Diet and physical activity in pregnancy to prevent gestational diabetes: Individual Participant Data (IPD) meta-analysis on the differential effects of interventions with economic evaluation

BACKGROUND

Mothers with gestational diabetes (glucose intolerance diagnosed for the first time in pregnancy) are at high risk of complications in pregnancy, and of type 2 diabetes in the long term. Diet and physical activity interventions have the potential to prevent gestational diabetes. However, numerous trials (> 60) and aggregate meta-analyses (>10) have not led to clear implementable recommendations due to variations in population, interventions and outcomes. Evidence synthesis using the individual participant data (IPD) can address these the limitations.

AIM

To determine using IPD meta-analysis, whether diet and physical activity-based interventions prevent gestational diabetes, and its complications, in all or in specific groups of women, and identify the costs and components of effective interventions through our NIHR-funded i-WIP (International Weight Management in Pregnancy) database.

OBJECTIVES Primary To assess the

1. overall effects across all interventions, and according to each type (diet, physical activity, mixed), on gestational diabetes as defined by NICE (National Institute of Health and Care Excellence) and by any criteria

2. differential effects on subgroups based on maternal body mass index (BMI), age, parity, ethnicity, and socioeconomic status Secondary To evaluate 3. the effects of interventions on critically important offspring and maternal outcomes 4. variation in effects by intervention components 5. the effects on gestational diabetes diagnosed using specific criteria

6. the cost-effectiveness of interventions METHODS IDENTIFICATION OF NEW TRIALS AND DATA ACQUISITION

We already have standardised relevant IPD of 37 trials (17,106 women, 16 countries) in our i-WIP database. A further 21 trials (7,660 women) have agreed to share IPD and are in the process of transferring the data. QUALITY ASSESSMENT AND DATA HARMONISATION

We will add new studies and variables to the existing i-WIP database, assess study quality, harmonise new data, and extract data on the diagnostic blood glucose values, intervention components, and outcomes. DATA SYNTHESIS

We will perform 1-stage and 2-stage random effect IPD meta-analyses to obtain the pooled intervention effect for primary and secondary outcomes. We will examine whether maternal characteristics (covariate values) modify effect by extending the meta-analysis framework to summarise treatment-covariate interaction terms. Our cost effectiveness analysis will be primarily based on outcomes for a hypothetical cohort of 10,000 pregnant women derived from the IPD meta-analysis. A secondary analysis will compare costs and outcomes for subgroups of women.

SAMPLE SIZE

Of the 27,538 women (71 studies) with eligible data, we have a sample size of 24,766 women (58 studies shared/agreed IPD). This provides us with 90% or higher power to detect any interactions between overall effects and most subgroups for a 30% reduction in odds of gestational diabetes; for a 25% reduction in gestational diabetes our IPD sample size accrued so far has over 80%, and around 75-80% for most other subgroups.

TIMELINE 18 months ANTICIPATED IMPACT AND DISSEMINATION

We will liaise with NHS England Directors, UK Chief Medical Officers, NICE Guideline leads, NHS providers, local authorities, Charities and patient groups to enable implementation of findings into practice. We plan to disseminate via journals, conferences, blogs, social media.