Quality of Life Measures in Epilepsy
How Well Can They Detect Change Over Time?
Published in: Neurology, v. 54, no. 9, May 2000, p. 1822-1827
Posted on RAND.org on January 01, 2000
OBJECTIVE: To evaluate the ability of health-related quality of life (HRQOL) measures to detect change over time in persons with epilepsy. BACKGROUND: The application of HRQOL measures in clinical trials has been limited by a dearth of information regarding their abilities to measure change over time (i.e., their responsiveness). To calculate responsiveness, one must categorize subjects as changed or 'unchanged by a priori criteria. METHODS: The authors analyzed data collected at baseline and at 28-week follow-up from an antiopileptic drug trial. Two different criteria for classifying subjects as changed or unchanged-one based on seizure frequency (where changed = attainment of seizure freedom) and one based on self-reported overall condition (where changed = improvement in overall condition)-were used. The authors compared responsiveness indices for two generic (Short Form [SF]-36 and SF-12) and two epilepsy-targeted (Quality of Life in Epilepsy [QOLIE]-89 and QOLIE-31) HRQOL measures. Two scoring procedures for the SF-36, one based on classic test theory and the other on item response theory (IRT), were compared. RESULTS: Effect sizes of the most responsive HRQOL measures were medium to large. The shorter epilepsy-targeted measure had similar responsiveness indices to those of the longer measure. Epilepsy-targeted measures were consistently more responsive than generic measures under the overall condition criterion, but for the seizure freedom criterion, IRT scoring of the SF-36 yielded responsiveness indices comparable to those of the epilepsy-targeted measures. CONCLUSION: Epilepsy-targeted health-related quality of life measures may be preferable to generic ones in longitudinal studies. Selection of a shorter epilepsy-targeted measure does not compromise responsiveness. Item response theory scoring should be applied to epilepsy- targeted HRQOL measures.