Prediction models on causing factors for delirium in intensive care unit patients

Abstract

This study intends to screen high-risk group for delirium in the intensive care unit by building optimal prediction models through multidimensional analysis of the factors causing delirium in intensive care unit patients, thereby providing the theoretical foundation for the development of protocols for delirium prevention and management in such patients. The data were collected by looking through the electronic medical records by 3,781 patients who were admitted to the medical/ surgical intensive care units. The research tools used were the questionnaire prepared by the researcher based on literature review, the results of previous studies, and the delirium screening tool developed by Schuumans et al. The collected data were analyzed using SPSS. In addition, the prediction model was built using the SPSS program to conduct decision-tree analysis technique and obtain the optimal models that predict delirium based on the explanatory variables related to delirium. The prediction model for delirium occurrence in intensive care unit patients appeared in seven paths. As a result of evaluating the model built in this study, the misclassification rate was 0.145, sensitivity was 0.857, specificity was 0.854, and positive predictive value was 0.780. Therefore, as the delirium occurrence prediction model proposed in this study provides the theoretical framework to prevent delirium by screening the high-risk group in the intensive care unit, it is expected that it can be beneficially used in clinical nursing practice. In addition, such model will contribute to improving quality of care, preventing and managing the delirium of intensive care unit patients, and reducing their medical care costs. Furthermore, continued studies for the development of effective protocols for delirium prevention and management, as well as for the development of standardized delirium assessment tools, are needed based on the results of this study.