PROPENSITY SCORE METHODS FOR COMPETING RISKS
Public DepositedAdd to collection
You do not have access to any existing collections. You may create a new collection.
Downloadable Content
Download PDFCitation
MLA
Aimyong, Natnaree. Propensity Score Methods For Competing Risks. Chapel Hill, NC: University of North Carolina at Chapel Hill Graduate School, 2014. https://doi.org/10.17615/ch1v-a404APA
Aimyong, N. (2014). PROPENSITY SCORE METHODS FOR COMPETING RISKS. Chapel Hill, NC: University of North Carolina at Chapel Hill Graduate School. https://doi.org/10.17615/ch1v-a404Chicago
Aimyong, Natnaree. 2014. Propensity Score Methods For Competing Risks. Chapel Hill, NC: University of North Carolina at Chapel Hill Graduate School. https://doi.org/10.17615/ch1v-a404- Last Modified
- March 19, 2019
- Creator
-
Aimyong, Natnaree
- Affiliation: Gillings School of Global Public Health, Department of Biostatistics
- Abstract
- Non-experimental studies have increasingly been used to examine the safety and effectiveness of medication. Challenges to this method include confounding, which may cause the estimator to be biased. Propensity score (PS), which is the conditional probability of receiving treatment given all confounders, may be used to control for confounding. Analysis of vulnerable populations may involve competing risks, which may occur before the event of interest. Statistical methods that account for competing risks are needed to obtain valid causal estimate. However, little knowledge attention has been given to this topic in the literature. The objective of this research was to investigate the performance of estimators under imple- mentation of various PS methods in competing risk survival analyses for estimating marginal and conditional treatment effects. The competing risk models were a cause-specific hazard model and subdistribution hazard model. According to simulation results, the weighted method produced efficient estimators for marginal treatment effects. However, it leads to an inflated variance when low incidence of event and strong confounder effects. A bootstrapping method can be used to estimate the variance under this scenario. For the conditional treatment effect, PS adjustment in the model performed the best for the null model. Depending on the sample size and the number of confounding variables, the subclassification and matching methods yield best performance under the alternative when treatment effect is non-null. Heterogeneity of treatment effect associated with statin therapy may be present in el- derly who experience myocardial infarction. Examining treatment effect across age groups and the revascularization procedure illustrated the heterogeneity of statin effects. Statins significantly reduce risks of heart failure among younger age groups. The combination of statins with revascularization procedures presents better treatment effects than occurs with statins alone. Application of propensity score methods to competing risks is illustrated in this study, with the analysis of treatment effects providing an improved understanding of the heterogeneity of the effects of statins therapy. The efficiency of implementing propensity score method to competing risks is illustrated in this study. Analyzing the treatment effects by subgroup and medical procedure contributes better results for estimating the heterogeneity treatment effect.
- Date of publication
- December 2014
- Keyword
- Subject
- DOI
- Identifier
- Resource type
- Rights statement
- In Copyright
- Advisor
- Ivanova, Anastasia
- Suchindran, Chirayath
- Brookhart, M. Alan
- Fine, Jason
- Schwartz, Todd
- Degree
- Doctor of Public Health
- Degree granting institution
- University of North Carolina at Chapel Hill Graduate School
- Graduation year
- 2014
- Language
- Publisher
- Place of publication
- Chapel Hill, NC
- Access right
- There are no restrictions to this item.
- Date uploaded
- April 22, 2015
Relations
- Parents:
This work has no parents.
Items
Thumbnail | Title | Date Uploaded | Visibility | Actions |
---|---|---|---|---|
Aimyong_unc_0153D_14886.pdf | 2019-04-12 | Public | Download |