The North-South Divide: Substance Use Risk, Care Engagement, and Viral Suppression Among Hospitalized Human Immunodeficiency Virus–Infected Patients in 11 US Cities

Abstract

Regional variability in human immunodeficiency virus (HIV) care engagement remains underexplored. Multiple logistic models compared HIV outcomes for participants from 5 Southern (n = 557) and 6 non-Southern (n = 670) sites. Southern participants were less likely to experience viral suppression (adjusted odds ratio [aOR], 0.52; 95% confidence interval [CI], .37–.72) and had a higher likelihood of a CD4+ count <200 cells/µL (aOR, 1.53; 95% CI, 1.17–2.00). HIV intervention and social safety net programs should be expanded.

Clinical Trials Registration

NCT01612169.

Improving care engagement and viral suppression among human immunodeficiency virus (HIV)–infected individuals is vitally important to reducing morbidity, mortality, and HIV transmission [1]. Certain populations, including racial/ethnic minorities, males, and adolescents, face extensive barriers to HIV care engagement and viral suppression [2, 3]. Substance use is also a major contributor to negative HIV-related outcomes [4].

Few studies have examined regional differences in access to HIV-related care and outcomes. The Centers for Disease Control and Prevention (CDC) has identified the South as the new epicenter of the HIV epidemic in the United States [5]. Data suggest that Southern states tend to have lower rates of care linkage and engagement than those in the North and Midwest [6, 7], and higher HIV-related mortality and poorer survival [8]. Identifying geographic differences in treatment and care outcomes could assist policy makers in tailoring interventions, updating policies, and determining areas of greatest need.

This secondary analysis used screening data from a randomized controlled trial to examine differences in antiretroviral (ART) history, care retention, HIV outcomes, and substance use treatment participation among hospitalized HIV-infected patients across 5 Southern compared with 6 non-Southern US sites.

METHODS

The study analyzed screening data from the National Institute on Drug Abuse Clinical Trials Network 0049 study (NCT01612169). From July 2012 to January 2014, 2291 HIV-infected patients were screened for participation in Hospital visit as an Opportunity for Prevention and Engagement for HIV Infected Drug Users (Project HOPE) [9], which assessed whether a patient navigation intervention (with or without contingency management) impacted HIV viral suppression. While all sites began by screening all HIV-infected patients in the hospital whom medical staff had cleared for approach, a few Project HOPE sites initiated prescreening at a later stage; this analysis includes participants recruited before the prescreening process (N = 1227).

Recruitment occurred in 11 hospitals across the United States that had a high (>200 unique patients per year) HIV inpatient census and high prevalence of substance use among HIV-infected patients; the majority of patients were admitted for HIV-, substance use–, or psychiatric-related reasons. The geographic classification of sites in the study as Southern vs non-Southern was based on the CDC definition [5]; there was no difference in the percentage of people approached for screening between the Southern and non-Southern sites. Institutional review boards at all participating institutions approved this research, and patients provided written informed consent. Southern sites included Atlanta, Georgia; Baltimore, Maryland; Birmingham, Alabama; Dallas, Texas; and Miami, Florida. Non-Southern sites included Boston, Massachusetts; Chicago, Illinois; Los Angeles, California; New York, New York; and Philadelphia and Pittsburgh, Pennsylvania. Of the 6 non-Southern hospitals, 3 were private, 2 were county/state government, and 1 was both state and private; of the 5 Southern hospitals, 2 were private and 3 were county/state government.

Measures

Sociodemographics, substance use treatment, health status, and healthcare utilization data were collected using standard measures (for more details, see Metsch et al [9]).

Statistical Analyses

To compare participants living in Southern vs non-Southern sites, we used Pearson χ² tests for categorical variables, a t test for age, and Wilcoxon signed-rank test for viral load and CD4⁺ counts. Participants in these analyses were nested within sites. To account for the potential site effect at a participant level, we used generalized estimating equation models to evaluate the association between living in a Southern vs non-Southern site, problematic drug/alcohol use, HIV care engagement, antiretroviral adherence, CD4⁺ count, and viral load while controlling for sociodemographic variables including age, gender, race, employment, relationship status, housing, health insurance, food security, and substance use. We report adjusted odds ratios (aORs) with 95% confidence intervals (CIs). Analyses were performed using SAS statistical software version 9.3 (SAS Institute, Cary, North Carolina). All tests were performed at a significance level of .05.

RESULTS

A total of 1227 participants were screened and included in these analyses: 557 in Southern sites and 670 in non-Southern sites. Just over two-thirds of the sample was male; 12% self-identified as Hispanic, 66% black, 16% white, and 6% as other racial/ethnic group. The median age was 48 years. Nearly two-thirds had a high school education or greater and had never married. Most had an annual income of $20000 or less (86%) and reported stable housing (72%) and health insurance (80%). Nearly two-thirds reported a lifetime history of incarceration and one-quarter reported insufficient food in the past month. Just over half reported having ever used drugs and, among those, nearly half had attended substance use treatment in their lifetime and 40% in the last year. Eighty-eight percent reported having ever been in HIV care and, among participants with a primary HIV provider, 90% had a past-year visit. The majority (80%) reported currently taking ART. Approximately half had a CD4⁺ count <200 cells/µL (49%) while 43% were virally suppressed.

There were significant sociodemographic differences between participants in Southern vs non-Southern sites. Participants in Southern sites were more likely to be younger (45.8 vs 47.5 years; P < .01), black (70.9% vs 62.6%; P < .001), disabled (55.7% vs 43.8%; P < .001), and uninsured (23.6% vs 17.1%; P < .01). They were less likely to report a hepatitis C diagnosis (31.6% vs 37.9%; P = .023), be unemployed (22.9% vs 37.3%; P < .001), or to have been incarcerated in the last 6 months (6.7% vs 10.6%; P = .017). Participants in Southern sites were also less likely to report any substance use (44.6% vs 59.4%; P < .001), to report problematic drug or alcohol use (45.8% vs 60%; P < .001), or to have ever attended a substance use treatment program (38.6% vs 50.5%; P < .001). Participants in Southern sites were more likely to have had their last HIV care visit >12 months ago (12.7% vs 8.2%; P = .016) and to have a CD4⁺ count of <200 cells/µL (55.1% vs 43.4%; P < .001); in contrast, they were less likely to be virally suppressed (35.3% vs 49.3%; P < .001).

In the multivariate regression models (Table 1), participants in Southern sites were less likely to report problematic drug or alcohol use (aOR, 0.55; 95% CI, .43–.71), currently take ART (aOR, 0.69; 95% CI, .48–.98), or be virally suppressed (aOR, 0.52; 95% CI, .37–.72). They were more likely to have had their last HIV visit >12 months ago (aOR, 1.69; 95% CI, 1.09–2.62) and to have a CD4⁺ count <200 cells/µL (aOR, 1.53; 95% CI, 1.17–2.00).

Table 1.

Adjusted Associations of Region and Other Factors With Access to Human Immunodeficiency Virus (HIV) Care and Biomarkers of Disease Status Among HIV-infected Inpatients Recruited From 11 US Hospitals

Characteristic	Problematic Drug/Alcohol Usea	Ever Linked to HIV Care	Last HIV Visit >12 mo	Ever Taken Antiretroviral	Currently Take Antiretrovirals	Viral Suppressionb	CD4 <200 Cells/µL
Age	0.99 (.98–1.01)	1.01 (.99–1.03)	0.98 (.96–1.00)	1.02 (1.01–1.04)	1.02 (1.01–1.04)	1.06 (1.04–1.08)	0.97 (.95–.98)
Male vs female	1.10 (.84–1.43)	0.68 (.43–1.08)	0.88 (.56–1.40)	1.12 (.77–1.62)	1.27 (.88–1.83)	0.78 (.55–1.10)	1.17(.88–1.55)
Race/ethnicity
Hispanic vs white	1.04 (.65–1.67)	0.54 (.24–1.21)	1.35 (.53–3.46)	0.65 (.34–1.22)	0.83 (.42–1.61)	0.49 (.27–.89)	1.65(1.01–2.70)
Black vs white	0.88 (.63–1.24)	0.44 (.23–.84)	2.14(1.05–4.36)	0.70 (.42–1.17)	0.64 (.39–1.05)	0.48 (.31–.74)	1.92(1.32–2.79)
Other vs white	1.00 (.56–1.80)	0.73 (.25–2.12)	2.97(1.13–7.76)	0.79 (.34–1.80)	0.73 (.33–1.64)	0.94 (.44–2.00)	1.56(.83–2.91)
Marriage
Nevermarriedvs married/cohabiting	1.64 (1.15–2.34)	1.00 (.56–1.79)	0.83(.44–1.54)	0.99 (.60–1.64)	1.03 (.63–1.68)	1.22 (.76–1.93)	0.88(.61–1.28)
Employment
Unemployed vs working	1.01 (.68–1.50)	1.19 (.70–2.03)	0.87 (.45–1.68)	0.73 (.45–1.19)	0.82 (.48–1.40)	0.83 (.49–1.40)	0.94 (.61–1.42)
Disabled vs working	1.17 (.80–1.70)	2.46 (1.39–4.34)	0.73 (.38–1.40)	1.86 (1.10–3.14)	1.45 (.86–2.42)	1.00 (.62–1.63)	0.95(.64–1.42)
Problematic drug/alcohol use	…	0.55 (.36–.82)	1.55 (.99–2.43)	0.53 (.37–.76)	0.43 (.30–.62)	0.43 (.31–.60)	1.23(.94–1.61)
Without medical insurance	1.33 (.96–1.83)	0.34 (.22–.52)	2.82(1.73–4.59)	0.43 (.30–.63)	0.36 (.24–.54)	0.54 (.34–.86)	2.00(1.42–2.83)
Unstable housing	2.55 (1.90–3.44)	0.97 (.61–1.53)	1.16 (.71–1.92)	1.01 (.68–1.50)	1.10 (.74–1.64)	0.64 (.44–.95)	1.07 (.79–1.45)
Insufficient food	1.50 (1.13–1.99)	1.17 (.74–1.84)	1.55 (.98–2.46)	0.91 (.62–1.33)	0.53 (.37–.76)	0.80 (.55–1.16)	1.57 (1.17–2.11)
South vs non-South	0.55 (.43–.71)	0.72 (.48–1.06)	1.69 (1.09–2.62)	1.10 (.77–1.55)	0.69 (.48–.98)	0.52 (.37–.72)	1.53 (1.17–2.00)

Data are presented as adjusted odds ratios with 95% confidence intervals. Values in bold are significant at the P < .05 level.

Abbreviation: HIV, human immunodeficiency virus.

^aDefined as Alcohol Use Disorders Identification Test (AUDIT-C) score of >3 for women and >4 for men or having used stimulants or opioids.

^bViral load of ≤200 copies/mL in the past 12 months.

DISCUSSION

This study among hospitalized HIV-infected participants across 11 sites in the United States found that while participants living in Southern sites had lower rates of problematic drug alcohol use, they were less likely to report care engagement and more likely to have worse HIV outcomes. Specifically, almost two-thirds of HIV-infected participants recruited from the Southern hospitals presented with a detectable viral load compared with half of participants in the non-Southern hospitals.

While we cannot infer the reasons for this difference in viral suppression, the majority of participants were low-income ethnic/racial minorities who faced multiple health disparities including unstable housing, food insecurity, unemployment, disability, and educational level. These problems may be particularly acute in Southern sites, where social safety net programs to address these issues may be especially underfunded or unavailable [8, 10]. Also, research suggests that individuals in the South report higher levels of HIV-related stigma, homophobia, and racism than those in the North [5, 8]; Southern states are also more likely to have anti-sodomy laws and abstinence-only education in schools [11]. These contextual factors have been shown to be associated with suboptimal healthcare engagement, medication adherence [12], and awareness of HIV prevention measures such as preexposure prophylaxis [13].

In addition to social-structural policy differences between the North and South, there are also regional variations between state and federal funding streams for HIV prevention and treatment that may have influenced study results. For instance, the amount the Ryan White Program provided per HIV-infected individual was, on average, $2912 in the Southern sites in this study and $3250 in the non-Southern sites; however, overhead and personnel costs are often lower in the South [14]. These discrepancies may be further exacerbated by the lack of Affordable Care Act–supported Medicaid expansion in most Southern states [15].

There were limitations to this study. This population represents a geographically diverse sample from 11 cities; however, this was a secondary analysis, so results are exploratory. While we controlled for many potential confounders, our findings may be influenced by unmeasured factors; as a cross-sectional analysis, causality cannot be inferred. Study sites were urban, further limiting the generalizability to areas such as the rural South. To enroll, participants had to be hospitalized at major medical centers and were therefore already accessing care, albeit emergency care. Variables related to risk factors and the use of HIV care, substance use disorders treatment, ART adherence, and kept medical appointments were based on self-report, but the screening process relied on medical records to document the presence or suppression of viremia. Participants represent a hospitalized population with more advanced HIV disease, limiting our ability to contribute generalizable data to the overall HIV continuum of care.

In conclusion, in this secondary analysis, HIV participants in the Southern sites were less likely to be engaged in care and to be virally suppressed compared with participants from non-Southern sites. These findings are consistent with other research that has shown HIV resources, Medicaid, and social safety-net programs that may benefit populations at high risk for HIV are generally underfunded in the South compared with other parts of the United States.

Notes

Financial support. This work was supported by the National Institute on Drug Abuse (NIDA) under the following awards: U10DA013720 and UG1DA013720 to Dr José Szapocznik and L. R. M.; U10DA013035 and UG1DA013035 to Drs John Rotrosen and Edward V. Nunes, Jr; U10DA013034 and UG1DA013034 to Drs Maxine Stitzer and Robert Schwartz; U10DA013727 and UG1DA013727 to Drs Kathleen T. Brady and Matthew Carpenter; U10DA020024 and UG1DA020024 to Dr Madhukar H. Trivedi; U10DA013732 and UG1DA013732 to Dr Theresa Winhusen; U10DA015831 and UG1DA015831 to Drs Roger D. Weiss and Kathleen Carroll; U10DA015815 and UG1DA015815 to Drs James L. Sorensen and Dennis McCarty; U10DA020036 to Dr Dennis Daley; U10DA013043 to Dr. George Woody; U10DA013045 to Dr Walter Ling; N01DA142237 to Dr Paul VanVeldhuisen; N01DA102221 to Dr Robert Lindblad; K01DA039804A to M. M. P.; and K24DA035684 and P30AI094189 to G. M. L. This work was also supported by the University of Miami Center for AIDS Research (CFAR) (award number P30AI073961 to Dr Savita Pahwa); the Emory University CFAR (award number P30AI050409 to C. d. R., Drs James W. Curran and Eric Hunter); and the Atlanta Clinical and Translational Science Institute (award number UL1TR000454 to Dr David Stephens). A. N. and R. D. have received grants from the National Institutes of Health. D. K. M. has received grants from the University of Pittsburgh. L. G. and J. M. J. have received grants from NIDA.

Potential conflicts of interest. E. S. D. receives research support from Gilead, Merck, and ViiV and is a consultant/advisor for Bristol-Myers Squibb, Gilead, Janssen, Merck, Theratechnologies, Teva, and ViiV. M. D. is an employee of Gilead Sciences and owns stock in the company. A. N. has received grants from Gilead. All other authors report no potential conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.