Rehabilitation before discharge: A retrospective service evaluation of group-based rehabilitation in an Italian acute inpatient mental health unit.

OBJECTIVE: Acute inpatient mental health units often prioritize containment, risk management, and pharmacological stabilization while deferring rehabilitation until after discharge. This retrospective service evaluation examined program reach and associations between group attendance and clinician-rated outcomes. METHODS: We analyzed 480 consecutive admissions to an Italian acute inpatient mental health unit during 2018-2019. Daily newspaper and fairy-tale groups and weekly music and cinema groups were embedded in care. Newspaper-group attendance was the primary exposure. The outcomes were the Clinical Global Impression-Severity, Clinical Global Impression-Improvement, and Health of the Nation Outcome Scales. Robust regression and length-of-stay sensitivity analyses were conducted. RESULTS: At least one activity was attended during 406 admissions (84.6%); newspaper-group attendance occurred during 384 (80.0%). No group-related adverse events were documented in retrospectively reviewed records. Sensitivity analyses showed that the Clinical Global Impression-Severity association attenuated to nonsignificance under several flexible length-of-stay specifications, whereas the Health of the Nation Outcome Scales and Clinical Global Impression-Improvement associations were comparatively stable. In primary adjusted models, each 10-percentage-point attendance increase was associated with Health of the Nation Outcome Scales change of -0.296 points (95% CI [-0.410, -0.182]), Clinical Global Impression-Improvement of -0.057 points (95% CI [-0.082, -0.033]), and Clinical Global Impression-Severity change of -0.035 points (95% CI [-0.058, -0.012]); the Clinical Global Impression-Severity estimate should therefore be interpreted cautiously. CONCLUSIONS AND IMPLICATIONS FOR PRACTICE: Early group-based rehabilitation can be integrated into acute inpatient care. The outcome associations are hypothesis generating, not causal, because attendance accrued during admission, ratings were unblinded, and participation may reflect stability or engagement. Prospective studies should use time-varying exposure, independent ratings, language-access measures, patient-reported outcomes, and longer term recovery endpoints. (PsycInfo Database Record (c) 2026 APA, all rights reserved).