Psychological measures: reliability in the assessment of stroke patients.
OBJECTIVE: To determine whether acute stroke patients can give reliable responses to standardized psychological measures. DESIGN: Survey design with retrospective review of Neurobehavioral Cognitive Status Examination (NCSE) scores. SETTING: A rehabilitation hospital in Philadelphia. PATIENTS: 106 consecutive admissions to stroke services at a rehabilitation hospital were evaluated according to the following inclusion criteria: Patients at least 65 years of age, English speaking, having a documented history of cerebrovascular accident (CVA), possession of receptive speech comprehension, and the ability to maintain attention adequate for participation in the study interview. Of the 47 eligible for participation, 10 patients refused participation and 13 were discharged before the interview could be completed. Thus 24 patients were interviewed, with one patient unable to complete the second half of the interview. MAIN OUTCOME MEASURES: In phase 1, the patient's performance on a structured interview (including the Multidimensional Health Locus of Control Scales (MHLC), Life Orientation Test (LOT), and Center for Epidemiological Studies Depression Scale (CES-D) scales) produced a Total Reliability Score, indicating the overall reliability of the patient's responses. In phase 2, NCSE scores were examined (retrospectively) for their ability to predict the patient's Total Reliability Score on the structured interview. RESULTS: Using Total Reliability Scores, 9 patients were categorized as "passing," 9 were categorized as "uncertain," and 5 were categorized as "failing" the interview. The NCSE was predictive for individuals who had a very low likelihood of being able to respond consistently. The NCSE failed to identify which individuals would respond in a reliable fashion. CONCLUSIONS: It is both possible and important to assess response reliability when using psychological measures soon after stroke. Future research will need to document other potential predictors of interview performance, including combinations of NCSE subscales.