Adherence to app-based dose guidance for once-weekly insulin icodec in insulin-naive individuals with type 2 diabetes: Post hoc analysis of ONWARDS 5.

In ONWARDS 5 (NCT04760626), a 52-week, Phase 3a trial in insulin-naive type 2 diabetes, 542/1085 participants were randomized to once-weekly insulin icodec with a dosing guide app to assist titration (icodec with app). This post hoc analysis of the icodec with app group assessed titration behaviours and associations between app-based dose guidance adherence and outcomes. App-based dose guidance and manual overrides were assessed. App guidance adherence was defined as dose guidance being requested and followed, and that automatic dose titration was possible; adherence subgroups were defined by mean fraction of weeks adherent across all weeks with app usage (per 4 weeks: 'low', ≤1 [n = 52]; 'medium', >1-≤3 [n = 215]; 'high', >3 [n = 239]). Observed glycated haemoglobin (HbA1c; baseline to week 52) and hypoglycaemia were described. Modelling analyses assessed associations between app adherence and estimated pre-breakfast self-measured blood glucose (SMBG) and weekly insulin dose. Few administered icodec doses (1053/21 244; 5.0%) differed from app guidance (dose changed by participant). Across 541 participants with app data, 56.7% had ≥1 app guidance override by a health care professional; most overrides (81.9%) were increases to app-recommended dose. Observed mean HbA1c decreased from baseline across all adherence subgroups. Across 52 weeks of app use, 'medium' and 'high' versus 'low' adherence was associated with significantly lower estimated mean pre-breakfast SMBG and higher estimated mean weekly insulin dose (both p < 0.0001). Clinically significant hypoglycaemia rates were low across subgroups, with no severe hypoglycaemia observed. Greater adherence to app-based dose guidance was associated with greater estimated improvements in pre-breakfast SMBG and higher weekly insulin dose. Trends indicated HbA1c improvements in all adherence subgroups. Findings support app use for real-world insulin titration.