Early Metformin Initiation and Glycemic Outcomes in Gestational Diabetes

Abstract

Importance Conventional approaches to analyzing composite end points in clinical trials may not fully capture clinically meaningful treatment effects. Objective To estimate treatment effects of early metformin initiation on clinically ordered glycemic outcomes (insulin initiation and fasting hyperglycemia at gestational weeks 32 and 38) using generalized pairwise comparisons (GPCs). Design, Setting, and Participants This secondary analysis of a double-masked, placebo-controlled randomized clinical trial (conducted from June 2017 through September 2022) at 2 centers in Ireland (1 tertiary hospital and 1 regional hospital) included participants with gestational diabetes based on World Health Organization 2013 criteria. Participants were followed up through 12 weeks post partum. The data analysis was performed between October 18, 2025, and February 2, 2026. Intervention Participants were randomized 1:1 to metformin (maximum dose, 2500 mg) or matching placebo, in addition to usual care. Main Outcomes and Measures Generalized pairwise comparisons were performed using (1) a prioritized hierarchy (insulin initiation, fasting hyperglycemia at week 32, and fasting hyperglycemia at week 38) and (2) nonprioritized analyses of each end point, with overall summaries using equal weights and prespecified clinical weights (0.50, 0.25, and 0.25, respectively). Treatment effect sizes were summarized using win ratio (WR), net treatment benefit (NTB), and win odds (WO). Results Of 535 randomized pregnancies, 526 (262 placebo [49.0%] and 264 metformin [51.0%]; mean [SD] age of participants, 34.4 [4.8] years) had primary outcome data. Prioritized GPCs favored metformin over placebo for the composite outcome (WR, 1.46 [95% CI, 1.13-1.89]; NTB, 13.7 percentage points [95% CI, 4.3-22.8 percentage points]; WO, 1.34 [95% CI, 1.10-1.64]) ( P = .004). The largest separation was observed for insulin initiation (WR, 1.68 [95% CI, 1.19-2.38]; NTB, 12.7 percentage points [95% CI, 4.2-21.0 percentage points]) ( P = .003). Nonprioritized overall summaries favored metformin (WR, 1.53 [95% CI, 1.16-2.01]; NTB, 7.5 percentage points [95% CI, 2.7-12.3 percentage points]; WO, 1.32 [95% CI, 1.10-1.57]) ( P = .002), with consistent clinically weighted results (WR, 1.57 [95% CI, 1.19-2.08]; NTB, 8.8 percentage points [95% CI, 3.4-14.2 percentage points]; WO, 1.39 [95% CI, 1.14-1.70]) ( P = .001). Conclusions and Relevance In this secondary analysis of a randomized clinical trial, prioritized and nonprioritized GPCs yielded statistically significant overall estimates favoring early metformin initiation over placebo across insulin initiation and fasting hyperglycemia outcomes. These findings complement the original primary analysis, provide a more clinically interpretable assessment of the composite outcome, and may inform prospective end point design in future trials. Trial Registration ClinicalTrials.gov Identifier: NCT02980276 ; EudraCT: 2016-001644-19

The authors' abstract, as published at the source. JAMA Network Open, 2026 · DOI ↗