Publication Date




Embargo Period


Degree Type


Degree Name

Doctor of Philosophy (PHD)


Public Health Sciences (Medicine)

Date of Defense


First Committee Member

Daniel J. Feaster

Second Committee Member

WayWay M. Hlaing

Third Committee Member

Adam W. Carrico

Fourth Committee Member

Lisa Rosen-Metsch


Engaging and retaining individuals with HIV who also use drugs or alcohol in care is essential to reducing HIV-related morbidity and mortality in the United States. HIV-infected individuals who use substances are less likely to access care, adhere to treatment plans, and achieve viral suppression compared to those who do not report substance use. This leads to uncontrolled infection and contributes to ongoing disease transmission. Multiple factors are associated with poor engagement in HIV care and often include a combination of syndemic psychosocial factors, comorbid conditions, and structural barriers to healthcare utilization. Patient Navigation (PN) is a client-centered intervention designed to reduce barriers to care, but its effectiveness is unclear among HIV-infected people who use drugs or alcohol. Through secondary analysis, this dissertation examined characteristics of a PN intervention to identify factors that contribute to its effectiveness in this population. Data were obtained from Project HOPE, a randomized, controlled trial evaluating the effect of PN on engagement in HIV care and viral suppression among people living with HIV who use drugs or alcohol (n=801). Three unique statistical methods were employed in this analysis. First, latent variable analysis was used to identify barrier domains and latent profiles of individuals within the study sample. The effects of PN on engagement in care and viral suppression were then compared across latent profiles. Second, a mediation analysis was conducted to understand the indirect effects of PN through two proposed mediators—treatment barrier reduction and self-efficacy. Third, Random Forest analysis was conducted to identify additional patient-level characteristics and PN administration factors predicting engagement in care and viral suppression. The results of these analyses revealed four domains of barriers to care—Psychosocial Conditions, Healthcare Avoidance, Financial Hardship, and System Barriers. In addition, there were three general subgroups of individuals categorized as having High, Medium, and Low levels across all barrier domains except System Barriers, which had different relationships to the other domains by race. Among Black participants, System Barriers were inversely related to the other domains. In contrast, there was no variation in System Barriers across profiles among White participants. PN interventions are likely to be most effective for individuals with low levels of healthcare barriers. Special attention should be given to individuals with a history of physical or sexual abuse, intimate partner violence, discrimination, or high suicidal risk. These individuals are least likely to benefit from PN alone and may require alternative interventions. PN works mainly through a direct effect but also has an indirect effect through self-efficacy when coupled with financial incentives. Barrier reduction did not have a significant mediating effect. Additional factors that may support positive health outcomes include provider trust and participatory decision making in treatment. Collectively, these findings help to inform the delivery of PN interventions for individuals with HIV who use drugs or alcohol. Additional research is needed to expand these findings to include new populations and additional intervention factors.


HIV; patient navigation; substance use; latent variable analysis; mediation analysis; random forest

Available for download on Saturday, December 11, 2021