Does Adaptive Cognitive Testing Combine Efficiency with Precision? Prospective Findings

Article type: Research Article

Authors: Wouters, Hans^{a; g; *} | van Campen, Jos^f | Appels, Bregje^e | Lindeboom, Robert^a | Buiter, Maarten^a | de Haan, Rob J.^c | Zwinderman, Aeilko H.^a | van Gool, Willem A.^b | Schmand, Ben^{b; d}

Affiliations: [a] Department of Clinical Epidemiology, Biostatistics and Bioinformatics Academic Medical Centre, Amsterdam, the Netherlands | [b] Department of Neurology Academic Medical Centre, Amsterdam, the Netherlands | [c] Clinical Research Unit, Academic Medical Centre, Amsterdam, the Netherlands | [d] Department of Psychology, University of Amsterdam, Amsterdam, the Netherlands | [e] Department of Medical Psychology Slotervaart Hospital, Amsterdam, the Netherlands | [f] Department of Geriatric Medicine, Slotervaart Hospital, Amsterdam, the Netherlands | [g] Department of Pharmacoepidemiology and Pharmacotherapy, Utrecht Institute for Pharmaceutical Sciences (UIPS), Utrecht University, Utrecht, the Netherlands

Correspondence: [*] Correspondence to: Hans Wouters, Department of Pharmacoepidemiology and Pharmacotherapy, Faculty of Science, Utrecht Institute for Pharmaceutical Sciences (UIPS), Utrecht University, 3584 CG Utrecht, The Netherlands. Tel.: +31 (0)6 22 73 60 73; E-mail: [email protected].

Abstract: Longer cognitive tests, such as the Alzheimer's disease assessment scale (ADAS-cog) or the Cambridge cognitive examination (CAMCOG), are more precise but less efficient than briefer tests, such as the Mini Mental State Examination (MMSE). We examined if computerized adaptive testing (CAT) of cognitive impairment can combine brevity with precision by tailoring a precise test to each individual patient. We conducted a prospective study of 84 participants [normal aging, n = 41; mild cognitive impairment (MCI), n = 21; dementia, n = 22]. CAT estimated a participant's ability during testing by selecting only items of appropriate difficulty from either the CAMCOG or the CAMCOG supplemented with ADAS-cog items and neuropsychological tests (the CAMCOG-Plus). After tailored testing with CAT, the remaining CAMCOG and CAMCOG-Plus items not selected by CAT were administered. The time needed to complete the CAT was compared to that needed for the whole CAMCOG and CAMCOG-Plus. Results showed that testing time reductions achieved with CAT were 37% or more compared to the whole CAMCOG and 55% or more compared to the whole CAMCOG-Plus. Estimated ability levels with CAT were in excellent agreement with those based on the whole CAMCOG and CAMCOG-Plus (intraclass correlations 0.99 and 0.98, respectively). Diagnostic accuracy of detecting mild dementia and MCI seemed better for the CAT administered tests than for the MMSE, but the differences were not significant. We conclude that adaptive testing combines brevity with precision, especially in grading the severity of cognitive impairment.

Keywords: Algorithms, Alzheimer's disease, amnesia, cognition disorders, dementia, logistic models, neuropsychology, vascular dementia

DOI: 10.3233/JAD-2011-101743

Journal: Journal of Alzheimer's Disease, vol. 25, no. 4, pp. 595-603, 2011