Development of a computerized adaptive test for assessing activities of daily living in outpatients with stroke

Background. An efficient, reliable, and valid measure for assessing activities of daily living (ADL) function is useful to improve the efficiency of patient management and outcome measurement. Objective. The purpose of this study was to construct a computerized adaptive testing (CAT) system for measuring ADL function in outpatients with stroke. Design. Two cohort studies were conducted at 6 hospitals in Taiwan. Methods. A candidate item bank (44 items) was developed, and 643 outpatients were interviewed. An item response theory model was fitted to the data and estimated the item parameters (eg, difficulty and discrimination) for developing the ADL CAT. Another sample of 51 outpatients was interviewed to examine the concurrent validity and efficiency of the CAT. The ADL CAT, as the outcome measure, and the Barthel index (BI) and Frenchay Activities index (FAI) were administered on the second group of participants. Results. Ten items did not satisfy the model's expectations and were deleted. Thirty-four items were included in the final item bank. Two stopping rules (ie, reliability coefficient >.9 and maximum test length of 7 items) were set for the CAT. The participants' ADL scores had an average reliability of.93. The CAT scores were highly associated with those of the full 34 items (Pearson r=.98). The scores of the CAT were closely correlated with those of the combined BI and FAI (r=.82). The time required to complete the CAT was about one fifth of the time used to administer both the BI and FAI. Limitations. The participants were outpatients living in the community. Further studies are needed to cross-validate the results. Conclusions. The results demonstrated that the ADL CAT is quick to administer, reliable, and valid in outpatients with stroke.