Estimation of newborn risk for child oradolescent obesity: lessons fromlongitudinal birth cohorts

Childhood obesity should be ideally prevented as early as possible during the child life. Assessing the inborn risk of obesity development would be the first step towards timely focused prevention. During my doctoral fellowship, I aimed at building clinical tools to predict the newborn risk for obesity development. To this purpose, I analyzed a unique prospective Northern Finland Birth Cohort 1986 (N=4032) (http://kelo.oulu.fi/NFBC) to assess the predictive value for child and adolescent obesity phenotypes of inborn traditional risk factors, like parental BMI, birth weight, maternal gestational weight, behaviour and social indicators, as well as of a genetic score issued from 39 BMI/obesity-associated polymorphisms. I also analysed a retrospective sample of 1,503 children aged 4-12 from Veneto, Italy, previously used in a survey assessing prevalence and risk factors of childhood obesity in the North of Italy, as a validation sample, in order to explore whether results issued from the NFBC1986 could be applied to a European paediatric cohort contemporary to the NFBC1986, with similar obesity prevalence (4%) but different cultural background. Finally, I analysed a prospective sample of 1032 children (7 years) from Massachusetts (United States) participating in the Project Viva (http://www.dacp.org/viva/index.html) as additional validation sample, in order to explore whether results issued from the NFBC1986 would remain valid when applied to a very recent U.S. child cohort, with higher obesity prevalence (8%) and very different cultural background. In the NFBC1986, I found a fair to good cumulative discrimination accuracy of traditional risk factors for the prediction of child obesity, adolescent obesity, and child obesity persistent into adolescence (0.75 < AUROCs < 0.85, p<0.001). The genetic score alone showed poor accuracy (0.55 < AUROCs < 0.60, p<0.001) for the prediction of obesity outcomes and, combined with clinical predictors, it only produced integrated discrimination improvements (IDI) < 1%. The version of the NFBC1986 equation for childhood obesity lacking gestational smoking and number of household members (not available in the Veneto dataset) had an AUROC = 0.70[0.63-0.77] (p < 0.001) when applied to the Veneto cohort versus an AUROC = 0.73[0.69 – 0.77] in the NFBC1986, with acceptable calibration accuracy (p for Hosmer-Lemeshow test > 0.05). The NFBC1986 equation for childhood obesity had an acceptable AUROC = 0.73[0.67-0.80] (p < 0.001) when applied to the project Viva children. However calibration in the Project Viva sample was not satisfactory (p for Hosmer-Lemeshow test = 0.02). The study I present here provides the first evidence that routinely available inborn variables may be combined into handy and cheap scores to estimate the newborn risk for obesity. An accurate estimation of the inborn risk to develop obesity may have important public health implications, being the basis for further efforts towards potential precocious prevention based on the “high risk approach”, i.e. focused on families of high risk newborns. The models I describe exploit variables usually easy to record retrospectively and strongly and consistently associated with obesity in several countries. Therefore, it is likely that reproduction/implementation or adaptation elsewhere can be successfully performed in a short time. This study also rules out the hypothesis that currently known polymorphisms may provide any useful contribution to accuracy of child obesity prediction.