Preventing childhood obesity with combined hypothetical interventions on interpregnancy maternal smoking and weight: applying the parametric g-formula to a UK population-based cohort

Abstract

BackgroundMaternal exposures during pregnancy have been linked to offspring adiposity, and the interpregnancy period is an opportunity to intervene on modifiable risk factors. We considered combinations of interpregnancy interventions and effects on rates of childhood obesity (≥95th centile) in second children. MethodsWe examined records of 5612 women from a population-based cohort of routinely collected antenatal health-care records at University Hospital Southampton (2003–2014), linked to measured child body-mass index (BMI) at age 4–5 years from the National Childhood Measurement Programme. We applied the parametric g-formula to estimate the effects of interventions between first (P1) and second (P2) pregnancies and to calculate estimated second child obesity and population risk ratio reduction (PRRR), compared with the estimated natural-course scenario (without intervention). Analysis was undertaken using the gfoRmula package in R This analysis is part of the SLOPE study which has been approved by the National Health Service Health Research Authority. FindingsObserved frequency of obesity in the second child at age 4–5 years was 8·5% (479 of 5612 children). We estimated that if all mothers started P2 with a BMI of 18·5–24·9 kg/m2 and all smokers stopped smoking by confirmation of P2, then, compared with a natural course estimate of 8·3% (95% CI 7·6–9·1), there would be a PRRR of second child obesity of 25·8% (21·0–30·8) to a second child obesity prevalence of 6·2% (5·5–7·0). For mothers who started P1 with BMI 18·5–24·9 kg/m2, if all smokers stopped smoking, then, compared with a natural course estimate of 5·9% (5·0–6·7), there would be a PRRR of second child obesity of 16·6% (10·0–23·2) to a prevalence of 4·9% (4·1–5·7). If BMI was also maintained between 18·5 and 24·9 kg/m2 there would be a PRRR of 21·3% (15·3–28·2) to a prevalence of 4·6% (3·9–5·4). For mothers who started P1 with BMI of 30 kg/m2 or greater, if weight was reduced to being either 18·5–24·9 kg/m2 or 25–29·9 kg/m2 by the start of P2, then, compared with a natural course estimate of 17·7% (15·1–20·9), there would be PRRRs of second child obesity of 29·4% (4·7–53·4) to a prevalence of 12·5% (8·3–17·4) and 17·6% (2·6–35·1) to a prevalence of 14·6% (11·1–17·7), respectively. InterpretationInterventions between pregnancies could reduce obesity rates in the second child, with the most effective interventions varying by maternal BMI during the first pregnancy. FundingNIHR Southampton Biomedical Research Centre, University of Southampton Primary Care, Population Sciences and Medical Education PhD studentship, and an Academy of Medical Sciences and Wellcome Trust grant.