Understanding the Multilevel Drivers of Liver Cancer Disparities
Project Summary/Abstract From 2000-2014, hepatocellular carcinoma, or HCC, incidence rates increased nearly 4% per year, while most cancers in the United States were on the decline. HCC disproportionately impacts minority racial/ethnic groups who are diagnosed at rates approximately twice that of non-Hispanic Whites. To inform primary prevention strategies that will reduce disparities in HCC risk, we need to determine the relative contribution of well- established and emerging (e.g., hepatitis B virus, hepatitis C virus, alcohol, smoking, cirrhosis, NAFLD, metabolic disorders, diabetes, HIV infection), and novel (e.g., medications, comorbidities, neighborhood attributes) risk factors to these disparities. To inform secondary and tertiary prevention strategies to reduce disparities in HCC burden, we need to understand the multilevel factors that contribute to HCC surveillance disparities. Answering these gaps in knowledge requires a robust high-quality study with a sample enriched for racial/ethnic minorities. Thus, we propose to leverage existing multi-disciplinary collaborations to develop an integrated dataset that includes electronic health records (EHR) data linked to population-based state cancer registry data and geospatial contextual data. This multilevel resource will include data on nearly 2.3 million individuals from three healthcare systems (mixed payer, integrated healthcare, federally qualified health centers) in California and Hawaii, thus providing diversity in healthcare settings and enrichment for racial/ethnic minorities: 59,400 are Black, 189,500 are Hispanic, and 441,700 are Asian American/Native Hawaiian/Pacific Islander (AANHPI). With this resource, we specifically aim to: (1) assess the relative importance of established and emerging examine the extent to which these factors independently and jointly contribute to racial/ethnic disparities in HCC risk; (2) discover novel risk factors and assess their relative importance to HCC risk; and (3) assess racial/ethnic disparities in adherence with surveillance for HCC as well as examine the extent to which these disparities are attributable to modifiable individual-, clinician-, system-, and neighborhood factors (Aim 3). For Aim 1, using prospective data, we will assess the relative importance of risk factors and their contribution to racial/ethnic disparities in HCC risk with causal inference methods. For Aim 2, we will apply innovative machine learning methods to identify novel factors and validate their associations with HCC risk using modeling strategy from Aim 1. For Aim 3, we will use multilevel generalized linear regression to investigate the patient, clinician, institutional and geographic factors that contribute to disparities in HCC surveillance. Given the importance of sex and age/birth cohort for HCC risk, these social determinants will be considered together with race/ethnicity using an intersectional approach. By applying a multilevel framework to understand how biological, clinical, and social factors at multiple levels contribute to HCC disparities in incidence and surveillance, the proposed study will identify modifiable factors that can be translated to the clinical and community settings to collaboratively identify strategies to ameliorate racial/ethnic disparities in HCC.