Predictors of acute confusional states in hospitalized elderly patients

Abstract

Development of two models for predicting acute confusional states in elderly postsurgical hip-fracture patients is described. Admission variables were used in the first model to predict confusion in the first five postoperative days; treatment and clinical progress data were added for the second model to predict on a day-to-day basis. The main sample was 170 patients from four hospitals (M age = 78.8) with no prior history of mental impairment. Over half (51.5%) evidenced some confusion postoperatively. Three variables were significant in the admission model: age, errors on a mental status test, and level of pre-injury activity. Age and test errors remained significant in the daily model and urine elimination problems were added, but the strongest predictor was the previous day's confusion score. When that effect was removed, pain, narcotics, and mobility emerged as important predictors. Validity testing suggested that the three admission factors were core predictors of acute confusional states.