Feasibility, psychometric performance, and stability across modes of administration of the CARES-SF

Abstract

The primary objective was to investigate the validity and reliability of the Cancer Rehabilitation Evaluation System-Short Form (the CARES-SF) as a quality of life instrument in clinical trials [10]. A heterogeneous sample of 485 cancer patients completed the CARES-SF before treatment (T1), one month later (T2), and three months following T2 (T3). At T3 the patients completed the questionnaire either by mail, in a telephone interview, or in the clinic. A sub sample of patients completed the CARES-SF a fourth time (T4) one week following T3, for purposes of test-retest reliability estimation. On average, the CARES-SF required 11 minutes for completion and could be completed by 90% of the patients without assistance. However, 82% of the patients re ported difficulty with at least one item. Multitrait scaling analysis and factor analysis generally confirmed the hypothe-sized scale structure. Internal consistency reliability coefficients exceeded the 0.70 criterion for four of six multi-item scales. The test-retest reliability coefficients of the six scales were above 0.70. No systematic differences were found in the psychometrics of the CARES-SF across administration conditions. In the mail condition the proportion of missing items was significantly higher, and patients reported having significantly more problems than in both the telephone and in- clinic condition. According to expectation, selective scales distinguished clearly between patients differing in disease stage, performance status, treatment modality and tumor response. Additionally, selective scales were responsive to changes in health status over time. These results lend support to the reliability and validity of the CARES-SF in assessing the quality of life of patients with cancer. At the same time, efforts to refine the questionnaire are recommended.