Arnold GK, Baranowski RA, Lipner RS. — American Board of Internal Medicine
Presented: International Conference on Health Policy Statistics, October 2011
Abstract: We compared patient ratings of primary care physicians on four composite measures related to provider communication skills (5 items), clinical care skills (4 items), shared decision-making (3 items) and overall satisfaction (11-point rating scale). Measures came from practice-based Clinical and Group (CG) CAHPS Surveys in English or Spanish, completed anonymously (minimum of 25 responses) using an automated phone or Web system. These surveys satisfy Self-Evaluation of Practice Performance requirements that physicians need to complete to maintain board certification in internal medicine. Responses differ among patients because measurement constructs do not translate well between languages. If constructs are interpreted by Spanish- and English-speaking patients similarly, then ratings of the same physicians would be indistinguishable after controlling for differences unrelated to language such as health status, education, etc. We compared patient ratings of 49 physicians who administered surveys to both Spanish- (n=183 cases) and English- (n=896 controls) speaking patients. A second control group included 1,153 ratings (English only) from another 49 physicians propensity-matched to each study physician based on 17 practice, demographic, training and achievement characteristics. This process simulated a randomized block design with measures repeated across cases and two controls. Differences in physician means between groups were assessed using signed-rank tests. Results indicated that ratings by English-speaking controls of study physicians did not differ significantly from propensity-matched controls, and so they were combined. Spanish-speaking patients gave higher satisfaction ratings (p=0.003) and higher shared decision-making ratings (0.002) than their English-speaking controls; however, communication skills (0.13) and clinical care (0.09) ratings were not significantly different. After adjusting comparisons for group differences in gender, age, education, Hispanic origin, health status and number of visits, none of the composites differed significantly: satisfaction (0.74), shared decision-making (0.15), communication (0.46) or clinical care (0.21). This study provides evidence that patients' ratings on four provider constructs yield similar outcomes from both Spanish- and English-speaking patients after adjustment for confounding factors such as education, health status, etc. The study also suggests that propensity matching helps increase sample sizes in matched case-control studies.
For more information about this presentation, please contact Research@abim.org.