Indiana University-Purdue University Indianapolis (IUPUI) / Joint models of longitudinal and survival outcomes have been used with increasing frequency in clinical investigations. Correct specification of fixed and random effects, as well as their functional forms is essential for practical data analysis. However, no existing methods have been developed to meet this need in a joint model setting. In this dissertation, I describe a penalized likelihood-based method with adaptive least absolute shrinkage and selection operator (ALASSO) penalty functions for model selection. By reparameterizing variance components through a Cholesky decomposition, I introduce a penalty function of group shrinkage; the penalized likelihood is approximated by Gaussian quadrature and optimized by an EM algorithm. The functional forms of the independent effects are determined through a procedure for structural discovery. Specifically, I first construct the model by penalized cubic B-spline and then decompose the B-spline to linear and nonlinear elements by spectral decomposition. The decomposition represents the model in a mixed-effects model format, and I then use the mixed-effects variable selection method to perform structural discovery. Simulation studies show excellent performance. A clinical application is described to illustrate the use of the proposed methods, and the analytical results demonstrate the usefulness of the methods.

Multiple-outcome longitudinal data are abundant in clinical investigations. For example, infections with different pathogenic organisms are often tested concurrently, and assessments are usually taken repeatedly over time. It is therefore natural to consider a multivariate modeling approach to accommodate the underlying interrelationship among the multiple longitudinally measured outcomes. This dissertation proposes a multivariate semiparametric modeling framework for such data. Relevant estimation and inference procedures as well as model selection tools are discussed within this modeling framework. The first part of this research focuses on the analytical issues concerning binary data. The second part extends the binary model to a more general situation for data from the exponential family of distributions. The proposed model accounts for the correlations across the outcomes as well as the temporal dependency among the repeated measures of each outcome within an individual. An important feature of the proposed model is the addition of a bivariate smooth function for the depiction of concurrent nonlinear and possibly interacting influences of two independent variables on each outcome. For model implementation, a general approach for parameter estimation is developed by using the maximum penalized likelihood method. For statistical inference, a likelihood-based resampling procedure is proposed to compare the bivariate nonlinear effect surfaces across the outcomes. The final part of the dissertation presents a variable selection tool to facilitate model development in practical data analysis. Using the adaptive least absolute shrinkage and selection operator (LASSO) penalty, the variable selection tool simultaneously identifies important fixed effects and random effects, determines the correlation structure of the outcomes, and selects the interaction effects in the bivariate smooth functions. Model selection and estimation are performed through a two-stage procedure based on an expectation-maximization (EM) algorithm. Simulation studies are conducted to evaluate the performance of the proposed methods. The utility of the methods is demonstrated through several clinical applications.

Indiana University-Purdue University Indianapolis (IUPUI) / Background: Recent literature on medical professionalism claims that self-awareness and the ability to reflect upon one’s experiences is a critical component of professionalism; however there is a paucity of empirical evidence to support this claim. This study employed a mixed methods approach to explore the utility of reflective writing and self- and peer assessments in explaining professionalism lapses among medical students. Methods: A retrospective case-control study was conducted using students from Indiana University School of Medicine (IUSM) who had been disciplined for unprofessional behavior between 2006-2013 (case group; n=70). A randomly selected control group (n=230) was used for comparison. Reflective ability was assessed using a validated rubric to score students’ professionalism journals. Mean reflection scores and assessment scores were compared using t-tests. Logistic regression analysis was used to determine the impact of reflection scores and self- and peer assessment scores on the likelihood of having been disciplined for unprofessional behavior. Subsequent qualitative analysis further explored when and how students learned professionalism during their clinical experiences. Results: The study found that students in the case group exhibited lower reflective ability than control students. Furthermore, reflective ability was a significant factor in explaining the odds that a student had been cited for professionalism lapses. There were no differences in self-assessment scores between the two groups, but students in the case group had significantly lower peer assessment scores than control students. Peer assessment scores also had the greatest influence on the odds that a student had been cited for professionalism deficiencies during medical school. Qualitative analysis revealed that students learn professionalism from role models who demonstrated altruism and respect (or lack thereof). Conclusions: These findings suggest that students should be provided with guidance and feedback on their reflective writing to promote higher levels of reflection, which may reduce the number of students who are cited for professionalism lapses. These findings also indicate that peer assessments can be used to provide students with insightful feedback regarding their professional development. Finally, role models have a strong influence on students’ professional development, and therefore must be cognizant of the implicit messages their behaviors convey.

Indiana University-Purdue University Indianapolis (IUPUI) / The objective of this research is to examine the development of the California alcohol outlets over time and the relationship between neighborhood characteristics and densities of the alcohol outlets. Two types of advanced analyses were done after the usual preliminary description of data. Firstly, fixed and random effects linear regression were used for the county panel data across time (1945-2010) with a dummy variable added to capture the change in law regarding limitations on alcohol outlets density. Secondly, a Bayesian spatio-temporal Poisson regression of the census tract panel data was conducted to capture recent availability of population characteristics affecting outlet density. The spatial Conditional Autoregressive model was embedded in the Poisson regression to detect spatial dependency of unexplained variance of alcohol outlet density. The results show that the alcohol outlets density reduced under the limitation law over time. However, it was no more effective in reducing the growth of alcohol outlets after the limitation was modified to be more restrictive. Poorer, higher vacancy rate and lower percentage of Black neighborhoods tend to have higher alcohol outlet density (numbers of alcohol outlets to population ratio) for both on-sale general and off-sale general. Other characteristics like percentage of Hispanics, percentage of Asians, percentage of younger population and median income of adjacency neighbors were associated with densities of on-sale general and off sale general alcohol outlets. Some regions like the San Francisco Bay area and the Greater Los Angeles area have more alcohol outlets than the predictions of neighborhood characteristics included in the model.

Indiana University-Purdue University Indianapolis (IUPUI) / Oral health is crucial to overall health and a focus of the U.S. Health Center program, which provides preventive dental services in medically underserved communities. Dental hygiene is an oral health profession whose practice is focused on dental disease prevention and oral health promotion. Variations in the practice and regulation of dental hygiene has been demonstrated to influence access to dental care at a state level; restrictive policies are associated lower rates of access to care. Understanding whether and to what extent policy variations affect availability and access to dental care and the oral health of medically underserved communities served by grantees of the U.S. Health Center program is the focus of this study. This longitudinal study examines dental service utilization at 1,135 health center grantees that received community health center funding from 2004 to 2011. The Dental Hygiene Professional Practice Index (DHPPI) was used as an indicator of the state policy environment. The influence of grantee and state level characteristics are also considered. Mixed effects models were used to account for correlations introduced by the multiple hierarchical structure of the data. Key findings of this study demonstrate that state policy environment is a predictor of the availability and access to dental care and the oral health status of medically underserved communities that received care at a grantee of the U.S. Health Center program. Grantees located in states with highly restrictive policy environments were 73% less likely to deliver dental services and, those that do, provided care to 7% fewer patients than those grantees located in states with the most supportive policy environments. Population’s served by grantees from the most restrictive states received less preventive care and had greater restorative and emergency dental care needs. State policy environment is a predictor of availability and access to dental care and the oral health status of medically underserved communities. This study has important implications for policy at the federal, state, and local levels. Findings demonstrate the need for policy and advocacy efforts at all levels, especially within states with restrictive policy environments.