Disparities in Initiation of Combination Antiretroviral Treatment and in Virologic Suppression Among Patients in the HIV Outpatient Study, 2000–2013

Objectives:The National HIV/AIDS Strategy emphasizes virologic suppression (VS) to reduce HIV incidence in the United States. We assessed temporal trends of and disparities in time to combination antiretroviral therapy (cART) initiation and HIV VS in a large demographically diverse cohort of HIV-infected patients. Design:We included antiretroviral-naive HIV Outpatient Study participants from 2000 to 2013 enrolled within 6 months of their HIV diagnosis who attended ≥2 HIV care–related visits. Methods:We evaluated time from HIV diagnosis to first use of cART, time from HIV diagnosis to VS, and time from first use of cART to VS. Kaplan–Meier time-to-event curves and Cox proportional hazards models were used to assess temporal trends and correlates of initiating cART and achieving HIV VS (<500 copies per milliliter). Results:Among 1156 HIV Outpatient Study patients [median age, 37 years; 43.2% non-Hispanic/Latino black (NHB), 14.1% Hispanic/Latino], estimated median times from HIV diagnosis to cART initiation and from HIV diagnosis to VS both shortened by >40% during the 13.5-year study period, reaching, respectively, 2.5 and 5.4 months. In multivariable analyses, NHB patients (as compared with non-Hispanic/Latino white) and those who had injected drugs (as compared with those who did not) initiated cART in a less timely fashion. After adjusting for CD4+ cell count and viral load at cART initiation, NHB patients and those aged <30 years (compared with ≥40 years) had lower rates of VS. Conclusions:Despite improvements in HIV treatment over time, patients who were NHB, younger, or used injection drugs had less favorable outcomes.


INTRODUCTION
The National HIV/AIDS Strategy in the United States emphasizes the importance of virologic suppression (VS) to improve the health of HIV-infected individuals and to reduce population-level transmission of HIV infection. 1 Modern combination antiretroviral therapy (cART) regimens are better tolerated and less complex than older regimens, increasing the likelihood of achieving VS. The success of HPTN 052, the HIV Prevention Trials Network randomized trial which demonstrated 96% reduction in HIV transmission with use of cART in HIV serodiscordant couples, lent support to the recommendation for universal treatment of persons with HIV infection in the United States, regardless of their CD4 + cell count. 2,3 Owing to improvements in the potency and tolerability of cART and changes in US guidelines to offer treatment to all patients, HIV-infected persons in the United States have been increasingly prescribed cART and achieving VS sooner after entry into HIV care. [4][5][6] Unfortunately, disparities in the continuum of HIV care in the United States persist and access to HIV treatment and subsequent clinical responses are not equitably distributed. [7][8][9] For example, despite the disproportionately higher diagnosis rates of HIV infection among non-Hispanic/Latino blacks (NHBs) compared with non-Hispanic/ Latino whites, 10 individuals in the former group have had less access to and uptake of cART and higher rates of virologic nonsuppression. 8,[11][12][13] We sought to evaluate temporal trends in cART initiation and VS in a large and demographically diverse cohort of HIV-infected patients to investigate potential risk factors and sociodemographic disparities. Understanding sociodemographic disparities is a first step toward identifying modifiable factors for interventions to improve the continuum of care for all HIV-infected persons in the United States.

The HIV Outpatient Study
The HIV Outpatient Study (HOPS) is an ongoing prospective observational cohort study of HIV-infected adults who have received care at any of the 9 participating HIV clinics (university-based, public, and private) in 6 US cities (Chicago, IL; Denver, CO; Stonybrook, NY; Philadelphia, PA; Tampa, FL; and Washington, DC). The HOPS, which started in 1993, is an open cohort; patients may enter the study at any time after a diagnosis of HIV infection regardless of treatment history and may leave the study at any time for any reason (eg, patient request, death, or loss to follow-up). 14 Since its inception, the HOPS protocol has been reviewed and approved annually by the institutional review boards at the Centers for Disease Control and Prevention (Atlanta, GA) and each of the local sites. Patient data, including sociodemographic characteristics, diagnoses, antiretroviral and other treatments, and laboratory values [including CD4 + T-lymphocyte count (CD4 cell count) and plasma HIV RNA viral load (HIV VL)], were abstracted from medical charts and entered into an electronic database for central processing and analysis.

Study Population
HOPS participants were included in this analysis if they were newly diagnosed with HIV during 2000-2013, joined the HOPS within 6 months of their diagnosis, were ART naive at the time of entry into HOPS, and attended at least 2 clinic visits at a participating site during the study period. Observation began at the time of HIV diagnosis and ended at the last HOPS contact, death, or June 30, 2013, whichever occurred first. We analyzed HOPS data collected during January 1, 2000, to June 30, 2013, using the HOPS data set updated through December 31, 2013, to account for any lags in medical records abstraction.

Variable Definitions
For analysis, 2 dates were of interest: date of first known cART use, defined per standard criteria, 15,16 and date of first VS, defined as first documented HIV VL ,500 copies per milliliter after HIV diagnosis. Patients were categorized according to their age at HIV diagnosis, grouped as ,25, 25-29, 30-39, and $40 years. Race/ethnicity was categorized as non-Hispanic/Latino white, NHB, Hispanic/Latino, and other/unknown. Patients were further classified by whether they were a person who injects drugs (IDU). We classified HOPS participants hierarchically according to their HIV transmission risk as gay, bisexual, and other men who have sex with men (collectively referred to as MSM), followed by women with heterosexual contact, men with heterosexual contact, and other/unknown. Insurance payor was defined as the primary insurance provider at the date of HIV diagnosis or earliest recorded payor thereafter and was categorized as private (ie, private insurance, preferred provider organization, health maintenance organization, point of service), public (ie, public insurance, Medicare, Medicaid, and Ryan White/AIDS Drug Assistance Program), and other/unknown (ie, self-pay, clinical study, other, unknown). We grouped HOPS clinic sites at which patients received care as either publicly or privately funded institutions. We stratified year of HIV diagnosis as 2000-2003, 2004-2007, 2008-2010, and 2011-2013. We used the laboratory values obtained closest to and within 3 months after HIV diagnosis to classify baseline CD4 cell count and HIV VL results. Similarly, for CD4 cell count and HIV VL at cART initiation, we analyzed the value obtained closest in time up to 6 months before the start of cART.

Statistical Analyses
We calculated percentages for categorical variables and medians and interquartile ranges for continuous variables. Subgroups of patients were compared using x 2 tests and Wilcoxon rank sum tests as appropriate; ordinal categorical variables were analyzed using the Cochran-Armitage test for trend.
We used Kaplan-Meier time-to-event methods and Cox proportional hazards models to assess temporal trends and to identify correlates of initiating cART and of achieving viral suppression after HIV diagnosis. Observation time for the primary analyses of the timing and rates of cART initiation and VS began on the date of HIV diagnosis (termed "index" date). In the secondary analyses of rates of VS, we reset the time of origin to the date of cART initiation and adjusted for laboratory values (CD4 cell count and HIV VL) at the time of cART initiation. We used univariate Cox models to examine the statistical association of these variables with age, race/ ethnicity, IDU, HIV risk group, primary insurance payor, type of institution, period of HIV diagnosis, and baseline CD4 cell count. We constructed multivariable Cox models that included all variables from the univariate analyses, regardless of their statistical significance, to facilitate comparison of model findings for cART initiation and VS and to control for residual confounding. Results from the Cox proportional hazards models are reported as hazard ratios (HRs) with associated 95% confidence intervals (CI). P values less than 0.05 were considered statistically significant. All analyses were performed using SAS version 9.3 (SAS Institute, Cary, NC).

RESULTS
The 1156 HOPS patients who met eligibility criteria had a median age of 37.0 years (interquartile range: 29.1-45.4), 74.8% were male, and the study sample was ethnically diverse: 39.0% were non-Hispanic/Latino white, 43.2% NHB, 14.1% Hispanic/Latino, and race/ethnicity was other/unknown for 3.7%; 52.4% of patients were MSM and 51.7% were privately insured (Table 1).

CD4 Cell Count at HIV Diagnosis
The median CD4 cell count at HIV diagnosis was somewhat higher in consecutive periods in the study: 284 cells per cubic millimeter in 2000-2003 compared with 337 cells per cubic millimeter in 2011-2013 (test for trend P = 0.064, not significant). The median CD4 cell count at HIV diagnosis varied markedly among patient subgroups: 177 cells per cubic millimeter for heterosexual men, 371 cells per cubic millimeter for MSM, and 314 cells per cubic millimeter for heterosexual women (P , 0.01). NHBs and Hispanics/Latinos were diagnosed with lower CD4 cell counts than whites: median

CD4 Cell Count at cART Initiation
The median CD4 cell count at cART initiation was higher for patients diagnosed in consecutive periods of the study: . MSM and heterosexual women had higher CD4 cell counts at cART initiation than did heterosexual men (P , 0.001; data not shown); additionally, patients of non-Hispanic/Latino white race/ethnicity had higher CD4 cell counts than those of non-Hispanic black and Hispanic/Latino race/ethnicity (P , 0.001; data not shown). There were no statistically significant differences by IDU group for CD4 cell count at cART initiation.

Time to cART Initiation After HIV Diagnosis
Overall, 926 (80.1%) of eligible patients began cART during the study period. Compared with patients who did not, patients who started cART were older (median age 37.7 vs. 34.8 years), more frequently of non-Hispanic/Latino white race/ethnicity (41.0% vs. 30.9%), and less frequently IDU participants (3.8% vs. 8.7%). They were also more likely to have been diagnosed with AIDS at the time of their HIV diagnosis (20.8% vs. 6.1%) and to have a lower CD4 cell count (median 271 vs. 517 cells per cubic millimeter) and higher HIV VL (median 4.8 vs. 4.2 log 10 copies per milliliter) ( Table 1; P , 0.05 for all). Patients who did vs. did not start cART during the study period did not differ significantly in terms of sex, HIV transmission risk group, insurance payor, institution, or calendar period of HIV diagnosis.
In univariate analyses, the estimated median duration from HIV diagnosis to cART initiation was progressively shorter for persons diagnosed with HIV in later calendar periods: the median time was 4.4 months for persons diagnosed in 2000-2003 and 2.5 months for persons diagnosed in 2011-2013 (log rank test P , 0.001, Fig. 1A). In univariate Cox proportional hazards models for 2000-2013, factors significantly (P , 0.05) associated with later cART initiation included age ,25 years (compared with $40 years) and IDU (Table 2). However, factors significantly associated with earlier cART initiation included being a heterosexual man or belonging to other/unknown risk group (compared with MSM), being diagnosed with HIV infection in later calendar periods, and having a lower CD4 cell count at diagnosis ( Table 2). After adjusting for CD4 cell count and all other variables displayed in Table 2, later cART initiation was now significant for persons of NHB race/ethnicity [adjusted HR (aHR) = 0.8, 95% CI: 0.7 to 0.9, as compared with non-Hispanic/Latino white; Fig. 2A Fig. 1B). In univariate Cox proportional hazards models for 2000-2013, factors significantly (P , 0.05) associated with later VS were age ,25 and 25-29 years (compared with $40 years), NHB race/ethnicity (compared with Hispanic/Latino white), and receiving care at a publicly funded institution (compared with private institutions), whereas having an HIV diagnosis in later calendar years and having lower CD4 cell count at diagnosis were associated with shorter time to VS. In

Achieving VS After cART Initiation
We performed a secondary analysis by resetting the time of origin in Cox regression analyses to the date of cART initiation (instead of the date of HIV diagnosis) to assess disparities in VS among those patients who were prescribed cART. Of the 1156 patients included in our main analysis,  Table 2 regarding time to achieve VS by age, race/ethnicity, institution, and period of HIV diagnosis (Figs. 1B, 2B), we observed additional disparities by HIV risk group and insurance payor, whereas age was no longer significant (Table 3). In multivariable analyses, participants aged ,25 years were less likely to achieve VS (aHR = 0.7, 95% CI: 0.6 to 0.9) and those aged 25-29 years (aHR = 0.8, 95% CI: 0.6 to 1.0) when compared with those aged $40 years and NHBs were less likely to achieve VS than non-Hispanic/Latino whites (aHR = 0.8, 95% CI: 0.7 to 1.0). The likelihood of achieving earlier VS 3). Patients with HIV VL $5 log 10 copies per milliliter took longer to achieve VS after cART initiation (aHR = 0.7, 95% CI: 0.6 to 0.8), compared with those having HIV VL from 3 to 5 log 10 copies per milliliter. Neither IDU nor CD4 cell count at cART initiation was independently associated with time to VS after cART initiation.

DISCUSSION
In this longitudinal observational cohort of HIVinfected US patients, time to initiation of cART and time to VS have improved markedly during 2000-2013 for patients who entered HIV care within 6 months after their diagnosis. Our results are similar to earlier findings from a large North American AIDS Cohort Collaboration on Research and Design 6 and somewhat exceed the estimates from the National HIV Surveillance System, which reported that among persons diagnosed with HIV infection in 2009 who entered care within the next 3 months, the median time from HIV diagnosis to viral suppression was approximately 11 months. 4 In the present report and in an earlier HOPS analysis through year 2009, 16 there were no statistically significant improvements over the years in CD4 cell counts at the time of HIV diagnosis. However, in the current analysis, we found a significant increase in CD4 counts at the time of cART initiation. From the vantage point of our analysis cohort, which includes only persons who successfully linked to HIV care at HOPS clinics, once patients were diagnosed and in HIV care, they initiated therapy increasingly more promptly, a finding consistent with changes in treatment guidelines recommending that treatment be offered to all patients regardless of CD4 cell count. 2 As a result of earlier cART initiation and likely improvements in potency and tolerability of cART over time, we observed more prompt VS after HIV diagnosis across the study intervals.
Despite overall improvements, we found that some patient subgroups lagged behind others in initiating cART and achieving VS, as has been shown in other populations. 5,7,12 Most notably, NHBs experienced delays in initiating cART as compared with non-Hispanic/Latino whites even after controlling for public insurance payor (a surrogate indicator of poverty) and being enrolled at publicly funded sites; both variables have been shown to be associated with poorer outcomes in the HOPS cohort. 15,16 The association of NHB race with later initiation of cART may stem from residual confounding by poverty, access to HIV care, or other structural or psychosocial factors that were not captured by our medical abstraction study. 6 Other studies addressing racial disparities in cART initiation or discontinuation identified stigma, fear of disclosure, 17 distrust of the medical establishment or providers, 18-23 low literacy, 24 poor access to case management, 25 and racial/ethnic discrimination 26,27 as contributing factors. Other factors associated with poverty, including living in high-crime neighborhoods and substance abuse, may also contribute to delayed access to effective care. 28 The health impact of incarceration and subsequent barriers to reentry into society are not only difficult to measure but also likely affect access to HIV care. 29 Heterosexual women were diagnosed with HIV at lower CD4 cell counts than MSM but did not experience delays in initiating cART and did not lag behind MSM in time to VS, in contrast to findings by others. 1,21 Of note, HIV among heterosexual men had significantly lower median CD4 counts at diagnosis compared with other risk groups, but once diagnosed, they initiated cART and achieved VS no later than MSM (Table 2). Later diagnosis among heterosexual men vs. MSM echoes findings from some European countries, which differ demographically from the United States. 30,31 A recent study from Florida also observed that heterosexual men were being diagnosed late (ie, at lower CD4 cell counts) in rural compared with urban settings. 32 Unlike the Florida study, the HOPS cohort is based in urban US settings. Heterosexual men may not perceive themselves at risk for HIV infection and therefore may not seek HIV testing, but just as heterosexual women are at risk, their heterosexual male counterparts are no less so. Perhaps, the most important finding from the current analysis was the age-related disparities, with patients ,25 years at HIV diagnosis or cART initiation and marked delays in achieving VS compared with patients $40 years, even after adjusting for CD4 cell counts and calendar year of HIV diagnosis. The incidence rate of new HIV infections has been increasing most rapidly among young adults, which is also the group with the greatest delays in accessing HIV care and poorest adherence to ART and VS. [33][34][35][36] Among the estimated 47,500 new HIV infections in the United States in 2010, about 29,800 (or 63%) occurred among MSM, of whom about 36% were NHB and 30% were 13-24 years old. 33 Nonadherence to treatment among youth is not unique to HIV and has been reported with other chronic illnesses. 37 Youth with HIV have poorer access to care, adherence to medication, and retention in care than adults (reviewed in Zanoni and Mayer 38 ).
Our study has some limitations. We studied patients who were linked to HIV care and enrolled in the HOPS study within 6 months of their HIV diagnosis; thus, our findings may reflect more favorable patterns in cART initiation and viral suppression than would be observed for all HIV-infected patients in the United States, an estimated 23% of whom do not promptly enter HIV care after being diagnosed. 39 We relied on routinely abstracted medical records data and did not capture information on some potential structural and psychosocial confounders, as noted above. Measurements of HIV viral load were conducted as part of routine HIV care, and because the schedule of viral load monitoring may vary by physician practice and individual patient's adherence to clinical visits, 40 we may have overestimated the median times to VS. The delay to VS after controlling for time to cART start could well reflect poor adherence to prescribed cART, which we did not have data to evaluate in this study. Failure to suppress HIV VL highly correlates with nonadherence. 41 Finally, we did not have available data on some patient-level risk factors for delayed cART start or adherence to care and cART, such as depression or alcohol dependence, or structural and socioeconomic factors as income, housing instability, or stigma and discrimination that could interfere with achieving optimal outcomes. 17,19,42 In conclusion, we found that over a 13.5-year period (2000-2013), although the CD4 count at HIV diagnosis has not improved significantly among enrollees in the HOPS, the timeliness of initiation of cART and subsequent VS have both improved. However, heterosexual men were diagnosed at significantly lower CD4 counts than all other subgroups. NHBs initiated treatment and achieved VS significantly later than all other subgroups. Young people aged ,25 years also experienced significantly later VS compared with persons aged $40 years. These observations highlight the disparities that persist within the US continuum of HIV care that must be adequately addressed as part of the National HIV/AIDS Strategy. 1,8