245Comparison of methods for classifying heterogeneity in gestational diabetes (GDM) and association with outcomes

Abstract

Abstract Background Clinical experience suggests that diverse clinical subtypes exist within the broader diagnosis of GDM. Analysis from a single centre recently outlined heterogeneity in GDM with respect to insulin secretion and sensitivity, defining four GDM subtypes: 1) GDMsecr (<25th centile HOMA-β (Hb) for non-GDM); 2) GDMsens (<25th centile Matsuda Index for non-GDM); 3) GDMmixed (both GDMsecr and GDMsens); 4) GDMND, no defect (neither GDMsecr and GDMsens). Classification using these subtypes is associated with adverse outcomes. Methods Following similar methodology, women with GDM were classified into four subtypes including comparison of Hb, insulinogenic index (II) and Stumvoll first-phase estimate (SV) for defining GDMsecr. Analyses compared neonatal outcomes with non-GDM women and between GDM groups using c2 tests and regression analyses adjusted for multiple confounders including maternal age, BMI and HAPO study centre. Results Hb, II and SV gave divergent results for GDMsecr, with only 19% concordance. In all analyses, GDMND (10% by Hb, 6% by II, 6% by SV) showed outcome frequencies similar to those of non-GDM women; groups 1-3 showed higher risks (p < 0.01 vs non GDM). These results persisted in the fully adjusted model (aOR generally >2.0). Conclusions Different clinical subtypes in GDM are associated with differing risks of adverse outcome. Key messages Determination of GDM subtype can assist in assessing GDM women at higher risk of adverse clinical outcome and help guide clinical practice.