EDITORIAL

‘A chart for all reasons’

Growth reference charts are the basic assessment tool of child health whether at an individual, community, or national level. The comparison of a child's height or weight with the distribution of heights or weights of a reference sample provides evidence of the normality or otherwise of the process of growth. A normal, healthy child will grow in parallel to the centile lines whilst a child being subjected to some form of growth constraint will ‘fall’ through the centiles and demonstrate ‘growth faltering’ as the process of growth is compromised by the insult. Interventions to reverse such a trend will depend on the severity and duration of the fall and the identification of any obvious cause. It follows that the ability of the reference chart to identify the child at risk depends on how accurately the chart reflects normal growth in terms of magnitude and rate of change. That accuracy depends in turn on the proximity of the source sample to the child being assessed; the greater the degree of similarity in factors known to affect the growth process the more likely it is that the growth chart will faithfully reflect average growth and its normal variability.

The holy grail of growth assessment is the development of a growth chart that accurately reflects the growth of all normal children. The debate about such a chart has tended to focus on the characteristics of the source sample that would reflect such a universal growth pattern. What, for instance, does ‘normal’ mean? Is ‘normal’ represented by the average child in a particular population who is subjected to the potential constraints upon growth common to children of average socio-economic and nutritional status? Or is ‘normal’ the growth pattern exhibited by the child growing in an environment free from potential growth constraints and thus most likely to be of high socio-economic status? Almost all growth reference charts prior to 2006 (e.g. NCHS, CDC, UK, etc.) adopted the former definition of ‘normal’ and used representative samples of children from North America or Europe as the source samples.

In 2006 the World Health Organization launched a new chart based on the Multicentre Growth Reference Study (WHO 2006). The study had the core aim of providing growth data from 0 to 5 years on a source sample of infants, breastfed according to WHO recommendations, from widely different ethnic backgrounds and cultural settings in Brazil, Ghana, India, Norway, Oman and the USA. The WHO state that, ‘The new growth curves are expected to provide a single international standard that represents the best description of physiological growth for all children from birth to 5 years of age and to establish the breastfed infant as the normative model for growth and development.’ (http://www.who.int/childgrowth/mgrs/en/).

These charts therefore chose the second definition of ‘normal’. The study populations

‘… lived in socioeconomic conditions favourable to growth and where mobility was low, =20% of mothers followed WHO feeding recommendations and breastfeeding support was available. Individual inclusion criteria were: no known health or environmental constraints to growth, mothers to follow MGRS feeding recommendations (exclusive or predominant breast-feeding for at least 4 months, introduction of complementary foods by 6 months of age and continued breastfeeding to at least 12 months of age), no maternal smoking before and after delivery, single term birth, and the absence of significant morbidity.)’ (de Onis et al. 2004).

These criteria meant that the sample to create the longitudinal data was composed of 17% of all those screened, i.e. 83% of infants (11 351 from 13 741) were excluded and 69% of children (>2 years) were excluded from the cross-sectional data (14 813 from 21 510). Thus these standards reflect the growth of 17% of infants and 31% of young children who live in unconstrained environments. It was expected that these exclusion rates would differ by country because it was likely that, for example, more American infants would fulfil the inclusion criteria than, say, Ghanaian children. This situation led to fewer American children being screened to reach the desired sample size for inclusion (398) than Ghanaian children (2057) but similar percentages of American and Ghanaian children in the final sample (275 (11.5%) and 376 (15.7%), respectively).

In addition to understanding the specific nature of the source sample it is also important to appreciate the significance of describing the chart as an ‘international standard’ as opposed to a ‘reference’. The distinction between a growth standard and a growth reference is fundamentally important when interpreting the pattern of growth of the individual child or sample of children.

The WHO standards are based on longitudinal data from 0 to 24 months and cross-sectional and/or mixed longitudinal data thereafter. All other growth references (CDC, NCHS, etc.) are based on cross-sectional data. Cross-sectional growth references, based on a source sample of ‘normal’ children, serve to provide a comparison of the growth of samples of children within say, a country, once it has been agreed that the reference used is accurately reflecting the normal pattern and variation of growth found in that country, i.e. that there are no significant differences between the growth of the source sample and the growth of children in the country of interest. However, the important point is that these are reference charts for use with samples and that they reflect growth ‘as is’ rather than ‘as it ought to be’.

Growth standards are a different tool – they are based on longitudinal data in which the source sample has been selected according to some pre-defined criteria, e.g. social, ethnic, etc. That pre-selection, and the longitudinal nature of the data, means that such charts only reflect the pattern of growth of children with similar characteristics to the source sample. In addition, the variation of the patterns is likely to be less in a longitudinal standard than in a cross-sectional reference because the participants in the former share similar characteristics. The general aim of identifying optimal growth (the growth pattern associated with greatest health) as a basis for a growth standard means that the selection process creates a ‘standard to which all children should aspire’. It does not reflect growth ‘as is’ but growth ‘as it ought to be’.

This distinction in interpretation between references and standards is probably the least understood and poorly appreciated aspect of growth assessment and in practice ‘… references often are mistakenly used as standards.’ (http://www.who.int/childgrowth/standards/Chap_1.pdf). The importance is in the fact that the growth patterns of children from unconstrained environments differ significantly from the growth patterns of children in constrained environments. The main growth characteristics of infants and children from unconstrained environments is relatively lower weight and greater length. Chronic constraints upon growth tend to have their effect through an increased prevalence of stunting and a resultant increase in weight for age. So, a child of average stature and weight from an average family living in average social and economic conditions would be expected to conform to the 50th centile of height and weight reference charts. This same child would not be expected to conform to the 50th centile of the WHO growth standard. It would be between the 50th and 75th centiles for weight and just below the 50th centile for height resulting in a BMI or weight-for-height consistently above the 50th centile. The average child on a growth reference would thus present as being short and heavy on the WHO growth standard.

The second important consideration is the difference in growth charts produced by cross-sectional as opposed to longitudinal data. Descriptive data from longitudinal studies tends to exhibit less variation than data from cross-sectional studies. Thus outer centiles of the distribution tend to be closer to the 50th centile in longitudinal standards than in cross-sectional references. Thus when data from individual children are plotted on a growth reference they are more likely to be extreme than when plotted on a growth standard. Similarly the descriptive data from samples of children are more likely to conform to a growth reference than to a growth standard. Appropriate growth assessment of individuals is thus more accurate when done using growth standards and assessment of samples is more accurate when using references. In both cases the similarity of the child or sample to the source sample of the chart is critical.

In this issue of the Annals of Human Biology we present two examples of the way in which the new 2006 WHO standards are being used (Myatt et al. 2009; Norris et al. 2009).

Mark Myatt and his colleagues have used the WHO standards in an investigation of the relationship between weight-for-height and body shape (sitting to standing-height ratio) and mid-upper arm circumference (MUAC) and body shape in different samples of children from Ethiopia. Their concern is with the different prevalences of acute malnutrition identified when using MUAC as opposed to weight-for-height and their paper concludes that because differences in body shape between samples lead to the potential for overestimation of acute malnutrition, MUAC might be a better indicator than weight-for-height in determining prevalence of acute malnutrition. Weight-for-height Z-score was calculated using both the NCHS reference (Hamill et al. 1979) and the 2006 WHO standard (de Onis et al. 2004). In this study similar results were produced by both sources. Myatt and colleagues include an important discussion on the use of a standard of ideal growth in a situation requiring an indicator of the need for emergency nutritional intervention.

The second paper, from Shane Norris and his colleagues, concerns the implications of using the 2006 WHO standard in the primary care setting of South Africa – a country in the midst of social and economic transition. Prevalences of stunting and wasting are changed by using the WHO standard as opposed to the NCHS reference and Norris et al. conclude that training of primary care staff to recognize and act on the differences is absolutely vital.

In light of the forgoing comments what recommendations should be proposed regarding the use of the new WHO standards?

Firstly, the WHO charts are standards not references and reflect the growth of children growing in unconstrained environments.

Secondly, because they are longitudinal standards the WHO charts (particularly from birth to 2 years) should be used specifically for individuals rather than samples.

Thirdly, they do not purport to represent the growth characteristics of the average child in a population and if used as a growth reference rather than a growth standard they will provide conflicting and erroneous information. This is particularly true when assessing weight-for-height and thus is critical in decisions regarding interventions to prevent overweight; the WHO growth standard will classify more children as being overweight than a growth reference.

When used appropriately the WHO growth standards have a critical role to play in the assessment of normal growth. Because they reflect growth in unconstrained environments they provide a clear picture of the magnitude of the difference between the morphological status of children living in average as opposed to unconstrained environments. They provide a target of what could be achieved if breastfeeding recommendations are followed and the critical factors constraining growth can be removed.