Risk assessment in liver transplantation patients: A shared frailty parametric approach

详细信息    查看全文

• 作者：Dilip C. Nath^a ; ^{dilipc.nath@gmail.com" class="auth_mail" title="E-mail the corresponding author} ; Atanu Bhattacharjee^b ; Ramesh K. Vishwakarma^a
• 关键词：Lognormal frailty ; Akaike Information Criteria ; Bayesian Information Criteria
• 刊名：Clinical Epidemiology and Global Health
• 出版年：2016
• 出版时间：March 2016
• 年：2016
• 卷：4
• 期：1
• 页码：1-15
• 全文大小：2142 K

文摘

The frailty model consists of two components; the first is baseline hazard. If baseline hazard is known in advance, then, the parametric estimation approach can be used advantageously. The second component of frailty model is frailty distribution. In this paper, we compared the risk of mortality by different parametric frailty models, namely gamma, inverse Gaussian, positive stable and lognormal, with respect to Akaike Information Criteria (AIC) and Bayesian Information Criteria (BIC) values on liver transplantation patients. We conclude that lognormal frailty model gives slightly lower frailty variance compared to the other frailty models when Weibull distribution is considered as baseline hazard. Therapy and MELD (model for end-stage liver disease) scores may have significant role in liver transplant patients when distribution of frailty is considered as lognormal with Weibull baseline hazard for fitting the model. Therefore, there is 4.6 times more risk in getting hazard of mortality for patients receiving conventional therapy with consent for participation as compared to patients receiving new therapy with consent for participation with 95%CI [1.059, 20.199], p-value 0.042.