Relative importance measures for reprioritization response shift

Response shift (RS), a change in the meaning of an individual’s self-evaluation of a target construct, such as health-related
quality of life (HRQOL), can affect the interpretation of change in measures of the construct collected over time. This study
proposes new statistical methods to test for reprioritization RS, in which the relative importance of HRQOL domains changes
over time.

The methods use descriptive discriminant analysis or logistic regression models and bootstrap inference to test for change
in relative importance weights (Method 1) or ranks (Method 2) for discriminating between patient groups at two occasions.
The methods are demonstrated using data from the Manitoba Inflammatory Bowel Disease (IBD) Cohort Study (n = 388). Reprioritization of domains from the IBD Questionnaire (IBDQ) and SF-36 was investigated for groups with active and
inactive disease symptoms.

The IBDQ bowel symptoms and SF-36 bodily pain domains had the highest ranks for group discrimination. Using Method 1, there
was evidence of reprioritization RS in the IBDQ social functioning domain and the SF-36 bodily pain and social functioning
domains. Method 2 did not detect change for any of the domains.

Compared to IBD patients without active disease symptoms, those with active symptoms were likely to change the meaning of
their self-evaluations of pain and social interactions. Further research is needed to compare these new RS detection methods
under a variety of data analytic conditions before recommendations about the optimal method can be made.