Oedema and nutritional surveys
oedema and nutritional surveys Michael Golden 02.09.99
Re: oedema and nutritional surveys Andre Briend 03.09.99

Date: Thu, 02 Sep 1999 19:09:38 +0100

To: ngonutatabdn.ac.uk

From: Michael Golden <m.goldenatabdn.ac.uk> Subject: oedema and nutritional surveys


Dear Ngonuts,

It has been suggested (more details : click here) that children with oedema should not be included in weight-for-height or weight-for-age surveys , presumably on the basis that these children are perceived to have a significantly higher weight than they would have if they were oedema free.

Some years ago I did an analysis of oedema assessment by experienced and unexperienced clinicians (Golden MH. The clinical assessment of oedema: implications for feeding the malnourished child. Eur.J.Clin.Nutr.

1989;43:581-582.) in which oedema was consistently over-assessed by those unused to looking at weights as oedema is lost - an uncommon skill. From this it was suggested that no allowance be made for oedema weight in calculating the dietary requirements of oedematous children. That study was done with small numbers of children in a metabolic ward setting where the intake was carefully controlled so that no tissue growth occurred whilst oedema was lost. Those with + oedema lost 2.8% of body weight, with ++ lost 4.2% and with +++ 8.2% of body weight.

I have now analysed the amount of weight lost between admission and minimum weight in children over 6 months who are 65 to 110 cm in height from Therapeutic Feeding centres in Africa. The weight loss was skewed, probably because of errors in the minimum weight, however the median weight loss is as given below, expressed as a % of body weight.

Oedema grade

wt loss%

  median mean -maras Number : Jamaica
none 0 1.8% 0 4786 ?
mild (+) 2.3% 4.4% 2.6% 1536 2.8%
Moderate (++) 3.9% 5.6% 3.8% 1757 4.2%
Severe (+++) 7.3% 8.9% 7.1% 782 8.2%

-maras is the mean percentage weight loss of the oedematous children in excess of the mean weight lost by marasmic children.

These figures are close to those obtained previously and consistent whether the mean or the median is considered - the difference are within the standard errors of the previous study - alternatively the difference of about 0.5%, if real, could be accounted for by the higher dietary intake, and hence tissue growth, whilst oedema is being lost in the therapeutic feeding centres. The weight changes are very much smaller than most people imagine. The present analysis includes sufficient numbers of children to make adjustments at a population level.

I strongly suggest that oedematous children be included in the surveys.

The difference in weight with loss of oedema will not make a major difference (i.e. 2.3% of a 10kg child is 230g which is not dissimilar from the accuracy of many of the scales used in the surveys!

Alternatively, a correction factor could easily be applied to the weights from the above figures.

It would be wrong if children who are randomly selected are then systematically rejected on the basis of oedema particularly as these children are, by definition, severely malnourished.

If the suggestion of omitting these children from surveys becomes common practice it will not be long before the problem of oedematous malnutrition is not considered in surveys at all - this would be tragic. These children must not only be counted but must also be classified as severely malnourished! Any survey that fails to count oedematous malnutrition is likely to underestimate the problem and thus not act when action is required, particularly in those countries where kwashiorkor is a common form of malnutrition.

The MICS surveys done by the UN agencies and National governments do not even record whether a child was oedematous or not so that these data must be viewed as giving a minimum prevalence of severe malnutrition, and could potentially be grossly in error. It is presumably for this reason that the WHO global database on childhood growth and nutrition ignores the presence of oedematous malnutrition in its tables. I suggest that, far from excluding these children, and designing software that allows some severely malnourished children to be ignored, that the MICS and other nutritional and anthropometric surveys should always assess the prevalence of nutritional oedema.

I understand that there is still a problem with epinut6 in terms of oedema (those who are not also wasted will not be counted if the include oedema option is chosen and if it is not chosen those with oedema will be eliminated from the count of the malnourished - if I understand correctly) Epinut 5 does not have this problem and enumerated oedematous children properly. At any rate, Pierre Nabeth is discussing this with Epicentre and CDC and has agreed to tell the membership of the outcome of these discussions.

Best wishes,


Mike Golden

Date: Fri, 03 Sep 1999 13:56:21 +0100

From: "Andre' Briend" <briendaatcnam.fr>

Subject: Re: oedema and nutritional surveys


Dear NGO nuts,

As far as I remember, this point about the effect of including children with oedema in nutritional surveys (or rather the lack of effect) is discussed at length in the paper below:

Franklin RR, Dikassa LN, Bertrand WE. The impact of oedema on anthropometric measurements in nutritional surveys: a case study from Zaire. Bull World Health Organ 1984;62(1):145-50.



Dr. André Briend


Groupe Nutrition Santé - ISTNA

5 rue du Vertbois, 75003 Paris, France

tel : 33-1-53 01 80 36 fax : 33-1-53 01 80 38


Note added by Mike Golden

Andre is quite correct - thank you for bringing this to our attention.

Franklin et al were looking at the relative prevalence of oedema in nutritional surveys from Zaire during the emergencies of 1978.

In their initial survey there was a very high prevalence of oedema in children that would not even be classified as moderately undernourished (ie >80% WFH). Their table is as follows with percentages in parentheses:

% WFH oedema no-oedema total
<80% 45(4.1) 97(8.7) 142(12.8)
>80% 114(10.3) 851(76.9) 965(87.2)
total 159(14.4) 948(85.6) 1107(100)

"...if calculation of the proportion of children with acute malnutrition has been based upon anthropometric criteria alone, the total would have been grossly underestimated; 72% of the children (114/159) would not have been considered malnourished". In fact the prevalence of less than 80% (ie moderate malnutrition) would have been judged to be only 12.8% of the population (the proportion less than 70% is not given but is likely to be 2-4%), and yet the true prevalence of SEVERE malnutrition was 14.4% plus those <70% (or -3Z of course) who did not have oedema. Without considering oedema a quite erroneous impression of this emergency would have led to inappropriate action.

In 5 subsequent surveys the percent below 80% WFH and the number above 80%WFH with oedema was as follows:

  %<80 no oedema %>80 oedema
Baseline 8.7 10.3
1 2.5 4.6
2 1.9 3.5
3 1.8 2.7
4 1.6 1.9
5 2.0 1.7

It is clear that with resolution of the famine the prevalence of undetected severe malnutrition remained just as high as the prevalence of moderate wasting.