Performance of Variables in Screening for Gestational Diabetes

Abstract

Abstract Introduction: Gestational diabetes mellitus (GDM) is associated with adverse pregnancy outcomes. The strategies used to screen for GDM vary both internationally and nationally. Therefore, we investigated the performance of the capillary random plasma glucose (RPG) test, maternal body mass index (BMI) and maternal age in predicting GDM. Methods: In a retrospective cohort study, we included pregnant women without pre-existing diabetes or metabolic disease who gave birth in Västernorrland County, Sweden, in 2015–2016. Values for RPG in gestational weeks 23–28 were obtained from obstetric medical records for each pregnancy. The development of GDM was confirmed by evaluating data in the obstetric records. The ability of RPG, maternal BMI, and age to predict GDM was assessed with receiver-operating characteristic curves. Results: In total, 4,698 pregnancies were included in the final statistical analysis. RPG was fairly effective in screening (area under the curve [AUC] 0.73; 95% confidence interval [CI] 0.66–0.80), and BMI performed slightly better (AUC 0.75; 95% CI 0.68–0.82), whereas maternal age performed poorly (AUC 0.61; 95% CI 0.53–0.68). Combining RPG ≥7 and BMI ≥27.9 yielded the best overall sensitivity (75.4%), specificity (70.1%), and AUC (0.75; 95% CI 0.68–0.82). Conclusions: Our results show that the sensitivity of capillary RPG alone in predicting GDM is fair. The combination of RPG with maternal BMI or age showed greater sensitivity. However, none of the screening factors (RPG, BMI, and maternal age), alone or combined, showed sufficiently good performance in predicting GDM.