Speaker: Dr Jennie Louise (University of Adelaide)
Dr Louise was once a philosopher (ethical theory /logic), but she has recovered now. Having retrained in biostatistics, she now works in the DDASS/AHTA, School of Public Health at the University of Adelaide. Most of her work involves providing statistical support for large clinical trials and longitudinal studies of pregnancy and early childhood. The topic of her talk was ‘Gestational weight gain: modifiable cause of adverse maternal/infant outcomes?’.
Dr Louise firstly introduced background regarding the 2009 Institute of Medicine (IOM) guidelines for weight gain in pregnancy, which recommended ranges for ‘optimal weight gain’ in pregnancy. The optimal ranges differ by pre-pregnancy BMI. The recommendations were based on observational evidence, and advised that women should ‘gain weight within the recommended range’, and focus on reducing gestational weight gain (GWG) in overweight and obese women. The guidelines have been widely adopted worldwide.
Dr Louise pointed out that although IOM recognised multiple determinants of GWG, e.g. physiological, societal/institutional, community and interpersonal, genetic and epigenetic, pre-existing health conditions, etc, the recommendations were focused on behavioural. There are three assumptions for such a focus (which we need to think again whether they are appropriate): 1) GWG is modifiable by behavioural change; 2) the association between GWG and adverse outcomes is causal; 3) the association between GWG and outcome is modified by pre-pregnancy BMI.
She investigated the results from two large randomised controlled trials (RCT), the LIMIT study and the UPBEAT study, and the IPD meta-analysis on 36 RCTs. What we could conclude from these studies are: 1) It is very difficult to change GWG with diet and lifestyle interventions. If effect was demonstrated, despite improvement in diet and physical activity, the effect was very modest. 2) Even when GWG is reduced, this does not improve maternal or neonatal outcomes. 3) The pre-pregnancy BMI does not appear to modify the effects of GWG.
Dr Louise then dug into this seemly straightforward, but indeed a very complex issue, with her exceptional logical thinking, examining these assumptions and tangled underlining associations using mediation models. Previously, the hypothesis was that lifestyle could have effect on GWG, and consequently on pregnancy outcomes. She thinks we need to consider other factors, such as common causes of both GWG and relevant outcomes. For answering questions, such as ‘Is the effect of BMI mediated through GWG, and if so, to what extent?’, a few mediation models of GWG, GA (gestational age), and GDM (gestational diabetes mellitus) on birthweight were tested.
The results show that: 1) GWG is an inconsistent mediator; 2) Direct effect of BMI is to increase birthweight; 3) Mediated effect is to decrease birthweight via decrease in GWG. However, the causal question regarding GWG is still unanswered. The further difficulties and next steps were discussed: 1) develop models for each ‘exposure’ and each outcome; 2) bias due to missing data; 3) investigate mediation models with random effects in larger datasets; 4) models for binary outcomes (cannot use this method with binary mediator); 4) interactions between exposures and between mediators (though no evidence for this so far).
By Rosie Meng