Development of a clinical score, PANDA, to predict delirium in stroke care unit

Background and purpose: Delirium frequently complicates acute stroke and worsens outcomes. Because delirium is potentially preventable, predicting its occurrence is essential. Although several prediction scores have been proposed, nurses need to quickly predict delirium in stroke care units (SCUs). We aimed to develop a simple tool for this purpose by examining a comprehensive set of potential predictors.

Methods: This is a prospective cohort study on acute stroke patients admitted to an SCU. Patients without stupor, coma, or delirium upon admission were eligible. Participants were followed for 5 days from admission. Delirium was defined as Intensive Care Delirium Screening Checklist ≥4 points. We examined 27 potential predictors, of which 13 predictors were used to developed a least absolute shrinkage and selection operator-penalized logistic regression model. Five variables with the largest coefficients were assigned one point each in the prediction score. The internal validation was performed by bootstrapping.

Results: Delirium occurred in 42 of the 387 participants. The score consisted of prior delirium, alcohol, NIHSS ≥5, dementia, and auditory/visual impairment (PANDA). The apparent AUC was 0.84 (95% confidence interval [CI], 0.78-0.89), and the optimism-corrected AUC was 0.81 (95% CI, 0.73-0.88). With a cutoff of ≥2 points, sensitivity was 0.78 (95% CI, 0.65-0.90), and specificity was 0.74 (95% CI, 0.70-0.79).

Conclusions: PANDA score is simple and predicts delirium in an SCU satisfactorily.