Validation of a Nurse-Based Delirium-Screening Tool for Hospitalized Patients

Background: Guidelines recommend daily delirium monitoring of hospitalized patients. Available delirium screening tools have not been validated for use by nurses among diverse inpatients. Objective: We sought to validate the Nursing Delirium-Screening Scale (NuDESC) under these circumstances. Methods: A blinded cross-sectional and quality-improvement study was conducted from August 2015 February 2016. Nurses' Nu-DESC scores were compared to delirium diagnosis according to Diagnostic and Statistical Manual of Mental Disorders-5 (DSM-5) criteria. A total of 405 consecutive hospitalized patients were included. Nu-DESC-positive (threshold score >= 2) patients were matched with equal numbers of Nu-DESC-negative patients, by sex, age, and nursing unit. Nurses recorded a Nu-DESC score for each patient on every 12-hour shift. A Nu-DESC-blinded evaluator interviewed patients for 2 consecutive days. Delirium diagnosis was determined by physicians using DSM-5 criteria applied to collected research data. Sensitivity and specificity of the NuDESC were calculated. In an exploratory analysis, the performance of the Nu-DESC was analyzed with the addition of bedside measures of attention. Results: The sensitivity of the Nu-DESC at a threshold of >= 2 was 42% (95% CI: 33-53%). Specificity was 98% (97 98%). At a threshold of >= 1, sensitivity was 67% ( 52 80%) and specificity 93% (90-95%). Similar results were found with the addition of attention tasks. Conclusion: The Nu-DESC is a specific delirium detection tool, but it is not sensitive at the usually proposed cut point of >= 2. Using a threshold of >= 1 or adding a test of attention increase sensitivity with a minor decrease in specificity.