They have done this type of analysis in a recent evaluation of the CMAM program in Niger, you may ask the report to Eric eckouam@gmail.com, I have a draft copy, not sure I can share.

Best,

Hi Dr. Oloyede,

There is a quick way way to roughly estimate deaths averted that can be derived from routine data. However, it is important to note that the method is based on several assumptions, based on MUAC only, assumes negligible relapse rate and may not be applicable to your specific context or data set.

Case finding and community mobilisation are still overlooked or given low priority in many CMAM programmes. Whatever method of calculation is used, the most important take-away message must be that vigorous case finding and early admission to treatment has a big effect on reducing mortality.

Research from 1986 to 1994 in various contexts (Briend & Zimicki, 1986; Briend, 1987; Vella et al, 1994) indicates, for example that the mortality rate for untreated SAM at a MUAC of 110mm varies between 10.5% and 21.1% (mean = 17.325%) of cases per year. Based on these estimates we can approximate the number of deaths averted for any given year in the CMAM programme.

Deaths averted = Number of admissions in one year x 17.325% x Cure Rate

E.g. For 20,000 admissions in one year with a 70% cure rate

Deaths averted = 20,000 x 0.17325 x 0.7 

SAM deaths averted in one year = 2425 (Range: 1,470 – 2954)

It is probable that a significant proportion of non-death negative outcomes also survive but these aren't included in this quick estimate above. Therefore interpret the result of this quick calculation very cautiously.

A fuller discussion of estimating lives saved, the various assumptions and objections to the method that need to be considered / accounted for can be found here:

https://www.ennonline.net/fex/50/deathsavertedcmamnigeria

I hope this helps,

Paul

Dear Paul,

Thank you very much for this simple rought estimation. I think we do not have much problem with relapse and defaulters and the cure rate was very high at 97% being a camp setting.

I sincerely  thank you.

James

James,

I think this will be useful.

There is a reasonably simple way of doing this. See this FEX article.

Much of the data required is often collected as routine program monitoring data (e.g. number treated and proportion cured).

Calculating expected mortality needs different data. You can estimate the case-fatality rate if you know the average MUAC at admission. You may need to collect this data from beneficiary registers (this should not be tooo expensive or diffiuclt in a camp setting). It is usually available from SQUEAC reports ... I have used these in cost-effectiveness analyses of CMAM in Niger and Burkina Faso. You will also need to use mortality estimates from historical cohort studies. You put these together using a linear interpolation procedure to find an expected mortality. You then adjust this to account for background mortality using commonly available U5MR estimates. The World Bank has a publicly available database  which reports U5MR for many countries.

The method described in the FEX article (sampling based estimation) uses R (free software). You could, as an alternative, use fuzzy triangular numbers to handle uncertainty.

This handbook  provides extract of the mortality estimates from historical cohort studies, a tutorial on how to work with fuzzy triangular numbers with numerous worked examples, and a link to an online fuzzy arithmetic calculator.

I hope this is some help.

Many thanks for sharing these thoughts and documents.

Dear James

Just to add that when using U5MR it is essential to have this broken down by age. The reason is that almost 50% of U5MR occurs in the neonatal period and much of the rest before the age of 12 months. I have seen people comapring their figures for populations with a mean age of 2 year with overall U5MR, which vastly over-estimates background mortality for the age group that was investigated.

Jay