Prediction of childhood obesity by infancy weight gain: an individual-level meta-analysis

To assess the predictive ability of infant weight gain on subsequent obesity we performed a meta-analysis of individual-level data on 47 661 participants from 10 cohort studies from the UK, France, Finland, Sweden, the US and Seychelles. For each individual, weight SD scores at birth and age 1 year were calculated using the same external reference (British 1990). Childhood obesity was defined by International Obesity Task Force criteria. Each +1 unit increase in weight SD scores between 0 and 1 year conferred a twofold higher risk of childhood obesity (odds ratio = 1.97 [95% confidence interval (CI) 1.83, 2.12]), and a 23% higher risk of adult obesity (odds ratio = 1.23 [1.16, 1.30]), adjusted for sex, age and birthweight. There was little heterogeneity between studies. A risk score for childhood obesity comprising weight gain 0-1 year, mother's body mass index, birthweight and sex was generated in a random 50% selection of individuals from general population cohorts with available information (n = 8236); this score showed moderate predictive ability in the remaining 50% sample (area under receiving operating curve = 77% [95% CI 74, 80%]). A separate risk score for childhood overweight showed similar predictive ability (area under receiving operating curve = 76% [73, 79%]). In conclusion, infant weight gain showed a consistent positive association with subsequent obesity. A risk score combining birthweight and infant weight gain (or simply infant weight), together with mother's body mass index and sex may allow early stratification of infants at risk of childhood obesity.