Diagnosis of postoperative delirium in older adults using the Confusion Assessment Method for the intensive care unit in non-intensive care unit settings: A test modification might improve its diagnostic performance

Aim: To evaluate the diagnostic performance of the Confusion Assessment Method for the intensive care unit (CAM-ICU) among postoperative older patients in non-ICU settings. Methods: The CAM-ICU was used by trained staff to prospectively evaluate postoperative patients for delirium. The patients were aged ≥60 years, were in general wards and had no critical illnesses. The assessments occurred for 7 consecutive days after surgery. The results were compared with delirium diagnoses obtained by geriatricians using Diagnostic and Statistical Manual of Mental Disorders 5th edition criteria as the reference standard. Results: The sensitivity of delirium detection for the CAM-ICU was 31.6% (95% confidence interval [CI] 12.6–56.6), while the specificity was 97.6% (95% CI 94.9–99.1), positive predictive value was 50.0% (95% CI 26.3–73.7) and negative predictive value was 95.0% (95% CI 93.3–96.3). Feature 4 (disorganized thinking) yielded the highest sensitivity (60%; 95% CI 14.7–94.7), whereas feature 2 (inattention) had low sensitivity (36.8%; 95% CI 16.3–61.6). Further analyses to explore the highest sensitive criteria showed that if CAM-ICU diagnoses were made by the presence of any two out of feature 1 (acute change or fluctuation of cognition), feature 3 (altered level of consciousness) or feature 4, the sensitivity increased substantially to 80.0% (95% CI 28.4–99.5), with a reasonably high specificity of 81.8% (95% CI 48.2–97.7). Conclusions: Modification of the flow of delirium diagnosis using the CAM-ICU appears to offer a better sensitivity for detecting delirium in non-ICU settings. Furthermore, changing feature 2 to evaluate patients’ attention levels over a longer period of time might yield a better diagnostic performance. Geriatr Gerontol Int 2019; 19: 762–767.