Test-Retest Reliability of a Computerized Adaptive Depression Screener
Abstract
Objective:
Computerized adaptive testing (CAT) provides improved precision and decreased test burden compared with traditional, fixed-length tests. Concerns have been raised regarding reliability of CAT-based measurements because the items administered vary both between and within individuals over time. The study measured test-retest reliability of the CAT Depression Inventory (CAT-DI) for assessment of depression in a screening setting where most scores fall in the normal range.
Methods:
A random sample of adults (N=101) at an academic emergency department (ED) was screened twice with the CAT-DI during their visit. Test-retest scores, bias, and reliability were assessed.
Results:
Fourteen percent of patients scored in the mild range for depression, 4% in the moderate range, and 3% in the severe range. Test-retest scores were without significant bias and had excellent reliability (r=.92).
Conclusions:
The CAT-DI provided reliable screening results among ED patients. Concerns about whether changes in item presentation during repeat testing would affect test-retest reliability were not supported.