Purpose: To examine prospectively the association between parity and subsequent incidence of non-insulin-dependent diabetes mellitus (NIDDM). Most previous studies have not controlled for potential confounding by age and obesity.
Patients and methods: In a prospective cohort study, 113,606 United States registered nurses aged 30 to 55 years and free of diagnosed diabetes, coronary heart disease, stroke, and cancer at baseline were followed for 12 years. Endpoint was incidence of confirmed NIDDM.
Results: During 1,278,188 person-years of follow-up, we confirmed 2,310 incident cases of NIDDM. An apparent association between parity and diabetes was observed in unadjusted analyses (relative risk = 1.56 [95% confidence interval (CI), 1.27 to 1.91] among women with six or more births compared with that in nulliparous women) (p, trend less than 0.00001). This association was attenuated after adjustment for age (relative risk = 1.19 [CI 0.97 to 1.48], p, trend = 0.06) and was completely abolished after adjustment for both age and body mass index (relative risk = 0.95 [CI 0.75 to 1.19], p, trend = 0.19). Multivariate adjustment for family history of diabetes, age at first birth, hormone use, and other variables did not materially alter these findings. There was no important modifying effect of family history of diabetes on these associations.
Conclusions: Despite a temporary diabetogenic effect of pregnancy, parity is not associated with an increased risk of subsequent clinical NIDDM. These data underscore the importance of control for confounding by age and obesity in evaluating these associations.