Structured psychiatric care and psychosocial support during placebo participation: association with violent and domestic-violence offending in the ReINVEST trial

Background: Participants in the ReINVEST randomised placebo-controlled trial of sertraline, conducted among men with high trait impulsivity and histories of violent offending, received structured clinical contact throughout the trial, including psychiatric assessments, nursing consultations, crisis support, and referrals to mental health and external services. We estimated the effect of placebo trial participation, compared with non-participation after baseline and single-blind run-in, on violent and domestic-violence reoffending. Methods: This prespecified secondary analysis included men from the ReINVEST trial pathway who completed baseline assessment and entered the single-blind run-in phase but did not proceed to randomisation, to inform the counterfactual. Violent and domestic-violence offences were identified from linked administrative records over 12- and 24-month follow-up periods. The adjusted difference in offending was estimated using two independent analytical approaches accounting for baseline differences. Additional analyses examined whether the effect varied by baseline clinical and criminal-history characteristics, whether pre-randomisation external referrals explained selection into placebo participation, and whether post-randomisation external referrals accounted for any part of the estimated effect. Results: Placebo trial participation was associated with lower offending across both outcome domains and follow-up periods. Placebo-standardised mean count differences for violent offending were -0.19 (95% confidence interval [CI] -0.38, -0.04) at 12 months and -0.22 (95% CI -0.51, -0.05) at 24 months. Corresponding differences for domestic-violence offending were -0.37 (95% CI -0.81, -0.14) at 12 months and -0.49 (95% CI -0.92, -0.22) at 24 months. The association was more apparent among men with a documented psychiatric history and, for domestic-violence offending, among those with higher baseline anger, irritability and aggression. Pre-randomisation referrals did not explain selection into placebo participation or materially alter the estimates. Post-randomisation referrals were observed in both groups, remained more common in the placebo group, and did not account for the observed association. Conclusion: Placebo participation in this trial involved sustained clinical contact and psychosocial support beyond exposure to inactive medication, and these non-pharmacological components may have contributed to lower reoffending. In placebo-controlled trials involving populations with high psychiatric morbidity and limited continuity of coordinated care, the clinical content of placebo participation should be explicitly characterised in trial design and interpretation.