A computerized adaptive test for patients with shoulder impairments produced responsive measures of function

Background. Computerized adaptive tests (CATs) promise efficient outcomes data collection with little loss of measurement precision. The shoulder CAT has not been assessed for administrative efficiency, nor have prospective studies been completed to evaluate the sensitivity to change or the responsiveness of CAT-based functional status (CAT-FS) measures. Objective. The purpose of this study was to evaluate the efficiency and responsiveness of the shoulder CAT. Design. This was a secondary analysis of prospectively collected data. Methods. Data were analyzed from patients with shoulder impairments who received outpatient rehabilitation in 518 clinics in 30 US states. Over the testing time, 30,987 patients completed the shoulder CAT at intake, and of these, 13, 805 completed the CAT at discharge (45% completion rate). The efficiency of routine CAT administration was evaluated, and the sensitivity to change and responsiveness of CAT-FS measures were assessed. Results. On average, patients took about 1.5 minutes (SD <2 minutes) to answer 6 CAT items (SD=2), which produced precise estimates of CAT-FS measures that adequately covered the content range and had negligible floor and minimal ceiling effects. Ninety-four percent of the patients had CAT-FS scores between 20 and 80, where upper-level 95% confidence interval (CI) standard errors were between 3.2 and 4.6 (out of 100). Of patients with both intake and discharge data, 79% had CAT-FS change scores greater than minimal detectable change, and 76% had changes greater than minimal clinically important improvement. Limitations. Because this study was a secondary analysis, the results may have been affected by patient selection bias. Future studies would benefit from more complete data. Conclusions. The results indicate the shoulder CAT was efficient and support the precision, sensitivity, and responsiveness of CAT-FS measures.