Quantifying child growth effects using height-age instead of height-for-age z-scores in a meta-analysis of small-quantity lipid-based nutrient supplement trials

Abstract

Abstract Height-age is the age at which growth-faltered children’s average observed height or length equals the median height or length of a child growth standard, corresponding to a length-for-age z-score (LAZ) of 0. In randomized controlled trials (RCTs) in low- and middle-income countries (LMICs), expression of linear growth outcomes using height-age may enhance the interpretability of intervention effects compared to conventional use of LAZ. Height-age can be used to derive the proportion of maximal benefit (PMB), whereby PMB = 0% indicates no effect and PMB = 100% indicates the intervention promoted growth at the rate expected for healthy children with the same starting height-age. In this proof-of-concept study, height-age and PMB were compared to LAZ in a meta-analysis of RCTs of small-quantity lipid-based nutrient supplements (SQ-LNS). Pooling across 15 trials in 10 LMICs, mean differences (MD; SQ-LNS minus control) in LAZ and height-age were 0.15 (95%CI: 0.12, 0.17) and 12 days (95%CI: 9, 14), respectively (N = 36,970). LAZ MD and height-age MD were highly correlated (rho = 0.74 overall and 0.94 upon exclusion of an outlier). The pooled PMB indicated that SQ-LNS achieves 11% of optimal growth potential (95% CI: [9.4, 12]; N = 19,768; 12 comparisons), but there was a substantial impact of between-trial heterogeneity (I 2  = 90%). In conclusion, the effect of SQ-LNS on linear growth can be alternatively expressed in terms of height-age instead of LAZ. The PMB may enhance the interpretability of effect estimates by quantifying the extent to which an intervention improves growth in relation to a biological threshold, but further research is required to establish its validity and usefulness for assessing and comparing intervention effectiveness.