Testing the construct validity of a health transition question using vignette-guided patient ratings of health

BACKGROUND A single-item transition question is often used to assess improvement or worsening in health, but its validity has not been tested extensively. The purpose of this study was to test the construct validity of a transition question by relating it to qualitative changes in patient's self-rating of health guided by clinical vignettes. METHODS We studied 169 patients with active rheumatoid arthritis (RA) before and after treatment escalation. At both assessments, patients scored their current health on a rating scale after first rating three vignettes describing mild, moderate, or severe RA. We classified patients into one of these three RA categories using a nearest-neighbor match. We then related the change in these self-rated categories between visits to responses to a transition question on visit 2. RESULTS Sixty patients improved their RA vignette category after treatment, 86 remained in the same vignette category, and 23 worsened categories. On the transition question, 101 patients reported improvement, 48 reported no change, and 20 reported worsening, representing a modest association with changes in RA vignette categories (polychoric correlation r = 0.19). The association was stronger if patients who were in the mild RA category at both visits were also classified as improved if their self-rating changed from below to above their mild vignette rating (r = 0.23) and when incorporating the importance of changes on the transition question (r = 0.26). CONCLUSION Changes in health states, guided by clinical vignettes, support the construct validity of the transition question.