We have described the differences in the cut-offs for acute malnutrition between the NCHS reference and WHO standard datasets. If WHO standards are adopted in nutrition programmes without critical review and careful consideration of consequences, several potentially serious implications are likely.
Firstly, making interpretations of the seriousness of a nutritional crisis and the need for a response will be difficult where comparable trend data are not available. Decisions should remain rooted in an understanding of seasonal, regional, and annual variations in the prevalence of malnutrition, and these are only realistically possible with data generated by the continued or parallel use of the NCHS reference.16
Given time, comparable trend data could be generated if a dual analysis approach is adopted. Data on the relative risk of mortality associated with cut-offs based on the new WHO standards are not currently available, and established risk models will need to be recalibrated using these data.
Secondly, overall estimates of the prevalence of global acute malnutrition, and particularly of severe acute malnutrition, obtained from surveys analysed with the WHO standards are likely to be higher than estimates obtained with the NCHS reference cut-offs. This raises the potential for the misdirection of resources between emergency situations if data generated using different diagnostic criteria start to be reported to governments and donors.
Possible implications for clinical admissions
The third potential problem concerns the admission of children to selective feeding programmes. If the WHO standards are introduced and used according to current practice, although the prevalence of acute malnutrition measured by surveys will be higher, admissions to selective feeding (therapeutic and supplementary) programmes will, if still done on the basis of percentage of the median, be lower than at present. A subgroup of children admitted under current admission criteria would not be admitted in the future, and those admitted would be discharged sooner than at present. Whether this would result in increased mortality in this population subgroup is unknown.
One option that operational agencies might choose is to switch admission criteria for feeding programmes from the NCHS reference percentage of the median to the WHO standard z score cut-offs. The figures presented in tables 1 and 2 show that, under this scenario, the number of children eligible for selective feeding may increase overall by 1.5 to 2.1 (<80% of the NCHS reference median v <−2 z scores of the WHO standard). The biggest proportionate changes are likely to be seen in the numbers eligible for therapeutic feeding. These estimates need to be confirmed by studying other datasets, but they do raise serious questions about the capacity of established programmes to cope with increased patient load and about the resourcing of new relief operations.
The fourth area of concern is the divergence between measurements reported using the z score and percentage of the median methods. Although this problem has existed for some years with the NCHS reference, if the WHO standards were used to calculate both z scores and percentage of the median indices a greater divergence would exist in the estimates of malnutrition. If these data were used for assessment and planning, the gap between needs assessment (routinely done with z scores) and admissions to feeding programmes (routinely done with percentage of the median) would increase. This increased mismatch would be likely to have a detrimental effect on the planning of programmes, allocation of resources, and effectiveness of interventions.
Opportunity for harmonisation
The introduction of the WHO standards presents an opportunity for the harmonisation of indicators used in prevalence surveys and admissions to and discharges from nutrition programmes. However, the harmonisation of these indicators will need planning, training, and resources. To complicate matters further, published data are lacking on whether field staff in emergencies will have the capacity to use z scores effectively and safely.
An associated problem with the use of the WHO standards is the release of software (WHO Anthro 2005) for the analysis of anthropometric data from individuals and surveys. Although otherwise an excellent software tool, the program fails to separate cases with oedema and account for them as a separate category of severe malnutrition in its summary statistics. If not explicitly recognised by the user, this may have the effect of falsely reducing the reported prevalence of nutritional oedema. Modification of the software, to ensure its consistency with standardised reporting formats, is a prerequisite for efficient analysis of surveys with the WHO standards.
In any interim transition period using a dual analysis system, potential problems will arise if communication efforts are not strengthened to ensure that decision makers are presented with consistent and comparable data. The potential for confusion and misuse exists, thereby risking a reduction in operational effectiveness and equity. Where percentage of the median and z score results are reported, decision makers may now be presented with four different estimates of prevalence to deal with. As the history of the kcal and kJ energy units illustrates well, the international nutrition world is sometimes slow to adapt to change and some changes may be never fully implemented. An intensified effort is needed if this track record is not to be repeated with the introduction of the WHO standards.
The practical implications of adopting the WHO standards need to be thoroughly assessed before operational agencies start to implement programmes that use their weight for height cut-offs. If adoption of the WHO standards in nutrition programmes is to proceed, it should not be piecemeal and haphazard. Implementation needs to be coordinated, and we propose that a body comprising major UN and non-governmental implementing agencies should be rapidly established to coordinate the response to this operational challenge.
What is already known on this topic
- The 1978 NCHS/WHO child growth reference curves are widely used but have some important limitations in their applicability to all populations
- New 2006 WHO growth standards were designed to be a global standard, reflecting optimal growth, nutrition, and development for all children in all countries
What this study adds
- Surveys that use the new WHO standards and a <−3 z score weight for height cut-off will markedly increase the number of children identified as having severe acute malnutrition
- The new standards may lead to a significant reduction in children admitted to feeding programmes, as usual practice is to use a <70% of the median weight for height cut-off for admissions