Ultra-brief Screeners for Detecting Delirium Superimposed on Dementia

Delirium superimposed on dementia (DSD) is common, morbid, and costly, yet frequently undiagnosed. Our study aimed to develop a brief screening test to improve health care worker recognition of DSD.

Older hospitalized adults with dementia were prospectively enrolled from medical and surgical inpatient units of 3 hospitals (2 in Pennsylvania, 1 in Tennessee).

The reference standard delirium assessment used Confusion Assessment Method (CAM) criteria and was based on a structured interview including the Mini-Mental State Examination, interviewer observations, and medical record review. To develop the screening test, 1-, 2-, and 3-item combinations from the reference standard assessment were analyzed to determine their sensitivity and specificity in diagnosing delirium presence in a dementia population compared to the reference standard. For multiple-item screeners, error on 1 or more items was considered a positive screen.

Overall, 391 older adults with dementia were enrolled (mean age: 83.9 years, 71.1% female), and 95 (24.4%) developed DSD during their hospitalization, based on the reference standard. The best single-item screen for DSD was "What day of the week is it?" with 84% sensitivity [95% confidence interval (CI): 0.75, 0.91] and 41% specificity (CI: 0.35, 0.47). The best 2-item screen was "list the days of the week backwards" and "What day of the week is it?" with 93% sensitivity (CI: 0.85, 0.97) and 30% specificity (CI: 0.25, 0.36). The best 3-item screen was "list the days of the week backwards," "What type of place is this? [hospital]" and "Does the patient appear sleepy?" with 94% sensitivity (CI: 0.87, 0.98) and 42% specificity (CI: 0.36, 0.48).

We identified a 3-item DSD screener with excellent sensitivity but limited specificity. This screener can be used to quickly rule out DSD in populations with a high prevalence of dementia and is a promising step toward developing efficient tools for DSD recognition among care providers.