Structured, exercise-only interventions reduce the odds of GDM, but motivated trial participants are likely unrepresentative. Studies of free living exercise in population-based cohorts are needed and causal inference methods, which emulate randomized trials under specific a priori assumptions, represent the best alternative for observational data. This study estimates the causal effects of first trimester exercise on the risks of abnormal screening after a 50-OGTT, GDM diagnosed by the Carpenter-Coustan criteria, and abnormal glucose according to the Carpenter-Coustan thresholds for the 100-g OGTT. Data come from PETALS, a prospectively followed pregnancy cohort (n= 2,070, 70% from minority racial-ethnic groups) receiving care at Kaiser Permanente Northern California in 2013-2017. A Pregnancy Physical Activity Questionnaire was completed at 12.8 weeks gestation, on average (SD 2.5). Targeted Maximum Likelihood with data-adaptive estimation (a.k.a., machine learning) of propensity scores and outcome regression with recognized confounders (e.g., prepregnancy BMI, diet, education) was used to obtain causal risk differences for all women having exercised versus not. Exercise was defined as performing any amount of vigorous intensity exercise or meeting the cohort’s 75th percentile for moderate to vigorous intensity exercise (MVPA; ≥ 13.3 MET hours per week). The prevalence of abnormal screening was 23.8% and GDM 6.2%. Any participation in vigorous exercise significantly decreased the risk of GDM by 1.6 fewer cases per 100 women, and abnormal 1-hour and 2-hour OGTT values by 2.7 and 1.8 fewer cases, respectively, per 100 women. Meeting the 75th percentile for MVPA significantly decreased the risks of abnormal screening and abnormal 2-hour OGTT values by 5.2 and 2.5 fewer cases per 100 women, respectively. Findings suggest that, during the first trimester, any amount of vigorous intensity exercise or a high volume of moderate to vigorous intensity exercise reduces hyperglycemia later in pregnancy.


S.F. Ehrlich: None. R. Neugebauer: None. J. Feng: None. M.M. Hedderson: None. A. Ferrara: None.


National Institute of Diabetes and Digestive and Kidney Diseases (K01DK105106, P30DK092924); National Institute of Environmental Health Sciences (R01ES019196)

Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at