Published Jan 17, 2013

Bristol, UK - “The benefits of new treatments in improving quality of life have been routinely under-estimated.” That is the conclusion from the first of two papers by Lu, Ades et al. from the University of Bristol. In the paper, “Mapping from Disease-Specific to Generic Health-Related Quality-of-Life Scales: A Common Factor Model,” published in Value in Health, they argue that the use of Ordinary Least Squares regression to estimate the “mapping” of treatment effects on disease-specific scales to generic scales, such as EQ-5D, inevitably under-estimates the EQ-5D benefits as it ignores measurement error. The second paper, “Which Health-Related Quality-of-Life Outcome When Planning Randomized Trials: Disease-Specific or Generic, or Both? A Common Factor Model,” also published in Value in Health, considers whether it is best to measure EQ-5D “directly” in trials, or to measure a disease-specific outcome (such as the Hamilton Depression scale, or the Health Activities Questionnaire in rheumatoid arthritis), and then “map” the results to obtain an “indirect” estimate of the treatment effect on EQ-5D. It is often assumed that “direct” measurement of EQ-5D must be more efficient, but, in this paper, Ades and Lu show there is a trade-off. Ultimately, the choice between “direct” and “indirect” estimates will depend on trial size, uncertainty in the mapping, and the relative responsiveness of the tests. The ideal approach is to measure both, and combine the information. Interestingly, it was discovered that it can sometimes be more efficient to power a trial on the combined generic outcome than on the more responsive disease-specific measure.

---