Impact of frailty in cure rate models: A systematic review.
Cure rate models dominate discussions on survival analysis as can be used to indicates the effectiveness of some drugs/techniques/methods etc. This is in addition to giving a better fit than those models that does not include the cure fraction. The influence of individual heterogeneity known as frai...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Conference or Workshop Item |
| Published: |
2023
|
| Subjects: | |
| Online Access: | http://eprints.utm.my/107975/ http://dx.doi.org/10.1063/5.0110359 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Universiti Teknologi Malaysia |
| Summary: | Cure rate models dominate discussions on survival analysis as can be used to indicates the effectiveness of some drugs/techniques/methods etc. This is in addition to giving a better fit than those models that does not include the cure fraction. The influence of individual heterogeneity known as frailty in cure fraction estimates has been recognized. But its role when dealing with different type of time-to-event data, type of cure rate model e.t.c., has not been systematically exploited. The objective of the study is to systematically review literature related to the role of frailty in cure rate models. We developed a search strategy to identify relevant literature using SCOPUS database. All searches cover the period from inception of data base to 2020 and included only journal articles published in English. The selection criteria were based on Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) methodology in which the search was tailored on mapping existing literature on frailty in cure rate models. Although many distributions were discovered to be used for frailty, yet Gamma is the dominant distribution irrespective of type of cure rate model, time-to-event data and whether baseline distribution is assumed or not. Frailty distributions were found to be slightly align to the type of cure model. It was also found that, what frailty control depends on nature of time to event data. In univariate survival data, frailty mostly accounts for individual heterogeneity. When the data is multivariate, it controls individual heterogeneity and accounts for different types of relationships such as correlations between subjects within cluster or heterogeneity between clusters. Introducing frailty in defective models leads to generate a new defective distribution. Generally introducing frailty in to the cure model improves the fit of the model. There is need to introduce frailty in zero survival group of cure models as well as to use distributions other that gamma in defective cure models. |
|---|