Abstract

Background

Sample size calculations for treatment trials that aim to assess health-related quality-of-life (HRQOL) outcomes are often difficult to perform. Researchers must select a target minimal clinically important difference (MCID) in HRQOL for the trial, estimate the effect size of the intervention, and then consider the responsiveness of different HRQOL measures for detecting improvements. Generic preference-based HRQOL measures are usually less sensitive to gains in HRQOL than are disease-specific measures, but are nonetheless recommended to quantify an impact on HRQOL that can be translated into quality-adjusted life-years during cost-effectiveness analyses. Mapping disease-specific measures onto generic measures is a proposed method for yielding more efficient sample size requirements while retaining the ability to generate utility weights for cost-effectiveness analyses.

Objectives

This study sought to test this mapping strategy to calculate and compare the effect on sample size of three different methods.

Methods

Three different methods were used for determining an MCID in HRQOL in patients with incontinence: 1) a global rating of improvement, 2) an incontinence-specific HRQOL instrument, and 3) a generic preference-based HRQOL instrument using mapping coefficients.

Results

The sample size required to detect a 20% difference in the MCID for the global rating of improvement was 52 per trial arm, 172 per arm for the incontinence-specific HRQOL outcome, and 500 per arm for the generic preference-based HRQOL outcome.

Conclusions

We caution that treatment trials of conditions for which improvements are not easy to measure on generic HRQOL instruments will still require significantly greater sample size even when mapping functions are used to try to gain efficiency.