Abstract
OBJECTIVE: This study aimed to determine the association between diabetes mellitus (DM) medication use and glycaemic control. DESIGN: This was a retrospective diabetes registry-based cohort study. SETTING: Singapore. PARTICIPANTS: Patients aged 18 and above with incident DM in the SingHealth Diabetes Registry from 2013 to 2020 were included. The entire study period included a 1 year baseline period, a 1 year observation period and a 3 month outcome period. OUTCOME MEASURES: Drug use was measured using the proportion of days covered (PDC), and the changes in glycated haemoglobin (HbA1c) between the outcome and baseline periods were assessed. The associations between baseline HbA1c and PDC ≥0.80 and between PDC and change in HbA1c were analysed using logistic regression and the Kruskal-Wallis test, respectively. RESULTS: Of 184 646 unique patients in the registry from 2013 to 2020, 36 314 met the inclusion and exclusion criteria and were included in the analysis. The median PDC for any DM drug, oral DM drugs and insulin during the observation period was 20.3%, 16.8% and 0%, respectively. Those who had good glycaemic control at baseline were less likely to receive DM drugs and those with poor baseline glycaemic control or missing baseline HbA1c were more likely to be consistent users (PDC >80%) (p<2.2 × 10(-16)). CONCLUSION: The relationship between DM drug use and glycaemic control is complex and non-monotonic. Higher PDC for any DM drug and oral DM drugs during the observation period was significantly associated with clinically relevant HbA1c improvements.