BACKGROUND: Comparative effectiveness research across cultures requires unbiased measures that accurately detect clinical differences between patient groups. OBJECTIVE: The purpose of this study was to assess the presence and impact of differential item functioning (DIF) in knee functional status (FS) items administered using computerized adaptive testing (CAT) as a possible cause for observed differences in outcomes between 2 cultural patient groups in a polyglot society. DESIGN: This study was a secondary analysis of prospectively collected data. METHODS: We evaluated data from 9,134 patients with knee impairments from outpatient physical therapy clinics in Israel. Items were analyzed for DIF related to sex, age, symptom acuity, surgical history, exercise history, and language used to complete the functional survey (Hebrew versus Russian). RESULTS: Several items exhibited DIF, but unadjusted FS estimates and FS estimates that accounted for DIF were essentially equal (intraclass correlation coefficient [2,1]>.999). No individual patient had a difference between unadjusted and adjusted FS estimates as large as the median standard error of the unadjusted estimates. Differences between groups defined by any of the covariates considered were essentially unchanged when using adjusted instead of unadjusted FS estimates. The greatest group-level impact was <0.3% of 1 standard deviation of the unadjusted FS estimates. LIMITATIONS: Complete data where patients answered all items in the scale would have been preferred for DIF analysis, but only CAT data were available. CONCLUSIONS: Differences in FS outcomes between groups of patients with knee impairments who answered the knee CAT in Hebrew or Russian in Israel most likely reflected true differences that may reflect societal disparities in this health outcome.
BACKGROUND: Comparative effectiveness research across cultures requires unbiased measures that accurately detect clinical differences between patient groups. OBJECTIVE: The purpose of this study was to assess the presence and impact of differential item functioning (DIF) in knee functional status (FS) items administered using computerized adaptive testing (CAT) as a possible cause for observed differences in outcomes between 2 cultural patient groups in a polyglot society. DESIGN: This study was a secondary analysis of prospectively collected data. METHODS: We evaluated data from 9,134 patients with knee impairments from outpatient physical therapy clinics in Israel. Items were analyzed for DIF related to sex, age, symptom acuity, surgical history, exercise history, and language used to complete the functional survey (Hebrew versus Russian). RESULTS: Several items exhibited DIF, but unadjusted FS estimates and FS estimates that accounted for DIF were essentially equal (intraclass correlation coefficient [2,1]>.999). No individual patient had a difference between unadjusted and adjusted FS estimates as large as the median standard error of the unadjusted estimates. Differences between groups defined by any of the covariates considered were essentially unchanged when using adjusted instead of unadjusted FS estimates. The greatest group-level impact was <0.3% of 1 standard deviation of the unadjusted FS estimates. LIMITATIONS: Complete data where patients answered all items in the scale would have been preferred for DIF analysis, but only CAT data were available. CONCLUSIONS: Differences in FS outcomes between groups of patients with knee impairments who answered the knee CAT in Hebrew or Russian in Israel most likely reflected true differences that may reflect societal disparities in this health outcome.