Determining levels of unawareness in dementia research.

Clinical methods used to determine unawareness in dementia exist; however, their applicability to empirical research is limited. The authors present a statistically derived approach to determining unawareness that addresses these limitations. Dementia patients (n=32) completed an awareness questionnaire. On an identical questionnaire, collateral sources (relatives or friends; n=32) provided their best estimate of participants' abilities. The authors compared cluster analysis, the proposed empirical approach, to a currently used standard deviation cutoff score approach. Cluster analysis included all participants, displayed sound statistical properties, and was more sensitive to between-group differences in psychotic symptoms than standard deviation cutoff. Cluster analysis appears more appropriate for understanding the overall spectrum of unawareness in dementia research.