To follow

Abstract

Residential harm reduction and overdose outcomes in Glasgow: A retrospective cohort study

Aims: Scotland has among Europe’s highest drug-related death rates. Non-fatal overdose signals overdose risk and an intervention opportunity. Residential services are core to treatment, but evidence for models prioritising harm reduction is limited. We aimed to evaluate overdose and healthcare outcomes associated with a 16-bed residential harm reduction service in Glasgow, and to explore post-discharge predictors including behavioural exposures, opioid agonist treatment, and gender.

Methods: Retrospective cohort study using routinely collected data from 447 admission episodes (2022–2025). Outcomes were measured for 12 weeks pre-admission and 12 weeks post-discharge. Primary outcome: non-fatal overdose (any non-fatal overdose; event counts). Secondary outcome: emergency department attendance (any; counts). Paired comparisons used McNemar’s test; incidence rate ratios summarised count changes. Binary outcomes were expressed as proportions. Multivariable logistic regression modelled post-discharge non-fatal overdose and emergency department attendance.

Results: Non-fatal overdoses fell from 14.5% pre-admission to 9.8% post-discharge among admission episodes (absolute −4.7 percentage points; relative −32.4%; p=0.007); total non-fatal overdose events −42.5% (IRR≈0.57). Emergency department attendance reduced: total events −42.1%, and any attendance 21.0% to 13.9% (p<0.001). Post-discharge non-fatal overdose: street benzodiazepines (aOR 7.84), injecting (aOR 3.53); long-acting buprenorphine protective (OR 0.19). Gender differences attenuated after adjustment; recovery-support engagement increased and reduced risk in minimally adjusted models.

Conclusions: Residential harm-reduction admission associated with significant 12-week reductions in non-fatal overdose and emergency department use, including fewer repeat overdoses and high-frequency emergency department attendance. Risk driven mainly by behavioural exposures (street benzodiazepines and injecting) rather than gender or planned discharge, supporting low-threshold care. Long-acting buprenorphine showed a protective association; recovery engagement increased. Observational design (no comparator; repeat episodes unclustered) limits causal inference; findings inform policy planning for services.