Costs and cost-effectiveness of stepped, evidence-based and integrated care for common and persistent severe mental illness: Results of the RECOVER trial.

RECOVER is a severity-stratified, evidence-based care model integrating multiple levels of mental illness. This analysis focuses on individuals with severe and persistent mental illness (SMI/SPMI) and complex care needs, who received outreach-based services through Assertive Community Treatment (ACT) and Crisis Resolution Teams (CRT) as core components of the RECOVER intervention. This study assesses cost-effectiveness from a societal perspective, informed by clinical outcomes. Three domains were analyzed: (1) societal costs over a 12-month follow-up were compared between RECOVER and treatment as usual (TAU) using adjusted generalized linear models; (2) cost-effectiveness was assessed by estimating the incremental cost-effectiveness ratio (ICER) and cost-effectiveness probabilities via net monetary benefit regressions; and (3) psychosocial functioning was assessed via linear mixed model for repeated measures. Among 891 randomized trial participants, 506 (56.8 %) met criteria for SMI/SPMI (RECOVER n = 252, TAU n = 254). RECOVER was associated with lower societal costs (-€6681; p < 0.01). Although outpatient costs increased (+€1593; p < 0.001), these were more than offset by inpatient cost savings (-€6382; p < 0.001), resulting in a substantial net cost reduction. The ICER revealed dominance of RECOVER compared to TAU; cost-effectiveness probabilities exceeded 99 % across varying willingness-to-pay thresholds. Effectiveness analyses showed improved psychosocial functioning in the RECOVER group relative to TAU. The RECOVER model, characterized by outreach-based care (ACT/CRT), demonstrated cost-effectiveness and reductions in inpatient service use. Results highlight the economic and structural benefits of shifting from inpatient to community-focused care for individuals with SMI/SPMI. RECOVER offers a scalable framework for sustainable mental health service reform and should inform future standard care models.