Almost 30% of women with gestational diabetes (GDM) fail to achieve glycemic control with lifestyle modification alone and require medication treatment. Currently, the American Diabetes Association recommends insulin as the first line medication treatment for women with GDM. However, oral agents have several advantages including ease of use and acceptance among patients. The safety and efficacy of glyburide for the treatment of GDM remains controversial. We aimed to compare the effectiveness of glyburide and insulin in improving perinatal outcomes. We used rigorous causal inference methods with time-varying exposures and machine learning to evaluate perinatal outcomes of exposure to glyburide versus insulin during pregnancy in a retrospective cohort study of 11,321 women diagnosed with GDM who required medication treatment to achieve glycemic control in Kaiser Permanente Northern California from 2007-2017. For each outcome, Inverse Probability Weighting (IPW) estimation was used to separately compare several perinatal outcomes between glyburide versus insulin initiators. This approach combined with Super Learning for propensity score estimation can account for both baseline and time-dependent confounding in per protocol analyses to evaluate the effect of continuous exposure to the same therapy. Women who used glyburide versus insulin had similar rates of macrosomia, hypoglycemia, hyperbilirubinemia, and cesarean delivery; significantly lower rates of a NICU admission (risk difference -0.03 [95%CI:-0.06;-0.002]) or respiratory distress syndrome (-0.02 [-0.04;-0.001]) and suggestive higher rates of an appropriately sized for gestational age infant (0.04 [-0.002;0.08]) after adjusting for baseline and time-varying covariates. Using data from a real-world clinical setting we find no evidence to suggest that adverse outcomes were higher among glyburide users compared with insulin users.


M.M. Hedderson: None. A. Ferrara: None. A. Regenstein: None. R. Neugebauer: None.


Permanente Medical Group

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