| چکیده انگلیسی مقاله |
Abstract: Background: The patients undergoing cardiac operation are routinely sent to the intensive care unit (ICU) after the surgery. Due to this sections critical nature and in order to reduce health care costs, it is necessary to determine the features affecting the length of stay at ICU after cardiac surgery. This study aimed to investigate the most influential factors effecting on patients length of stay at ICU after cardiac surgery. Materials and Methods: Firstly, a literature review conducted in four main databases including Web of Science (ISI), PubMed, SCOPUS, PROQUEST using defined search strategy. Having applied the inclusion/exclusion criteria, 15 relevant papers were found revealing 32 factors effecting on the outcome variable of length of stay at ICU after cardiac surgery. Secondly, related data for found variables were collected in three ICU of hospitals of Shiraz/Iran including Namazi, Shahid Faghihi, and Al-Zahra Heart Hospital within the second quarter of 2016 through a non-random convenience sampling method. Data of 311 operated patients collected manually prospectively from the form available in the patients’ files, including Perfusion Data Sheet and Anesthesia Record Sheet. Those data not recorded in the patients’ files were collected through interview with the patients. After data collection and preparation, C5 CART decision tree were developed through dividing the entire datasets into two sets randomly: a training set consisted of 70% of the data for model development, and a test set consisted of the remaining (30%) for model validation. . Results: Among 32 variables cleared in literature review, finally 23 influential factors determined as important Features. For collected data, it was showed that the mean age and BMI of studied patients were 49.38 and 23.44 respectively. Meanwhile, the duration of aorta clamp time was 46.53 and CPB was 71.7 minute; the mean value of patients' hematocrit was 35.58%. Decision tree developed using training set (N=217) and evaluated by testing set (N=94). The results of model evaluation were sensitivity and G-Mean of 86% and 0.88% respectively. It achieved good discrimination power with AUC above 0.80. These variables came across the developed CART decision tree. There are as follow: age, Gender, Surgery type, hematocrit, Type of operation, Duration CPB, Clamp time, IVEF, Renal disease, Reoperation, Hypertension, OPCAB, CPB, sinus rhythm, Myocardia infraction, Mild valve pathology, NYHA, Creatinine, MIDCAB, HVS, hyper cholesterol, Preoperative infection, BMI. Conclusion: Our study emphasizes the utility of DT for exploring interactions between variables influencing on length of stay at ICU after cardiac surgery. Three main categories including demographic and individual characteristics (4 factors), surgical features (10 factors) and clinical features (18 factors) were defined. This study addressed more specialized factors including CLAMP TIME, cardiogenic shock, embolic history, left ventricular ejection volume, left ventricular dysfunction, coronary artery bypass through PUMP-OFF method, cardiopulmonary bypass and Heart valve surgery. |
