Volume 18, Issue 61 (7-2015)                   jha 2015, 18(61): 57-68 | Back to browse issues page


XML Persian Abstract Print


Download citation:
BibTeX | RIS | EndNote | Medlars | ProCite | Reference Manager | RefWorks
Send citation to:

Karim H, Etminani K, Tara M, Mardani M. Identifying Factors Associated with Length of Hospital Stay Using Decision Tree. jha. 2015; 18 (61) :57-68
URL: http://jha.iums.ac.ir/article-1-1770-en.html

1- MSc Student of Medical Informatics, Faculty of Medicine, Mashhad University of Medical Sciences, Mashhad, Iran Mashad University of medical Sciences , Karim.hesam@gmail.com
2- Assistant Professor of Medical Informatics Department, Faculty of Medicine, Mashhad University of Medical Sciences, Mashhad, Iran Mashad University of medical Sciences
3- BSc of Medical Record, ShohadayeDehaghan Hospital, Isfahan University of Medical Sciences, Isfahan, Iran Shohadaye Dehaghan Hospital, Isfahan
Abstract:   (1192 Views)
Introduction: One of the main factors of hospital costs resulting in hospital resources constraints is Length of Stay (LOS). Recognizing factors associated with this index can result in optimized utilization of resources and reducing LOS. The present study has been conducted with the aim of investigating LOS-related factors using decision tree applied on hospital admission data. 
 Methods: In this descriptive-retrospective study, 188068 patient data were extracted from computerized patient records including 16 variables in all wards in Ghaem Hospital in Mashhad, Iran, during 2009-2014. After data cleaning, in order to determine the factors associated with LOS, statistical procedures including univariate analysis, chi-Square statistical test, one-way ANOVA, and CHAID algorithm were applied on our data using SPSS Statistical Software version 19. 
Results:
The mean and median of LOS in Ghaem hospital were 6. 5 and 4 days, respectively. Out of 15 independent variables entered the model, 11 variables including the ward, the patient’s age, the admission reason, the doctor, the referral status, the patient’s job, the marital status, the status of the residential place, the caregiver(s), admission day of the week, and the insurance type were associated with LOS. Conclusion: The results of data mining techniques can be different based on the input data. Therefore, to identify the factors related to LOS, it is necessary to enter the related data of individual hospitals into the model and to use the obtained results at the same hospital.
Full-Text [PDF 627 kb]   (1007 Downloads)    
Type of Study: Research | Subject: General
Received: 2015/01/6 | Accepted: 2015/08/22 | Published: 2015/08/22

Send email to the article author


© 2015 All Rights Reserved | Journal of Health Administration

Designed & Developed by : Yektaweb