The Effect of Parity on the Later Development of Diabetes

Publisher Website
Google Scholar
Abstract
Kritz-Silverstein et al. (Nov. 2 issue)1 have made a valuable contribution to the literature on the effects of pregnancy on the risk of diabetes mellitus. The results that they present, however, are not entirely consistent with their conclusion of a 1.16-fold and a 1.10-fold increase per pregnancy in the risks of non-insulin-dependent diabetes mellitus (NIDDM) and impaired glucose tolerance, respectively. The data in Figure 1 of their article suggest that the increase in the risk of NIDDM does not appear until the number of pregnancies exceeds five. Among women who have had fewer than six pregnancies, the age-adjusted prevalence of NIDDM does not appear to differ substantially according to the number oT pregnancies. Therefore, it does not seem appropriate for Kritz-Silverstein et al. to conclude that the risk of diabetes increases with each pregnancy, since this effect appears to be confined to highly parous women. The age-adjusted prevalence of impaired glucose tolerance also does not appear to increase with each pregnancy (as shown in Figure 1 of their article). In fact, the age-adjusted prevalence of impaired glucose tolerance was lowest among the women who had had six or more pregnancies, suggesting a protective effect of high parity for this condition. On the basis of these observations, the inclusion of a linear term for parity in the logistic model used by Kritz-Silverstein et al. does not seem appropriate. It would be helpful if the authors would do their logistic regression analysis by modeling the risk of NIDDM and impaired glucose tolerance as a function of parity defined as a set of discrete categorical (dummy) variables to account for the apparent nonlinear relation between these metabolic disorders and parity, and to develop confidence limits for the risk of diabetes associated with the occurrence of more than five pregnancies.

This publication has 1 reference indexed in Scilit: