Williams LS (Roudebush VAMC)
Bakas T (Indiana University)
Brizendine E (Indiana University)
Tu W (Indiana University)
Plue L (Roudebush VAMC)
Kronke K (Indiana University)
Proxy respondents are required to measure functional and health-related QOL (HRQOL) outcomes in at least 25% of stroke survivors. In other conditions, proxies systematically rate HRQOL lower than patients, but whether this difference is observed in stroke and how it might affect outcome assessment in clinical trials is not well studied.
We compared patient and family proxy HRQOL responses in patients (Pt) enrolled in a clinical trial for post-stroke depression. Half the patients were depressed and half were non-depressed and all had an enrolled family caregiver as the proxy (Pr, N = 182 Pt-Pr pairs) who saw the patient at least four days per week and performed at least two specific caregiving tasks. We used paired t-tests to determine if there was a difference between Pt and Pr responses at study entry on the seven domains of the SS-QOL, a stroke-specific QOL scale with domain scores ranging from 1.0 (worst) to 5.0 (best). We used the ICC statistic to assess the strength of agreement between Pt and Pr domain scores and the overall SS-QOL score. We also compared Pt-Pr agreement between depressed and non-depressed Pts.
Pts were older (62 vs. 53 years) and less often female (48% vs. 74%) than Prs. Prs rated all domains of HRQOL slightly worse than pts except for Role function (Pts 3.0 vs. Pr 2.9). The Mood, Energy, and Thinking domains had the greatest disparity with mean Pt-Pr difference of 0.5 points. The ICC for each domain ranged from 0.3 (Role function and Thinking) to 0.5 (Physical function). Proxy overall SS-QOL score was also lower (worse) than patient score (3.4 vs. 3.7, p < 0.001) with ICC of 0.4. Although depressed Pts and their Prs had more similar domain scores than non-depressed Pt-Pr pairs, agreement was still only modest (ICCs 0.3 - 0.6).
Proxies systematically report more dysfunction in multiple aspects of HRQOL than stroke patients themselves. Agreement between patient and proxy HRQOL domain scores is highest for assessment of physical function but is still modest at best. This suggests that adjusted Pr HRQOL scores may not be confidently substituted for Pt HRQOL scores.
Family proxy reports of patient status may be necessary, but if used should be adjusted to more accurately reflect patient scores. Assessment of agreement should be done prior to using proxy reports to evaluate treatments. Further work on what influences disagreement between patients and proxies is needed.