| Abstrakt: |
The Length of Stay (LOS) is an important indicator that can be useful for financial and management of hospital planning. If identify the patients after admission who will have a long stay at hospital, suitable resources can be available in order to accelerating health care at the primary time. This research is a kind of descriptive retrospective study. The samples are consist of 4120 computerized patients' medical records (each data had 12 variables) from 2014 February to 2015 March at Sina Hospital and Maternity in Mashhad, Iran. After data cleaning, for determination of correlated ingredients to LOS, we use the single variable analysis consist of Chi-square tests, unilateral variance analysis and so use CHAID Decision Tree algorithm in SPSS software, version 19. The average and median of LOS in the mentioned hospital were 2.3 and 1 day respectively. Between 12 variables that entered to decision Tree, 8 of them recognized as correlated ingredients to LOS. These variables consist of primary diagnosis, age, occupation, attending physician, habitat place, marriage status, previous hospitalization history and clinical insurance. According to clinical and demographic variables at the time of patient admission, could predict patient Length of Stay, however the result of Decision Tree can be different due to entered data. Therefore, in order to create a LOS prediction model, we have to use special hospital data separately from each other in Decision Tree and use the result of them in the same care center. [ABSTRACT FROM AUTHOR] |
