AIDS RESEARCH AND HUMAN RETROVIRUSES

Volume 26, Number 2, 2010

ª Mary Ann Liebert, Inc.
DOI: 10.1089=aid.2009.0001

Predictors of Success with Highly Active Antiretroviral

Therapy in an Antiretroviral-Naive Urban Population

Elisa Zaragoza-Macias,1 Dominique Cosco,1 Minh Ly Nguyen,1,2 Carlos del Rio,1,2 and Jeffrey Lennox1,2

Abstract

Predictors of successful virologic, immunologic, and clinical response with combined antiretroviral therapy
(cART) containing a boosted protease inhibitor or a nonnucleoside reverse transcriptase inhibitor were analyzed
among an antiretroviral naive (ARV-naive) urban cohort. Measures of success included virologic suppression
[HIV-1 viral load (VL) <400 copies=ml], an increase in CD4þ T cells from baseline of >100 cells=ml, and lack of
development of an AIDS-defining illness at 24 and 48 weeks after cART initiation. Two hundred and eighty-
seven ARV-naive patients were included in this cohort, of which 76.7% were male and 86.8% were nonwhite. At
the time of cART initiation their median age was 39 years, the geometric mean CD4þ count was 42 cells=ml, and
the mean viral load was 5.3 log10 copies=ml. At 48 weeks, 72% of patients achieved virologic suppression, with

90% adherence and high school graduation predicting viral undetectability at 48 weeks. Baseline VL 100,000
copies=ml and a CD4þ cell count >100 cells=ml were associated with viral suppression at 24 weeks [OR (95%
CI) ¼ 3.55 (1.29–9.81) and 3.96 (1.19–13.15), respectively]; female gender was associated with a greater increase in
CD4þ cell counts [OR (95% CI) ¼ 7.41 (2.48–22.1)]. CDC stage A1–C2 at baseline predicted lack of clinical
progression at 48 weeks. The results of this analysis of an ARV-naive cohort comprised predominantly of
indigent, minority patients suggest that men who did not have a high school education and who had advanced
HIV infection are less likely to have therapeutic success after cART initiation.

Introduction the first time (antiretroviral naive) between 2003 and 2005
at an urban clinic located in Atlanta, Georgia. Our objec-

W ith the widespread use of highly combined anti-

retroviral therapy (cART) since 1996, AIDS-related
death has decreased dramatically in the United States.1,2 To
tive was to evaluate individual patient characteristics and
their association with successful response to therapy, de-
fined as having an HIV-1 RNA viral load less than 400
optimize therapeutic success, and minimize morbidity, timely copies=ml at 24 and 48 weeks. In addition, we also ex-
identification of patients who are at risk of failing cART is plored the effect of specific antiretroviral combinations on
essential. Previous literature describing predictors of success viral load suppression, CD4þ cell counts change, and lack
to highly active antiretroviral therapy (HAART) have the of AIDS clinical progression.
limitation of including patients who were taking regimens not
recommended by current antiretroviral guidelines [non-
Materials and Methods
boosted protease inhibitors (PI)].3–12 Furthermore, the ma-
jority of patients in these studies are males of European This study was conducted at the Grady Infectious Disease
descent, which does not represent current trends in the U.S. Clinic, a Ryan White funded clinic that provides compre-
AIDS epidemic. The U.S. HIV-1 epidemic has became more hensive HIV care to over 4000 HIV-1-infected patients annu-
concentrated in the southern states, and has an increasingly ally. Among all patients seen in the clinic 72% of the patients
large number of female and minority patients who present are African American and 71% are below 300% of the federal
with advanced immune suppression.13,14 poverty level. Additionally, the majority of patients seen in
We conducted a retrospective cohort analysis of patients this clinic are severely immunosuppressed (CD4þ cell counts
with HIV infection who started antiretroviral therapy for of less than 200 cells=ml).

1
Department of Medicine and 2Division of Infectious Diseases, Department of Medicine, Emory University School of Medicine, Atlanta,
Georgia 30303.

133
134 ZARAGOZA-MACIAS ET AL.

Patients patient characteristics were compared between those with

successful response and those who did not achieve a suc-
We used pharmacy records to identify potential treatment-
cessful response. Subsequently, logistic regression analysis of
naive patients who had started either a boosted-protease
outcome predictors was conducted. The following variables
inhibitor (PI=r) or a nonnucleoside reverse transcriptase in-
were included a priori in all models: type of cART (PI=r vs.
hibitor (NNRTI)-based cART regimen between January 2,
NNRTI), gender (female vs. male), age (50 vs. >50 years),
2003 and December 31, 2005. During this time period our
ethnicity (whites vs. African Americans, black Africans, and
clinic pharmacy served as the Georgia State pharmacy for the
Latinos), baseline HIV-1 viral load (5 vs. >5 log10 cop-
AIDS Drug Assistance Program (ADAP) and for patients re-
ies=ml), baseline CD4þ count (100 vs. >100 cells=ml), and
ceiving medications through Medicaid or the Ryan White
adherence (100–90% vs. <90%). Additionally, variables
grant program. In later years a variety of pharmacies supplied
reaching a p value of <0.10 in univariate analyses were en-
medications for these programs. Between 2003 and 2005 we
tered into multivariate models, and those still statistically
were therefore able to verify exact dates of prescription refill
significant at 0.05 were kept in the final multivariate model.
for all patients served in our program. We confirmed that
All analysis used a two-tailed p value with a significance
patients were cART naive by reviewing the patient’s chart,
level of <0.05. Statistical analyses were performed using SAS
where providers were required to document the date and type
statistical package version 9 (SAS Institute, Cary, NC).
of specific ART that the patient had ever taken. Those patients
This study was approved by the Institutional Review Board
who were 18 years old and older, were not pregnant, had a
(IRB) of Emory University and the Grady Research Oversight
CD4þ count and viral load at baseline, and at least 24  3
Committee. The IRB committee waived the need for informed
weeks follow up were included in this analysis. Collected data
consent.
included demographics, continent of origin (for immigrants),
highest level of education attained, type and date of initial
cART, body mass index (BMI), and active drug use (cocaine, Results
amphetamine, and intravenous drug use) use documented by
From pharmacy records 512 potential patients were iden-
health care provider, HIV risk behaviors [men who have sex
tified, of whom 287 patients (56%) met criteria to be included
with men (MSM), heterosexual, injection drug users (IDUs)],
in this study. A diagram showing the percent and reasons for
opportunistic infections (OI) at baseline and follow up, clinical
exclusion of patients is found in Fig. 1.
stage of HIV disease (as by CDC1993 modified classification),
Baseline characteristics stratified by baseline cART regimen
major psychiatric illnesses (schizophrenia and bipolar disease
are shown in Table 1. The cohort consisted of mostly males
depression documented by health care provider), hepatitis B
(77%) and nonwhites (87%) with a median age of 39 years
surface antigen, hepatitis C antibody, CD4þ cell count, and
(standard deviation 9.7). Fifteen percent had active hepatitis B
HIV plasma viral load at baseline and during follow up.
infection and 11% had positive serology for hepatitis C. At the
Due to the retrospective nature of this study, patient ad-
time of initiation of antiretroviral therapy the geometric mean
herence was assessed through monthly pharmacy anti-
CD4þ count was 42 cells=ml (range 1–354 cells= ml) with 67%
retroviral (cART) pick up. Pharmacy records for the majority
having a CD4þ count <100 cells=ml. The mean log10 HIV-1
of the clinic patients were monitored by a centralized phar-
viral load was 5.3 copies=ml, with 63.5% having an HIV-1
macy located at the Grady Health system. The pharmacy
viral load >100,000 copies=ml. Fifty-six percent of the indi-
provides only 1 month of medication after which the patient
viduals were started on an NNRTI-based regimen and 44%
has to return for a refill. Pharmacy records include data on
with a PI=r-based regimen. Patients who were prescribed a
patients who did not pick up cART for the month or picked up
boosted PI were more likely to have had a baseline CD4þ
medications later than the expected duration of pills dis-
count <100 cells=ml than were those prescribed an NNRTI
pensed for the prior month. A late pick up was considered if
(75% vs. 60%, respectively, p ¼ 0.003). Table 1 presents base-
15 days or more had passed from due date of pick up. Ad-
line characteristics by gender. Women had a higher baseline
herence was calculated at 24 and 48 weeks by dividing the
BMI and were less likely to be prescribed an NNRTI regimen
number of months individuals picked up medications on time
(18% in females vs. 81% in males, p ¼ 0.02).
by the total number of months of follow up then multiplied by
Of the 287 patients included in the study cohort, follow-up
100.15–17
data at 48 weeks were available for 238 (83%). Those with
information at 48 weeks were more likely to have had ad-
Outcomes herence of 90–100% in the first 24 weeks of follow up
( p ¼ 0.04).
The primary end point was virologic treatment success,
At 24 weeks, 73% of the patients achieved an HIV-1 viral
defined as an HIV-1 RNA less than 400 copies=ml at 24 and 48
load of <400 copies=ml and 72% achieved virologic sup-
weeks. Secondary end points included increase of baseline
pression (<400 copies=ml) at 48 week. If all 58 patients who
CD4þ T cells by 100 or more cells=ml at 24 and 48 weeks,
did not return for follow-up visits are assumed to meet the
respectively, and lack of AIDS clinical progression (presen-
criteria for virologic failure, then the proportion of the cohort
tation of AIDS defining illness) during the 48 weeks of follow
who achieved 48 week virologic suppression would fall to
up.
61%. CD4þ cell recovery by more than 100 cells=ml from
baseline was observed in slightly over half (56%) of the pa-
Statistical analysis
tients at 24 weeks and in 72% of patients at 48 weeks. Clinical
Due to their highly skewed distribution, HIV-1 viral load progression to a new or recurrent AIDS-defining opportu-
and CD4þ counts were transformed (log10 and geometric nistic infection developed in 34 (12%) of the study cohort; 29 of
mean, respectively). Using Fisher’s exact test and t test, these events occurred by week 24 of follow up. No patient
PREDICTORS OF SUCCESS IN AN ART-NAIVE URBAN POPULATION 135

Lack of clinical progression was observed among those

with CDC stage A1–C2 disease at initiation of therapy, having
at least a high school education, and a viral load of <100,000
copies=ml

Multivariate analysis
The type of cART regimen, gender, age less than 50 years,
and ethnicity were not found to be significant predictors for
virologic suppression by multivariate analysis at 24 or 48
weeks (Table 2). Significant predictors of virologic suppres-
sion at 24 weeks were a baseline viral load of <100,000
copies=ml [OR (95% CI) ¼ 3.55 (1.29–9.81)] and baseline CD4þ

100 cells=ml count [OR (95% CI) ¼ 3.96 (1.19–13.15)] and at
48 weeks adherence of 90–100% [OR (95% CI) ¼ 2.70 (1.17–
6.24)] and having at least a high school education [OR (95%
CI) ¼ 4.98 (1.11–22.40)].
Female gender and high school education or above were
associated with increased CD4þ cell count by
100 cells=ml
[OR (95% CI) ¼ 7.41 (2.48–22.1) and 4.98 (1.01–24.7), respec-
tively]. Patients of black African ethnicity compared to Afri-
can Americans were less likely to have CD4þ cell recovery at
48 weeks [OR (95% CI) ¼ 0.23 (0.06–0.97)] (Table 3).
For the analysis of clinical progression, only CDC stage A1–
C2 remained significant in the multivariate model [OR (95%
CI) ¼ 7.72 (1.47–40.6)] (Table 4). However, there was a trend
that favored NNRTI-based regimens in decreasing likelihood
of clinical progression [OR (95% CI) ¼ 2.57 (0.76–8.71)]. Pa-
tients having a concordant response (virologic suppression
and an increase in CD4þ count) were 6.88 (95% CI 2.28 to 20.8)
times less likely to have clinical progression than those who
had a discordant response (viral suppression without an in-
crease in CD4þ counts or increase in CD4þ counts without
achieving virologic suppression).

Discussion
This study provides insight into patient characteristics that
help predict successful response to cART when either boosted
protease inhibitor-based or NNRTI-based regimens are used
in urban dwelling patients residing in the southeastern United
States. Overall, patients included in this cohort started cART
FIG. 1. Total patients included in the study and reasons for in the late stages of the disease (51% had stage C3, 64% had an
exclusions.
HIV-1 viral load of >100,000 copies=ml, and 69% had a CD4þ
cell count of <100 cells=ml) and at a median age of 39 years. In
comparison to similar southern U.S. cohorts, individuals in
this study had lower CD4þ cell counts at baseline.11 Re-
who at baseline had achieved a high school or higher level of
markably, even with such advanced disease, up to 72% of
education developed a new opportunistic infection.
patients who remained in care achieved virologic suppression
at 24 and 48 weeks.
Univariate analysis
Cohort studies have shown that viral load and CD4þ
There was an association between viral suppression (<400 counts at baseline are associated with HAART out-
copies=ml) and having a baseline CD4þ cells >100 cells=ml comes.9,11,18–20 In our study, the degree of immunosuppres-
[OR (95% CI) ¼ 3.96 (1.19–13.15)] and a baseline viral load sion as well as the baseline viral load predicted a favorable
<100,000 copies=ml [OR (95% CI) ¼ 3.13 (1.52–6.44)]. Vir- response to HAART at 24 weeks, while at 48 weeks patients’
ologic suppression was also associated with being Latino [OR adherence and a higher education level were the most im-
(95% CI) ¼ 4.79 (1.24–18.55)] or black African [OR (95% portant predictors of response to HAART. This finding em-
CI) ¼ 4.58 (1.28–17.79)] as well as adherence with medication phasizes the dynamic processes influencing response to
pick up of more than 90% [OR (95% CI) ¼ 2.63 (1.80–6.15)]. A HAART and also reflects our clinical practice: in the first
higher percentage of Latinos and patients immigrating from period of HAART initiation, patients are followed very clo-
Africa achieved virologic suppression at 48 weeks compared sely to monitor for side effects, tolerability, and compliance,
to African Americans (85% Latinos, 86% black Africans, 71% while at 48 weeks, the follow up is not as stringent and
African Americans, p ¼ 0.02 for trend). compliance to medication would play a more important role.
136 ZARAGOZA-MACIAS ET AL.

Table 1. Baseline Characteristics of the Study Patients by Type of cART and by Gender

NNRTI PI=r Males Females

Characteristics (n ¼ 161) (n ¼ 126) p (n ¼ 220) (n ¼ 67) p

Male gender (%) 130 (81) 85 (68) 0.02

Female gender (%) 29 (18) 37 (30)
cART ¼ NNRTI 130 (60) 29 (44) 0.02
High school or higher education (%) 113 (70) 91 (72) 0.92 166 (76) 42 (63) 0.14
Mean age (SD) 39 (9.7) 40 (9.6) 0.88 38 (9.5) 41 (9.9) 0.06
Mean BMI (SD) 24 (4.4) 24 (4.1) 0.43 23 (4.1) 25 (5.3) 0.02
Ethnicity 0.18 <0.0001
African American (%) 103 (64) 74 (59) 142 (65) 39 (59)
White (%) 18 (11) 20 (16) 33 (15) 5 (8)
Black African (%) 14 (9) 15 (12) 13 (6) 17 (26)
Latino (%) 24 (15) 11 (9) 31 (14) 4 (6)
HIV transmission risk factor 0.47 <0.0001
MSM (%) 74 (46) 50 (40) 124 (57) NA
IVDU (%) 14 (9) 15 (12) 19 (9) 10 (15)
Heterosexual (%) 70 (44) 57 (46) 71 (33) 56 (84)
HIVRNA <100,000 copies=ml (%) 36 (22) 37 (29) 0.4 54 (27) 19 (28) 0.89
Mean CD4þ cells=mm3 47 37 0.65 38 53 0.06
CD4þ <100 cell=mm3 (%) 96 (60) 94 (75) 0.003 148 (68) 42 (63) 0.29
Drug abuse (%) 30 (19) 31 (25) 0.19 43 (20) 21 (32) 0.04
Major psychiatric illness (%) 7 (4) 4 (3) 0.76 10 (5) 1 (2) 0.26
Depression (%) 36 (23) 34 (27) 0.32 51 (24) 20 (30) 0.24
Hepatitis C antibody positive (%) 17 (11) 15 (12) 0.64 21 (10) 11 (17) 0.12
Hepatitis B surface antigen positive (%) 28 (18) 17 (14) 0.4 38 (17) 7 (11) 0.17
OI at baseline (%) 81 (50) 66 (53) 0.6 107 (50) 40 (60) 0.14
CDC Stage C3 (%) 80 (50) 64 (51) 0.67 107 (49) 38 (57) 0.23
NRTI used 0.7 0.19
Zidovudine=lamivudine (%) 73 (46) 66 (53) 101 (46) 38 (57)
Tenofovir=emtricitabine (%) 65 (41) 34 (27) 82 (37) 17 (26)
Other (%) 21 (13) 22 (18) 38 (17) 11 (17)

Table 2. Multivariate Logistic Regression Similarly, a cohort of HIV-infected Brazilian individuals with
for Undetectable Viral Load (<400 Copies) CD4þ counts less than 200 cells=ml who started therapy after
among the Cohort Patientsa 1999 revealed that compliance and number of years of formal
education were predictors of successful response to
Variable OR 24 weeks OR 48 weeks HAART.21
For the outcome of clinical progression, HIV disease stage
Baseline ART regimen at baseline was the only predictive characteristic reaching
NNRTI versus bPI 1.08 0.46–2.54 0.85 0.34–2.18
statistical significance in the multivariate model. This result is
Baseline viral load
<100,000 copies=ml 3.55 1.29–9.81b 0.87 0.34–2.23 expected due to the relationship of HIV disease stage with
Baseline CD4þ count severity and degree of immunosuppression. A remarkable

100 cells=ml 3.96 1.19–13.15b 2.34 0.76–7.20 finding was that no clinical events occurred among patients
Gender with an education of high school or higher. This finding is in
Female 0.68 0.23–1.99 1.34 0.44–4.09 concordance with good outcomes in patients with higher
Age education shown in other studies, and highlights the impor-
<50 2.79 0.91–8.6 0.73 0.16–3.21 tance of education in health outcomes.4,21 It also suggests that
Ethnicity physicians and clinical treatment programs should assess
African American — — — — education and comprehension levels among their patients,
White 2.51 0.61–10.3 0.30 0.07–1.18
and devote additional time for adherence training for those
Black African 3.16 0.60–16.73 1.34 0.44–4.09
Latino 0.91 0.23–3.57 1.18 0.22–6.30 with less educational attainment. As HIV increasingly affects
Adherence patients of racial or ethnic minorities in whom lower rates of
100–90% 1.75 0.74–4.14 3.47 1.36–8.88c high school education completion may be observed, achiev-
Education ing successful outcomes with cART may become more chal-
High school or 0.59 0.10–3.41 4.98 1.11–22.4b lenging.
higher education As previous studies have shown, the type of cART regimen
a used was not a significant predictor of success for any
NNRTI, nonnucleoside reverse transcriptase inhibitor; ART,
antiretroviral treatment; bPI, boosted protease inhibitor. outcome in this study, despite a trend that favored NNRTI-
b
Significant at 0.01. based regimens in decreasing likelihood of clinical progres-
c
Significant at 0.001. sion.4,22–24
PREDICTORS OF SUCCESS IN AN ART-NAIVE URBAN POPULATION 137

Table 3. Multivariate Logistic Regression from baseline at 48 weeks compared to African Americans
for Change in Baseline CD4 Counts by More Than (OR ¼ 0.23; 95% CI, 0.06 to 0.97). This finding could not be
100 CD4 Cells among the Cohort Patientsa attributed to black Africans having had lower initial CD4þ
counts at initiation of therapy, since there were no statistical
Variable OR 24 weeks OR 48 weeks
differences in baseline T cells among ethnic groups (CD4þ T
Baseline ART regimen cells ¼ 48.8 cells=ml for black Africans vs. 38.4 cells=ml for other
NNRTI 0.72 0.31–1.65 1.03 0.40–2.63 racial=ethnic groups; p ¼ 0.25).
Baseline viral load Finally, women had a better immunological response at 24
<100,000 0.57 0.24–1.38 0.81 0.32–2.09 weeks. This might be explained by the fact that at baseline
Baseline CD4þ count women had higher CD4þ cell counts. Nonetheless, the dif-
<100 1.97 0.74–5.29 3.53 1.30–9.56b ference in baseline CD4þ cell counts between males and fe-
Gender males did not reach statistical significance (geometric mean
Female 7.38 2.25–24.17c 2.20 0.69–7.08
for women 37.5 vs. 53.1 in males; p ¼ 0.06), and the multi-
Age
<50 1.82 0.56–5.93 0.99 0.24–4.06 variate analysis adjusting for baseline CD4þ cell count still
Ethnicity favored immunologic recovery in women. Previous literature
African American — — — — has recognized that females have a tendency to less clinical
White 0.76 0.23–2.56 0.34 0.09–1.29 progression than males.27
Black African 0.09 0.02–0.45c 0.23 0.06–0.97d This study has several limitations. Due to the retrospective
Latino 7.57 1.14–50.13b 2.02 0.33–12.3 nature of the study there was a lack of randomization of
Adherence treatment assignments; this leads to some confounding be-
100–90% 0.91 0.39–2.15 1.39 0.54–3.58 tween groups that cannot be fully accounted for even by lo-
Education gistic regression. Furthermore, patients for whom viral load
High school 4.98 1.01–24.70d 1.55 0.31–7.78
or CD4þ counts were not available 24 weeks after treatment
or higher education
initiation and patients who failed to return after an initial visit
a
NNRTI, nonnucleoside reverse transcriptase inhibitor; ART, were not included in the analysis. This is important as non-
antiretroviral treatment. compliance with appointments and treatment follow up has
b
Significant at 0.01. been associated with poor outcomes in other studies.3,5,20,28,29
c
Significant at 0.001.
d
Significant at 0.05.
In addition, not all patients had an HIV-1 viral load and a
CD4þ count measured at exactly 24  3 and=or 48  3 weeks
and thus could not be analyzed. Third, we did not include a
Ethnic disparities in access and response to HAART have variable for socioeconomic status (SES), which has been re-
been previously reported.25,26 In our cohort, Latinos and black lated to clinical outcomes.30 However, as a public clinic the
Africans were more likely than African Americans and whites SES of the population at our clinic is relatively homogeneous
to achieve virologic suppression at 24 weeks in univariate and this reduces the potential confounding of this vari-
analysis, but not in multivariate analysis. Black Africans were able.14,31 Finally, other confounding risk factors could be
also less likely to have an increase of more than 100 CD4þ cells missing on the analysis, such as family=social support, co-
morbid diseases, and other pharmacological agents taken
(such as prophylactic antimicrobial agents, antivirals other
than cART, and lipid-lowering drugs).
Table 4. Multivariate Logistic Regression of Clinical In conclusion, this retrospective cohort study suggests that
Progression at 48 Weeksa potential predictors of success to currently approved cART
combinations for initial therapy in antiretroviral naive pa-
Variable OR 95%CI
tients are a baseline HIV-1 viral load of <100,000 copies=ml, a
Baseline ART regimen CD4þ count of >100 cells=ml, a high school or greater educa-
NNRTI 2.57 0.76–8.71 tion, and female gender. For treatment regimens that include
Baseline viral load an NNRTI or a boosted PI, 90–100% adherence also predicts
<100,000 copies=ml 2.07 0.32–3.58 long-term virologic success.
Baseline CD4þ count
<100 0.75 0.13–4.34
Gender Acknowledgments
Female 1.48 0.36–6.15
The project was supported in part by the NIH=FIC=AIDS
Age
<50 0.42 0.05–3.78 International Training and Research Program of Emory Uni-
Ethnicity versity (D43 TW01042) and the NIH=NIAID Center for AIDS
African American 0.21 0.02–1.96 Research of Emory University (2P30 AI 50409-04A1).
Black African 0.09 0.01–1.40
Adherence Author Disclosure Statement
100–90% 1.08 0.32–3.71
Stage No competing financial interests exist.
A1–C2 6.72 1.31–34.8b
References
a
NNRTI, nonnucleoside reverse transcriptase inhibitor; ART,
antiretroviral treatment. 1. Palella FJ, Jr., Delaney KM, Moorman AC, et al.: Declining
b
Significant at 0.01. morbidity and mortality among patients with advanced
138 ZARAGOZA-MACIAS ET AL.

human immunodeficiency virus infection. HIV Outpatient infected women. AIDS Res Hum Retroviruses 2006;22(3):
Study Investigators. N Engl J Med 1998;338(13):853–860. 222–231.
2. Palella FJ, Jr., Baker RK, Moorman AC, et al.: Mortality in the 19. Egger M, May M, Chene G, et al.: Prognosis of HIV-1-
highly active antiretroviral therapy era: Changing causes of infected patients starting highly active antiretroviral ther-
death and disease in the HIV outpatient study. J Acquir apy: A collaborative analysis of prospective studies. Lancet
Immune Defic Syndr 2006;43(1):27–34. 2002;360(9327):119–129.
3. Rastegar DA, Fingerhood MI, and Jasinski DRC: Highly 20. Mugavero MJ, Lin HY, Willig JH, et al.: Missed visits and
active antiretroviral therapy outcomes in a primary care mortality among patients establishing initial outpatient HIV
clinic. AIDS Care 2003;15(2):231–237. treatment. Clin Infect Dis 2009;48(2):248–256.
4. Friedl AC, Ledergerber B, Flepp M, et al.: Response to first 21. Tuboi SH, Harrison LH, Sprinz E, Albernaz RK, and
protease inhibitor- and efavirenz-containing antiretroviral Schechter M: Predictors of virologic failure in HIV-1-infected
combination therapy. The Swiss HIV Cohort Study. AIDS patients starting highly active antiretroviral therapy in Porto
2001;15(14):1793–1800. Alegre, Brazil. J Acquir Immune Defic Syndr 2005;40(3):324–
5. Robbins GK, Daniels B, Zheng H, Chueh H, Meigs JB, and 328.
Freedberg KA: Predictors of antiretroviral treatment failure 22. Waters L, Stebbing J, Jones R, et al.: A comparison of the CD4
in an urban HIV clinic. J Acquir Immune Defic Syndr 2007; response to antiretroviral regimens in patients commencing
44(1):30–37. therapy with low CD4 counts. J Antimicrob Chemother
6. Lucas GM, Chaisson RE, and Moore RD: Highly active an- 2004;54(2):503–507.
tiretroviral therapy in a large urban clinic: Risk factors for 23. Bartlett JA, Fath MJ, Demasi R, et al.: An updated systematic
virologic failure and adverse drug reactions. Ann Intern overview of triple combination therapy in antiretroviral-
Med 1999;131(2):81–87. naive HIV-infected adults. AIDS 2006;20(16):2051–2064.
7. Paris D, Ledergerber B, Weber R, et al.: Incidence and pre- 24. MacArthur RD, Novak RM, Peng G, et al.: A comparison of
dictors of virologic failure of antiretroviral triple-drug ther- three highly active antiretroviral treatment strategies con-
apy in a community-based cohort. AIDS Res Hum sisting of non-nucleoside reverse transcriptase inhibitors,
Retroviruses 1999;15(18):1631–1638. protease inhibitors, or both in the presence of nucleoside
8. Palella FJ, Jr., Chmiel JS, Moorman AC, and Holmberg SD: reverse transcriptase inhibitors as initial therapy (CPCRA
Durability and predictors of success of highly active anti- 058 FIRST Study): A long-term randomised trial. Lancet
retroviral therapy for ambulatory HIV-infected patients. 2006;368(9553):2125–2135.
AIDS 2002;16(12):1617–1626. 25. Hall HI, Byers RH, Ling Q, and Espinoza L: Racial=ethnic
9. Bosch RJ, Bennett K, Collier AC, Zackin R, and Benson CA: and age disparities in HIV prevalence and disease progres-
Pretreatment factors associated with 3-year (144-week) vi- sion among men who have sex with men in the United
rologic and immunologic responses to potent antiretroviral States. Am J Public Health 2007;97(6):1060–1066.
therapy. J Acquir Immune Defic Syndr 2007;44(3):268–277. 26. Easterbrook PJ, Farzadegan H, Hoover DR, et al.: Racial
10. Hammer SM, Eron JJ, Jr., Reiss P, et al.: Antiretroviral differences in rate of CD4 decline in HIV-1-infected homo-
treatment of adult HIV infection: 2008 recommendations of sexual men. AIDS 1996;10(10):1147–1155.
the International AIDS Society-USA panel. JAMA 27. Nicastri E, Angeletti C, Palmisano L, et al.: Gender differ-
2008;300(5):555–570. ences in clinical progression of HIV-1-infected individuals
11. Mugavero MJ, Pence BW, Whetten K, et al.: Predictors of during long-term highly active antiretroviral therapy. AIDS
AIDS-related morbidity and mortality in a southern U.S. 2005;19(6):577–583.
Cohort. AIDS Patient Care STDS 2007;21(9):681–690. 28. Bangsberg DR, Perry S, Charlebois ED, et al.: Non-adherence
12. DHSS: Guidelines for the use of antiretroviral agents in HIV- to highly active antiretroviral therapy predicts progression
1-infected adults and adolescents: Department of Health to AIDS. AIDS 2001;15(9):1181–1183.
and Human Services; 2008:<http:==www.aidsinfo.nih.gov= 29. Mugavero MJ, Lin HY, Allison JJ, et al.: Racial disparities in
ContentFiles=AdultandAdolescentGL.pdf>. HIV virologic failure: Do missed visits matter? J Acquir
13. Hall HI, Geduld J, Boulos D, et al.: Epidemiology of HIV in Immune Defic Syndr 2009;50(1):100–108.
the United States and Canada: Current status and ongoing 30. Cunningham WE, Hays RD, Duan N, et al.: The effect of
challenges. J Acquir Immune Defic Syndr 2009;51(Suppl 1): socioeconomic status on the survival of people receiving
S13–20. care for HIV infection in the United States. J Health Care
14. Epidemiology of HIV=AIDS––United States, 1981–2005. Poor Underserved 2005;16(4):655–676.
MMWR Morb Mortal Wkly Rep 2006;55(21):589–592. 31. Ashman JJ, Marconi KM, McKinney MM, and O’Neill JF:
15. Steiner JF and Prochazka AV: The assessment of refill com- Who receives Ryan White CARE Act services? A demogra-
pliance using pharmacy records: Methods, validity, and phic comparison of CARE act clients and the general AIDS
applications. J Clin Epidemiol 1997;50(1):105–116. population. AIDS Patient Care STDS 2000;14(10):561–565.
16. Low-Beer S, Yip B, O’Shaughnessy MV, Hogg RS, and
Montaner JS: Adherence to triple therapy and viral load Address correspondence to:
response. J Acquir Immune Defic Syndr 2000;23(4):360–361. Jeffrey Lennox
17. Delgado J, Heath KV, Yip B, et al.: Highly active anti- Department of Medicine
retroviral therapy: Physician experience and enhanced ad- Emory University School of Medicine
herence to prescription refill. Antivir Ther 2003;8(5):471–478. 69 Jesse Hill Jr. Drive
18. Vaamonde CM, Hoover DR, Anastos K, et al.: Factors Atlanta, Georgia 30303
associated with poor immunologic response to virologic
suppression by highly active antiretroviral therapy in HIV- E-mail: [email protected]