The impact of using different imputation methods for missing quality of life scores on the estimation of the cost‐effectiveness of lung‐volume‐reduction surgery

Abstract

A post hoc analysis of data from a prospective cost-effectiveness analysis (CEA) conducted alongside a randomized controlled trial (National Emphysema Treatment Trial - NETT) was used to assess the impact of using different imputation methods for missing quality of life data on the estimation of the incremental cost-effectiveness ratio (ICER). The NETT compared lung-volume-reduction surgery plus medical therapy with medical therapy alone in patients with severe chronic obstructive pulmonary disease due to emphysema. One thousand sixty-six patients were followed for up to 3 years after randomization. The cost per quality-adjusted life-year gained was obtained, computing costs from a societal perspective and using the self-administered Quality of Well Being questionnaire to measure quality of life. Different methods of imputation resulted in substantial differences in ICERs as well as differences in estimates of the uncertainty in the point estimates as reflected in the CEA acceptability curves. Paradoxically, the use of a conservative single imputation method resulted in relatively less uncertainty (anticonservative) about the ICER. Owing to the effects of different imputation methods for missing quality of life data on the estimation of the ICER, we recommend use of a minimum of two imputation methods that always include multiple imputation.