Considering Care-Seeking Behaviors Reveals Important Differences Among HIV-Positive Women Not Engaged in Care: Implications for Intervention

Introduction

Many women living with HIV in the United States (US) face substantial challenges to engaging in HIV medical care. ^1,2 Only 40% of HIV-positive women are engaged in ongoing HIV care (e.g., regular visits with an HIV medical care provider) and merely one-quarter of HIV-positive women are virally suppressed. ³ As a result, many are not receiving the full benefits of HIV treatment. Consequently, they are at high risk for poor health outcomes and for transmitting HIV to others. ^{4 –10}

Although prior studies have identified numerous factors associated with poor engagement in care (e.g., lack of transportation ^{11 –13} and health insurance, ^{14 –19} caregiving responsibilities, ^13,20 depression, ^13,16,21 and substance use ^{14,16,19,21 –25} ), these studies have been limited in two important respects. First, many have viewed those who are not engaged in care as a single, homogenous group, failing to recognize potentially nuanced differences among women that may help to inform the development of targeted interventions. Second, these studies have not routinely taken into account care-seeking behaviors among those who are not engaged in care. Care-seeking behaviors are a significant determinant of health care utilization ²⁶ and, therefore, vital to consider when attempting to understand potential differences among individuals not engaged in care. Specifically, there may be differences in characteristics and barriers to HIV medical care among those who are not engaged in care, but who may be actively seeking care as compared with those who are not engaged in care and not seeking care. The latter group may be at greater risk for poor health outcomes, as they are probably less likely to interface with the health care system and, therefore, have fewer opportunities to receive care.

Currently, there are only two published, rigorously evaluated interventions focused on retaining women in HIV medical care in the US. ²⁷ Therefore, understanding differences that may exist among HIV-positive women who are not engaged in care, by taking into account care-seeking behavior, may be helpful in the development of future interventions. Therefore, as has been traditionally done, we first examined differences between women engaged in care and those not engaged in care. To explore additional potential differences that may be targets for tailored interventions, we then compared three distinct categories of women, which take care-seeking behaviors into account: women engaged in care; women not engaged in care, but seeking care; and women not engaged in care and not seeking care.

Methods

Results

Discussion

Few published, rigorously tested interventions exist that are specifically focused on retaining HIV-positive women in medical care. To help inform the future development of targeted retention-in-care interventions for HIV-positive women, we sought to identify potential differences among women not engaged in care by taking into account care-seeking behaviors. By considering care-seeking behaviors, we identified important differences in characteristics between seekers (not engaged in care, but seeking care) and non-seekers (not engaged in care and not seeking care), differences that were not evident when the women were traditionally categorized as one homogenous group. When examined as a single group, women not engaged in care were less likely to be black and more likely to be uninsured, to report fair or poor health status, and report high-risk drug use behaviors. However, taking care-seeking behaviors into account revealed important distinctions among women not engaged in care. Compared to women in care, seekers were more likely to be uninsured and report fair or poor health status. In contrast, non-seekers were not only more likely to be uninsured, but to also report high-risk drug use and sexual behaviors and less likely to report transportation as a barrier to care.

Prior studies of women not engaged in HV medical care have not explicitly examined potential differences among this group nor have they routinely taken into account role of care-seeking behaviors. Our findings suggest that among HIV-positive women who are not engaged in care, considering care-seeking behaviors uncovers key differences. Similar to previous studies, we found that regardless of care-seeking behaviors, women who are out of care were more likely to be uninsured. ^{14,16 –19} However, despite this finding, we found that, compared with women in care, both seekers and non-seekers were no more likely to perceive lack of health insurance as a barrier to care. Although prior studies have found drug use to be associated with poor engagement in care, ^{14,16,19,21 –25} we found that, compared with women in care, only non-seekers were more likely to report high-risk drug use and sexual behaviors. As this subgroup was less likely to report transportation as a barrier to care and was no more likely to report other structural barriers to care as women in care, our findings suggest that this subgroup of women may not report barriers because they are not actively seeking HIV medical care. Taken together, these findings indicate that women who are not engaged in care and not seeking care are a particularly vulnerable subgroup of HIV-positive women who are involved in high-risk behaviors and who may have limited to no interaction with the health care system.

Interventions focused on women who are not engaged in care and not seeking care may need to focus on risk behaviors, including drug use, and be community-based. As previously mentioned, substance use is a well-known factor associated with poor engagement in HIV care including among women. Effective interventions may need to incorporate substance use treatment or provide active referral to treatment programs. Efforts to engage women who are not seeking care must recognize that some women who are not engaged in care may not necessarily be actively seeking HIV medical care. As such, interventions targeting this subgroup may need to be based at venues or locales in which women who are not engaged in care and not seeking care may be more likely to frequent such as community-based organizations or homeless shelters. For instance, a recent systematic review identified few rigorously evaluated interventions focused on retention in care among HIV-positive individuals that were community-based and only two interventions that focused specifically on HIV-positive women, suggesting a substantial need for community-based approaches to engaging women. ³¹

As engagement in HIV care remains a persistent challenge, our findings underscore the importance of the research and public health communities taking into account the potential heterogeneity among those not engaged in care. Past studies that have treated those who are not engaged in care as one homogenous group may have missed important nuanced differences among subpopulations such as those observed in our study. Understanding key differences among distinct subgroups can guide development of targeted interventions that aim to improve engagement in care for vulnerable subgroups, thus, filling the existing void of rigorously evaluated interventions to promote engagement in care.

Our study has several limitations. First, one eligibility criterion focused on previous clinical visits at only one specific site. As such, it is possible that some participants may have been engaged in care at another clinical site and therefore, could have been misclassified in our analyses. Second, our categorization of care status and care-seeking behavior relied on self-report. Therefore, assessment of receipt of medical care could have been inaccurate. Third, we dichotomized care-seeking behavior, a health behavior which is complex and dynamic. ^{32 –34} Yet, we believe this analysis serves as a formative step to begin investigating novel ways of thinking about engagement in care that can reveal important distinctions. Fourth, diagnoses of mental health and co-morbid medical conditions were based on participant self-report and, as such, may have been under-reported. Therefore, we are unable to fully explore their contributions to our findings. Lastly, we did not have access to participants' CD4 cell count and HIV viral load data at the time of enrollment. Consequently, we were unable to determine to what extent there may have been differences in immunologic and virologic control between groups and how these variables might be associated with care status.

In a multisite study of HIV-positive women, we found that women who are not engaged in HIV medical care are a heterogeneous group. Specifically, compared to women in care, women who were not engaged in care and not seeking care were more likely to be uninsured and report current high-risk behaviors and less likely to report transportation as a barrier. Our findings suggest that a single, uniform approach to women who are not engaged in HIV medical care will likely be insufficient given the heterogeneity in key characteristics, particularly risk behaviors, among women not engaged in care. Moreover, they highlight the need for those developing targeted interventions to think creatively about how to engage HIV-positive women who may have limited interaction with health care system and are at high risk for poor health outcomes.