Letter on WHO 2006 Growth Standards, by Marko Kerac and Andrew Seal

This new 2006 WHO Growth standards: What will they mean for emergency nutrition programmes?

Dear Editor

Whilst welcoming the principles which have driven the development of the new 2006 WHO growth standards (see news piece this page), we wish to draw attention to important practical implications for emergency nutrition programmes. We think it is important that these are explored and discussed in detail before the new standards are implemented in operational settings.

The need for new growth standards

An internationally valid, 'gold standard' range against which child growth can be assessed has long been needed. There are several reasons why the previous NCHS (National Centre for Health Statistics)/ WHO Reference data fell short of this ideal:

1. It was constructed on a cohort of North American children, from a single community and a single ethnic group of European ancestry.
2. Data was gathered from 1929- 1975, a long period during which nutrition varied greatly. The main issue of concern was that infants were pre-dominantly bottle-fed rather than breastfed, as is considered ideal today.
3. Statistical methods have advanced significantly since the original NCHS/WHO growth curves were constructed in the 1970's. Applying better statistical techniques to the same dataset was what led to the CDC 2000 growth references.
4. Increasing numbers of studies in both developed and developing country settings found that apparently healthy, breastfed children were being labelled as abnormal according to the NCHS/WHO References.

MGRS (Multi-centre Growth Reference Study)

The MGRS¹ ran from 1997-2003 and was explicitly designed to generate a growth standard to show how children should grow, rather than just a reference that allows comparison. Following extensive screening to select only those children free of health or environmental (socio-economic/ nutritional) constraints to growth, a total of 8,440 children were observed at six international sites (Brazil, Ghana, India, Oman, Norway, USA). The study had two components: longitudinal work followed children from birth to 24 months; a cross-sectional study observed children from 18-71 months. State-of-the-art statistical techniques were chosen to construct growth curves from this data. Key outcomes from the MGRS are:

1. The strongest evidence yet that a single international child growth standard is valid. Free of environmental and nutritional constraints, children of very different ethnic groups all grew the same: only 3% of length variance was due to inter-site differences.
2. New z-score and percentile refer ences charts/tables for weightforage, length/height-for-age, and weight-for-length/height.
3. Additional standards not present in NCHS/WHO Reference: Body Mass Index (BMI); Mid upper arm circumference (MUAC); skin-fold thickness; and motor developmental milestones.
4. A devoted website with extensive literature relating to MGRS and the new standards.
5. Free downloadable software which may, in the future, enable both individual and population anthropometric status to be calculated using either NCHS/WHO Reference or WHO Standard data.

Differences between the old and new growth curves

There are important differences between the old references and the new standards. There is however no easy or consistent way of transforming anthropometric measures between the two: the growth lines do not run in parallel with simple shifts up or down. Factors affecting the magnitude and direction of differences between old and new cut-offs include: a child's age; a child's length/height; which measure (i.e. WHZ; WAZ or HAZ) is being considered; whether the child under consideration is above or below median; and whether the z-score or % of median is being considered. As an example, shown below are the weight-for-age percentile lines (P) for boys between 0 and 36 months. The curves cross, sometimes more than once, illustrating that the magnitude and direction of the difference between the NCHS/WHO Reference and the WHO Standards is dependent on the age of the children and his location on the distribution.

In short, the net effect of the new standards on the measurement and diagnosis of growth and malnutrition is complex!

Implications for emergency nutrition assessments and feeding programmes

1. Comparability and interpretation of nutrition data

Interpreting trends in nutritional status and setting agreed thresholds for action are important for emergency nutrition programmes. With the adoption of the new WHO standards the ability to easily compare the results of current surveys with previous data will be lost, and this will make new data more difficult to interpret.

This problem could be overcome by allowing for a period of dual-analysis of survey data. If results from surveys are analysed using both the new WHO Standards and the currently used NCHS/WHO Reference, then sufficient data and experience may be built up with the new system whilst assuring 'backwards compatibility'. Though potentially complex and confusing for non-specialist policy-makers, this approach would eventually enable trend and risk models to be recalibrated and appropriate new action thresholds set. However, a note of caution must be added. Although software is available from the WHO web site that can be used to analyse surveys (WHO Anthro 2005), at the moment it does not deal with cases of oedema in the standard way, making calculation of the correct estimates of Global Acute Malnutrition (GAM) and Severe Acute Malnutrition (SAM) difficult.

2. Prevalence assessments using z-scores

Weight-for-length/height is a key anthropometric measure for emergency feeding, widely used in malnutrition prevalence surveys to assess the need for, or effect of, a nutrition programme. It is therefore important to know what are the expected effects of the WHO Standards on the measured prevalence of GAM and SAM.

. Effect on SAM ~ a marked increase
Overall, the new WHO standards will increase the measured prevalence of SAM through increasing the value of the weight for height <-3 z-scores cut-off. The difference is maximal (1kg) for infants with lengths around 60 cm. As length increases, this difference narrows and from 76.5- 86cm length, the NCHS/WHO Reference cut-off is temporarily higher by about 0.2 kg. With yet further height increase, the WHO Standard cut-off again becomes greater. There is a second peak difference of 0.6 kg for children above 100 cm.

. Effect on GAM ~ small but unpredictable
For GAM the weight for height cut-off is <- 2 z-scores. The pattern of differences for -2 z-score follows that for -3 z-scores but, due to their different magnitudes, may lead to an increase or decrease in prevalence depending on the height profile of the surveyed populations and the relative contribution of SAM to GAM.

3. Feeding programme enrolment and discharge

. Feeding programme enrolment if using % of median
The percentage of the median has long been the de facto field measurement for the admission of children to selective feeding programmes (therapeutic for SAM; supplementary for moderate acute malnutrition). As yet, WHO do not provide % of the median tables for use with the new Growth Standards. However, if these are tabulated something unexpected is observed. Paradoxically, in contrast to what we saw with z-scores, when the new standards are applied in percentage of the median measurements, there is an overall decrease in the cut-offs for acute malnutrition. This would therefore lead to a decrease in the numbers of children eligible for selective feeding. Both the 70% and 80% curve follow the same pattern.

. Feeding programme discharges if using % of median
Discharge from selective feeding programmes typically takes place when patients achieve 80 or 85% of their median weight for height. With application of the WHO Standards, these cut-offs will be reached at a lighter weight so children will, on average, be discharged earlier. The effects of this change on relapse, re-admission or case fatality rates are currently unknown.

. Feeding programme enrolment if using z-scores
If agencies move to using z-scores based on the WHO Standards as entry criteria for selective feeding programmes what changes in admissions can be expected? We mentioned above that the -3 z-score cut-offs have increased with the new standards meaning that more children will be diagnosed as severely malnourished and therefore eligible for admission to therapeutic feeding. To assess what magnitude of difference this could entail we looked at historical data from 3 refugee operations in Africa and Asia. The numbers of children eligible for admission to therapeutic feeding increased by between 500 and 600%². If programmes plan to use the new WHO Standard z-score cut-offs they may need to plan for at least a 5 fold increase in patient load.

. Feeding programme discharges if using z-scores
If patients are discharged at -2 z-scores then the impact on average treatment duration in any programme will be variable, depending on the particular height profile of the population. Conclusions The new WHO standards represent significant theoretical advantages over the old NCHS/WHO growth references. However, their introduction poses a number of potentially serious operational challenges, which, in the opinion of the authors, have not yet been adequately discussed or addressed.

In emergency settings, the likely effect is a great increase in the diagnosis of SAM, and a possible increase in GAM - if assessed by <-3 and <-2 Z-scores respectively. This might be seen as a great opportunity to enrol more children in therapeutic feeding programmes (TFPs). However, if this line is pursued the funding implications and possible diversion of resources away from food security, livelihoods, and other public health interventions need careful consideration.

There is however another, perhaps more serious possibility: if % of median remains the field programme admission criteria, significantly less children might be admitted to programmes. This risks confusion between different programmes, misallocation of resources, and potentially harmful impacts on clinical care and public health outcomes. It is crucial that operational agencies work to achieve a consensus on the way ahead.

Recommendations

We believe that the new WHO standards represent both great opportunities but also great challenges for emergency nutrition. They should not be implemented in haste. We call for a body comprising UN and NGO implementing agencies to be rapidly established to coordinate a response to this operational challenge.

Sincerely,

Marko Kerac, Valid International and College of Medicine, Blantyre, Malawi, and Andrew Seal, Institute of Child Health, London, UK

¹www.who.int/childgrowth/en/

²unpublished data