ABSTRACT
BACKGROUND: Risk factors for delirium are well-described, yet there is no widely used tool to predict the development of delirium upon admission in hospitalized medical patients. OBJECTIVE: To develop and validate a tool to predict the likelihood of developing delirium during hospitalization. DESIGN: Prospective cohort study with derivation (May 2010-November 2010) and validation (October 2011-March 2012) cohorts. SETTING: Two academic medical centers and 1 Veterans Affairs medical center. PATIENTS: Consecutive medical inpatients (209 in the derivation and 165 in the validation cohort) over age 50 years without delirium at the time of admission. MEASUREMENTS: Delirium assessed daily for up to 6 days using the Confusion Assessment Method. RESULTS: The AWOL prediction rule was derived by assigning 1 point to each of 4 items assessed upon enrollment that were independently associated with the development of delirium (Age ≥ 80 years, failure to spell "World" backward, disOrientation to place, and higher nurse-rated iLlness severity). Higher scores were associated with higher rates of delirium in the derivation and validation cohorts (P for trend < 0.001 and 0.025, respectively). Rates of delirium according to score in the combined population were: 0(1/50, 2%), 1(5/141, 4%), 2(15/107, 14%), 3(10/50, 20%), and 4(7/11, 64%) (P for trend < 0.001). Area under the receiver operating characteristic curve for the derivation and validation cohorts was 0.81 (0.73-0.90) and 0.69 (0.54-0.83) respectively. CONCLUSIONS: The AWOL prediction rule characterizes medical patients' risk for delirium at the time of hospital admission and could be used for clinical stratification and in trials of delirium prevention.