Change in quality of life of people with stroke over time: true change or response shift?

In many studies, assessments of change in self-report measures such as health-related quality of life must account for potential response shift, including reconceptualization and changes in internal standards of measurement.

The objective of our study was to compare healthy controls and individuals with stroke on the extent to which changes in internal standards and reconceptualization of health related quality of life (HRQL) occurs over the first 6 months post-stroke.

Confirmatory factor analysis was used to assess invariance of the SF-36 measurement model over time among 238 individuals with stroke and 392 controls, separately. This procedure assessed changes over time in the factor loadings, variances, and covariances of responses, and compared the extent of change between individuals with stroke and those in the control group. In addition a multisample comparison was made between individuals with stroke and members of the control group at the first evaluation in order to assess invariance of the SF-36 measurement model between the groups. The controls were considered to be a 'proxy' for the stroke cohort prior to the stroke.

We found no evidence of reconceptualization and changes in internal standards over time when the groups were assessed separately. There was a significant difference in the factor covariances (reconceptualization) between the two groups at the time of the first evaluation. However, measurement error was also significant for this comparison.

This study indicates that the improvement in HRQL over time is real rather than a result of reconceptualization or a recalibration. If response shift does occur with stroke it is likely to be mediated by the event itself and not the recovery process.