Evaluation of screening strategies for gestational diabetes mellitus: Risk factor-based v. universal approaches in women with and without HIV

Background. Gestational diabetes mellitus (GDM) poses a substantial risk of adverse maternal, fetal and child health outcomes, and may be influenced by HIV infection. In South Africa (SA), risk-based GDM screening is used owing to resource constraints, but this approach may miss some cases. There is a need to optimise screening methods to efficiently identify pregnant women with GDM. Objectives. To compare risk factor-based screening, the current standard of care and screening with fasting plasma glucose (FPG) to the gold standard of universal GDM screening with an oral glucose tolerance test (OGTT). Methods. We universally screened 399 women with and without HIV attending antenatal care in Gugulethu Community Health Centre, Cape Town, SA, for GDM at 24 - 28 weeks’ gestation using a 2-hour 75 g OGTT. GDM was diagnosed based on the International Association for Diabetes in Pregnancy Study Group criteria. We used logistic regression and receiver operating characteristic curves to compare the screening performance of (i) risk factor-based screening, based on the SA National Department of Health Guidelines for maternity care; and (ii) FPG with and without clinical risk factors including age, body mass index, HIV status and family history of diabetes, with an OGTT (the gold standard). Results. The median (interquartile range) age of participants was 30 (25 - 34) years, and 81% were multiparous. Using OGTT, the prevalence of GDM was 6.3% (9.0% in women with HIV and 3.5% in women without HIV). Risk factor-based screening was poor and showed no discrimination (area under the curve (AUC) 0.53, sensitivity 0.52, specificity 0.55, positive predictive value (PPV) 0.07 and negative predictive value (NPV) 0.94). FPG (AUC 0.91, sensitivity 0.50, specificity 0.98, PPV 0.25 and NPV 0.99) and FPG plus clinical risk factors (AUC 0.98, sensitivity 0.83, specificity 0.97, PPV 0.31, NPV 1.00) showed strong discrimination. FPG had low sensitivity, but this improved greatly when combined with clinical risk factors. Conclusion. Risk factor-based screening, which is the current standard of care, did not discriminate between those with and without GDM. Screening for GDM with FPG and clinical risk factors improved discrimination, and had high sensitivity and specificity, but low PPV. These findings suggest that combining FPG with risk factors may be a practical approach to identify women who should undergo a full OGTT for GDM screening in resource-constrained settings.