Abstract
Outpatient care for people living with HIV is delivered in diverse settings. Differences in setting may impact HIV outcomes. We evaluated HIV-infected adults in care at Ryan White-funded clinics in Philadelphia, PA, between 2008 and 2011 to determine how setting of care (hospital versus community-based) influenced HIV outcomes. Clinics were categorized as hospital-based if they were located onsite at a hospital. The composite outcome was completion of the final three steps of the HIV care continuum: (1) retention in care; (2) use of antiretroviral therapy (ART); and (3) viral suppression. Mixed-effects logistic regression, accounting for patient and clinic factors, examined the relationship between care setting and the outcome. In total, 12,637 patients, contributing 32,515 patient-years, received care at 25 clinics (12 hospital-based). Women, non-Hispanic blacks, those with private insurance, and individuals with higher household incomes more commonly attended hospital-based clinics (p<0.05). Of the 12,962 patient-years (40%) during which patients attended community-based clinics, 59% met the outcome. Similarly, 59% of the 19,553 patient-years (60%) in which patients attended hospital-based clinics met the outcome. Adjusting for patient and clinic factors, setting was not associated with the outcome (adjusted odds ratio=1.24, 95% CI=0.84–1.84). In summary, demographics differ among patients visiting hospital and community-based clinics. Completion of the final three steps of the HIV care continuum did not vary between hospital and community-based clinics, which may reflect advances in HIV therapy and the wide availability of HIV care resources.
Introduction
T
Numerous factors contribute to successfully completing continuum steps, including patient characteristics, quality of the patient–provider relationship, and the structure of healthcare delivery. 13 –18 However, most prior studies have focused on patient-level factors, such as sociodemographic, clinical, and behavioral characteristics, with few examining the effects of the healthcare environment on HIV outcomes. 13,19 –22
Outpatient care for PLWH is delivered in diverse settings. While many clinics offer similar medical services, including delivery of HIV-specific care and prescription of ART, they often have unique identities that emerge from different care models, organizational structures, academic affiliations, and funding sources. 23,24 Many HIV-infected persons receive care at community-based HIV clinics that often meet the definition of a community health center (CHC): a private, nonprofit organization that receives public funding and provides comprehensive primary health services to residents of a defined geographic area that is medically underserved. 25 Alternatively, PLWH may engage in care at clinics located within large ambulatory practices, health systems, or academic health centers. The Ryan White Program (RWP) funds outpatient medical care for many PLWH across the US and thus supports a diverse range of HIV care delivery settings and models. 26
Differences in clinic settings may impact health utilization and outcomes. In a nationally representative survey of non-federal, office-based physicians, patients seen at CHCs had more visits to their provider than those seen at private outpatient offices. 27 Similarly, analyses of three national primary care surveys demonstrated lower continuity of care in hospital outpatient departments than either physicians' offices or CHCs. 28 In a retrospective study of 854 HIV-infected patients, Chu et al. compared rates of viral suppression and immunologic success (defined as a 100 cell/mm3 increase in CD4 count) among those who received care at a hospital-based specialty practice to those seen at a community-located primary care clinic. 29 No significant differences in viral suppression and immunologic success were observed. However, this study was limited by its small sample size and use of data from only two clinical sites within the same healthcare network.
To better understand how the setting of care influences completion of the final three steps of the HIV care continuum (retention in care, use of ART, and viral suppression) in hospital versus community-based HIV clinics, we used data from a large, geographically based cohort of HIV clinics. An appreciation of the relative strengths of hospital and community-based HIV clinics is important and will help identify targets for improving HIV outcomes in each setting.
Methods
Study design and participants
We conducted a retrospective cohort study of HIV-infected adults in care at RWP funded clinics in the Philadelphia, PA, eligible metropolitan area (EMA) between 2008 and 2011. Twenty-five HIV clinics treated adults and received RWP funding during the study period, representing approximately 71% of all PLWH in care in the City (unpublished data, City of Philadelphia Department of Public Health). All HIV-infected adults (age ≥18 years) engaged in care, were eligible for inclusion. Patients were engaged in care if they had at least one primary HIV visit and one CD4 test in a calendar year, between January 1, 2008 and December 31, 2011, consistent with criteria in prior studies. 7,30 –32 Thus, each individual could contribute 1, 2, 3, or 4 years of observation to the study analysis.
Data collection
Data were extracted from CAREWare, a Health Resources and Services Administration (HRSA) recommended data management system containing demographic, laboratory, pharmacy, and health service utilization information for all patients seen at RWP funded clinics in the Philadelphia EMA. Clinics abstract patient-level information from medical records of all patients in care, not only those covered under the RWP. After quality control and verification, data are sent to the City of Philadelphia Department of Public Health AIDS Activities Coordinating Office (AACO) and combined across clinics to produce a uniform database. Each patient in the database has a unique identifier, independent of personal information and site of care. Periodic chart reviews and site visits are undertaken to verify the accuracy and completeness of data abstraction and entry.
Clinic-level data annually collected by the City of Philadelphia Department of Public Health AACO, in combination with responses to a structured questionnaire completed by administrators at all adult Philadelphia RWP funded clinics in February 2013, were used to determine the type of care setting and availability of onsite HIV case management and onsite clinical pharmacist(s) during the study period. The study was approved by the Institutional Review Boards of the University of Pennsylvania and the City of Philadelphia Department of Public Health.
Patient sociodemographic and clinical variables
For each calendar year of observation, patients' age as of January 1 was divided into four groups: 18–29, 30–39, 40–49, and 50 years or older. Race/ethnicity was categorized as non-Hispanic white, non-Hispanic black, Hispanic, and other/unknown. Self-reported HIV transmission behavior was grouped into heterosexual, men who had sex with men (MSM), injection drug use (IDU), and other/unknown. Patients who had IDU in combination with another risk factor (e.g., MSM, heterosexual transmission) were classified as IDU. Insurance coverage in each year was categorized as private, Medicaid, Medicare (including those with dual eligibility), uninsured, or other/unknown. Patients whose care was funded through RWP were considered to be uninsured. Annual household income was divided into <$10,000 and ≥$10,000 according to CDC classifications. 33 Median CD4 cell count, based on all collected values in each year, was grouped as ≤350, 351–500, or >500 cells/mm3. 34 These categories were selected based on differential indications for initiating ART, and to account for changes in HIV treatment guidelines over the 4-year study period. 34 –36
Clinic variables
Clinics were categorized as hospital-based if they were located onsite at a hospital or hospital campus and designated as offering case management and access to a clinical pharmacist if these services were located onsite.
Outcome variable
The outcome of interest was completion of all of the final three steps of the HIV care continuum: retention in care, use of ART, and viral suppression. Retention in care was based on the US National HIV/AIDS Strategy metric, which defines retention as having two or more outpatient visits separated by ≥90 days during a calendar year. 37 Patients were designated as using ART if they received at least three antiretroviral drugs (excluding ritonavir), a definition consistent with prior literature, at the last outpatient visit in the calendar year. 7,38,39 Fixed-dose combination pills were counted as the number of individual drug components. Outpatient visits refer only to primary HIV care appointments and do not include nursing, pharmacy, laboratory, or other types of visits. HIV viral suppression was classified as last annual HIV-1 RNA ≤200 copies/mL. Patients had to meet each of these steps to be assigned as completing the care continuum.
Statistical analyses
The patient-year was the unit of analysis, reflecting the common practice of measuring retention in care, use of ART, and viral suppression on a calendar year basis. Each patient could contribute one observation per calendar year. Analyses were limited to patient-years in which the patient was at least 18 years and in care, defined by having at least one primary HIV visit and one CD4 test in the year. Thus, the number of patient-years was not constant across patients or years.
We excluded the first calendar year in care for 5400 new patients and the calendar year of death for 430 patients, as they did not provide adequate time to measure the outcome. After excluding these patient-years, 24% of the sample contributed 1 year of data; 22% contributed 2 years; 24% contributed 3 years; and 30% contributed 4 years.
Statistical comparisons of demographic and clinical characteristics of the sample across calendar years were conducted using the X 2 test for trend. Multivariate logistic regression examined demographic (age, sex, race/ethnicity, HIV transmission behavior, insurance coverage, annual household income) and clinical factors (median CD4 cell count) associated with care setting (attending a hospital-based vs. community-based clinic), adjusting for calendar year.
To assess the relationship between care setting and the outcome of interest (completion of the final three steps of the HIV care continuum), mixed-effects logistic regression was used. 40 The model included random effects for clinic and patient to account for within patient and within clinic correlation. Fixed effects included care setting and potential confounders: age, sex, race/ethnicity, HIV transmission behavior, insurance coverage, annual household income, median CD4 cell count, calendar year, and availability of onsite case management and clinical pharmacy services.
We identified 429 patient-years, 1.3% of the sample, in which patients had no HIV-1 RNA test reported; these observations were excluded from the primary analysis. A sensitivity analysis was conducted assuming all patient-years with missing HIV-1 RNA were not suppressed (HIV-1 RNA >200 copies/mL). Two-sided testing was used, with a p value of<0.05 considered significant. Analyses were conducted using STATA 12.1 (College Station, TX).
Results
Between 2008 and 2011, 12,637 patients, contributing a total 32,515 patient-years, were in care at Philadelphia EMA RWP clinics (Table 1). The number of patients in care increased from 6406 patients in 2008 to 9388 in 2011. In each year, the majority of patients was male, of minority race/ethnicity, and had heterosexual contact as their HIV mode of transmission. Most patients had Medicaid insurance and yearly income less than $10,000. Across all 4 years, the proportion of patients retained in care and on ART increased from 79% to 85% (p<0.01) and 83% to 88% (p<0.01), respectively. The percent of patients with median CD4 count above 500 cell/mm3 increased from 43% to 51% (p<0.01), while the proportion virally suppressed increased from 64% to 77% (p<0.01).
ART, antiretroviral therapy; HET, heterosexual transmission; HIV, human immunodeficiency virus; IDU, injection drug use; MSM, men who have sex with men.
In total, patients visited hospital-based clinics in 60% of patient-years (Table 2). In multivariate analyses, women (adjusted odds ratio 1.45, 95% confidence interval 1.34–1.57) and non-Hispanic blacks (1.07, 1.01–1.13) were significantly more likely to visit a hospital-based clinic than men and non-Hispanic whites, respectively. Likewise, individuals with a higher yearly income more commonly attended hospital-based clinics (1.13, 1.11–1.15). Patients with Medicaid (0.95, 0.93–0.97), Medicare (0.96, 0.93–0.98), or who were uninsured (0.82, 0.80–0.84) were less likely to attend hospital-based clinics than those with private insurances.
AOR, adjusted odds ratio; CI, Confidence Interval; HET, heterosexual transmission; HIV, human immunodeficiency virus; IDU, injection drug use; MSM, men who have sex with men; PY, patient-year.
Of the 25 clinics, 12 (48%) were hospital-based and 13 (52%) were community-based. Twelve clinics (48%) offered onsite clinical pharmacists (eight hospital-based and four community-based) and the vast majority (92%) had case management services (twelve hospital-based and eleven community-based).
The continuum of care for HIV-infected patients engaged in care at community and hospital-based clinics is presented in Fig. 1. Among the 12,962 patient-years in which patients attended community-based clinics, 85% had patients who were retained in care, 74% had patients who were retained in care and on ART, and 59% had patients who were retained in care, on ART, and virally suppressed. In comparison, among the 19,553 patient-years in which patients attended hospital-based clinics, 82% had patients who were retained in care, 72% had patients who were retained in care and on ART, and 59% had patients who were retained in care, on ART, and virally suppressed.
The care continuum for HIV-infected patients in community-based and hospital-based clinics. ART, antiretroviral therapy. *Among patients retained in care; **Among patients retained in care and on ART.
Table 3 presents the multilevel mixed-effects regression model characterizing the association between care setting and completion of the final three steps of the HIV care continuum (i.e., attainment of viral suppression among people retained in care and on ART). Adjusting for clinic and patient factors, patients seen at hospital-based clinics were equally likely to complete the final three steps of the continuum compared to those seen at community-based clinics (1.24, 0.84–1.84). Patients seen at clinics with onsite pharmacists were significantly more likely to meet the outcome compared to those without onsite pharmacists (1.52, 1.01–2.30). Availability of onsite case management was not associated with the outcome (1.33, 0.62–2.89).
AOR, adjusted odds ratio; CI, Confidence Interval; HET, heterosexual transmission; HIV, human immunodeficiency virus; IDU, injection drug use; MSM, men who have sex with men.
Older age, higher household income, higher median CD4 cell count, and receiving care in later years were all significantly (p<0.05) associated with completion of the final three steps of the HIV care continuum. Females, non-Hispanic blacks (vs. non-Hispanic whites), individuals with IDU transmission risk (vs. heterosexual risk), and those covered by Medicare or Medicaid (vs. private insurance) were less likely (p<0.05) to meet the outcome (Table 3). Sensitivity analyses categorizing those with missing viral load data as “not suppressed” yielded similar results (results not shown).
Discussion
These results, from a large, geographically based cohort of HIV clinics, confirm those of an earlier study, which found no differences in the rates of virologic suppression and immunologic recovery between patients receiving care at a community-based primary care clinic and those in care at a hospital-based specialty care clinic in Bronx, NY. 29 Our findings are encouraging, suggesting that PLWH can achieve similar outcomes independent of clinic setting.
A number of factors may explain our findings, including advances in HIV therapy and the wide availability of HIV care resources. Improvements in the treatment of HIV have led to the advent of effective, better tolerated, and more convenient therapy. 41,42 As a result, adherence to ART has improved, the proportion of patients with HIV viral suppression has increased, and HIV-related complications have decreased. 43 –45 Similarly, data indicate that compliance with HIV treatment guidelines is associated with viral suppression and improved survival. 46 The number of HIV education and care resources has increased in recent years and are widely accessible to clinics. One example is the HRSA-sponsored in+care campaign, a national quality improvement effort aimed at optimizing retention in care. 47,48 This initiative offers clinics the opportunity to participate in educational webinars, receive individualized coaching, and interact with national HIV experts. Similarly, many online training centers can assist clinics in designing and implementing interventions to improve outcomes along the HIV care continuum. 49,50 Among clinics in this study, nine participated in the in+care campaign, suggesting that these resources are frequently but not universally utilized.
Additionally, the comparable outcomes we observed across hospital and community-based clinics may be influenced by the fact that all clinics in our study received RWP funding and interact closely with Department of Public Health. Consequently, clinics are encouraged to track HIV care and outcomes using standardized HRSA HIV/AIDS Bureau performance measures, including retention in care, use of ART, and viral suppression. 51 This uniform focus on quality improvement may lead to improved HIV outcomes regardless of clinic setting. 52
Consistent with prior research, the presence of clinical pharmacists was strongly associated with the outcome. 53 –57 One study of 75 patients on ART found improvements in ART adherence, CD4 count and rate of viral suppression following interventions by a clinical pharmacist. 53 Another retrospective analysis examining 14,128 patients on ART showed that patients who used HIV-specialized pharmacies were more likely to be adherent to ART than those who visited traditional pharmacies. 55 Altogether, pharmacists at various levels of clinical care may improve patient outcomes.
Our study demonstrated significant patient demographic variations by clinic setting, with individuals with low income and public or no health insurance more commonly attending community-based clinics. Most of the community-based clinics in our analysis function as CHCs. Accordingly, the sociodemographics of patients attending community-based clinics are consistent with those seen at CHCs and with their mission. 25,27,28,58,59 Members of these traditionally underserved populations have also been shown to have worse HIV outcomes than their counterparts. 60,61 In our study, community-based clinics, which were more likely to serve these patient groups, had similar outcomes to hospital-based clinics, demonstrating the utility of the CHC model.
Drivers impacting the selection of one clinic setting over another vary. Patients may select a clinic due to actual or perceived access to specialty services, including mental health and substance abuse treatment. Some CHCs have struggled with providing specialty services to their patients. 62,63 Although prior studies have reported that 90% of CHCs have onsite behavioral health services, it is unclear whether this includes comprehensive psychiatric and substance abuse care. 63 Clinics that integrate mental health services with HIV care have been reported to have improved ART adherence, decreased high-risk sexual behavior and severity of addiction and increased patient satisfaction. 64 –66 A recent qualitative study of 212 patients (60% of whom were HIV infected) participating in an integrated primary care and substance abuse program found that patients appreciated the consolidation of services and perceived improvements in their quality of life. 66
The current analysis has several limitations. First, this study involved patients receiving care at primarily urban RWP funded clinics in the Philadelphia EMA. While some non-urban clinics were included, the generalizability of results to rural or suburban settings may be limited. Second, data on many potentially relevant services (e.g., care outreach initiatives, co-location of mental health and substance abuse treatment) were not collected. Additionally, we assessed use of ART at the last visit of the year; this definition may underestimate individuals on ART whose regimens were on hold at that time. Third, our sample size precluded an accurate assessment of how academic affiliation or teaching status may have affected the outcome. Fourth, almost all the clinics in the study offered onsite case management; this lack of variability may explain why case management was not associated with the outcome. Lastly, our study was not designed to collect qualitative data.
Improving health outcomes for PLWH requires a multifaceted approach. 67,68 Interventions should be targeted not only to patients, but also to clinics and health systems that deliver HIV care. Our study contributes new knowledge on how the setting of care delivery affects clinical outcomes, noting that PLWH seen at either hospital or community-based clinics had a similar likelihood of completing the final three steps of the HIV care continuum. Further studies, including qualitative research, are needed to identify how changes in the health care environment can improve HIV outcomes.
Footnotes
Acknowledgments
We are grateful to all the patients, physicians, investigators, and staff involved in the Philadelphia Ryan White System. We would like to acknowledge the staff of the Philadelphia Department of Health AIDS Activities Coordinating Office including Jane Baker, Coleman Terrell, Michael Eberhart, Marlene Matosky, and Ethan Schofer.
Funding: This study was supported by the National Institutes of Health (K23-MH097647-01A to BRY).
Disclaimer: The views expressed in this article are those of the authors. No official endorsement by the Philadelphia Department of Public Health or the National Institutes of Health is intended or should be inferred.
Author Disclosure Statement
BRY received investigator-initiated research support (to the university of Pennsylvania) and consulting fees from Gilead Sciences.