Health disparities are preventable differences in health status and outcomes that adversely affect certain populations, and are generally attributable to unjust social or environmental influences. Mitigating health disparities is crucial toward preventing unnecessary and avoidable human suffering, and as such there has been a significant increase in health disparities research and funding. However, existing health disparities publications are geographically-constrained to specific institutions or populations, and often rely on disease definitions that cannot be easily applied elsewhere. While more recent publications have begun identifying differences utilizing larger datasets, for most diseases, differences in prevalence, age of onset, and time to diagnosis differences remain unstudied and unknown. This dissertation leverages informatics solutions built atop observational health datasets to enable high-throughput, reproducible assessments of disparities across subgroups, conditions, and datasets. In the first aim, this dissertation examines the literature to identify how health disparities in disease diagnosis are measured, computed, and reported. It then proposes an iterative approach for generating fair phenotype definitions that are more inclusive of subgroups of interest by utilizing algorithmic fairness measurements translated to epidemiological measures. In the second aim, this dissertation conducts large-scale characterizations of disease diagnosis patterns across subgroups (gender and race), conditions, and datasets. In particular, this dissertation conducts a prevalence-based assessment of disease diagnosis by computing prevalence differences, risk ratios, and age of onset differences across diseases and datasets. The dissertation then conducts a scalable assessment of time to diagnosis differences across 122 disease phenotypes. Finally, in the third aim, this dissertation moves from quantifying differences to identifying disparities in diagnosis. To do so, the dissertation applies a framework for causal fairness to decompose observed time to diagnosis differences into direct, indirect, and spurious effects. In conclusion, this dissertation's primary contributions are providing a systematic, scalable approach for identifying health differences and then quantifying health disparities at-scale across large-scale observational health datasets. The dissertation (1) proposes an iterative approach for systematically assessing the fairness of phenotypes used in observational health research, (2) systematically characterizes differential patterns of disease diagnosis across diseases and observational datasets, and (3) causally decomposes differences into quantifiable effects that suggest the presence of potential health disparities.

Mental health care, other than being a critical medical service, is a centrally important cultural institution, shaping individuals’ understanding of their lives and problems. This dissertation consists of a multi-method investigation examining the U.S. mental health care field from these two intersecting perspectives. To achieve this, it examines (1) how structural conditions determine different social groups’ access to different mental health care services; (2) what effects these disparities in access have on individuals’ experiences when seeking care and on the outcomes they face; (3) how the stratification of the mental health care field exposes different Americans to different sets of cultural scripts and understandings, generating correspondence between sociodemographics and intimately-held meanings and normative action scripts for mental well-being. Its four chapters thus tackle mental health care’s social significance from multiple angles, using geospatial, computational text analysis, and audit methods, seeking to provide actionable, public health policy-relevant empirical research concerning inequalities in access to care, as well as to address key theoretical problems regarding the segregation of meaning-making with which sociology has long contended.

The main research questions examined in this thesis concern the extent to which social class influence individuals' health, and how and whether individuals' occupation, education and lifestyles mediate between class and health. The conclusions drawn from the analysis of these empirical questions cast further light on the widening health inequalities seen in developed societies in recent decades. In particular, this research suggests that, employment conditions as well as educational levels are variables that need to be taken into account when planning policies aimed at tackling differences in health outcomes. Lifestyle variables, on the other hand, would appear to be almost irrelevant when explaining why the members of the more privileged social classes not only live longer than those in other classes, but also enjoy significantly better health over the course of their lives. In trying to understand the association between class and health, I define a theoretical framework that specifies the mechanisms through which class is linked to health. Social structure influences health by distributing certain factors such as material resources or some health-related behaviour that ultimately result in individuals having different living conditions. Educational attainment also affects the way these resources are employed and, therefore, lifestyles. A fundamental element of a social class is occupation: individuals' employment and working conditions also affect their health. Furthermore, the nature of a social structure has an effect on health at the aggregate level of analysis since social policies are partly the result of the structure of class interests. Four mechanisms are specified in order to systematically test this theoretical framework. Mechanisms (2) and (3), those that relate class and health through education and lifestyle lie at the heart of the empirical analysis. This analysis employs individual-level data drawn from health surveys carried out during the first half of the 1990s in the two countries selected for the analysis, United Kingdom and Spain. These countries are treated as contexts in which to test the theoretical explanation. The main results of the analysis reveal the importance of social class in determining health outcomes. Indeed, individuals from different classes enjoy distinct degrees of health. Specifically, individuals in the most privileged class categories have persistently better health than those in the other class categories. Differences exist in terms of both objective and subjective or self-perceived health. Moving on from observation to explanation, the analysis suggests that the distribution of certain resources across classes accounts for some of the variance in health outcomes. Hence, education is identified as a significant variable to comprehend part of the health inequalities in developed societies. Lifestyle, on the other hand, does not appear relevant in accounting for health outcomes. The small differences found between the United Kingdom and Spain in the mechanisms that link class and health suggest that the process through which class affects health is essentially similar in developed societies.

Public health and social policies are often debated, designed, and adopted without implementation, sustainability, or equity in mind, which can generate profound uncertainty about how to equitably deliver them initially and over time. Although sustainability and equity considerations are sometimes considered in post-hoc policy analysis and evaluation, little is known about how to plan for and track planned and unplanned adaptations to policy implementation, as well as the ways that key sustainability factors and strategies can relate to the equitable delivery or relative effectiveness of policies on the ground and in community settings. The purpose of this dissertation was to explore the long-term sustainability and equity of tobacco control policies and programs in New York City to understand and contextualize their limited reach and impact on persistent smoking and tobacco-related health disparities in underserved Asian American communities. The specific aims were to: 1) conceptually specify an operational definition of policy sustainability and its key dimensions, including health equity, and to refine this using a mixed methods single case study of tobacco policies in New York City; 2) use the case study approach to describe the extent to which tobacco policies have been sustained and adapted in New York City; and 3) to use the case study to identify key multi-level factors that influence the long-term sustainability and equity of tobacco policies in New York City. Using a single, in-depth, convergent-parallel mixed methods case study design, data were collected, analyzed, and integrated across five key primary and secondary sources: 1) Policymaking documents – text of key tobacco bills and statutes, as well as transcripts from when they were first proposed, amended, debated, and adopted; 2) Local newspaper coverage – articles from a database of 29 major newspapers in New York State on the policies and their impacts on communities and businesses over time; 3) Key informant interviews – conducted with community members and community leaders at local health and advocacy organizations in New York City that primarily serve Asian American and immigrant communities (n = 21); 4) Direct observation periods – conducted within and around the health and advocacy organizations, as well as in majority Asian neighborhoods and Asian ethnic enclaves (n = 15); and 5) the New York City Community Health Survey (2012-2017) – conducted annually by the New York City Department of Health and Mental Hygiene. The integrated study findings point to the importance of understanding policy sustainability not as a static end goal, but rather as a dynamic set of processes and outcomes that impact health and health equity. Findings from this case study clustered across three key themes: 1) since the initial adoption of comprehensive local tobacco control measures in New York City in 2002, broad “one-size-fits-all” approaches to policy implementation and monitoring have been sustained, which have had and continue to have limited reach and impact within underserved Asian American and immigrant communities; 2) two delayed adaptation efforts were made by policymakers during the sustainability phase, one in 2012 and another in 2018, were intended to improve on prior uneven implementation to better reach Chinese-speaking communities, with the 2018 adaptation demonstrating significant improvements from the 2012 effort; and 3) community-based organizations have played a direct role in functioning not just as key stakeholders but also as key implementers to ensure that tobacco and other health policies are reaching communities that the designated or official implementers cannot reach. This suggests the need for further study of unofficial implementers in implementation science – those who have not been formally designated as the ones responsible for ensuring that implementation takes place, but are still delivering implementation strategies to ensure adoption, integration, and sustainment. Overall, this case study points to the potential for policy sustainability research to advance health equity by identifying factors and mechanisms that can be improved to maximize and sustain the equitable reach and impact of social and public health policies. By focusing on dynamic contextual factors and sustainability as a set of processes and outcomes, the findings from this case study raise critical questions about the criteria typically used to evaluate whether policy interventions are deemed evidence-based and effective by asking: 1) effective for whom?; 2) based on what evidence?; and 3) what happens as dynamic populations and contexts change over time? These questions highlight how the tobacco control success story was largely constructed around broad population-wide implementation and benefits, while overlooking underserved Asian American communities who continue to disproportionately bear the brunt of smoking and tobacco-related health disparities in New York City.

Real-world evidence (RWE) plays a crucial role in understanding the impact of medical interventions and uncovering disparities in clinical practice. However, confounding bias, especially unmeasured confounding, poses challenges to inferring causal relationships from observational data, such as estimating treatment effects and treatment responses. Various methods have been developed to reduce confounding bias, including methods specific for detecting and adjusting for unmeasured confounding. However, these methods typically rely on assumptions that are either untestable or too strong to hold in practice. Some methods also require domain knowledge that is rarely available in medicine. Despite recent advances in method development, the challenge of unmeasured confounding in observational studies persists. This dissertation provides insights in adjusting for unmeasured confounding by exploiting correlations within electronic health records (EHRs). In Aim 1, we demonstrate a novel use of probabilistic model for inferring unmeasured confounders from drug co-prescription pattern. In Aim 2, we provide theoretical justifications and empirical evidence that adjusting for all (pre-treatment) covariates without explicitly selecting for confounders, as implemented in the large-scale propensity score (LSPS) method, offers a more robust approach to mitigating unmeasured confounding. In Aim 3, we shift focus to the problem of evaluating fairness of treatment allocation in clinical practice from a causal perspective. We develop a causal fairness algorithm for assessing treatment allocation. By applying this fairness analysis method to a cohort of patients with coronary artery disease from EHR data, we uncover disparities in treatment allocation based on gender and race, highlighting the importance of addressing fairness concerns in clinical practice. Furthermore, we demonstrate that social determinants of health, variables that are often unavailable in EHR databases and are potential unmeasured confounders, do not significantly impact the estimation of treatment responses when conditioned on clinical features from EHR data, shedding light on the intricate relationship between EHR features and social determinants of health. Collectively, this dissertation contributes valuable insights into addressing unmeasured confounding in the context of evidence generation from EHRs. These findings have significant implications for improving the reliability of observational studies and promoting equitable healthcare practices.

Problem to be addressed: Stillbirth is a major public health problem. The stillbirth burden is on a par with newborn deaths. The stillbirth rate measures not only a substantial portion of the global and national burden of mortality, but also equity and quality of care for women’s and children’s health. Reducing the numbers of these deaths requires an understanding of why they occur, yet approximately one-third of stillbirths are unexplained, even in settings with high-quality autopsy and placental examination, while deaths considered to be explained are usually ascribed to single, proximal causes. An important limiting factor for efforts to reduce the large and inequitable stillbirth burden has been insufficient research into conditions that could inform prevention strategies and reduce inequity.1 2 Substantial evidence exists for associations between structural racism, maternal stress, and adverse pregnancy outcomes, yet research focusing on stillbirth is sparse, particularly at the ends of the causal spectrum—macro-level structural conditions and mechanisms. Several studies have called for research on possible biological mechanisms by which racism, racism-related stress, and stillbirth may be associated, including epigenetic mechanisms.3-6 The most recent review of causes of racial disparities in stillbirth rates in the U.S. recommended that researchers take a multi-domain approach, considering not just individual-level risk factors, which have been relatively well-studied, but also upstream factors such as institutional racism, and biological mechanisms such as epigenetic modification. The objective of this dissertation was to explore evidence that could help to explain persistent racial disparities in stillbirth. The specific aims were: 1. To review the literature on racial disparity in stillbirth rates; 2. To assess whether structural racism can help to explain racial disparity in stillbirth rates in New York City; and 3. To assess whether maternal stress is associated with stillbirth, whether stress is associated with methylation of stress-related genes, whether methylation is associated with stillbirth, and whether there is evidence that methylation of stress-related genes mediates associations between stress and stillbirth. Materials and methods used: For Aim 1, we carried out a scoping review of the literature in five databases (PubMed, Scopus, Cinahl, Embase, PsycInfo) to identify all reports including stillbirth rates stratified by race in the U.S., mapping exposures and effect modifiers (“domains of analysis”) and authors’ comments on racial disparity in stillbirths (“domains of explanation”) into one of eight domains (race, genetic, fetal, maternal, family, community, healthcare system, and structural). We defined Stillbirth Disparity Ratios (SDRs) as the ratio of the stillbirth rate in a racial/ethnic minority group to the stillbirth rate in white individuals. Selected SDRs were extracted from each report, as were all SDRs for Black/white comparisons. For Aim 2, we modelled associations between four measures of structural racism and stillbirth in all non-Hispanic (NH) Black and white singleton births in New York City between 2009 and 2018. Exposures were four Public Use Microdata Area (PUMA)-level measures of structural racism (Indices of Dissimilarity, Isolation, and Concentration at the Extremes (ICE), and an Educational Inequity Ratio) constructed from U.S. Census American Community Survey data. Using multilevel logistic regression, we first tested for interaction between race and structural racism in relation to stillbirth. For structural racism measures that interacted with race, we estimated odds ratios for stillbirth separately in 221,925 NH Black and 325,058 NH white births. Race-specific models were further stratified by maternal age. For Aim 3, we assessed associations between maternal stressors and stillbirth in 183 non-anomalous full-term singleton births (63 stillbirths and 120 livebirths) from the U.S. Stillbirth Collaborative Research Network. Measuring maternal stress with two hypothesized stressors, an Index of Significant Life Events and an Index of Disadvantage, we assessed associations between maternal stressors and stillbirth in our sample, and then whether maternal stressors and stillbirth were associated with differential methylation of 1,191 CpGs on five stress-related genes (BDNF, FKBP5, HSD11B2, IGF2, and NR3C1). Finally, we assessed whether methylation mediates associations between stressors and stillbirth. Conclusions reached: For Aim 1, we found 95 reports presenting stillbirth rates stratified by race/ethnicity in the U.S. We found evidence of increased risk of stillbirth in Black as compared to white births in the majority of the 83 reports with the necessary data. Among the 1143 Black-white SDRs that we extracted, the median SDR was 1.67, with 74% of SDRs showing evidence of disparity. Family and community factors, healthcare system factors, and structural factors were commonly used as domains of explanation (20-38% of reports), but rarely (family/community, structural, 4-5%) or never (healthcare system) used in analysis. The most commonly used domains of analysis—fetal and maternal factors including gestational age, maternal age, education, and prenatal care—do not appear able to explain the observed racial disparities. Gaps in the literature include a paucity of studies examining the possible role of health system, community, and structural factors in Black-white disparity in stillbirth rates, and limited data on other types of racial disparities in stillbirth rates, including Hispanic and Native American births. For Aim 2, we found that structural racism as measured by ICE and Isolation was associated with stillbirth in NH Black but not NH white mothers. This would seem consistent with our hypothesis that structural racism may help to explain racial disparity in stillbirth rates; however, the associations we observed were not in the expected direction. Specifically, NH Black mothers living in PUMAs with a high concentration of privilege had 90% greater odds of stillbirth in comparison to those living in PUMAs with a high concentration of disadvantage (ICE quintile 5 vs 1), and NH Black mothers living in PUMAs that were the most isolated had 40% lower odds of stillbirth in comparison to those living in PUMAs that were the least isolated (Isolation tertile 3 vs 1). We suggest that while the measures we used (ICE and Isolation) do help to explain the Black-white disparity in stillbirth rates, our results raise questions about the way these measures operationalize structural racism, meriting further investigation. For Aim 3, we found that having two or more vs no items in the Index of Disadvantage (“Disadvantage”) was associated with more than fourfold greater odds of stillbirth (95% CI 1.58, 12.93). We found no association between the Index of Significant Life Events and stillbirth. We found that 32 out of 1,191 CpGs on five stress-related genes were differentially methylated with respect to stillbirth, and six CpGs were differentially methylated with respect to Disadvantage. Methylation at two CpGs on IGF2 and one on HSD11B2 (cg02097792, cg12283393, and cg19413291, respectively) mediated the association between Disadvantage and stillbirth. Research on causes is a critical component of stillbirth prevention and reducing the inequitable distribution of this public health burden. Limited understanding of causes at both “ends of the spectrum”, from upstream distal factors to mechanisms, has likely contributed to slow progress on prevention.7 8 This dissertation contributes to science and public health by providing researchers with data to support new lines of inquiry, e.g., into associations between structural racism and stillbirth, and for methylation as a mechanism of effect, that should help to improve our understanding of causes. Our research may also support health policy makers who now have additional data to illustrate the adverse health outcomes of structural racism in the U.S. Finally, it may help the parents and other family members of stillborn babies who continually seek to understand “why”.

This dissertation investigates the effectiveness of and disparities in timely goals of care (TGOC) discussions in nursing homes (NHs), as well as hospitalizations due to infections. Chapter One provided an overview of the dissertation, including its aims and conceptual frameworks. Chapter Two is a systematic review of the current evidence concerning the timing of goals of care discussions in NHs and their impact on end-of-life outcomes. Chapter Three examines the differences in TGOC discussions in NHs for Black. Chapter Four analyzes the association between TGOC discussions and hospitalizations due to infections. Finally, Chapter Five synthesizes all findings. Key findings demonstrate that while effective TGOC discussions are associated with reduced hospitalizations due to infections in NH residents, significant racial and ethnic disparities exist, particularly among NHs with higher proportions of Black residents and those located in socioeconomically disadvantaged areas. These findings highlight the need for improved TGOC practices and targeted interventions to address systemic barriers and ensure equitable end-of-life care.

Over the past 60 years, the immigrant population in the United States (U.S.; i.e., those born outside of the U.S.) has grown significantly. Included in this group are newcomer immigrant adolescents (NIA), who immigrated to the U.S. within the past 5 years. Upon arrival in the U.S., NIA experience a multitude of unique systemic and psychosocial stressors and protective factors, placing them at risk for negative social emotional and academic outcomes. Despite this, culturally-informed, evidence-based interventions addressing the needs of NIA are limited due to the widespread underrepresentation of this marginalized group in research, underscoring a need to understand and conduct research with NIA. Despite multiple calls and efforts to improve representation of minoritized populations, including NIA, in research, researchers have noted challenges in recruiting NIA. Literature on barriers to participation, motivation to research participation, and strategies to navigate barriers to participation exist but have yet to be explored with NIA. This is despite research underscoring the need to consider research participants’ culture and developmental level when developing informed strategies to improve recruitment efforts. Thus, the current study qualitatively explored the barriers to participation in research among NIA, examined their motivations to participate in research, and elicited recommendations from NIA and relevant stakeholders (i.e., community partners who work in NIA serving organizations) on how to increase their research participation. Twenty-six semi-structured interviews were conducted with NIA from Latin America (n = 3), South Asia (n = 1), Southeast Asia (n = 7), and West Africa (n = 3) and community partners who serve NIA (n = 12). Implications for stakeholders, researchers, and health equity are discussed.

Luo, Weixiang. / Thesis (Ph.D.)--Chinese University of Hong Kong, 2013. / Includes bibliographical references (leaves 90-105). / Electronic reproduction. Hong Kong : Chinese University of Hong Kong, [2012] System requirements: Adobe Acrobat Reader. Available via World Wide Web. / Abstracts also in Chinese.

Existing research on health disparities in breast cancer is heavily focused on outcomes for poor or low-income women. Little is known about the experience of privately insured Black breast cancer patients that have moderate to high SES. As a result, the present study was conducted to learn more about their experiences. It examines differences in physician prescribing of two breast cancer pain treatments, peripheral nerve block (PNB) and opioids, for Black and White women with like levels of health insurance coverage and socioeconomic status (SES). Three specific questions are addressed: 1. What, if any, race-based disparities exist in usage of PNBs at time of total mastectomy? 2. What, if any, race based disparities exist in the prescription of opioids for postoperative pain following total mastectomy? 3. What, if any, changes have occurred in the frequency of orders placed for PNBs and prescription opioids over time, to treat postoperative pain resulting from mastectomy? A cross-sectional designed was used relying on an existing national dataset, Optum Clinformatics Data Mart. The study period was January 1, 2012, through December 31, 2019. Study results revealed that while moderate to higher SES Black women have equitable access to PNB and opioids - a kind of shield from long established physician bias against Black women – this protection is quite porous. They still do not have open and ready access to PNB as a more advanced pain treatment. Nor do they have assurance that they are protected from the overprescribing of opioids, a class of drugs with serious and well-known safety risks. Therefore, on the surface, it appears that equity and racial inclusion are hallmarks of physician prescribing of postoperative breast cancer pain treatment. However, further interrogation reveals that ‘separate and somewhat equal’ is a more accurate characterization of their prescribing practices, based both on race and SES.