Measuring the effects of medical interventions.

This article reviews a number of important issues in selecting health-related quality-of-life measures in studies of the effects of medical interventions. Concerning the selection of domains, a plea is made for a greater use of qualitative discovery methods than is currently the case. It is also argued that measures of specific outcomes are inadequate to describe fully the effects of medical interventions, and must be coupled with more general outcome measures. Concerning the selection of measures, an argument is made that psychometric evaluation should be less concerned with the internal consistency reliability of short outcome measures and more with the ability of short-form measures to reproduce total scale variance. It is also noted that precision of measurement has been a neglected topic in the psychometric evaluation of outcome measures in this area of research. Future work should consider precision within the range of the outcome where effects are expected much more seriously. Finally, future methodologic work is called for using methods based on item response theory to study precision and bias.