1The CT Lamont Centre, Élisabeth-Bruyère Research Institute, Ottawa. 2The Division of Geriatric Medicine, Department of Internal Medicine, University of Ottawa. 3The Clinical Epidemiology Program, the University of Ottawa Health Research Institute, Ottawa. 4DementiaNET—a CIHR New Emerging Team. 5The Department of Epidemiology and Community Medicine, University of Ottawa. 6Regional Geriatric Program of Eastern Ontario.
Objective: To derive and validate simple screens of 2 to 4 questions to identify cognitive impairment.
Design: Cross-sectional database analysis.
Setting and Participants: Community-dwelling participants in the Canadian Study of Health and Aging with mild to moderate cognitive impairment (N = 958), or normal cognition (N = 602).
Measurements: Cognitive questions not requiring paper, pen, cue cards, props or more than 30 seconds to answer were selected from the Modified Mini-Mental State Examination (3MS). A sequential weighting approach was applied to logistic regression analyses to create scales of equally-weighted questions using the first 2, 3 or 4 questions from the regression equations.
Sensitivities and specificities were calculated for all cutoffs. Two sets of questions, which approximated the psychometric properties of the 3MS, were validated in a second database.
Results: The two tests whose properties approached those of the 3MS (sensitivity 84%, specificity 62%) were the Ottawa
3D test: Day, Date, DLROW (sensitivity 76%, specificity 62%) and the Ottawa 3DY test: Day, Date, DLROW, Year (sensitivity 80%, specificity 61%).
Conclusions: The Ottawa 3D and 3DY tests show promising psychometric properties and are easy enough to employ to promote widespread use but must be revalidated in the target groups for which they are intended.
The full potential value of the Ottawa 3D and 3DY tests can best be understood in the context of ‘serial trichotomization’ cognitive screening or case-finding algorithms. Dementia researchers focusing on biomarkers and neuroimaging should consider similar trichotomization approaches as these may improve the sensitivities and specificities of their tests.
PDF (307.06 KB PDF FORMAT)
RIS citation (ENDNOTE, REFERENCE MANAGER, PROCITE, REFWORKS)
BibTex citation (BIBDESK, LATEX)
I must say publishing with Evolutionary Bioinformatics was a great and partially unexpected experience. It allowed me to understand the name 'Libertas Academica' is not just a name but a real issue. The reviewers offered us stringent, rigorous but totally unprejudiced comments, the editor supported us and the paper visibility was very good. Thank you again.