Stunted child/overweight mother pairs represent a statistical artifact, not a distinct entity

Stunted child/overweight mother (SCOWT) pairs have been viewed as a distinct phenomenon and much effort has been expended to characterize the environments that create these dual burden households. We hypothesized that the prevalence of SCOWT pairs is not independent of the prevalence of overweight mothers and stunted children in the general population. We analyzed data from the Demographic and Health Surveys (DHS) from 1991 to 2009. Datasets were included if the maternal BMI and the height-for-age Z-scores for children were reported. Mothers were included if they had a living child between 2 and 5 y old and were not currently pregnant. In 121 datasets from 54 countries, 339,202 households met the inclusion criteria. The median prevalences of maternal overweight, childhood stunting, and SCOWT pairs were 19.6% (range 1.6-70.7%), 27.3% (range 6.65-50.8%), and 3.3% (range 0.5-16.0%), respectively. The mean difference between the observed and expected prevalence of SCOWT pairs was -1.18% (95% CI -1.32%, -1.04%). Only two datasets had an observed prevalence of SCOWT pairs that was higher than the expected prevalence, but both were within the 95% CI for the observed prevalence. SCOWT prevalence was more strongly associated with maternal overweight than with child stunting. SCOWT pairs are not independent and their prevalence depends primarily on the prevalence of maternal overweight.