality amongst untreated children from the classic cohort studies of about 30 years ago. We can think of mortality as the opposite of recovery. It's not exactly that because it is possible to not recover and not die but it is probably safe to make the simplifying assumption that a SAM child either recovers or dies. I just did a very quick review of some of the original reports and found a mortality rate for children with a MUAC of 110 mm to be between 100 and 250 deaths per thousand per year. Turn this about and you get 75% - 90% spontaneous recovery. As MUAC gets lower the mortality increases. At 100 mm the range was 300 to 600 deaths per thousand per year. Turn this about and you get 40% - 70% recovery. Looking at these figures ... the 20:1 you quote seems to be wrong. The SAM definition used for this is probably using W/H which is not so strongly associated with mortality as MUAC. This is just a quick answer. The point I am trying to make is that you can use the mortality estimates from the classic cohorts to answer this question. The problem that you may find is that it was common to define "severe PEM" as < 80% W/H ... this is close to what we now call GAM. To get answers that fit with current case-deifntions would probably require data to be re-analysed. Some of the data may be available for re-analysis. I hope this helps.

As always, happy to help.

Let us thank Severine. I have put the bundle of articles in this ZIP archive. Severine also found this online FNB article which is also useful.