Do pregnancy outcomes differ by mode of insulin delivery for type 1 diabetes mellitus?


      Type 1 diabetes mellitus (T1DM) in pregnancy increases maternal and neonatal complications. These complications are related to degree of hyperglycemia and can be reduced with insulin treatment, via multiple daily injections (MDI) or continuous subcutaneous insulin infusions (CSII). CSII appears superior to MDI in non-pregnant patients, but it is unclear whether CSII confers clinical benefits in pregnancy. This study investigated whether mode of insulin delivery affects risk of complications among pregnant individuals with T1DM, in a contemporary cohort accounting for disease severity.

      Study Design

      All T1DM singleton pregnancies delivered at UCSF from 01/2008 to 07/2020 were included. The primary predictor was mode of insulin delivery, the primary maternal outcome was primary cesarean delivery (PCD), and the primary neonatal outcome was NICU admit. Baseline characteristics were compared using Fisher’s exact and Kruskall-Wallis tests. Odds ratios (OR) were obtained using a generalized estimating equations model to account for individuals with > 1 pregnancy during the study period. Early A1c and time since T1DM diagnosis were included in adjusted analyses as markers of disease severity.


      129 individuals with 153 pregnancies were included, with 40% using MDI and 60% using CSII. There was no difference in PCD between the groups (51% CSII vs 49% MDI). CSII was associated with a significant reduction in the odds of NICU admit (57% vs 87%, p < 0.01), even after adjusting for early A1c, time since diagnosis, preterm birth, and year of delivery (adjusted OR 0.33, 95% CI 0.12-0.90).


      When used for T1DM in pregnancy, CSII reduced the odds of NICU admit by 67%. Currently, shared decision-making is utilized to determine best mode of insulin delivery, as there is no compelling evidence to promote use of one mode over the other in pregnancy. These findings provide preliminary support for improved neonatal outcomes with CSII and, if replicated in larger, multicenter cohorts, may encourage use of CSII over MDI.
      Figure thumbnail fx1
      Figure thumbnail fx2