Properties of the EQ-5D-3L index distribution when longitudinal data from 27,328 spine surgery procedures are applied to nine national EQ-5D-3L value sets.

Purpose: The purpose of the current study was to apply a single large longitudinal EQ-5D-3L data set to several national EQ-5D-3L value sets and explore differences in EQ-5D-3L index density functions and effect sizes before and after treatment.

Methods: Patients, surgically treated for lumbar spinal stenosis or lumbar disk herniation between 2007 and 2017, were recruited from the national Swedish spine register. A total of 27,328 procedures were eligible for analysis. The EQ-5D health states were coded to EQ-5D-3L summary indices using value sets for 9 countries: Argentina, Australia, Canada, China, Germany, Italy, Sweden, the UK, and the US. The EQ-5D-3L summary index distributions were then estimated with kernel density estimation. The change in EQ-5D-3L index before and after treatment was evaluated with the standardized response mean (SRM).

Results: There was a high variability in the resulting EQ-5D-3L index density functions. There were also considerable differences in EQ-5D-3L index density functions before and after treatment using the same value set. Effect sizes of 2-year change (SRM), however, were similar when the 9 value sets were applied on pre- and post-treatment data.

Conclusions: We found a marked variability in EQ-5D-3L index density functions when a single large data set was applied to 9 national EQ-5D-3L value sets. Consequently, studies that aggregate international data, e.g. meta-analyses, may produce misleading results if the underlying differences in EQ-5D-3L index density functions are inadequately handled. On the basis of the results of our study, we recommend against pooling of different national EQ-5D-3L index data.