Interpreting clinically significant changes in patient-reported outcomes.

Abstract

The goal of this study was to determine what magnitude of change in a patient-reported outcome score is clinically meaningful, so a clinicians' guide may be provided for estimating the minimal important difference (MID) when empiric estimates are not available. Consecutive laryngeal cancer patients (n = 98) rated their quality of life (QOL) relative to other patients. These comparisons were contrasted with arithmetic differences in scores on the Functional Assessment of Cancer Therapy-Head and Neck (FACT-H&N) scale, Functional Assessment of Cancer Therapy-General (FACT-G) scale, 2 utility measures (the time tradeoff [TTO] and Daily Active Time Exchange [DATE]), and performance status (Karnofsky) scores. The FACT-H&N score needed to differ by 4% for average patients to rate themselves as "a little bit better" relative to other patients (95% CI, 1%-8%) and by 9% to rate themselves as "a little bit worse" relative to others (95% CI, 4%-13%). The corresponding values for other measures were FACT-G 4% (1%-7%) and 8% (95% CI, 5%-11%); TTO 5% (95% CI, 0%-11%) and 6% (95% CI, 0%-10%); DATE 5% (95%CI, 2%-9%) and 14% (95% CI, 0%-5%); Karnofsky 4% (95% CI, 1%-6%) and 10% (95% CI, 7%-13%). In each case, the minimal important difference (MID) was about 5% to 10% of the instrument range. CONCLUSIONS. One rule of thumb for interpreting a difference in QOL scores is a benchmark of about 10% of the instrument range. Patients appear to be more sensitive to favorable differences, so an improvement of 5% may be meaningful. This simple benchmark may be useful as a rough guide to meaningful change.