Background and Aim: Length of stay (LOS) in a hospital is one of the best hospital indicators that can be used for various purposes. In this survey, we studied the hospital LOS and its associated factors in Tehran University of Medical Sciences Women's Hospital (a teaching hospital) in Tehran using the Cox proportional hazards semi parametric model and compared the results with the results obtained using the multiple linear regression.

Materials and Methods: This was a descriptive-analytical study in which we reviewed 3421 files of inpatients hospitalized in, and those discharged from, the oncology, surgery and obstetrics units in 2008. The required data were collected using a data collection sheet and inpatient interviews. A P<0.05 was considered statistically significant.

Results: The median of patients' LOS in the hospital was 50.8 hours, that in the obstetrics, surgical and oncology units being 48.5, 54.4 and 94.2 hours, respectively. Of all the patients, 2632 (76.9%) had been discharged with a recovery status and the rest (23.1%) with a no-recovery status. Results of the Cox proportional hazards model showed that the following variables had increased LOS: a distance longer than 200 km between a patent's residence and the hospital, hospitalization in the surgery and oncology units, admission on a Thursday, admission by an internist, hospitalization for neoplastic, endocrine, nutritional, or genitourinary system diseases (P<0.005), as well as a high number of diagnostic laboratory tests, radiographies or sonographies (P<0.001). Patients admitted and hospitalized as an emergency case had a shorter LOS (P<0.001) than others. On the other hand, based on the multiple linear regression model results, some occupations (being a worker, a farmer, a stockbreeder, or a retired spouse) admission on a Thursday, (The first day of the weekend in Iran), suffering from a neoplastic disease, and a high number of diagnostic tests or radiographies or sonographies increased, and admission by a resident decreased, patients' LOS (P<0.05).

Conclusion: Considering having censored data, the Cox proportional hazards model is a more suitable model than the multiple linear regression models for identifying factors influencing patients' LOS in a hospital. From among the LOS Cox model's associated factors as identified in this study, policy-makers and managers can only change admission days and the number of diagnostic tests. That is to say, they should try to prevent admission on a Thursday (unless emergency cases) and also perform the required primary diagnostic tests before admitting a patient into the hospital, which would lead to a more effective utilization of hospital beds and other resources.