Reliability of a structured assessment for nonclinicians to detect delirium among new admissions to postacute care

Objective: To evaluate the interrater reliability of a structured delirium assessment method for nonclinician interviewers in elderly patients newly admitted for postacute care. Design: Prospective assessment using dyads of nonclinician raters. Setting: Postacute (Medicare) units at 6 skilled nursing facilities. Participants: Forty elderly patients newly admitted for postacute care from medical or surgical units at acute care hospitals. Measurements: Subjects underwent dual delirium assessments within 5 days of admission. The standardized delirium assessment included the Mini-Mental Status Exam and Digit Span to assess overall cognitive function, the Delirium Symptom Interview to elicit specific delirium symptoms, the Memorial Delirium Assessment Scale to measure the severity of delirium, and the Confusion Assessment Method (CAM) to make the diagnosis of delirium. A coding protocol that linked observations to specific coding was used to improve reliability. Results: The structured delirium assessment process produced very high interobserver agreement for all instruments. Kappa for agreement on delirium diagnosis was 0.95. Conclusions: Nonclinician interviewers using a structured delirium assessment achieved reliability that rivaled or exceeded that of trained clinical assessors in other studies. Nonclinicians may offer an effective alternative for the assessment of delirium among postacute patients in skilled nursing facilities.