Likelihood-Based Methods of Mediation Analysis in the Context of Health Disparities

Usher, Therri Alexandria

Likelihood-Based Methods of Mediation Analysis in the Context of Health Disparities

dc.contributor.advisor	Fallin, Daniele (Dani)
dc.contributor.committeeMember	Rohde, Charles A.
dc.contributor.committeeMember	Bandeen-Roche, Karen J.
dc.contributor.committeeMember	Thorpe, Roland J.
dc.creator	Usher, Therri Alexandria
dc.date.accessioned	2017-04-19T12:26:47Z
dc.date.available	2017-04-19T12:26:47Z
dc.date.created	2016-08
dc.date.issued	2016-07-20
dc.date.submitted	August 2016
dc.date.updated	2017-04-19T12:26:47Z
dc.description.abstract	African-Americans experience higher incidences of death and disability compared to non-Hispanic whites. Much of the existing research has focused on identifying the existence of health disparities, as methodological issues have hampered the development of health disparities research. In order to create solutions to eliminate health disparities, researchers must understand the mechanisms powering their existence. Existing causal inference tools are not suitable for studying racial health disparities because race cannot be manipulated or changed, which makes it difficult to define appropriate counterfactuals. For the same reason, mediators stand to be useful in creating avenues to intervene on existing health disparities. Structural equation modeling (SEM) may be a more promising tool for quantifying the causal framework of health disparities and assessing mediation, viewed as the indirect effect. One of the most widely-used tests for assessing mediation is the Sobel test (Sobel, 1982; MacKinnon et al., 2007). However, it has disadvantages, which include low power, particularly at smaller sample sizes. Therefore, this work focuses on three varying methods for assessing mediation and compares their performance to the Sobel test. The first method is an adjustment of the Sobel test that accounts for the random nature of the mediator when estimating standard errors. The second method utilizes the joint distribution of the mediator and the outcome to determine the profile likelihood for the estimands of interest, which is then used to define an approximate, asymptotic distribution for the indirect effect. Finally, the third method utilizes Bayesian modeling techniques to fit the structural equation models and assess the indirect effect. Each method was assessed through simulations. All three methods demonstrated comparable estimated statistical power when compared to the Sobel test, often showcasing superior power at smaller sample sizes. Each method serves as a new tool of inference into the presence of mediation. The methods were applied to assess whether caloric intake mediates the relationship between race and blood pressure in non-Hispanic black and white subjects in the National Health and Nutrition Examination Survey (NHANES) from 1999-2004.
dc.format.mimetype	application/pdf
dc.identifier.uri	http://jhir.library.jhu.edu/handle/1774.2/40342
dc.language.iso	en_US
dc.publisher	Johns Hopkins University
dc.publisher.country	USA
dc.subject	Mediation Analysis
dc.subject	Health Disparities
dc.subject	Likelihood
dc.subject	Structural Equation Modeling
dc.subject	Residential Segregation
dc.subject	Hypertension
dc.subject	Race
dc.subject	Older Adults
dc.title	Likelihood-Based Methods of Mediation Analysis in the Context of Health Disparities
dc.type	Thesis
dc.type.material	text
thesis.degree.department	Biostatistics
thesis.degree.discipline	Biostatistics
thesis.degree.grantor	Johns Hopkins University
thesis.degree.grantor	Bloomberg School of Public Health
thesis.degree.level	Doctoral
thesis.degree.name	Ph.D.