To explore the impact of applying different non-standardized analytical choices for quality of life measurement to obtain quality-adjusted life years (QALYs). In addition to more widely discussed issues such as the choice of instrument (e.g. EQ-5D or SF-6D?) researchers must also choose between different recall periods, scoring algorithms and interpolations between points of measurement.
A prospective survey was made among 114 Belgian patients with acute hepatitis A illness. Using non-parametric tests and generalized linear models (GLM's), we compared four different methods to estimate QALY losses, two based on the EQ-5D (administered during the period of illness without recall period) and two based on the SF-6D (administered after illness with 4 weeks recall period).
We found statistically significant differences between all methods, with the non-parametric SF-6D-based method yielding the highest median QALY impact (0.032 QALYs). This is more than five times as high as the EQ-5D-based method with linear health improvement, which yields the lowest median QALY impact (0.006 QALYs).
Economic evaluations of health care technologies predominantly use QALYs to quantify health benefits. Non-standardised analytical choices can have a decision-changing impact on cost-effectiveness results, particularly if morbidity takes up a substantial part of the total QALY loss. Yet these choices are rarely subjected to sensitivity analysis. Researchers and decision makers should be aware of the influence of these somewhat arbitrary choices on their results.