Aims/hypothesis <p>Connected insulin pens and smart caps automatically capture dosing data and can prompt patients about missed prandial boluses, but independent randomised evidence in type 1 diabetes is scarce. We aimed to assess whether a connected insulin pen cap improves glycaemic management compared with an otherwise identical non-connected cap.</p> Methods <p>We conducted an investigator-initiated, randomised, open-label, parallel-group trial in a European public hospital over 8 weeks. Adults with type 1 diabetes treated with multiple daily injections and who had suboptimal glycaemic management (baseline time above range [&gt;10.0 mmol/l, &gt;180 mg/dl] &gt;25%) were randomly assigned (1:1) to a connected pen cap (Insulclock&#xa0;2.0; Insulcloud, Madrid, Spain) or an identical cap for which the Bluetooth connectivity had been disabled (disconnected). Allocation was performed using opaque sealed envelopes, and no masking was applied. The primary outcome was time above range, with two thresholds (&gt;10.0 mmol/l, &gt;180 mg/dl and &gt;13.9 mmol/l, &gt;250 mg/dl) based on intermittently scanned CGM downloads. Analyses followed the intention-to-treat principle, using longitudinal mixed models and multiple imputation. Secondary outcomes included time in range, glycaemic variability (standard deviation and coefficient of variation), HbA_1c and patient-reported outcomes (fear of hypoglycaemia).</p> Results <p>Forty-two participants were randomised for inclusion in the study (21 per group: 25 women, 17 men). One withdrew before baseline, leaving 41 in the intention-to-treat analysis. Compared with the control group, the connected-cap group had lower time above range &gt;13.9 mmol/l (mean difference −4.8 percentage points; 95% CI −9.5, −0.1; <i>p</i>=0.045) and reduced glucose standard deviation (−0.35 mmol/l; 95% CI −0.64, −0.06; <i>p</i>=0.018). HbA_1c showed a borderline reduction of −3.5 mmol/mol (−0.32%) (95% CI −7.1, 0.0 mmol/mol; −0.65, 0.00%; <i>p</i>=0.050). Per-protocol analyses suggested an 8% absolute increase in time in range among adherent users. No adverse events or device-related serious adverse events occurred. No between-group differences were observed in avoidance behaviours based on the hypoglycaemia fear survey significantly decreased in the treatment group (β=−2.44; 95% CI −4.45, −0.43; <i>p</i>=0.019).</p> Conclusions/interpretation <p>In routine clinical care, use of a connected pen cap reduced severe hyperglycaemia, glycaemic variability and hypoglycaemia avoidance behaviours in adults with type 1 diabetes and suboptimal glycaemic management. These findings support the integration of low-burden, data-driven tools into public diabetes care. Larger and longer trials should evaluate the durability and cost-effectiveness of these interventions.</p> Trial registration <p>ClinicalTrials.gov NCT06845891</p> Funding <p>Investigator-initiated study without industry funding; devices were procured by the hospital. Funders had no role in study design, conduct, analysis or reporting</p> Graphical Abstract <p></p>