Share this article on:

Mortality Among Antiretroviral-Eligible Patients in an Urban Public Clinic

Dowdy, David W MD, PhD*; Geng, Elvin H MD, PhD; Christopoulos, Katerina A MD; Kahn, James S MD; Hare, C Bradley MD; Wlodarczyk, Daniel MD; Havlir, Diane V MD

JAIDS Journal of Acquired Immune Deficiency Syndromes: August 1st, 2011 - Volume 57 - Issue 4 - p 297-300
doi: 10.1097/QAI.0b013e31822233aa
Brief Report: Clinical Science

Background: Advances in antiretroviral therapy (ART) over the last decade have improved clinical outcomes for people living with human immunodeficiency virus (HIV), but whether these improvements are experienced by disadvantaged urban populations is less clear.

Methods: We evaluated mortality among a clinical cohort in a public safety-net HIV specialty clinic in San Francisco, California.

Results: Among 1651 ART-eligible patients attending an urban US HIV clinic, 4-year mortality was 10.0% in 2000-2004 and 11.0% in 2005-2009. Despite universal ART availability, only 72 (42%) of 172 patients who died, compared with 69% of survivors, ever achieved an HIV viral load, 400 copies per cubic millimeter. The leading causes of death were acquired immunodeficiency syndrome (56%), violence/overdose (16%), and pulmonary disease (6%).

Conclusions: Disadvantaged subpopulations in the developed world can experience high mortality rates despite accessing specialty HIV clinical services with full ART availability. New strategies are needed to improve the outcomes in these populations.

From the *Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD; and †San Francisco General Hospital HIV/AIDS Division, University of California, San Francisco, CA.

Received for publication January 21, 2011; accepted April 29, 2011.

Supported in part by the National Institutes of Health (grant numbers A067039-01A1, MH64384 to J.S.K.).

Presented as an oral abstract at AIDS 2010, Vienna, Austria, July 2010.

The authors have no conflicts of interest to disclose.

Correspondence to: David Dowdy, MD, PhD, Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD (e-mail:

Back to Top | Article Outline


Over the last decade, new classes of antiretroviral medications and more potent, tolerable drugs have made suppression of human immunodeficiency virus (HIV) attainable in even extensively treatment-experienced patients.1 Improved clinical outcomes for HIV-infected patients are in large part attributed to these advances in antiretroviral therapy (ART). Among participants across 19 cohort studies of patients initiating ART in Europe and North America, adult life expectancy increased by 8 years between 2000 and 2005,2 and less than half of all deaths are now related to acquired immunodeficiency syndrome (AIDS)-defining diagnoses.3 Whether the availability of these options has translated into improved clinical effectiveness among all patient populations—including inner-city populations with a high burden of mental illness, substance abuse, and homelessness—is less clear.4,5 Indeed, the studies that examine the actual practice of evidence-based interventions in real-world clinical care (ie, effectiveness) often reveal heterogeneity in outcomes. We therefore evaluated the effect of recent advances in HIV treatment by examining the mortality trends between 2000 and 2009 in a clinic-based cohort of urban uninsured and publicly insured HIV-infected patients who were successfully linked to outpatient HIV care and eligible for ART.

Back to Top | Article Outline


Study Population

We evaluated a clinical cohort in a public safety-net HIV specialty clinic in San Francisco, CA. All the patients were either publicly insured or uninsured, but ART was universally available, regardless of their ability to pay. We included consecutive adult (≥18 years) HIV-infected patients enrolled in a larger multicenter cohort6 who had a CD4+ T-cell count nadir of ≤350 cells per cubic millimeter and at least 2 subsequent primary-care visits within any 12-month period from April 1, 2000, to August 31, 2009. Thus, all the patients were enrolled at a time when they were eligible for ART by contemporary criteria. Both ART-naive patients and those with prior ART experience (eg, transfers from another clinic) were included. The patients were observed until death or until the end of the study (August 31, 2009). The study was approved by the Institutional Review Board of the University of California, San Francisco.

Back to Top | Article Outline

Measurement of Study Variables

Baseline characteristics were abstracted from the clinic's electronic medical record at the time of entry into the cohort (usually the patient's second primary-care clinic visit). We measured the following variables at baseline: age, gender, race, ethnicity, HIV risk factor(s), CD4+ T-cell count, HIV viral load, prior exposure to ART, presence of a mental health diagnosis, alcohol abuse, and hepatitis C status. HIV risk factors were abstracted from historical data recorded during the patients' initial clinical visits. Diagnoses were recorded as International Classification of Diseases, 9th edition (ICD-9) codes in medical charts. The dates and values of all CD4+ T-cell counts and viral load measurements obtained after study entry were also abstracted from the medical record. We ascertained mortality by linkage of electronic medical records to the US Social Security Death Index.

For patients who died, we ascertained the cause of death from chart review using the Cause of Death project protocol ( “Likely AIDS-related” deaths were those due to AIDS-defining illness or nonsudden deaths of unknown cause in patients with CD4+ T-cell counts <50 cells per cubic millimeter. “Possibly AIDS-related” deaths included sudden deaths of unknown cause in patients with CD4+ T-cell counts <50 cells per cubic millimeter and nonsudden deaths of unknown cause in patients with CD4+ T-cell counts between 50 and 199 cells per cubic millimeter. Causes of death were ascertained by a single physician reviewer (D.W.D.) and verified against the data from the National Death Index. Disagreements between initial chart review and the National Death Index were resolved by consensus with a second physician reviewer (K.A.C.).

Back to Top | Article Outline

Statistical Analyses

Our primary analysis was the comparison of mortality rates in 2000-2004 vs. 2005-2009. January 1, 2005, was chosen a priori as the midpoint of the analytic frame. We assessed crude mortality rates by the Kaplan-Meier method and performed multivariable Cox proportional hazards regression to evaluate the differences in time to death across the 2 time periods. We evaluated the proportional hazards assumption using visual inspection of −log[−log(survival)] plots and scaled Schoenfeld residuals. All the baseline variables were included in multivariable models as categorical variables, with the exceptions of age (continuous), CD4+ T-cell count (nonparametric continuous, plus additional category for <50 cells per cubic millimeter), and viral load (continuous on logarithmic scale). Interaction terms with time (2000-2004 vs. 2005-2009) were tested for all the variables. Prespecified secondary analyses included comparison of mortality in 5 higher-risk subgroups (injecting drug users, non-white race, female or transgender, alcohol abuse, mental health diagnosis), year-by-year analysis of mortality, and assessment of viral suppression (defined as an HIV viral load <400 copies per milliliter) at any time during the study period.

Back to Top | Article Outline


A total of 1651 patients were enrolled, of whom 172 (10.4%) died over a median 3.7 [interquartile range (IQR), 1.7-5.9] years of follow-up. The population consisted of 1432 (87%) men and 779 (47%) whites; only 58 patients (3.5%) were white women. Regarding historical HIV risk factors assessed on the initial visit, 981 (59%) patients reported being men who have sex with men, 408 (25%) injecting drug users, and 419 (25%) at heterosexual HIV risk. At study entry, the median age was 49 (IQR 43-56), the median CD4+ T-cell count was 205 cells per cubic millimeter (IQR 78-289), and the median HIV viral load was 35,000 copies per milliliter (IQR 4700-130,000). ART had been prescribed before study entry for 672 patients (41%), and by the end of the study for 1283 patients (78%). Regarding comorbidities as documented in medical charts, 486 patients (31%) had hepatitis C infection, 664 (40%) a mental illness, and 168 (10%) alcohol abuse. Among patients with a documented mental illness, 348 (52%) had a mood disorder (ICD-9 296 or 311, primarily depression), 151 (23%) had a neurotic disorder (ICD-9 300, primarily anxiety), and 32 (5%) had both.

Mean all-cause mortality was 2.65% annually, with no significant year-over-year trend (P = 0.27). In the primary analysis, we observed no difference in mortality over 2000- 2004, as compared with 2005-2009 [Fig. 1; multivariable hazard ratio (HR) 1.40; 95% confidence interval (CI) 0.82, 2.39]. Baseline variables significantly associated with shorter time to death included younger age (HR 0.72 per 10 years; 95% CI: 0.61, 0.86) and CD4+ T-cell count <50 cells per cubic millimeter (HR 1.79; 95% CI: 1.11, 2.87). Time to death was not associated with prior ART prescription, measured either as a baseline variable at enrollment (adjusted HR 1.24, 95% CI: 0.91, 1.70) or as a time-dependent variable during the study period (HR 0.86, 95% CI: 0.55, 1.36).



Among all 5 preselected higher-risk subgroups (nonmale gender, nonwhite race, mental illness, alcohol abuse, injecting drug use), adjusted mortality was at least 70% higher (ie, multivariable HR ≥1.70) in 2005-2009 than in 2000-2004, although this difference was statistically significant only for injecting drug use (HR 4.15; 95% CI: 1.41, 12.2). Crude 4-year Kaplan-Meier mortality in injecting drug users was 6.0% (95% CI: 2.1%, 16.5%) in 2000-2004 vs. 17.9% (95% CI: 13.7%, 23.3%) in 2005-2009.

The cause of death was likely AIDS related in 96 patients (56%) (Fig. 2). Over half of non-AIDS-related deaths were due to either violence (including suicide) and drug use (27 patients) or smoking-related illness including lung cancer and chronic obstructive pulmonary disease (11 patients). Only 29 deaths (17%) were due to other known causes, primarily liver disease (6%), cardiovascular disease (4%), and nonlung, non-AIDS-related cancer (4%). Causes of death were discordant between initial chart review and National Death Index in 21 patients (12%).



Among the 172 patients who died, only 73 (42%) achieved a viral load <400 copies per milliliter at any point during the study period. Median CD4+ T-cell counts remained statistically unchanged in this group, from 128 cells per cubic millimeter (IQR 33-262) at baseline to 101 cells per cubic millimeter (IQR 29-267) at the last measurement before death (P = 0.24 for change). By contrast, among 1479 survivors, 1027 (69%) achieved an undetectable viral load, and median CD4+ T-cell counts increased from 209 cells per cubic millimeter (IQR 87-290) at baseline to 324 cells per cubic millimeter (IQR 188-488) at the last measurement (P < 0.001).

Back to Top | Article Outline


In this cohort of 1651 patients in an urban safety-net HIV clinic in the United States, mortality was high, largely AIDS related, and unchanged over the past decade. Four-year mortality was >10%, over twice the mortality rate in representative developed-country cohorts of patients initiating ART.2 In contrast to those cohorts' recent mortality reductions of >30%, mortality in this population did not change over the last decade. Mortality was dominated by AIDS-related causes, followed by sequelae of violence and drug use. Despite attending a specialty HIV clinic with universal ART availability at a time when they were ART eligible, the majority of patients who died never suppressed their viral loads. Advances in the ability of ART to suppress HIV RNA replication have not translated into reduced mortality among this urban, socially disadvantaged patient population.

Our finding of stable mortality and inadequate viral suppression, despite improvements in HIV care including availability of more potent, tolerable, and coformulated ART options, demands an explanation. In this urban safety-net population, linkage to specialty HIV care continues to occur late in patients' disease course (as evidenced by low baseline CD4 count), and access to care alone remains insufficient to avert substantial mortality. This study was neither designed nor powered to ascertain the specific causes for this phenomenon. However, the translation of clinical innovations into patient mortality benefit requires more than drug efficacy,7 and populations with social instability, low health literacy, mistrust of the health care system, and mental illness may not benefit from medical advances to the same extent as the general population of people living with HIV/AIDS. Reducing HIV-related mortality in these disadvantaged populations may require more resources to provide structural supports, including stable housing,8 food security,9 services for injecting drug users,10 and appropriate case management.11 Of particular importance is adequate psychiatric care12; 40% of patients had an ICD-9 coded mental illness, and clinical records may greatly underestimate the true burden of both psychiatric disease and substance use. Interventions that enable disadvantaged populations to reap equitable benefits from pharmacological advances must be sought and evaluated, as recommended in the National HIV/AIDS Strategy for the United States.13 Failure to do so may result in persistently high mortality rates in these populations and a lost opportunity to reduce HIV transmission.14

This study has several limitations: First, by using an observational design, we are unable to assess the causal nature of the associations detected. Second, our ability to ascertain causes of death using national mortality indices and chart review from a single electronic medical record is limited although we did review the charts of all the patients who died. Third, as a single-center study, our number of deaths is relatively small (n = 172), although the CIs on our estimates of 4-year mortality exclude equivalence with published rates from a large cohort collaboration.1 Fourth, given limited data on medication use, we were unable to assess time-dependent associations between ART adherence and mortality.

In conclusion, we found stable and high (>10% per 4 years) mortality rates in this population of uninsured or marginally insured patients who accessed specialty HIV care. Causes of death were dominated by AIDS, violence, and drug use. Most patients who died never suppressed their viral loads, despite being ART eligible on entry. Access to advanced HIV care alone is insufficient to universally reduce the mortality among people living with HIV/AIDS; more intensive attention is needed to populations whose mortality remains high despite linkage to care.

Back to Top | Article Outline


We would like to thank the patients and providers of Ward 86 and students from the UCSF School of Nursing who performed detailed analyses of mortality in the clinic population leading up to this analysis.

Back to Top | Article Outline


1. Volberding PA, Deeks SG. Antiretroviral therapy and management of HIV infection. Lancet. 2010;376:49-62.
2. Antiretroviral Therapy Cohort Collaboration. Life expectancy of individuals on combination antiretroviral therapy in high-income countries: a collaborative analysis of 14 cohort studies. Lancet. 2008;372:293-299.
3. Antiretroviral Therapy Cohort Collaboration. Causes of death in HIV-1-infected patients treated with antiretroviral therapy, 1996-2006: collaborative analysis of 13 HIV cohort studies. Clin Infect Dis. 2010;50:1387-1396.
4. Study group on death rates at high CD4 count in antiretroviral naive patients. Death rates in HIV-positive antiretroviral-naive patients with CD4 count greater than 350 cells per microL in Europe and North America: a pooled cohort observational study. Lancet. 2010;376:340-345.
5. Sackoff JE, Hanna DB, Pfeiffer MR, et al. Causes of death among persons with AIDS in the era of highly active antiretroviral therapy: New York city. Ann Intern Med. 2006;145:397-406.
6. Kitahata MM, Rodriguez B, Haubrich R, et al. Cohort profile: the centers for AIDS research network of integrated clinical systems. Int J Epidemiol. 2008;37:948-955.
7. Moss AR, Hahn JA, Perry S, et al. Adherence to highly active antiretroviral therapy in the homeless population in San Francisco: a prospective study. Clin Infect Dis. 2004;39:1190-1198.
8. Kushel MB, Hahn JA, Evans JL, et al. Revolving doors: imprisonment among the homeless and marginally housed population. Am J Public Health. 2005;95:1747-1752.
9. Weiser SD, Fernandes KA, Brandson EK, et al.The association between food insecurity and mortality among HIV-infected individuals on HAART. J Acquir Immune Defic Syndr. 2009;52:342-349.
10. Kimber J, Copeland L, Hickman M, et al. Survival and cessation in injecting drug users: prospective observational study of outcomes and effect of opiate substitution treatment. BMJ. 2010;341:c3172.
11. Kushel MB, Colfax G, Ragland K, et al. Case management is associated with improved antiretroviral adherence and CD4+ cell counts in homeless and marginally housed individuals with HIV infection. Clin Infect Dis. 2006;43:234-242.
12. Altice FL, Kamarulzaman A, Soriano VV, et al. Treatment of medical, psychiatric, and substance-use comorbidities in people infected with HIV who use drugs. Lancet. 2010;376:367-387.
13. Office of National AIDS Policy. National HIV/AIDS strategy for the United States. Washington, DC: The White House, July 13, 2010. Available at: Accessed January 18, 2011.
14. Das M, Chu PL, Santos GM, et al. Decreases in community viral load are accompanied by reductions in new HIV infections in San Francisco. PLoS One. 2010;5:e11068.

mortality; HIV; urban health; delivery of health care

© 2011 Lippincott Williams & Wilkins, Inc.