**4. Discussion**

This study provides important insights about the differential effects of PN interventions for engaging PLWH-SU in care. It suggests that PN, offered with or without CM, is most effective for individuals with relatively low levels of healthcare barriers. Of the three barrier profiles identified in this analysis, the LB group had the greatest response to PN, with higher 6- and 12-month rates of engagemen<sup>t</sup> in care and viral suppression than the TAU group. The positive intervention effects observed for the LB group may be explained by the absence of extreme healthcare barriers that would delay or compete with the need to engage in care. For example, if a patient has an overwhelming and immediate need to address an aspect of their wellbeing, such as a severe mental health condition or unstable housing, the patient may prioritize such a need over HIV care. The navigator would need to help resolve these other issues before the patient is ready to focus on HIV care. If, however, a patient is stable and does not require other assistance, the patient navigator can focus on linking the individual directly to care.

Conversely, individuals with a history of abuse, IPV, and discrimination are not likely to benefit from stand-alone PN interventions. In this analysis, the HB-AV group had only a partial response to the PN and PN+CM interventions. These individuals had higher odds of being engaged in care at 6 months, but these effects were not sustained at 12 months and did not lead to viral suppression. Finally, the HB-DAV group did not respond to either of the PN interventions. These results imply that PN (with or without CM) is not sufficient for all patient populations and underscore the importance of a thorough assessment of patients' needs when recommending behavioral interventions. In an era of precision medicine, the development of personalized interventions is becoming increasingly more valuable in prevention science.

These findings contribute to the current science of healthcare utilization among PLWH-SU by identifying high-risk barrier profiles. Specifically, a history of abuse, intimate partner violence, and/or discrimination are important indicators of a high overall level of healthcare barriers. In both profiles characterized by abuse and IPV, nearly all other barriers were present at higher rates compared to the profile without abuse and IPV. This is consistent with other work, most notably Singer's work on the SAVA syndemic of substance abuse, violence, and HIV/AIDS, indicating that these factors are likely to co-occur [25,26,53,69,80]. Individuals who experience IPV and/or abuse are more likely to suffer from depression and other psychiatric disorders [54] Substance use in this context further perpetuates violence and abuse. Healthcare personnel should be cognizant of these factors, incorporate screening for multiple conditions into practice, and be prepared to link patients to the appropriate programs or provide appropriate co-located services.

This study also provides insights about the potential impacts of trauma and abuse on the effectiveness of health interventions. PN interventions designed to engage PLWH-SU in care were not found to be effective for individuals with a history of IPV, physical or sexual abuse, or discrimination. This finding may be related to the possibility that individuals are still in abusive relationships at the time PN interventions are administered. Individuals in such situations may lack the resources and/or the autonomy to independently seek healthcare or suffer from fear or anxiety about being in a healthcare setting where the abuse may be discovered. Even if an individual is not actively in an abusive situation, the harms from past events may have lingering mental health effects that influence one's decision to seek care. Additionally, if a person experienced abuse or discrimination in a healthcare setting, this could deter that person from seeking care in the future. Thus, the identification of psychosocial barriers to care is an important part of a routine needs assessment, and it is especially important to determine if there is a history of abuse or IPV, with or without discrimination. Alternatives to PN, or PN delivered in combination with other interventions, may be required to result in positive health outcomes for these individuals.

Another noteworthy finding is that among the subgroups that had positive responses to the PN interventions, the effects were stronger when CM was added to PN, compared to PN alone. It could be that the combination of PN and CM interventions targets both intrinsic and extrinsic motivations for behavior change [70]. Alternatively, the financial incentives may have enhanced the effect of PN by encouraging individuals to attend more PN sessions [81]. Additional research is needed to evaluate the effect of this combined approach on this study's population and other patient populations.

The results of this study should be considered with the following limitations. First, there may have been some degree of measurement error associated with the barriers to care included in the analysis. All measures were self-reported and some of the barriers were measured indirectly. For example, information about transportation barriers was derived from responses about how participants go<sup>t</sup> to their clinic appointment. A better way would be to specifically ask about transportation barriers to healthcare. Measurement tools designed specifically to evaluate barriers to HIV care, such as the Kalichman's Barriers to Medical Care instrument [82], should be considered for future studies. Second, this analysis considered only individual and relationship-level barriers to care. A more comprehensive examination that includes higher level barriers, such as system and policy factors, may

reveal other distinct profiles that impact the response to interventions and should be explored in future analysis. Finally, the results of this work are limited to a specific population and may not be generalizable to other populations of PLWH. This study's population was a highly disadvantaged group of individuals with advanced HIV disease. This may have reduced the variability in observed healthcare barriers, as most of the CTN-0049 participants suffered from multiple barriers. Further research is needed to determine if similar barrier profiles exist in other populations.

Despite these limitations, this study has significant implications for public health practice. It underscores the importance of screening PLWH-SU for a history of abuse, IPV, and discrimination. Not only are they indicators of particularly vulnerable individuals, but they may also reduce the effectiveness of otherwise beneficial interventions. If these conditions are present, protocols should be initiated to make the appropriate referrals to mental health or social services. Screenings and follow-up assessments should be an ongoing part of interventions, not just part of the baseline evaluation. While this study was conducted in a U.S. population enrolled in a clinical trial, there may be important considerations for other PLWH populations with co-occurring drug use to identify and meet needs at the complex intersection of substance use and HIV services. A global review of studies assessing the integration of HIV and substance-use services showed that increased service integration can improve patient outcomes among this population across a variety of service models, both in and outside the U.S. [83]. Additionally, strategies to integrate treatment for mental health and substance use disorders among PLWH have been implemented in low-to-middle income countries [84]. Finally, this study builds on existing work by describing the complexities of how healthcare barriers group together. It suggests that, in addition to the number of barriers to care an individual faces, there are specific-barriers profiles that can differentially impact care. As a next step, it would be useful to conduct a direct comparison of latent variable approaches using barrier profiles with the composite-risk score method used in previous studies.
