Volume 8, Issue 2 (12-2019)                   IEJM 2019, 8(2): 0-0 | Back to browse issues page

XML Print

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

Karimi N, Safari M, Mirzaei M, Kassaeian A, Roshanaei G, Omidi T. Determining the Factors Affecting the Survival of HIV Patients: Comparison of Cox Model and the Random Survival Forest Method. IEJM. 2019; 8 (2)
URL: http://iejm.hums.ac.ir/article-1-200-en.html
Modeling of Noncommunicable Diseases Research Center, School of Health, Department of Statistics, Hamadan University of Medical Sciences, Hamadan, Iran.
Abstract:   (76 Views)
Background: In recent years, sexually transmitted diseases such as AIDS have become an epidemic and are growing rapidly. Given the importance of controlling the disease in recent years, the awareness of the most important risk factors associated with patient survival is important. Therefore, this study aimed to determine the most important factors affecting the survival of HIV patients using the random survival forest (RSF) method.
Materials and Methods: In this retrospective study, medical records of 769 HIV patients in Hamadan Health Center from 1997 to 2017 were used to determine the most important factors in patient survival using Cox proportional hazards model and RSF method. The Brier score and C-index were applied to compare the Cox model and RSF method.
Results: Based on the results, 662 (86.1%) patients were male. The mean ± SD diagnosis age was 33.83 ± 9.63 years. Using Cox model, variables such as injection history, co-injection history, tuberculosis (TB) status, the first CD4 cell count, and the time of disease diagnosis until TB were determined to be variables affecting the survival of patients. According to the hazard ratio (HR), the risk of death for those with a history of injections was 12.328 times greater than that of non-injectors, and for those with TB, it was 13.565 times greater than that of non-TB patients. An increase in CD4 cell counts was associated with a decline in the risk of mortality. Based on the log-rank model, the variables such as the time until diagnosis of TB, the first CD4 cell count, ART, and history of co-injection had the highest impact on predicting the survival of HIV+ patients, respectively.
Conclusion: In case of the presence of many risk factors and the relationship between risk factors, the use of RSF offers a better performance in determining the influential survival factors as compared to Cox model which has limiting presumptions.
Type of Study: Original Article | Subject: Special
Received: 2019/10/20 | Accepted: 2019/10/2

Add your comments about this article : Your username or Email:

Send email to the article author

© 2020 All Rights Reserved | International Electronic Journal of Medicine

Designed & Developed by : Yektaweb