Abstract
STUDY OBJECTIVES: To compare prospective and retrospective measurements of change in health status. DESIGN: Health status was measured using a French language version of the short form 36 (SF-36) health survey on two occasions one year apart--in 1992 and 1993. Differences in SF-36 scores measured prospectively were compared with the patients' single item retrospective evaluation of change in health (transition item). SETTING: This was a community based study among members of two health insurance plans in Geneva, Switzerland. PARTICIPANTS: Altogether 831 young adults (mean age 30 years at baseline). MAIN RESULTS: Health status remained stable on average during the study period. The retrospective rating correlated well with changes in health measured prospectively: those who said in 1993 that their current health was "much worse" than in 1992 experienced an average decrease of 1.06 SD on the eight SF-36 scales, while those who said that their health was "much better" recorded an average improvement of 0.43 SD. The associations between prospective and retrospective assessments of change were approximately linear for all scales but physical functioning. The transition item also discriminated between time periods: transition reported for 1991-92 did not correlate with changes recorded for 1992-93. Relative validity analyses indicated that the transition item was better suited to capture changes in general health than changes in purely physical or mental aspects of health. CONCLUSIONS: The concordance between retrospective and prospective measures of change in health suggests that both are sensitive, to some extent, to true changes in health status. Using both types of assessment may improve the reliability of measurements of change.