Oral Presentation ANZOS-Breakthrough Discoveries Joint Annual Scientific Meeting 2018

Predicting early childhood obesity at infancy: a model for the New Zealand population (#72)

Éadaoin M Butler 1 2 , José G B Derraik 1 2 3 , Rachael W Taylor 2 4 , Susan M B Morton 2 5 , Marewa Glover 2 6 , El-Shadan Tautolo 2 7 , Wayne S Cutfield 1 2
  1. Liggins Institute, University of Auckland, Auckland, New Zealand
  2. A Better Start - National Science Challenge, New Zealand
  3. Department of Women's and Children's Health, Uppsala University, Uppsala, Sweden
  4. Dunedin School of Medicine, University of Otago, Dunedin, New Zealand
  5. Centre for Longitudinal Research–He Ara ki Mua, University of Auckland, Auckland, New Zealand
  6. School of Public Health, College of Health, Massey University, Auckland, New Zealand
  7. Department of Public Health and Psychosocial Studies, Auckland University of Technology, Auckland, New Zealand

Background

One in three children in New Zealand are overweight or obese by the time they start school. Internationally, several prediction models of early childhood obesity have been developed, but none exist for New Zealand's diverse population. Thus, the aim of this study was to develop, and validate, an early childhood obesity prediction model using data obtained in infancy, for use in the New Zealand population

Methods

A prediction model was developed using data from the Growing up in New Zealand (GUiNZ) prospective cohort (N=6,853). The GUiNZ cohort was randomly split into derivation (70%), and validation (30%) populations. The model was also externally validated in two different longitudinal New Zealand-based cohorts: Prevention of Overweight in Infancy (POI) and the Pacific Islands Families Study (PIF).

Results

The derivation population consisted of 1,731 children. The following parameters were included in the final model: gestational age, maternal smoking during pregnancy, birth weight, maternal and partner BMI, and accelerated infancy weight gain. The model's discrimination was adequate, with an area under the receiving operating characteristic curve (AUROC) of 0.75 (0.72-0.78). The model's sensitivity was 70.1%, specificity was 65.6%, positive predictive value (PPV) was 27.7%, and negative predictive value (NPV) was 92.1%. 713 children were used for internal validation, and the AUROC produced was also adequate at 0.73 (0.68-0.78). Discrimination remained adequate for PIF (AUROC= 0.74 [0.66-0.82]), but improved for POI (AUROC= 0.83 [0.71-0.90]).

Conclusions

We have developed and validated a model, using birth and infancy data, for the prediction of early childhood obesity in the New Zealand population. The model's parameters are easily obtainable, thus we propose that use of this model could support targeted interventions to prevent early childhood obesity in New Zealand.