Concerns on methodological assumptions in the health economic evaluation of return-to-work interventions

High sickness absence and disability claims create major societal and budget pressures, making return-to-work (RTW) interventions a central policy priority. Because decisions often hinge on whether productivity gains are counted, robust long-horizon economic models are influential - but also vulnerable to strong extrapolation assumptions.

This Key Paper Evaluation examines a recent modeling study of RTWinterventions comparing a 3½-week inpatient multimodal occupational rehabilitation (I-MORE) with outpatient Acceptance and Commitment Therapy (ACT). The authors link a randomized trial to 7 years of registry follow-up and project costs and effects over a 25-year horizon from healthcare and societal perspectives.

The analysis is transparent and well executed, but the study's headline finding - lifetime dominance of I-MORE when productivity is counted - hinges on structural assumptions that effectively extend a 3½-week inpatient program into persistent productivity gains over a 25-year horizon: transition intensities are frozen beyond 4 years, and within-state absence-day differences are held constant thereafter, generating persistent productivity gains even after state prevalences converge. Health-related quality of life is observed only to 14 months and then modeled as state-specific constants, compressing incremental quality-adjusted life years and leaving the healthcare-only incremental cost-effectiveness ratio unfavorable. We argue that lifetime savings should be treated as upper-bound estimates and that decision relevance would be strengthened by waning-effect scenarios, mid-term horizons, and tests of the absorbing disability state. We also highlight the need for diagnosis-by-treatment analyses and stepped-care sequences given the heterogeneous population and fixed-dose ACT comparator.