Research on the determinants of injection drug use behavior has traditionally concentrated on factors operating at the individual level. However, more recent studies have found that behaviors surrounding injection drug use are shaped, not only by individual-level characteristics, but also by the environment in which they occur. The risk environment paradigm, proposed by Rhodes and colleagues, describes how factors exogenous to the individual influence high-risk injection behavior and blood borne virus (BBV) transmission among people who inject drugs (PWID). To date, few elements of the risk environment have been evaluated as potential determinants of high-risk injection behavior. The purpose of this dissertation was to study the influence of two elements of the risk environment on unsafe injection practices among PWID – neighborhood socioeconomic disadvantage and social discomfort surrounding the acquisition of sterile syringes from syringe exchange programs (SEPs) and pharmacies. To this end, a systematic literature review was conducted on the relation between neighborhood context and injection drug use behavior. Research gaps and methodological challenges identified in this review were used to design analyses exploring relations among neighborhood disadvantage, social discomfort, and high-risk injection behavior. These analyses were conducted using data collected from 484 PWID enrolled in the Pharmacists as Resources Making Links to Community Services (PHARM-Link) study, combined with data from the American Community Survey. Poisson regression with robust error variance was used to estimate associations between measures of neighborhood socioeconomic disadvantage and high-risk injection behavior. SEP accessibility and drug-related police activity were evaluated as potential modifiers of these relations. Similar methods were used to estimate associations between measures of social discomfort and high-risk injection behavior, including neighborhood socioeconomic disadvantage as a potential effect modifier. The systematic literature review on neighborhood context and injection drug use behavior identified few articles pertaining to this relation (n=22). Selected studies primarily investigated the influence of structural aspects of the neighborhood environment on behaviors surrounding injection drug use, while aspects of the social environment and potential modifiers of neighborhood-behavior relations were understudied. Subsequent quantitative analyses revealed that neighborhood socioeconomic disadvantage was associated with safer injection behaviors among PWID. Injectors in disadvantaged neighborhoods reported less receptive syringe sharing and less unsterile syringe use than their counterparts in relatively better off neighborhoods. Drug-related police activity attenuated associations between neighborhood disadvantage and unsterile syringe use, while the direction of associations between neighborhood disadvantage and the use of unsafe syringe sources varied with levels of SEP accessibility. In neighborhoods with high SEP accessibility, neighborhood disadvantage was associated with decreased use of unsafe syringe sources, while in neighborhoods with low SEP accessibility, neighborhood disadvantage was associated with increased use of unsafe syringe sources. Social discomfort was not associated with high-risk injection behavior, but effect modification was detected between neighborhood disadvantage and two items measuring the quality of relationships between participants and syringe staff: “Pharmacists care about my health and well-being” and “The staff at syringe exchange programs seems to care about my health and well-being.” In disadvantaged neighborhoods, participants who reported positive relationships with syringe staff were less likely to engage in receptive syringe sharing. However, in relatively better off neighborhoods, positive relationships with syringe staff were associated with increased receptive syringe sharing. Overall, the results of this dissertation support the validity of the risk environment paradigm in shaping high-risk injection behavior among PWID. Future studies should continue to investigate contextual factors as determinants of behavior surrounding injection drug use. Understanding how aspects of local-area environments influence injection risk behavior will be essential to eliminating the transmission of BBVs among PWID.

Financial strain and substance use are widespread in the United States (US), as many adults struggle to meet basic financial needs. There are many contributing causes of financial strain, such as unemployment or job loss and poverty or income loss. However, financial strain is distinct from its sources, as the experience of strain (i.e., inability to pay for bills, housing, healthcare, or food) may be necessary to produce a stress response that prompts substance use as a coping mechanism. Studies examining the relationship between financial strain and substance use have predominantly focused on the association between financial strain and alcohol use or acute financial strain due to brief changes in employment or income. However, there is a need to examine whether the relationship between financial strain and substance use varies by substance and the duration of financial strain (e.g., acute or brief financial strain, chronic or persistent financial strain, or intermittent financial strain due to repeated periods of financial strain). Several social safety net programs may mitigate financial strain, such as Unemployment Insurance, the Temporary Assistance for Needy Families program, the Supplemental Nutritional Assistance Program, and economic policies such as the federal Earned Income Tax Credit (EITC). Anti-poverty programs such as the federal EITC may provide an opportunity to reduce financial strain and subsequent substance use by providing financial support to low-income working populations. The purpose of this dissertation was to provide novel insights into the associations between acute, chronic, or intermittent financial strain and substance use and whether federal EITC eligibility altered these relationships. To achieve these goals, I had four aims. First, I sought to understand the associations between substance use and financial strain and its contributing causes (e.g., unemployment, poverty, etc.) by conducting a scoping review of the substance use literature (Aim 1; Chapter 2). Second, I examined the associations between acute, chronic, and intermittent financial strain and past month binge alcohol, cannabis, central nervous system (CNS) depressant (i.e., painkillers, sedatives, or tranquilizers), stimulant,and poly-drug use and whether these associations varied by sex (Aim 2.1; Chapter 3.1). Third, I examined the relationships between any financial strain and chronic financial strain on past month cannabis, CNS depressant, stimulant, and poly-drug use, adjusting for the time-dependent relationship between financial strain and drug use (Aim 2.2; Chapter 3.2). Fourth, I examined whether federal EITC eligibility was associated with short-term changes in financial strain, cannabis use, and CNS depressant use and whether this varied by state EITC policies or sex (Aim 3; Chapter 4). I conducted the scoping review presented in Chapter 2 in July-August 2023, searching for literature on the US, published since 2001 in PubMed, EBSCOhost [APA PsycINFO Database, EconLit with Full Text Database, SocINDEX with Full Text Database, Social Sciences Full Text Database], Web of Science, and Scopus. Two reviewers screened each abstract and title and conducted the full-text review. I extracted study characteristics and synthesized and evaluated evidence of the relationships between financial strain and its contributing causes and substance use. I found that more than one-third of studies examined employment-related measures (i.e., unemployment, job loss, or duration of unemployment) as indicators of financial strain, approximately 31% examined income-related measures (e.g., poverty, income loss) as indicators of financial strain, and only one-fifth of studies directly assessed financial strain, with more than half of those studies using a single-item question to assess financial strain. Job loss and duration of unemployment were generally associated with increased tobacco, binge alcohol, cannabis, opioid, drug, and substance use. Income-related indicators of financial strain were positively associated with tobacco, binge alcohol, and opioid use but inversely associated with any alcohol and cannabis use. Most studies found that financial strain was positively associated with tobacco and binge alcohol use. There was also evidence of a bi-directional relationship between alcohol and drug use and disorder with unemployment. My empirical aims leveraged individual- and state-level data from 5 waves (2013-2019) of the Population Assessment of Tobacco and Health study, a longitudinal cohort of >30,000 US adults. In Chapter 3.1, I characterized financial strain exposure patterns (i.e., none, acute, chronic, and intermittent) across Waves 1-5 and used survey-weighted multinomial logistic regression models to estimate the adjusted relative risk ratio of past month binge alcohol, cannabis, CNS depressants, stimulants, and poly-drug use at Wave 5 by financial strain exposure patterns, and whether this varied by sex. Consistent with prior research, I found that the relationship between financial strain and substance use varied by substance. Acute and intermittent financial strain were associated with an increased likelihood of binge alcohol, stimulant, and poly-drug use, whereas chronic financial strain was associated with an increased likelihood of cannabis or CNS depressant use. I found modest or no sex differences in the relationship between financial strain duration and binge alcohol, cannabis, CNS depressant, stimulant, or poly-drug use. Findings were robust in sensitivity analyses when I varied the number of waves considered chronic financial strain. Together, findings suggest that individuals experiencing a period of financial strain may be at greater risk of past month binge alcohol or drug use, and the risk for cannabis and CNS depressant use may increase with financial strain duration. In Chapter 3.2, I used longitudinal targeted maximum likelihood estimation methods to account for time- dependent confounding between financial strain and drug use and to estimate the association between any financial strain or chronic financial strain and cannabis, CNS depressants, stimulants, and poly-drug use across Waves 1-5. This doubly robust method allowed me to predict the average expected outcomes if everyone had never experienced financial strain across the study period (i.e., a never financial strain scenario) and if everyone had experienced financial strain at every time point across the study period (i.e., a chronic financial strain scenario). I then compared the expected outcomes under the never financial strain scenario to outcomes in the observed data to estimate the association between any financial strain and drug use. I found that if everyone had never experienced financial strain, the prevalence of past month cannabis use would be lower than the prevalence of past month cannabis use observed in the data. When I compared the expected outcomes under the never financial strain scenario to the expected outcomes in the chronic financial strain scenario, I found that the likelihood of past month cannabis or CNS depressant use would be significantly higher if everyone had experienced chronic financial strain across all five waves of data collection. These findings confirm the results in Chapter 3.1, suggesting that the association between chronic financial strain and drug use remained elevated after accounting for the potentially reinforcing relationship between financial strain and drug use over time. These analyses additionally revealed that the risk of cannabis use would have been slightly lower if no one had ever experienced financial strain compared with any financial strain. In Chapter 4, I examined changes in financial strain, cannabis use, and CNS depressant use associated with federal EITC eligibility during the EITC disbursement period. EITC could be considered a short-term intervention for financial strain and could provide temporary relief to people experiencing chronic financial strain, helping them transition from chronic to intermittent financial strain and lowering the likelihood of cannabis and CNS depressant use. I used survey participants’ interview dates to assess if EITC-eligible people who were randomly interviewed during the EITC disbursement period (February-April) had a lower risk of financial strain compared with EITC-eligible people interviewed outside the disbursement period (May-January) and EITC-ineligible people, using linear binomial models with a two-way interaction term for EITC eligibility and the EITC disbursement period. I then assessed whether changes in financial strain, cannabis use, and CNS depressant use associated with EITC eligibility during the EITC disbursement period varied by state refundable EITC policies. To do this, I used linear binomial models with a three-way interaction term between EITC eligibility, EITC disbursement period, and refundable state EITC policies. Finally, I conducted stratified analyses by sex to examine whether changes in financial strain, cannabis use, and CNS depressant use associated with EITC eligibility during the EITC disbursement period varied by men and women. Results indicated that receiving an EITC refund of at least $500 or more was associated with decreased financial strain, particularly among women. However, EITC eligibility during the EITC disbursement period was not significantly associated with past month cannabis or CNS depressant (i.e., painkiller, sedative, or tranquilizer) use overall or by sex at Wave 1. Changes in financial strain, cannabis use, and CNS depressant use associated with federal EITC eligibility during the disbursement period did not vary by whether the participant’s state of residence offered an additional refundable EITC. Findings from this dissertation provide empirical support for financial strain as a potential predictor of binge alcohol, cannabis, CNS depressant, stimulant, and poly-drug use. This relationship varied by duration of strain, and the association between chronic financial strain and drug use remained elevated when adjusting for potential time-varying confounding in this relationship. I also found that the likelihood of cannabis use would decrease if no one had ever experienced financial strain. When I examined the federal EITC as a potential short-term intervention for financial strain, I found that refunds of at least $500 or more were associated with decreased financial strain without increasing cannabis or CNS depressant use in the overall population. Thus, expanded and more generous income support policies such as the EITC may be effective tools to intervene on financial strain.