Alternative metrics for tracking population-level trends in child linear growth
This is a summary of the following paper: Aimone AM, Bassani DG, Qamar H et al (2023) Complementary and alternative metrics for tracking population-level trends in child linear growth. PLOS Global Public Health, 3, 4, e0001766.https://doi.org/10.1371/journal.pgph.0001766
This study explored a range of child linear growth indicators, based on height distributions from anthropometric surveys, which were treated as potential alternatives or complementary metrics to stunting prevalence1 – commonly used to track population-level trends in child nutritional status. Although stunting prevalence is straightforward to estimate, it has several limitations, including that the cut-off threshold has no biological basis and that it does not capture the whole population shift in linear growth faltering common in low- and middle-income countries (LMICs). Therefore, this study aimed to evaluate alternative linear growth metrics based on their correlations with stunting prevalence and 1) under-five mortality, 2) gross domestic product, and 3) maternal education.
The data used for this analysis came from 156 demographic and health surveys conducted in 63 LMICs between 2000 and 2020. The median sample size was 5,461, with sample sizes ranging from 1,290 to 239,588.
The study used Spearman's rank correlation coefficient (r) to identify indicators as alternatives to stunting if they were strongly correlated with stunting (r > 0.95) and at least as strongly correlated with the selected population indicators as stunting. Indicators were considered complementary if they were less strongly correlated with stunting (r < 0.95) but still correlated with population indicators.
The study identified several indicators as potential alternatives to stunting prevalence, including stunting at ages two to five years, mean height-for-age z-score (HAZ), and the 25th percentile HAZ. Six indicators were considered complementary to stunting, such as SITAR-IP2, predicted HAZ at age two and five years, HAZ and height-for-age difference slopes from one month to two years, and growth delay slopes from one month to two years and from two to five years. Three other metrics including predicted HAZ at birth had weak correlations with population indicators (r < 0.43).
Among its limitations, this approach had a narrow scope, and the use of other indicators of health, socioeconomic status, or within-country inequalities may have yielded different conclusions. However, the availability of population-representative measures of some of the other relevant domains (e.g., cognitive development) was limited both in demographic and health surveys and from other data sources. The influence of regression modelling on model-derived metrics, the low number of surveys, and the lack of examination of variations in survey quality present additional limitations.
Future research should focus on the acceptability and interpretability of these metrics by stakeholders. Stunting prevalence, despite its limitations, is well known to policymakers. Adopting alternative and complementary indicators would require defining benchmarks and establishing guidance for their use – but they may offer conceptional advantages.
1 Stunting prevalence is defined as the proportion of children in a population with a HAZ more than 2 standard deviations below the median of the World Health Organization Child Growth Standard.
2 The Super-Imposition by Translation and Rotation Intensity Parameter (SITAR-IP) is a scaling factor that reflects the velocity of a linear growth curve relative to the mean velocity.