An improved algorithm to harmonize child overweight and obesity prevalence rates

BACKGROUND: Prevalence rates of child overweight and obesity for a group of children vary depending on the BMI reference and cut‐off used. Previously we developed an algorithm to convert prevalence rates based on one reference to those based on another. OBJECTIVE: To improve the algorithm by combining information on overweight and obesity prevalence. METHODS: The original algorithm assumed that prevalence according to two different cut‐offs A and B differed by a constant amount [Formula: see text] on the z‐score scale. However the results showed that the z‐score difference tended to be greater in the upper tail of the distribution and was better represented by [Formula: see text] , where [Formula: see text] was a constant that varied by group. The improved algorithm uses paired prevalence rates of overweight and obesity to estimate [Formula: see text] for each group. Prevalence based on cut‐off A is then transformed to a z‐score, adjusted up or down according to [Formula: see text] and back‐transformed, and this predicts prevalence based on cut‐off B. The algorithm's performance was tested on 228 groups of children aged 6–17 years from 20 countries. RESULTS: The revised algorithm performed much better than the original. The standard deviation (SD) of residuals, the difference between observed and predicted prevalence, was 0.8% (n = 2320 comparisons), while the SD of the difference between pairs of the original prevalence rates was 4.3%, meaning that the algorithm explained 96.7% of the baseline variance (88.2% with original algorithm). CONCLUSIONS: The improved algorithm appears to be effective at harmonizing prevalence rates of child overweight and obesity based on different references.