Diagnosis of delirium and evaluation of sleep quality in patients hospitalized in the anesthesia and reanimation intensive care unit

INTRODUCTION AND AIM: Delirium is a complex process that causes mood and behavioural changes in intensive care patients, affects the sleep patterns of patients in the later stages, and can be prevented with good nursing care; when necessary, precautions are taken with rapid intervention. This study was conducted to diagnose delirium in patients hospitalised in an anaesthesia and reanimation intensive care unit and to examine the effect of delirium on sleep quality. MATERIALS AND METHODS: This study, planned as a descriptive and cross-sectional one, was conducted with the participation of 107 patients hospitalised in the Anesthesia and Reanimation Intensive Care Unit of Sakarya Training and Research Hospital. Patient Information Form, Nursing Delirium Screening Scale, Richards-Campbell Sleep Scale, Richmond Agitation-Sedation Scale and Glasgow Coma Scale were used as data collection tools. The data obtained in the study were evaluated using IBM SPSS Statistics 26 program, R programming language version 4.1.3 and G*Power 3.1. RESULTS: Among the patients who participated in the study, 49.9% were female, 55,1% were male, 43% were between 56-75 years of age 48,6% were primary school graduates, and 89.7% did not exercise. It was understood that the Glasgow Coma Scale scores of the patients were 14,07±0,80 and their Richmond Agitation-Sedation Scale scores were -0,38±1,06 indicating that the patients were not in a coma and could be evaluated for delirium. It was found that 36,4% of the study group had delirium, and the scores of the Nursing Delirium Screening Scale of patients with delirium were 3,10±1,27 the scores of the Richards-Campbell Sleep Scale were 39,30±18,62 and their sleep quality was lower than patients without delirium (p<0.05). There was no statistically significant difference between individuals diagnosed with delirium's Richards-Campbell Sleep Scale scores and socio-demographic characteristics (p>0.05). SHAP values show each variable's contribution or importance in the model's estimation. The essential variables in the model to predict the Delirium Status variable were determined. CONCLUSION: Patients hospitalised in the anaesthesia and reanimation intensive care unit and diagnosed with delirium have lower sleep quality than patients without delirium. Delirium diagnosis and observation tools designed for nurses in these units will ensure that necessary precautions are taken before patients enter delirium. Therefore, patients' sleep quality will improve, their recovery process will accelerate, and their quality of life will increase. Machine learning, which is used in many different fields today, is one of the subfields of artificial intelligence. As a result of the analysis for prediction with the machine learning approach used in this study, it is seen that the essential variables that should be in the model to predict the Delirium Status variable are Glasgow Coma Scale, Richards Campbell Sleep Scale and Richmond Agitation Sedation Scale, respectively. Conducting studies with new models for prediction with machine learning approaches is recommended. Keyword: Delirium, Intensive Care, Machine Learning, Nursing, Sleep Quality