Comment on: Including infants in nutrition surveys
Experiences of ACF in Kabul city
By Michael Golden, Department of Medicine and Therapeutics, University of Aberdeen, AB25 2ZD, Scotland.
Perinatal mortality and birth-weight reflect the quality of the intrauterine environment and complications of birth. Neonatal mortality results mainly from early infection and neonatal care. Most children with severe obstetric problems die before their first birthday. The difference in mortality rates between rich and poor countries, although substantial, is not as startling in infancy as in the 1-5 year age range. Consequently, Ken Standard, Jellife and others, in the 1960's, suggested that the 1-5 year mortality rate should be used to assess the degree of nutritional deprivation of a population. This mortality statistic was subsequently adopted by the international community, and a great deal of data collected and collated. In designing the original cluster survey methodology to examine nutritional status, this age was naturally also chosen, although the range was extended to include the 6-12 month old child.
Even though the original concept was to use measurements of nutritional status of this age group as an indicator to assess the nutritional status of the whole community, services recommended as a result of nutrition surveys were targeted almost exclusively at the 6-60 month age group. All the other age groups, the infant, the older child, the adolescent, the mother and father and the elderly have been ignored because data have not been collected and the original concept of using this group as a mirror of the whole population has been lost. Now we are realising that other age groups may indeed be at risk and need help in certain situations.
ACF is to be congratulated for collecting data on infants and attempting to address the problem of the infant in Emergency situations. Unless data, such as these are routinely collected and reported the problems of the young infant will continue to be ignored. Claudine Proudhon's report highlights that there is a problem to be addressed. However, she recognises that there are difficulties in assessment and interpretation of the results. We also do not yet know the best way to manage the malnourished infant. In the West a whole medical speciality devotes itself to these patients: they have their own dedicated wards, laboratories, equipment, nurses, doctors and consultants, community nurses and health visitors. In the emergency situation infants are largely left to fend for themselves, in the hope that mothers will universally breast feed. It is an article of faith of many that every breastfed infant is axiomatically well nourished. This is clearly not the case; anyone who has spent time with impoverished people recognises this all too clearly, but those at the 'coal face' rarely have a voice and there seems to be a taboo about questioning the received wisdom. At any rate, none of us want to give any comfort to the multinational breastmilk substitute purveyors, so the problems are ignored out of a false sense of solidarity. Recognising the failing infant who needs help, investigating the causes, treatments and prevention, generating protocols and policies, will give no comfort to the multinationals, for they have no place to play in solving this problem, and have often had a place in its creation.
There are nine 6-month periods between 6 and 59 months of age - if 900 such children are being surveyed then one would expect the number found in the ACF survey. 132 infants is not a trivial sample. The survey teams are visiting the houses; the extra effort required to collect these data is trivial compared to their value.
Let me now address some of the problems highlighted in Claudine Prudhon's article.
The use of a scale with divisions of 100g
Figures 1 & 2 show the error that is incurred in computing the weight-for-height if the weight is recorded with an error of 100g, the height with an error of 1 cm or if both errors compound one another. For an infant of 55cm the computed weight-forheight can vary from 64 to 76% (Z-score from -2.5 to 3.5). If such an error was systematic then a true prevalence of malnutrition of about 6% could be computed as 15% or 2%. Clearly, the weights and heights have to be taken to a greater degree of precision with small children. The confidence intervals Claudine has computed do not include a factor for these errors - she has reported the computed statistical confidence interval from sampling. With a large enough sample these become very small. Systematic error does not get smaller with sample size (such as routinely rounding up the measurement to the next 100g or cm). We assume that in Claudine's survey there was no systematic rounding or digit preference and positive and negative errors cancel each other out.
Figures 1 and 2 below show: The effect of 100g error in weight or 1 cm error in height on the computed % of median, and Z-score of malnourished children with an actual value of 70% or -3Zscore units.
In table 1 I give the last digits for weight and height from a survey in Monrovia of infants and children. It is clear that there is digit preference - in this case for weights ending in 0 and 6 and heights ending in 0.0 cm and 6.0 cm. This bias will potentially affect all surveys - but will be particularly significant in the measurement of young infants. Perhaps digit preference should be reported with surveys?
If the measurements are being used for screening, so that individual children are being chosen to enter a program then errors of this magnitude are much more important than for a survey where there is a tendency for the errors to cancel each other out. In screening, many of the non-malnourished infants will be included and many malnourished ones excluded from programs.
Table 1 | ||||||
Digit | Weight 0.1kg | Height 0.1 cm | Height 1 cm | |||
# | % | # | % | # | % | |
0 | 173 | 15.1 | 758 | 66.4 | 172 | 15.1 |
1 | 101 | 8.8 | 19 | 1.7 | 89 | 7.8 |
2 | 96 | 8.4 | 21 | 1.8 | 71 | 6.2 |
3 | 105 | 9.2 | 15 | 1.3 | 109 | 9.5 |
4 | 108 | 9.5 | 9 | 0.8 | 102 | 8.9 |
5 | 114 | 10.0 | 274 | 24.0 | 121 | 10.6 |
6 | 132 | 11.6 | 11 | 1.0 | 148 | 13.0 |
7 | 104 | 9.1 | 22 | 1.9 | 80 | 7.0 |
8 | 104 | 9.1 | 7 | 0.6 | 115 | 10.1 |
9 | 105 | 9.2 | 6 | 0.5 | 135 | 11.8 |
Evidence of digit preference in nutritional surveys. If the digits are recorded at random then each digit should occur 10% of the time on average.
The weight was measured to the nearest 100g, in 15% of cases the recorded number ended in x.0 kg - there was also a preference for the digit 6.
Height should have been measured to the nearest 0.1 cm. In 66% of cases this was rounded to the nearest whole centimetre and in 24% of cases to the nearest 0.5 cm.
There is even evidence that the height was rounded up and down to the nearest 10 cm as 15% of heights ended in x0.0 cm and again 6 was a preferred number with 13% ending in x6.0 cm.
This is evidence that the effect of 100g and 1 cm errors on the computed prevalence of malnutrition in infants might be an underestimate of the true error in terms of height. The graph shows that an error of 1 cm in height has a much larger effect on the result than an error of 100g in weight.
The causes of malnutrition in this age group
An analysis I was involved in some years ago in Jamaica on marasmic infants and children, where we had accurate birth-weights, showed that the influence of birth-weight on current weight-for-age remained significant until after 18 months of age. In other words, many of the over 6 month old children will be 'malnourished' partly as a result of the intrauterine environment that they experienced. This is particularly so for weight-for-age and height-for-age but not as much for weight-for-height. The extent to which weight-for-height of an infant, of say 3 months, is influenced by anti-natal events and by post-natal nutrition in developing countries is not clear. Concomitant records of weight for crown-heellength at birth (as well as birth weight) could be taken from the obstetric services when anthropometric surveys on infants are carried out.
Risk of Death
Malnourished infants undoubtedly have a high mortality rate. In the therapeutic feeding centres the following mortality rates by age group were found:
Age | Dead/Admitted | %Mortality |
0-5 | 97/565 | 17.2 |
6-11 | 155/1288 | 12.0 |
12-17 | 142/1414 | 8.3 |
18-23 | 53/899 | 5.9 |
24-29 | 142/1778 | 8.0 |
30-35 | 73/900 | 8.1 |
36-47 | 77/1029 | 7.5 |
48-59 | 41/747 | 5.5 |
total | 780/8620 | 9.0 |
This shocking mortality statistic reflects the inherent vulnerability of the young infant. They are much more susceptible to hypothermia, hypoglycaemia and infection than the older infant and child. The margins for error in their treatment are much smaller. The staff of a normal TFC in an emergency programme are not trained and equipped to look after infants - they should be, especially where 10% of admissions are for these infants.
There is also a problem with the standards, as Claudine points out, and new standards are being generated by WHO. BUT this will not solve the problem - it may even exacerbate it! This is because the present standards derived from predominantly bottle fed American infants are higher than most normal breast-fed infants. So if the normal standard is lowered, then the cut off points will also be more stringent and fewer infants will be classified as malnourished. But then how do we account for the very high mortality risk of the infants that are being admitted at the moment with the present criteria! When the new WHO standards are introduced many of these infants will not be admitted because they are no longer classified as malnourished. They will then die at home instead of in the TFC! Probably they will be given a different diagnosis because the changed standards will then mean that they are not malnourished. But they will be just as dead.
There is another major problem with the present survey methodology. In the MSF guidelines it says that children that are measured and weighed should be 6-59 months OR 65 - 110 cm. In many surveys the height cut off is used as ages are not generally known. But many societies have a major degree of stunting. So that even children of 12 months can be less than 65cm. For this reason WHO recommends (Field guide to rapid nutritional assessment in Emergency, 1995) using 60-110 cm where the population is 'stunted' (just how stunted the population has to be to change the cut-off is not stated). When both criteria are used those in the population that are over 6 months but less than 65 (MSF) or 60 (WHO) cm will be systematically excluded from the sample. They will also now be excluded from the infant survey because they are over 6 months.
Figure 3 shows the height-for-age of the children in a nutritional survey from Monrovia, Liberia, and the concomitantly measured 0 to 5 month old infants (and 11 six month old infants who were excluded from the main survey because they were short) from the same households (Corbett M, Grellety Y and Golden M, unpublished). The same methods that Claudine described from Kabul were used. In this survey there were 950 children and 192 infants. The graph clearly shows that if a cut off point of 65 cm is used then all the 6 month old children are of normal height (by selection) and it appears as if stunting starts very rapidly at this point. When the infants are examined it is clear that the population is shorter than normal at birth, and becomes progressively worse from birth with a deterioration from 4 months. When all the children are included in a single analysis the dashed line is obtained. This gives quite a different impression of when the problems of these infants start and their prevalence by height grouping. It is likely that stunted children are also more at risk of wasting. Thus, using height AND age will bias the results. One or other criterion should be used, but not both.
Figure 3 and 4:Height for age of infants (grey) and children (black) from Monrovia, Liberia, 1998
Polynomial regression of height-for-age against age- survey results.
However, unlike the results from Kabul, in this survey moderate malnutrition prevalence was quite different for young infants and older children although the prevalence of severe malnutrition by % of the median was very close.
N | <-2Z | <80% | <-3Z | <70% | |
0-5 | 192 | 3.6% | 5.7% | 1.0% | 1.0% |
6-59 | 950 | 24.5% | 18.2% | 2.9% | 0.8% |
Neither the pattern seen in Kabul nor in Monrovia is necessarily repeated elsewhere. For this reason data from other emergencies should be collected and analysed. Data on other age groups apart from the 6- 59 month old child should also be collected and reported to improve our understanding of the extent of infant malnutrition.
It should be noted that the 70% of the median and - 3Z score lines, when plotted on a chart cross at about 65cm. Above this height more children will be admitted to a program using the Z-score criteria than the % of median. Below this height more children will be admitted with the % of median than the Zscore. The % of median line is closer to the risk of death line in this respect, and is one of the main reasons for continuing to use the % of median rather than the Z-score for admission to therapeutic or supplementary feeding programs, notwithstanding the mathematical elegance of the Z-score. In terms of risk of death in the community both measures are untested.
In conclusion, I applaud ACF's effort in collecting and reporting these data, hope that others follow suit and that RNIS reports the data. There should be a cooperative operational research program using the resources of the many NGOs collecting data. The raw data should be made freely available for collective analysis by those trying to tackle the problems of the infant in emergencies; that way we could move forward on a broad front and much more rapidly than at present.
Note: Theoretical calculations of error and also the Monrovia data was collected Mary Corbett during her MSc Program - this study was done in collaboration with Yvonne Grellety and the team in ACF Liberia.
More like this
FEX: Including infants in nutrition surveys
Experiences of ACF in Kabul city Claudine Prudhon is the head of the nutrition department in ACF HQ. Claudine is interested in improving the nutritional management of the under...
en-net: Digit preference
I'm just back from Ethiopia and speaking with some key people in the relevant government departments noticed a great emphasis being place on posterior (i.e. after survey)...
FEX: Implications of 65 cm height cut-off as age proxy in Bangladesh
Summary of published research1 Location: Bangladesh What we know: Age data for children aged 6 to 59 months is often not available in poor, migrant or conflict affected...
en-net: Using length/height to estimate Age of 6-59 month old children in high Stunting level countries
WHO recommended to use 60 cm as an estimate for 6 month old infants and 100 cm for the 59-60 month old children in countries with high stunting rates in 1995 Nutrition...
FEX: Diagnostic criteria for severe acute malnutrition among infants under six months of age
Summary of research1 Location: Kenya What we know: Diagnosis of acute malnutrition in infants under six months old (U6M) is currently based on weight-for-length z score...
en-net: Tableau de l'OMS pour les enfants de moins de cinq ans
Hi, I have few questions about WHO charts indicated below. 1. Why are Weight-for-length and weight-for-height charts separate? 2. Why does weight-for-length chart range...
en-net: Urgent: Weight for Height Z-scores: Unisex Z-score table used in West African countries vs. calculated Z score values
In the 2011 Guidelines for THE INTEGRATED MANAGEMENT OF SEVERE ACUTE MALNUTRITION published by ACF, Annex 4 depicts a Weight-for-Height table. This same table has been adopted...
en-net: Precision of equipment to assess infants in surveys
I have a question from an NGO regarding including infants in an anthropometric survey. What precision is needed for anthropometirc assessment of infants, including those less...
en-net: SMART surveys and interpretation of quality
Few questions 1. To what extent should the results of ENA plausibility tests be used to reject surveys which collect more than anthropometric data,in the first place was...
en-net: Prevalence of moderate malnutrition in infants
Are there any data on the prevalence of moderate malnutrition in infants and if so how do these rates compare with prevalence amongst 6-59 month cohorts? Marie jogged my...
en-net: New version of R package 'zscorer' available
Ernest and I have been working on the zscorer package for the R Language for Data Analysis and Graphics. v0.3.1 of the zscorer package calculates and adds nutritional...
FEX: Practical challenges of evaluating BSFP in northern Kenya
Summary of published study1 Beneficiaries of BSFP in Kenya A mass or ‘blanket’ supplementary feeding programme (BSFP) was implemented by the World Food Programme (WFP) and...
en-net: What is the lowest WAZ score that would be "compatible with human life
Hello, I am passing on a question from a colleague. I am keeping his wording: "What is the lowest WAZ score that would be "compatible with human life"? I know that Epi-Info...
FEX: Adding proteins to lipid-based nutrient supplements in Uganda: Which was best?
View this article as a pdf This is a summary of the following paper: Mbabazi J, Pesu H, Mutumba R et al. (2023) Effect of milk protein and whey permeate in large quantity...
en-net: Anthropometric survey target group: 0-59 months vs. 6-59 months and implications
Greetings EN-NET, When I learned to do anthro surveys years ago, I was taught to sample children 6-59 months old. The rationale includes this group being most vulnerable to...
FEX: Prevalence of moderate malnutrition in infants
Summary of en-net discussion By Tamsin Walters, en-net coordinator Contributors: Nina Berry, Mike Golden, Mark Myatt, Marie McGrath A recent discussion on en-net concerning...
en-net: Recording low birth weight and considering it in SAM diagnosis of infants under 6 months
From Massimo Serventi: Dear All, I apologize if I submit a question that has been already answered in the past. I currently work in North Sudan, children with severe SAM are...
FEX: Older people, nutrition and emergencies in Ethiopia
By Vanessa Tilstone Vanessa Tilstone has worked for HelpAge International in Ethiopia for the last 3 years as the Country Programme Director and has worked previously in...
FEX: An investigation of anthropometric training by NGOs
By Naomi Tilley Naomi has just completed her MSc in Public Health Nutrition at the LSHTM. A qualified nurse, she has previously worked with MSF in Ethiopia and Sudan. A...
en-net: Planning of CMAM services
The number of the children who need CMAM services is based on the prevalence data from nutrition surveys that indicate the numbers of children with SAM/MAM at a given time. For...
Reference this page
Michael Golden (). Comment on: Including infants in nutrition surveys. Field Exchange 9, March 2000. p15. www.ennonline.net/fex/9/comment
(ENN_3422)