I have seen this done. I have just read a paper using <= -1.5 z-scores as the upper limit for MAM. The thresholds you use will depend on context. If you have resources and or low prevalence then you might want to add an "at risk" category or increase the case-defining threshold for MAM. Beware that small increases in case-defining thresholds lead to large increases in case numbers (e.g. using -1.5 rather than -2 will probably triple case numbers). Also, in some settings the effect of body-shape (long legs and short trunks) on WHZ will mean that you will select a very large number of healthy children (i.e. not reasonably "at-risk"). You are much "safer" using MUAC.

Thanks! Actually, I haven't seem like this before, except MUAC. By the way, could you please provide the paper you have seem. I do want to keep as document. Best, Ncik

It is just a paper that I happened to be reading when I saw your message. It does show that we can have flexibility with case-definitions. I have seen (e.g.) MSF programs in the Sahel using MUAC < 120 mm and/or oedema delivering the CMAM protocol. Anyway ... click [url=http://www.brixtonhealth.com/Purwestri2012.pdf]here[/url] for the paper.

I think defining as somehow abnormal those children whose weight-for-height z-score is between -2 and -1 is dangerous and incorrect. In a normally nourished population, this category would include 13.6% of the population. Are you willing to label 1 in 7 children in a nutritionally normal population "at risk"? Moreover, this category is not standard. As a result, including such a category in a report can result in misunderstandings. There is the famous case some years ago in which the prevalence of acute malnutrition was labeled as 25% in a country which previously had little acute malnutrition. It caused a panic among epidemiologically naive readers. It turns the report's authors included "at risk" children with z-scores between -2 and -1 in their definition of acute malnutrition. These children are only slightly thinner than other children in the population, but should not be labeled as abnormal.

Thanks for your clarification. If we can not say those children (WFH <-1 and >= - 2) as at risk children, how can we present the nutrition risk children with technically safely? Even Mark had already mentioned as MUAC is more safer in this case, is there any other tools or indicator for nutritional risk group children?

You might use other indicators. In terms of outcome (i.e. mortality) W/H performs worse than any other indicator (this is a very consistent finding). You would be better with low H/A or low W/A. These require age to ascertained with accuracy and that can be a problem. There are composite indicators (i.e. using W/H, H/A, and W/A) that might be of use. I'd prefer W/A or MUAC. Just to add to Woody's comments ... Since a lot of linear growth in children is in the limbs (thing of the body shape of a 6 month old and the body shape of a four year old) and long limbs reduces W/H you may, using WHZ < -1 in some populations, select the most healthy children as being "at risk".

Yes, I got it. Now, we can manage to get accurate DOB in field level and then, calculate age by ENA. It is easy to get WFA and HFA results. Technically, WFA <-2 reflect both chronic and acute malnutrition and HFA go for stunting . So, what value of threshold are more suitable to use as low HFA and WFA for this case. Hope, this might be last question from me for this topic! Thanks :)

All of the "-for-age" indicators (W/A, H/A, MUAC/A, BMI/A) are sensitive to errors in age. If you can get accurate age or DOB than you should not have problems. The W/H, W/A, H/A indicators are all standardised (i.e. to the "standard normal" so you can use the -3 z and -2 z thresholds. Note that H/A in a cross-sectional survey measures "stuntedness" not "stunting". Both low W/A and low H/A are more predictive of near term mortality than W/H. W/A is better than H/A. MUAC is as good or better than W/A. I'd stick with MUAC as it is quick, cheap, accurate, reliable, acceptable, and has good predictive value.