Reporting of MUAC data from nutrition and mortality surveys

I am not a user of ENA for SMART ... as far as I know it doesn't run on the computers (Apple) that I use. At the risk of making some grossly misinformed comments ...

I think it is past time that MUAC was included in the ENA tools since it is now the basis for needs estimation for CMAM services.

Some technical issues ...

My preference is _not_ to collect raw MUAC data (i.e. measurement recorded in cm or mm) but to collect MUAC class (i.e. red, orange, green) - see below. In this case the quality checking component during training will not use the method of Habicht (1974) but something like Cohen's Kappa (1960) as extended by Fleiss (1971) for multiple categories and multiple raters.

I can see advantages in collecting raw MUAC data as this will allow for some checking for normality and outliers but the (limited) data available (i.e. Feeney 2004) indicates that error is reduced when using banded straps.

Just my tuppence.

I have to take issue with this statement :

"The research on MUAC and body shape is very interesting. However, there is still a long way to go before MUAC becomes the accepted tool for estimating needs at a population level."

I am in Zambia at present. There are two integrated CMAM programs running here. Both find cases and admit on MUAC and do _not_ use W/H. I'd like to know, then, how I can assess need using an indicator (i.e. W/H) that is not used as a case-defintion in these programs?

It's not just Zambia ... CMAM program throughout the world use MUAC as the primary admission criteria. Many use only MUAC but some use a mixture of MUAC and W/H. In these programs you need both MUAC and W/H to assess need.

And this :

"The purpose of SMART is "to standardise methodologies for determining comparative needs based on nutritional status, mortality rates, and food security". At the moment, the internationally accepted method for determining the prevalence of malnutrition for the purposes of planning or assessment of need is weight-for-height (see Sphere)."

If that is the case then how do you explain the fact that emergency needs assessments performed by the Red Cross use MUAC? The SMART and W/H thing often lags behind the Red Cross assessments. To me this suggests that MUAC is the primary needs assessment tool in many emergencies. Are we to assume that SMART do not consider The Red Cross to be a reputable organisation that produces "internally accepted" early-phase needs assessments?

And this :

"While there is no reason why individual agencies shouldn't collect MUAC along with weight for height data"

The fact that SMART ENA does not make this easy for them to do is a reason. As for MUAC "add[ing] time to doing surveys". The time required to take MUAC is, at least, a couple of orders of magnitude shorter then that required to take weight and height.

Let us put MUAC into SMART and ENA as soon as possible.

In response to :

"I push for collecting heights, weights and MUAC for all women in the household also. This has been a problem as MUAC tapes for women are harder to find. Does anyone have a good source?"

TALC:

http://www.talcuk.org/

have a range of straps and MUAC training AIDS. See:

http://tng.brixtonhealth.com/node/24

For links to MUAC-specific products for TALC.

The VALID / Brixton Health strap:

http://tng.brixtonhealth.com/node/33

has marks at 210 mm and 220 mm although the strap is short and will not fit around arms with MUACs above about 230 mm.

Pictures of some of these straps are at:

http://tng.brixtonhealth.com/node/30

I hope this helps.

GBP 1.00 for a MUAC strap seems very high. That is the price for the long insertion tape used for measurements such as waist:hip ratio and chest circumference (an interesting alternative to birth weight). The standard TALC straps:

http://www.talcuk.org/accessories/small-coloured-insertion-tape-muac.htm

are 390 mm long and suited to use with both children and adults. They cost GBP 0.25 for a single strap. You should get a discount for a large order.

We need to be careful about treating W/H as a gold-standard since it appears that rather than MUAC overestimating the prevalence of wasting it is W/H that overestimates in pastoralist populations. W/H is just one way of measuring wasting. MUAC is another.

A agree that MUAC is needed to estimate need for CMAM / CTC programs and that collection adds little burden to surveys. I see no good reason why SMART / ENA does not analyse MUAC.

Another use for MUAC is as a first-stage measure in a two-stage screening process (MUAC then W/H). Finding an appropriate MUAC cut-point is an analysis that could be added to ENA for SMART. This is done in the old program:

http://www.brixtonhealth.com/Screen100Setup.exe

This is the latest advise from the WHO and UNICEF:

"To improve planning, it is therefore vital that the same criteria are used for estimating caseload as are being used for admission into programmes. This means that in settings where MUAC will be used as the admission criterion for therapeutic feeding, especially at the community level, it is important to include MUAC assessment in the nutritional prevalence surveys."

from their new guideline "WHO child growth standards and the identification of severe acute malnutrition in infants and children : A Joint Statement by the World Health Organization and the United Nations Children's Fund".

Given this, I think it is now time that SMART supported MUAC in its documentation and software.

I assume this is with the 125 mm and 110 mm thresholds. These are high prevalences of both GAM and SAM by MUAC. I would expect an elevated incidence of mortality in this context. Perhaps that is about to happen.

The predictive power (i.e. for near-term death) of MUAC has been demonstrate by several cohort (follow-up) study of untreated children in several (almost) completely enumerated populations (e.g. population laboratories). This is generally considered as epidemiological gold-standard evidence.

The problem with the 1.5 / 10000 / day figure is that it probably comes from a cluster-sampled retrospective cross-sectional survey. Such surveys are notorious for, unless very well done (and even then), being open to many sources of bias and have key methodological flaws (the big one, IMO, is that they attempts to measure a clustered phenomena with a clustered sample). They will have a large downward bias if the same sample as the nutritional anthropometry is used. This is an extremely common mistake.