Prognostic accuracy of WHO growth standards to predict mortality in a large-scale nutritional program in Niger.

BACKGROUND

Important differences exist in the diagnosis of malnutrition when comparing the 2006 World Health Organization (WHO) Child Growth Standards and the 1977 National Center for Health Statistics (NCHS) reference. However, their relationship with mortality has not been studied. Here, we assessed the accuracy of the WHO standards and the NCHS reference in predicting death in a population of malnourished children in a large nutritional program in Niger.

METHODS AND FINDINGS

We analyzed data from 64,484 children aged 6-59 mo admitted with malnutrition (<80% weight-for-height percentage of the median [WH]% [NCHS] and/or mid-upper arm circumference [MUAC] <110 mm and/or presence of edema) in 2006 into the Médecins Sans Frontières (MSF) nutritional program in Maradi, Niger. Sensitivity and specificity of weight-for-height in terms of Z score (WHZ) and WH% for both WHO standards and NCHS reference were calculated using mortality as the gold standard. Sensitivity and specificity of MUAC were also calculated. The receiver operating characteristic (ROC) curve was traced for these cutoffs and its area under curve (AUC) estimated. In predicting mortality, WHZ (NCHS) and WH% (NCHS) showed AUC values of 0.63 (95% confidence interval [CI] 0.60-0.66) and 0.71 (CI 0.68-0.74), respectively. WHZ (WHO) and WH% (WHO) appeared to provide higher accuracy with AUC values of 0.76 (CI 0.75-0.80) and 0.77 (CI 0.75-0.80), respectively. The relationship between MUAC and mortality risk appeared to be relatively weak, with AUC = 0.63 (CI 0.60-0.67). Analyses stratified by sex and age yielded similar results.

CONCLUSIONS

These results suggest that in this population of children being treated for malnutrition, WH indicators calculated using WHO standards were more accurate for predicting mortality risk than those calculated using the NCHS reference. The findings are valid for a population of already malnourished children and are not necessarily generalizable to a population of children being screened for malnutrition. Future work is needed to assess which criteria are best for admission purposes to identify children most likely to benefit from therapeutic or supplementary feeding programs.