Introduction: Minimally important differences (MIDs) are estimated to help interpret what changes on a patient reported outcome (PRO) measure are meaningful. The overall objective of this dissertation was to identify how MID estimates can differ by baseline health, direction of change, and scale recalibration response shift.

Methods: Two data sources were utilized for this research. The impact of baseline health and direction of change on MID estimation was assessed using SF-36 data collected during a two-year longitudinal study of 6,932 members of a California managed care organization. Health utilities were estimated from the SF-36 data. MID estimates for health utility were derived using an anchor-based method and validated using distribution-based methods. Differences in MID estimates were tested between patients reporting minimal gains and minimal losses. Patients were also grouped by baseline health utility scores to estimate the MID and results were compared. Global ratings of change and a thentest visual analog scale (VAS) were included in a prospective longitudinal study of 730 prostate cancer undergoing treatment to identify patients who underwent a response shift. Separate MIDs were estimated for patients with and without response shift and tested for differences.

Results: Results indicated that MID estimates were different for those reporting minimal gains and losses. Baseline health was also highly significant in explaining the variation in MID estimates. Finally, MID estimates in gains were impacted by patients who were identified as undergoing a scale recalibration response shift.

Conclusion: MID estimates for a PRO are impacted by patient baseline health, whether patients were improving or declining, and if patients changed their standards of how they evaluate their health state. Therefore, investigators who seek to utilize a published MID should ensure that their patient population is similar to the population that was originally used to estimate that MID.