Fig. 1.

Adjusted relative risk of in-hospital mortality and morbidities among MPT infants by maternal insulin receipt and timing of maternal insulin use*. *Generalized linear regression models were used for adjusted analysis of continuous outcomes. Poisson regression models with robust variance estimators were used for adjusted analysis of categorical outcomes. All models were adjusted for study center, infant gender, gestational age at birth, antenatal steroid use, maternal age, and maternal race/ethnicity, unless mentioned otherwise. Model for PVL not adjusted for center because of convergence issue. IBP, insulin before pregnancy; IDP, insulin during pregnancy; IVH, intraventricular hemorrhage; NEC, necrotizing enterocolitis; No IDDM, no insulin-dependent diabetes mellitus; PDA, patent ductus arteriosus; PMA, postmenstrual age; PVL, periventricular leukomalacia.