Article

Journal of the International

Association of Providers of AIDS Care
Zidovudine- versus Tenofovir-Based 1–5
ª The Author(s) 2017
Reprints and permission:
Antiretroviral Therapy for the Initial sagepub.com/journalsPermissions.nav
DOI: 10.1177/2325957416686190
Treatment of HIV Infection in the Ethnic journals.sagepub.com/home/jia

Minority Region of Liangshan Prefecture,

Sichuan Province, China: An Observational
Cohort Study

Cedric P. Cheung, MD1, Wen Hong Lai, MD2, and Jonathan Shuter, MD3

Abstract
Background: Tenofovir (TDF)-based highly active antiretroviral therapy (HAART), as recommended by the World Health
Organization guidelines for HIV-naive patients, has been limited in resource-constrained settings. The aim of this study was to
evaluate the effectiveness of zidovudine-(ZDV) versus TDF-based HAART in the Yi minority region of Sichuan Province, China at
a single HIV treatment center. Methods: The primary end point was the attainment of an HIV viral load <50 copies/mL.
Secondary end points included change in CD4 level, adverse reactions, mortality, and sustained virologic suppression.
Results: Of the 361 total participants, recipients of TDF-based HAART were more likely to achieve viral load <50 copies/mL
(60% versus 46%, odds ratio [OR] ¼ 1.7, P ¼ .016) as well as sustained virologic suppression (61% versus 28%, OR ¼ 3.4,
P ¼ .001). Tenofovir (adjusted odds ratio [ORadj] ¼ 1.71, P ¼ .025) and female sex (ORadj ¼ 1.93, P ¼ .003) were identified as
independent predictors of achieving HIV viral load <50 copies/mL in the multivariate logistic regression analysis. Conclusion:
Among Chinese Yi minority HIV-infected participants, TDF-based HAART was superior to ZDV-based HAART for initial
treatment of HIV infection, suggesting TDF-based HAART should be the regimen of choice in China.

Keywords
HIV, HAART, tenofovir, zidovudine, HIV viral load

Introduction ZDV-based HAART, data on virologic suppression have

Since its inception in 2002, the China National Free Antiretro-
viral Treatment Program has made tremendous progress in
providing highly active antiretroviral therapy (HAART) to over
220 000 patients and has markedly decreased the 5 year mor-
tality rate of Chinese HIV-infected patients.1 Current Chinese
guidelines recommend zidovudine (ZDV) or tenofovir diso-
proxil fumarate (TDF) plus lamivudine (3TC) plus efavirenz
(EFV) or nevirapine (NVP) for the initial treatment of HIV.2
The 2016 World Health Organization (WHO) guidelines list
TDF plus 3TC or emtricitabine as the preferred nucleoside
reverse transcriptase inhibitor backbone.3 In response to these
guidelines, the China National Free Antiretroviral Treatment
Program has expanded the use of TDF. In Sichuan Province,
44% of all HIV-infected patients prescribed HAART in 2013
used a regimen containing TDF and 3TC plus EFV or NVP
(Sichuan Center for Disease Control and Prevention, personal
communication, May 10, 2014). Although studies comparing
TDF- versus ZDV-based regimens for treatment of HIV
infection have demonstrated superior safety of TDF- over
yielded mixed results.4–10
In 2010, areas of the Liangshan Prefecture located in
Sichuan Province had an 11% prevalence of HIV among adults,
making this the highest prevalence region in China.11 In a pilot
project starting in 2011, our center located in Liangshan Pre-
fecture, replaced ZDV with TDF as part of the preferred
HAART regimen for treatment-naive HIV-infected patients.
This permitted us to conduct a comparison of TDF- versus
1
MSI Professional Services, San Po Kong, Kowloon, Hong Kong
2
Sichuan Center for Disease Control and Prevention, Chengdu, Sichuan
Province, China
3
Albert Einstein College of Medicine, Montefiore Medical Center, Bronx,
NY, USA
Corresponding Author:
Cedric P. Cheung, MSI Professional Services, Room 702, Win Plaza, 9 Sheung
Hei Street, San Po Kong, Kowloon, Hong Kong.
Email: cedcheung@yahoo.com
2 Journal of the International Association of Providers of AIDS Care
ZDV-based HAART in antiretroviral naive HIV-infected indi-
viduals in a high-prevalence area of China.
Methods
Setting
The HIV treatment center is part of a general county-level
hospital located in a rural area of Liangshan Prefecture, where
over 95% of the population is comprised of the Yi ethnic
minority group, with the majority employed as farmers. It
serves as the referral treatment center for a population of
approximately 270 000 people and has over 3500 HIV-
infected outpatients currently under treatment.
Study Design and Data Collection
Data were extracted from patient charts stored in the HIV treat-
ment center department of records. All participants enrolled in
the clinic and started HAART before December 31, 2012 were
considered for entry into the retrospective cohort, including
those who were lost to follow-up or died before December
31, 2012. Participants surviving beyond this date who received
care in the clinic were additionally observed prospectively for
any subsequent end points until December 31, 2013. In order to
be included in the study cohort participants must have had (1)
documented HIV seropositivity, (2) no prior antiretroviral ther-
apy (ART) exposure, (3) received ZDV/3TC or TDF/3TC in
combination with EFV or NVP for at least 3 months, and (4) at
least 1 HIV viral load measurement after ART initiation. The
primary end point was attaining an HIV viral load <50 copies/
mL. Secondary end points included CD4 response defined as
change in CD4 count level from baseline to peak CD4 count
level achieved, change of ART due to intolerance of ZDV or
TDF, and all-cause mortality. Sustained virologic suppression
was also analyzed in the subset of participants who had more
than 1 viral load measurement and was defined as achievement
of all HIV viral loads <50 copies/mL. HIV viral load testing
was determined by VERSANT 440 Molecular Systems (Bayer
Healthcare, Siemens Healthcare Diagnostics, Germany) in
which the lowest level of detection was <50 copies/mL and
the BD FACSCount system (Becton, Dickenson and Company,
USA) was used to measure CD4 count. This study was
approved by the HIV treatment center ethics committee.
Analysis and Statistics
Categorical and continuous data were analyzed by the
Cochran–Mantel–Haenszel test and the Wilcoxon rank sum
test, respectively. Differences in mortality rates were analyzed
by Kaplan-Meier survival analysis. Factors found to be associ-
ated with the primary outcome in univariate analyses were
entered into a multivariate logistic regression model. A back-
ward hierarchical strategy was employed, sequentially
removing covariates without a significant effect on the model.
Variables demonstrating associations with the primary end
point (P < .10) after adjustment for the other covariates were
retained in the final model. All reported P values were 2-sided
with a P < .05 considered statistically significant. All statistical
analyses were performed using SPSS (version 22).
Results
Demographic and Clinical Characteristics
Of the 361 participants meeting the inclusion criteria, 240
(66%) were started on a TDF-based regimen and 121 (34%)
on a ZDV-based regimen (Table 1). The 2 groups did not differ
with respect to gender, history of intravenous drug use, coin-
fection with hepatitis C, or in the mean number of HIV viral
loads measured per year. In participants who had a baseline
CD4 count, the TDF group had higher mean baseline CD4
count than the ZDV group (399 + 230 versus 302 + 139,
mean + standard deviation [SD] cells/mm3, P < .001). Com-
pared to those on ZDV-based therapy, those starting TDF-
based therapy were slightly older (35.9 + 6.9 versus
32.5+9.1 years, P < .001), had fewer mean number of CD4
measurements per year (2.1 + 1 versus 2.3 + 1.1, P ¼ .015), were
earlier in mean time to first viral load measurement after start-
ing HAART (331 + 138 versus 389 + 195 days, P ¼ .004), had
shorter mean length of follow-up (565 + 163 versus 834 + 240
days, P < .001), and were paired less with NVP (5% versus 21%,
P < .001).
Primary End Point
A total of 144 (60%) of the 240 participants in the TDF group
and 56 (46%) of the 121 participants in the ZDV group
achieved an HIV viral load <50 copies/mL (odds ratio [OR]
¼ 1.7, 95% confidence interval [CI] ¼ 1.1-2.66, P ¼ .016). To
examine if dissimilar rates of NVP use between the 2 treatment
groups had a substantive effect on the primary end point, EFV-
only containing regimens were analyzed in both treatment
groups. The OR of achieving the primary end point still trended
toward favoring the TDF group (OR ¼ 1.6, 95% CI ¼ 0.98 to
2.6, P ¼ .057).
Secondary End Points
There was no difference in change of CD4 counts from baseline
to peak between the 2 treatment groups (TDF group, N ¼ 109,
þ184 CD4 cells/mm3; ZDV group, N ¼ 95, þ153 CD4 cells/
mm3, P ¼ .31). There were no differences with respect to age,
history of intravenous drug use, hepatitis C virus status, or
treatment group in those who had and did not have a baseline
CD4 count measurement. There were 26 adverse events that led
to study drug substitution, 22 (18% of all ZDV group partici-
pants) in the ZDV group and 4 (2% of all TDF group partici-
pants) in the TDF group (OR ¼ 13, 95% CI ¼ 4-38, P < .001).
Although the raw mortality rate in the ZDV group was higher
than the TDF group (2.2 deaths/100 person years [PY] versus
0.4 deaths/100 PY, respectively), the study was not powered for
this end point, and it did not achieve statistical significance in
the Kaplan-Meier survival analysis. In participants who had
Cheung et al 3

Table 1. Baseline and Follow-up Characteristics.

All Participants, ZDV-Based Regimen, TDF-Based Regimen,

Variable N ¼ 361 (Mean + SD) n ¼ 121 (Mean + SD) n ¼ 240 (Mean + SD) P Value

Mean age, years 34.8 + 7.9 32.5 + 9.1 35.9 + 6.9 <.001
Female 49% 47% 51% .41
History of intravenous drug use 61% 58% 63% .44
Positive hepatitis C serostatus 45% 44% 46% .66
Mean baseline CD4 count, cells/mm3 n ¼ 206, 354 + 199 n ¼ 96, 302 + 139 n ¼ 110, 399 + 230 <.001
Mean CD4 measurements per year 2.2 + 1.1 2.3 + 1.1 2.1 + 1 .015
Mean HIV viral load measurements per year 0.89 + 0.5 0.89 + 0.53 0.89 + 0.47 .94
Mean time to first viral load after starting 350 + 161 389 + 195 331 + 138 .004
HAART, days
Mean length of follow-up, days 655 + 230 834 + 240 565 + 163 <.001
Received NVP as NNRTI 10% 21% 5% <.001
Abbreviations: HAART, highly active antiretroviral therapy; NVP, nevirapine; SD, standard deviation; TDF, tenofovir; ZDV, zidovudine; NNRTI, non-nucleoside
reverse transcriptase inhibitor.

more than 1 viral load measurement, 31 (57%) of the 54 in the Table 2. Factors Associated with Achieving HIV Viral Load
TDF group and 20 (28%) of the 71 participants in the ZDV <50 copies/mL on Univariate and Multivariate Analysis.
group achieved sustained undetectable viral load (OR ¼ 3.4, Multivariate
95% CI ¼ 1.6-7.3, P ¼ .001). There were more mean number Univariate Analysis, P Analysis, Adjusted P
of viral loads measured for the ZDV group than the TDF group Factor OR (95% CI) Value OR (95% CI) Value
(2.3 versus 2.1, 95% CI for difference in means 0.03-0.34,
P ¼ .028). Regimen
TDF 1.71 (1.1-2.66) .016 1.71 (1.07-2.72) .025
ZDV 1 1
Univariate and Multivariate Analysis of Factors Gender
Associated with Achieving HIV Viral Load <50 Copies/mL Female 1.96 (1.29-2.99) .002 1.93 (1.24-2.99) .003
Age not applicable 1.01 (0.98-1.04) .4
Factors associated with achieving HIV viral load <50 copies/ IVDU
mL in the univariate analyses were regimen (OR ¼ 1.71 favor- Negative 1.7 (1.1-2.61) .017 1.28 (0.45-1.38) .4
ing TDF group, 95% CI ¼ 1.1-2.66, P ¼ .016), gender (OR ¼ Positive 1 1
1.96 favoring females, 95% CI ¼ 1.29-2.99, P ¼ .002), and not HCV
Negative 1.17 (0.76-1.81) .46 1.19 (0.88-2.09) .4
having a history of intravenous drug use (OR ¼ 1.7, 95% CI ¼
Positive 1 1
1.1-2.61, P ¼ .017). In the multivariate logistic regression
model, only regimen (adjusted odds ratio [ORadj] ¼ 1.71 favor- Abbreviations: CI, confidence interval; IVDU, intravenous drug user; HCV,
ing TDF group, 95% CI ¼ 1.07-2.72, P ¼ .025) and gender hepatitis C virus; OR, odds ratio; TDF, tenofovir; ZDV, zidovudine.
(OR adj ¼ 1.93 favoring females, 95% CI ¼ 1.24-2.99,
P ¼ .003) were associated with achieving HIV viral load <50
copies/mL (Table 2). FTC-TDF) with EFV plus coformulated 3TC/ZDV which
found TDF-based therapy superior to ZDV-based therapy in
achieving and maintaining HIV viral load <400 copies/mL
(71% versus 58%, respectively, P ¼ .004) after 144 weeks of
Discussion treatment.4 In a cohort study from India, fixed-dose TDF-based
This observational study from a county-level HIV treatment HAART also showed significantly lower virologic failure as
center is the first to our knowledge to analyze the use of TDF compared to fixed-dose ZDV or stavudine-based HAART.5
in Chinese HIV-infected ART-naive patients as initial Likewise, another prospective cohort study from South Africa
HAART. The TDF treatment group was significantly superior showed in subgroup analysis that single tablet EFV-LMV-TDF
to the ZDV treatment group in achieving an HIV viral load <50 was superior to EFV plus fixed-dose LMV-ZDV in achieving
copies/mL and sustained virologic response. The association of an undetectable viral load.6 In contrast, 1 large randomized
TDF with improved virologic outcomes was also observed in clinical trial from low-, intermediate-, and high-income coun-
the logistic regression model when the analysis was adjusted tries compared EFV plus coformulated FTC/TDF to EFV plus
for several other clinically relevant factors. coformulated 3TC/ZDV,5 and 2 observational cohort studies
These data are consistent with the findings from the Study from South Africa and 1 from China examining EFV or NVP,
934 Group, a randomized clinical trial from the United States 3TC, and TDF versus EFV or NVP, 3TC, and ZDV found
and Europe comparing EFV–FTC (emtricitabine)–TDF (after no difference in virologic suppression between treatment
96 weeks the regimen was changed to EFV plus co-formulated groups.8–10 One possible explanation for the discrepancy in the
4 Journal of the International Association of Providers of AIDS Care
results could be that the use of coformulated EFV/FTC/TDF in
the Study 934 Group conferred an advantage over ZDV-based
HAART that the other studies could not detect using nonsingle
pill coformulated TDF-based HAART. It is possible that even
though our study did not use coformulated regimens, medica-
tion adherence to once daily TDF-based HAART was higher
than twice daily ZDV-based HAART in our patient population
with high rates of poverty and drug use, resulting in superior
virologic outcomes in the TDF group.
Interestingly, subgroup analysis of the Labhardt study from
South Africa showed no difference in virologic outcomes
between TDF and ZDV based regimens which used NVP as
the NNRTI.6 Additionally, a small randomized clinical trial
from France12 and a large retrospective cohort study from
Nigeria13 showed NVP-LMV-TDF was actually inferior to
NVP-LMV-ZDV. In the small subset of patients receiving
NVP as the NNRTI in our study, there was no significant
difference in obtaining undetectable viral load between TDF-
and ZDV-based regimens, though our study was not powered to
detect a difference. Similar to our data, the aforementioned
studies also found better tolerability and less drug substitutions
with TDF-based regimens than with ZDV-based regimens.4–9
Our study demonstrated no difference in increase of CD4 count
from baseline, which is consistent with the above-mentioned
studies. As in our study, there was no difference found in
mortality between TDF- and ZDV-based regimens in observa-
tional studies from South Africa8 and Zambia.14
Multiple studies from China using non–TDF-based regi-
mens in the majority of Han Chinese ethnicity participants
have found higher proportions of participants achieving vir-
ologic suppression (72%-92%) than that observed in our
cohort.10,15–19 However, a cross-sectional survey set in the
same locality and with participants closely matching our
cohort demographic had similar virologic responses to our
study using non–TDF-based regimens. Yi ethnic minority,
intravenous drug use, and illiteracy were all correlated with
poor outcomes.20 The percentage of intravenous drug users in
our study was much higher (61%) than that reported in a
countrywide HIV-infected patient registry (22%), which
found intravenous drug use a strong predictor of delayed
HAART initiation and death.21 It is likely that the high levels
of substance use and poverty in our population contributed to
the inferior outcomes in both treatment groups.
A larger proportion of female participants achieved HIV
viral load <50 copies/mL than male participants regardless of
treatment group. Data from other cohort studies have also
shown that female HIV-infected patients had better virologic
and immunologic response to HAART than males.22–24 A pos-
sible explanation for the gender discrepancy in our cohort is
that males were much more likely to have a history of intrave-
nous drug use (91% of males versus 31% of females) and
therefore likely had poorer general health and a more disor-
dered lifestyle leading to lower treatment adherence.
Our study had several limitations worthy of mention. As
with any unrandomized observational cohort study, there is the
possibility of unmeasured confounders and bias that could have
affected outcomes. There were many participants with missing
baseline CD4 count data which may have affected the power to
detect a difference in the change in CD4 level from baseline.
Mortality statistics were dependent on family reporting and
were not linked to any public database, thus our study may
have underestimated true mortality rate. Because the majority
of participants in the ZDV treatment group initiated HAART at
an earlier calendar date than participants in the TDF treatment
group, there is the potential for temporal bias. It is possible that
improvements in adherence counseling and management of
adverse reactions at a later time could have favored TDF. Our
analysis was from a single center, with all participants belong-
ing to the Yi ethnic minority, thus potentially limiting the gen-
eralizability of our results.
Our study also has several strengths. It describes one of the
largest published Chinese HIV HAART-naive single-center
cohorts, and to our knowledge is the first to report on use of
TDF treatment in China. The study was conducted in a ‘‘real-
world’’ setting, where participants were part of a large clinical
HIV treatment program and were managed according to stan-
dard care. Selection bias was minimized in our study because
prior to October 2011, ZDV-based HAART was the preferred
regimen, whereas after that date TDF-based HAART became a
preferred therapy. Additionally, due to the practice environ-
ment, we had a complete accounting of patients on HAART.
In conclusion, we have demonstrated that in a HAART-
naive Chinese patient cohort, TDF-based HAART is clinically
superior to ZDV-based HAART both in terms of efficacy and
tolerability. We believe our study supports the use of TDF as a
component of initial HAART in China and adds to the data to
support the use of TDF in resource-constrained countries as
outlined in the WHO guidelines.
Acknowledgments
The authors are grateful for the support of Lewu Xuewen (HIV
Treatment Center Director) and Ma Fanghua (HIV Treatment Center
Outpatient Department Medical Director). The authors also thank the
doctors, counselors, staff, and patients and their families at the HIV
Treatment Center who made this research possible.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to
the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for
the research, authorship, and/or publication of this article: Jonathan
Shuter was supported, in part, by the Clinical Core of the Center for
AIDS Research at the Albert Einstein College of Medicine and
Montefiore Medical Center funded by the National Institutes of Health
(NIH AI-51519).
References
1. 2014 China AIDS Response Progress Report. National Health and
Family Planning Commission of The People’s Republic of China.
June, 2014. http://www.unaids.org/sites/default/files/documents/
CHN_narrative_report_2014.pdf. Accessed April 5, 2015.
Cheung et al
2. Zhang FJ (Editor in chief). China National Free Antiretroviral
Program Treatment Guide (3rd ed.). Beijing, China: People’s
Medical Publishing House; 2012.
3. Consolidated Guidelines on the Use of Antiretroviral Drugs for
Treating and Preventing HIV Infection: Recommendations for a
Public Health Approach. World Health Organization, June 2016.
http://apps.who.int/iris/bitstream/10665/85321/1/9789241505727_
eng.pdf?ua¼1. Accessed October 30, 2016.
4. Arribas JR, Pozniak AL, Gallant JE, et al. Tenofovir disoproxil
fumarate, emtricitabine, and efavirenz compared with zidovudine/
lamivudine and efavirenz in treatment-naive patients, 144-week
analysis. J Acquir Immune Defic Syndr. 2008;47(1):74–78.
5. Shet A, Neogi U, Kumarasamy N, et al. Virological efficacy with
first-line antiretroviral treatment in India: predictors of viral fail-
ure and evidence of viral resuppression. Trop Med Int Health.
2015;20(11):1462–1472.
6. Labhardt ND, Bader J, Lejone TI, et al. Is zidovudine first-line
therapy virologically comparable to tenofovir in resource-limited
settings? Trop Med Int Health. 2015;20(7):914–918.
7. Campbell TB, Smeaton LM, Kumarasamy N, et al. Efficacy and
safety three antiretroviral regimens for initial treatment of HIV-1:
a randomized clinical trial in diverse multinational settings. PLoS
Med. 2012;9(8):e1001290.
8. Velen K, Lewis JJ, Charalambous S, Grant AD, Churchyard GJ,
Hoffmann CJ. Comparison of tenofovir, zidovudine, or stavudine
as part of first-line antiretroviral therapy in a resource-limited-
setting: a cohort study. PLoS One. 2013;8(5):e64459.
9. Njuguna C, Orrell C, Kaplan R, et al. Rates of switching antire-
troviral drugs in a primary care service in South Africa before and
after introduction of tenofovir. PLoS One. 2013;8(5):e63596.
10. Lu ZZ, Su QJ, Li X, et al. A comparative analysis of the efficacy
of HAART with TDF and with AZT. The Journal of Practical
Medicine. 2015;9.
11. Dong C, Huang ZJ, Martin MC, et al. The impact of social factor
on human immunodeficiency virus and hepatitis C virus
co-infection in a minority region of Sichuan, the People’s Repub-
lic of China: a population-based survey and testing study. PLoS
One. 2014;9(7):e101241.
12. Rey D, Hoen B, Chavanet P, et al. High rate of early viological
failure with once-daily tenofovir/lamivudine/nevirapine combina-
tion innaive HIV-1-infected patients. J Antimicrob Chemother.
2009;63:380–388.
5
13. Scarsi K, Eisen G, Darin KM, et al. Superior Effectiveness of
Zidovudine Compared With Tenofovir When Combined With
Nevirapine-based Antiretroviral Therapy in a Large Nigerian
Cohort. Clin Infect Dis. 2016;62(4):512–518.
14. Chi BH, Mwango A, Giganti M, et al. Early clinical and pro-
grammatic outcomes with tenofovir-based antiretroviral
therapy in Zambia. J Acquir Immune Defic Syndr. 2010;
54(1):63–70.
15. Zhou HY, Zheng YH, He Y, et al. Evaluation of a 6-year highly
active antiretroviral therapy in Chinese HIV-1-infected patients.
Intervirology. 2010;53(4):240–246.
16. Deng XJ, He Y, Yang L, et al. The evaluation of four-year highly
active antiretroviral therapy in HIV-1 infected patients. Chin J
Intern Med. 2011;50(3):230–234.
17. Chen XJ, Tang XP, Cai WP, et al. The therapeutic effects of
highly active anti-retroviral therapy in 74 treatment-naive patients
with AIDS in China. Chin J Intern Med. 2011;50(1):59–62.
18. Ruan YH, Xing H, Wang X, et al. Virologic outcomes of first-line
HAART and associated factors among Chinese patients with HIV
in three sentinel antiretroviral treatment sites. Trop Med Int
Health. 2010;15(11):1357–1363.
19. Li TS, Dai Y, Kuang JQ, et al. Three generic nevirapine-based
antiretroviral treatments in Chinese HIV/AIDS patients: multi-
centric observation cohort. PLoS One. 2008;3(12):e3918.
20. Wang QX, Wang X, Chen B, et al. Drug resistance and associated
factors on HIV in Liangshan prefecture, Sichuan province.
Chin J Epidemiol. 2011;32(11):1082–1086.
21. Zhang FZ, Dou ZH, Ma Y, et al. Effect of earlier initiation of
antiretroviral treatment and increased treatment coverage on HIV-
related mortality in China: a national observational cohort study.
Lancet Infect Dis. 2011;11(7):516–524.
22. Collazos J, Aseni V, Caron JA. Sex differences in the clinical,
immunological and virological parameters of HIV-infected
patients treated with HAART. AIDS. 2007;21(7):835–843.
23. Dou ZH, Xu JH, Jiao JH, et al. Gender difference in 2- year
mortality and immunological response to ART in an HIV-
infected Chinese population, 2006-2008. PLoS One. 2011;6(8):
e22707.
24. Alibhai A, Kipp W, Saunders LD, et al. Gender-related mor-
tality for HIV-infected patients on highly active antiretroviral
therapy (HAART) in rural Uganda. Int J Womens Health.
2010;2:45–52.