LAUSR

OBJECTIVE To describe the adaptive measurement of change (AMC) as a means to identify psychometrically significant change in reported function of hospitalized patients and to reduce respondent burden on follow-up assessments. DESIGN The AMC method uses multivariate computerized adaptive testing (CAT) and psychometric hypothesis tests based in item response theory to more efficiently measure intra-individual change using the responses of a single patient over two or more testing occasions. Illustrations of the utility of AMC in clinical care and estimates of AMC-based item reduction are provided using the FAMCAT, a newly developed functional multidimensional CAT-based (MCAT) measurement of basic mobility, daily activities, and applied cognition. SETTING Two quaternary hospitals in the Upper Midwest. PARTICIPANTS 495 hospitalized patients who completed the FAMCAT on two to four occasions during their hospital stay. INTERVENTION N/A. RESULTS Of the 495 patients who completed more than one FAMCAT, 72% completed two, 13% three, and 15% four sessions, with 22.1%, 23.4%, and 23.0%, respectively, exhibiting significant multivariate change. Use of the AMC in conjunction with the FAMCAT reduced respondent burden from that of the FAMCAT alone for follow-up assessments. On average, when used without the AMC, 22.7 items (range 20.4 to 24.4) were administered during FAMCAT sessions. Post-hoc analyses determined that when the AMC was used with the FAMCAT a mean (SD) reduction in FAMCAT number of items of 13.6(11.1), 13.1(9.8), and 18.1(10.8) would occur during the second, third, and fourth sessions, respectively, which corresponded to a reduction in test duration of 3.0(2.4), 3.0(2.8), and 4.7(2.6) minutes. Analysis showed that the AMC requires no assumptions about the nature of change and provides data that are potentially actionable for patient care. Various patterns of significant univariate and multivariate change are illustrated. CONCLUSIONS The AMC method is an effective and parsimonious approach to identifying significant change in patients' measured CAT scores. The AMC approach reduced FAMCAT sessions by an average of 12.6 items (55%) and 2.9 (53%) minutes among patients with psychometrically significant score changes.