Introduction and objectiveThe role of metformin in gestational diabetes mellitus (GDM) is also increasing. However, almost half of metformin-treated women required additional insulin. Therefore, identifying the characteristics of these women may help define optimal therapeutic strategy. MethodsThis is a retrospective cohort study done in a District General Hospital, UK. GDM was diagnosed by 75 g OGTT test between 24 and 28 weeks of gestation with fasting levels of ≥6.1 mmol/l and/or 2 h postprandial (PP) level of ≥7.8 mmol/l. Logistic regression and receiver operator curves (ROC) were performed to identify the predictors of metformin failure. ResultsOut of 228 women with GDM included, 46/228 (20.2%) and 151/228 (66.2%) received insulin and metformin as first-line medication respectively. Among the metformin-treated, 13 stopped treatment and were excluded from analysis. Of the included 138 metformin-treated women, 77 (55.8%) required supplementary insulin (metformin failure). Metformin failure group had higher maternal age and fasting glucose level at OGTT, HbA1c at OGTT and earlier gestational age (GA) at medication initiation. Metformin failure was predicted if fasting OGTT level >4.8 mmol/l (69% sensitivity and 62% specificity). If the fasting levels of IADPSG (International Association of Diabetes and Pregnancy Study Groups) criteria and NICE (National Institute of Health and Care Excellence) were used, the positive predictive value was 78% and 77% respectively. ConclusionAs women with higher fasting glucose levels have higher chance of necessitating insulin in later pregnancies, appropriate addition of insulin at metformin initiation for these women could help better glycaemic control throughout pregnancy.