A cost-utility analysis for return-to-work interventions comparing alternative methods for handling missing health-related quality of life data.

Abstract

Perform a cost-utility analysis for return-to-work interventions with missing health-related quality-of-life (HRQoL) data while transparently demonstrating the impact of different methods of handling missing data on outcomes. The costs and quality-adjusted life-years over a 2-year period were estimated for 2 return-to-work interventions, inpatient multimodal occupational rehabilitation (I-MORE) and outpatient acceptance and commitment therapy (O-ACT), using a healthcare perspective and a limited societal perspective. Four methods were used to handle the missing HRQoL data: complete case analysis, single imputation, multiple imputation, and linear mixed models. The cost-effectiveness outcomes were expressed as incremental net monetary benefit. The average incremental quality-adjusted life-years comparing I-MORE with O-ACT ranged between -0.001 and 0.330 depending on missingness method. From a healthcare perspective, I-MORE was consistently not cost-effective (incremental net monetary benefits ranged from -€7,094 to -€9,363) while from a limited societal perspective, I-MORE was consistently cost-effective (incremental net monetary benefits ranged from €1,293 to €16,277). While cost-effectiveness findings remained consistent within each analytical perspective, the choice of different missingness methods led to variations in incremental quality-adjusted life-years. Multiple imputation is recommended to handle missing HRQoL data as it is transparent and flexible. How-ever, a thorough investigation of the missing data mechanism should still be conducted.