Epidemiology and Biostatistics

Research

Projects

Identifying Determinants of Health Enhancing Physical Activity among Adults with Schizophrenia (Research to Promote Health Equity in Indiana grant from SPH-B)

Andrea Chomistek

Adults with schizophrenia experience significant health disparity in the form of early mortality by as much as 25 years.  The majority of this early mortality is due to cardiovascular disease secondary to smoking, obesity and a sedentary lifestyle.  The purpose of this study is to identify multi-level determinants of physical activity (PA) and sedentary behavior among this population for the purpose of developing a pilot intervention to increase health enhancing PA and reduce sedentary behavior.

Statistical models for multi-modal brain imaging studies of HIV-associated cognitive decline 

Jaroslaw Harezlak

The study of neurocognitive impairment and its progression in HIV patients is a multifaceted endeavor involving complex data structures. These include multi-modal brain imaging measures as well as longitudinal host and viral markers. Exploring the full potential of these data requires the application and development of advanced statistical analysis methods. This proposal will address scientific questions motivated by the data obtained in Dr Ances' laboratory at Washington University in St. Louis. His extensive studies provide a unique opportunity to make inferences about the diagnosis and progression of neurocognitive impairment (NCI) among HIV-infected individuals. In this study, however, majority of the data is in the form of multi-modal brain images. Hence, the statistical analysis requires an innovative perspective that goes beyond mass univariate analyses or simple regression methods. Analysis of these data requires innovative methods which take a holistic view of these neuroimaging data with a mathematical and analytical framework that succeeds by incorporating adjunct co-informative data and clinical knowledge. This project answers these challenges with four aims focusing on determining imaging markers discriminating between HIV-infected patients and matched HIV-negative controls and establishing specific biomarkers of NCI diagnosis and progression among HIV-infected individuals. In Aim 1, we determine which structural and functional brain imaging markers discriminate between HIV- infected patients and matched HIV-negative controls. In Aim 2, we determine which structural and functional imaging markers predict neurocognitive impairment (NCI) severity in HIV-infected patients, whereas in Aim 3, we establish the imaging biomarkers predicting future NCI. Aim 4 is devoted to elucidating synergistic effects of aging and HIV infection on the NCI presence and progression. Thirty five million people worldwide are HIV-infected with over 40% having neurological or cognitive impairment. Despite the availability of combination antiretroviral therapy (cART), these individuals are at risk of accelerated brain degeneration, despite disease controlled in terms of undetectable virus. Our proposed methods directly address the study of brain degeneration and consequent cognitive impairment and thus have significant clinical and scientific impact. The developed methods are broadly applicable in a wide variety of complex data settings.

Trace element levels and risk of stroke 

Ka He

Despite minor geographic shifts, the "Stroke Belt", a region of highest stroke mortality in the Southeastern US identified a half century ago, still persists today. Recent studies have found that the stroke mortality is also higher among children in the Southeast, thus environmental factors have been hypothesized contributing to the "Stroke Belt" in addition to other hypotheses on lifestyle and social-economic status. For decades, it has been demonstrated outside the US that geographic variations in trace elements may play critical roles in the development of cardiovascular diseases. However, the geographic variation of trace element levels in relation to stroke risk remains unclear. The overall objectives of this project are to examine the associations between trace element levels and stroke risk and to investigate whether geographic variation of trace element levels is related to the "Stroke Belt". We propose a case-cohort study, which will include incident ischemic stroke cases (n=~620) and a sub-cohort (n=~2500) of non-stroke cases randomly sampled within region-race-sex stratum from the REasons for Geographic And Racial Differences in Stroke (REGARDS) Study, an ongoing US national population-based, general population cohort of 30,239 African American and Caucasian adults, aged 45 and older at baseline (2003-2007). The proposed study will focus on arsenic, cadmium, mercury, magnesium and selenium. Specifically, we will characterize the distribution of trace element levels measured in urine or serum according to demographic and geographic characteristics of study participants; and to examine prospective associations between trace element levels and risk of stroke. This research will help identify at-risk individuals for stroke, thus providing important data identifying whether stroke risk can be reduced by dietary, supplemental, lifestyle or environmental interventions that modify trace element patterns.

High-dimensional statistical genetic methods for dental caries and orofacial clefts 

Nianjun Liu

Dental caries (i.e., dental cavities or tooth decay) and orofacial clefts cause tremendous public health burden. In addition to the costs of treatment, these conditions greatly affect the quality of life of patients and their families. The etiology of caries and clefts has been studied extensively, and both environmental and genetic risk factors have been implicated. Despite some success in the study of genetic factors of caries and oral clefts, there is still a large portion of "missing heritability" (i.e., the identified genetic variants explain only a small proportion of the estimated heritability). One important reason for this missing heritability is the lack of powerful statistical methods for the efficient use of data. In addition, translating the knowledge gained from scientific studies into clinical practice has been recognized as an essential step toward future personalized healthcare, but even more is lacking in this area. Such a lack of research on the risk prediction of dental caries and orofacial clefts needs to be addressed urgently given the critical role of the mouth and teeth in our daily lives. The long-term goals are to improve understanding of the mechanisms leading to dental and craniofacial disorders and to use the scientific findings to aid clinical practice. The objectives of this application are to develop new statistical methods to facilitate the identification of novel genetic variants contributing to dental caries and orofacial clefts and the prediction of the risk of their occurring using genome-wide data, with the following aims: (1) To develop a novel family-based statistical method to identify the joint effects of multiple genetic markers on multiple phenotypes with cross-sectional or longitudinal observations, and to apply it to dental caries GWAS data. Dental disorders usually show a strong familial aggregation. Complex diseases are often multifaceted and the joint analysis of these correlated phenotypes can increase power in risk gene discovery. To our knowledge, there are no such methods available. This aim intends to fill this gap. (2) To develop and validate high-dimensional, unified Bayesian models to identify genetic variants underlying dental caries and orofacial clefts and to predict the risks of them in family and unrelated samples.  We will develop a flexible, multiple-marker-based, hierarchical modeling framework that can handle different types of traits (e.g. discrete and quantitative traits) and analyze both family and unrelated samples, for both disease-variant identification and risk prediction. This is the first work to estimate and test both group effect and the effect of individual variants for family data and the first high-dimensional, family- based risk-prediction models for orofacial clefts and dental caries. We will analyze multiple GWAS datasets obtained from dbGaP and our collaborators. The successful completion of the Aims will lead to powerful statistical models and helpful software, and new discoveries on the etiology of dental caries and orofacial clefts. These expected results will expand our understanding and ultimately enhance our ability to decipher the genetic basis of dental caries and orofacial clefts and help us to efficiently prevent, diagnose, and treat them.

Secondary analysis and risk prediction of dental caries and oral clefts 

Nianjun Liu

Dental caries (i.e., dental cavities or tooth decay) and orofacial clefts cause tremendous public health burdens for patients, their families, and society. In addition to direct costs from treatment, the quality of life of the affected individuals and ther families is greatly affected. The etiology of dental caries and orofacial clefts has been studied extensively, and both environmental and genetic factors have been implicated in their risks. Despite some success in the study of genetic factors of caries and oral clefts, there is still a large portion of the heritability "missing" (i.e., not explained by the identified genetic variants. In addition, translating the knowledge gained from scientific studies into clinical practice has been recognized as an essential step toward future personalized healthcare, but even more is lacking in this area. Such a lack of research on the risk prediction of dental caries and orofacial clefts needs to be addressed urgently given the critical role of the mouth and teeth in our daily lives. The long-term goals are to improve understanding of the mechanisms leading to the development of dental and craniofacial disorders and to use the scientific findings to aid clinical practice. The objectives of this application are to identify novel genetic variants contributing to dental caries and orofacial clefting and to use the results (from our study and others in the literature) to predict the risk of their occurring. The planned specific aims are as follows. (1) Identify new genetic variants associated with dental caries and orofacial clefts by analyzing existing data from genome-wide association studies using novel statistical methods. Based on a literature review and our own previous experience, we postulate that there are as yet unidentified genetic variants affecting the risks of dental caries and orofacial clefts and that these variants can be identified by using powerful statistical models. (2) Develop and validate high- dimensional, family-based Bayesian models to predict the risks of dental caries and orofacial clefts. Currently there is a great need for-but lack of-approaches that can accurately predict disease risk in clinical practice. This aim is to address this critical need for dental cares and orofacial clefts. With respect to expected outcomes, the successful completion of Aim 1 will lead to new discoveries about the etiology of dental caries and orofacial clefts, and in particular about novel genes and gene x environment interactions that affect risk of their occurrence. The work proposed in Aim 2 is expected to provide powerful statistical models to predict the risks of dental caries and orofacial clefts that can aid decision making in clinical practice. Such results are expected to have an important positive impact because the expected results will expand our understanding and ultimately enhance our ability to decipher the genetic basis of dental caries and orofacial clefts and help us to better understand, accurately predict, efficiently prevent, diagnose, and treat them.

Social, structural, and economic influences of HIV risk in young South African women

Molly Rosenberg

Social, structural, and economic influences of HIV risk in young South African women: Young women in sub-Saharan Africa (SSA) are at incredibly high risk of HIV infection with 76% of all new infections among young people occurring in young women and 77% of HIV+ women living in sub-Saharan Africa. Partnering with the Agincourt Health and Socio-demographic Surveillance Site and with several ongoing randomized control trials and observational studies of young women in the site, we are identifying and testing structural intervention targets for HIV prevention and sexual health promotion. Examples of previous, ongoing, and future project focus as they relate to HIV risk in young women are: conditional cash transfers, alcohol outlets, executive function, and multi-level educational interventions.

Representation of campus sexual assault data

Molly Rosenberg

Representation of campus sexual assault data: We are conducting a systematic review of current data collected on campus sexual assault to identify whether the results represent the student population from which they are drawn. The primary objectives of this systematic review are to 1) quantify the amount of missing data in recent campus sexual assault surveys, and 2) understand how missing data may be influencing the conclusions drawn from the surveys. We are collecting data from all available campus sexual assault surveys conducted since 2010 from all 62 school members of the AAU. From each survey available, we are abstracting data on prevalence of sexual assault reported and survey response rates, as well as other school and survey-specific characteristics. Long-term, we hope to use the results from this project to improve the measurement of campus sexual assault in future surveys.

The influence of a microfinance intervention on major infectious diseases in Haiti

Molly Rosenberg

The influence of a microfinance intervention on major infectious diseases in Haiti (contingent on funding): Haitians carry a disproportionate public health burden from infectious diseases like cholera, malaria, and HIV. Although poverty is consistently identified as a distal cause of many infectious diseases, little is known about whether poverty alleviation programs like microfinance could reduce infectious disease risk. Our central hypothesis is that exposure to a microfinance intervention will reduce susceptibility to infectious diseases. We will conduct a cross-sectional, tablet-based survey on a random sample of 300 Fonkoze microfinance clients in southern Haiti, querying them about the length of their microfinance membership (primary exposure, Aim 1), major infectious disease outcomes and prevention behaviors (primary outcomes, Aim 1), and feasibility and acceptability of infectious disease outcome assessments (primary outcomes, Aim 2). The proposed study is significant because it will provide one of the first estimates of the association between microfinance and a range of specific infectious disease outcomes in Haiti.