Viability and effectiveness of large-scale HIV
treatment initiatives in sub-Saharan Africa:
experience from western Kenya
Kara Wools-Kaloustiana,d, Silvester Kimaiyod, Lameck Dierod, Abraham
Siikad, John Sidleb,d, Constantin T. Yiannoutsosc, Beverly Musickc,
Robert Einterzb, Kenneth H. Fifeaand William M. Tierneyb
Objectives: Todeterminetheclinical andimmunologicaloutcomesofa cohortofHIV-
infected patients receiving antiretroviral therapy.
Design: Retrospective study of prospectively collected data from consecutively
enrolled adult HIV-infected patients in eight HIV clinics in western Kenya.
Methods: CD4 cell counts, weight, mortality, loss to follow-up and adherence to
antiretroviral therapy were collected for the 2059 HIV-positive non-pregnant adult
patients treated with antiretroviral drugs between November 2001 and February 2005.
Results: Median duration of follow-up after initiation of antiretroviral therapy was 40
weeks (95% confidence interval, 38–43); 111 patients (5.4%) were documented as
deceased and 505 (24.5%) were lost to follow-up. Among 1766 (86%) evaluated for
Although patients with and without perfect adherence gained weight, patients with less
than perfect adherence gained 1.04 kg less weight than those reporting perfect adher-
ence (P ¼ 0.059). CD4 cell counts increased by a mean of 109 cells/ml during the first 6
weeks of therapy and increased more slowly thereafter, resulting in overall CD4 cell
countincreases of160, 225 and297 cells/ml at 12, 24, and36 monthsrespectively. At 1
year, a mean increase of 170 cells/ml was seen among patients reporting perfect
adherence compared with 123 cells/ml among those reporting some missed doses
(P < 0.001).
Conclusions: Antiretroviral treatment of adult Kenyans in this cohort resulted in
significant and persistent clinical and immunological benefit. These findings document
the viability and effectiveness of large-scale HIV treatment initiatives in resource-
? 2006 Lippincott Williams & Wilkins
AIDS 2006, 20:41–48
Keywords: HIV, AIDS, antiretroviral therapy, computerized
medical records system, sub-Saharan Africa
UNAIDS estimates that 2 million of Kenya’s 29.5 million
people are currently infected with HIV and that 1.5
million have already succumbed to the disease, resulting
trend is reflected in most of sub-Saharan Africa .
HAART has been proven to decrease HIV viral load,
increase CD4 cell counts and stem the progression to
From theaDivision of Infectious Diseases, thebDivision of General Internal Medicine, thecDivision of Biostatistics, Department of
Medicine, Indiana University School of Medicine, Indianapolis, Indiana, USA, and thedDepartment of Medicine, Moi University
Faculty of Health Sciences, Eldoret, Kenya.
Correspondence to Dr K. K. Wools-Kaloustian, 1001 W. 10th Street–OPW 430, Indianapolis, IN 46202, USA.
Received: 9 June 2005; revised: 29 August 2005; accepted: 7 September 2005.
ISSN 0269-9370 Q 2006 Lippincott Williams & Wilkins
5]. Studies from Senegal, Uganda and South Africa con-
firm that viral suppression and improvement in immune
status can be achieved in African populations [6–8].
Despite these findings, introduction of antiretroviral
drugs into sub-Saharan Africa has been slow because of
concerns about treatment adherence, emergence of resis-
tance and the impact of antiretroviral therapy on risk
behaviors. However, this delay is mostly attributable to
inadequate funds for antiretroviral drugs and absence of
sufficient infrastructure to dispense and monitor therapy
The Indiana University School of Medicine has had a
collaborative partnership with Moi University in Eldoret
Kenya since 1990 . In response to a significant incr-
ease in HIV prevalence noted in 2001 at the Moi Teach-
ing and Referral Hospital (Moi Hospital), the Academic
Model for the Prevention and Treatment of HIV/AIDS
University, Moi Hospital and the Indiana University
was to establish an HIV care system to serve the needs of
both urban and rural patients and to assess the barriers to
and outcomes of antiretroviral therapy. Details of the
development of this system have been described else-
where . The first urban and rural HIV clinics were
opened in November 2001, with six additional clinics
opening between January and July of 2004. A computer-
ized medical record system was developed to support
clinical care . This study determined the outcomes of
patients treated in this care system through analysis of data
routinely collected and stored in patients’ electronic
This study was approved by the Indiana University
School of Medicine Institutional Review Board and the
Moi University Institutional Research and Ethics
Committee. This study used data of prospectively
enrolled HIV-infected adult patients treated in the
AMPATH clinics. The data were stored in computerized
medical records and all patient identifiers were removed.
Patients were eligible if they were seen in any of the
AMPATH clinicsbetween November 2001 and February
2005, were at least 18 years of age, and initiated
antiretroviral drug treatment prior to September 2004.
excluded because it is our standard practice to provide a
regardless of CD4 cell count.
The urban clinic is located at Moi Hospital in Eldoret,
Kenya, a national referral hospital serving a catchment
area of approximately 13 million people. Comprehensive
HIV care serviceswere added to the primary care services
Three of these satellite clinics are 25–35 km from Eldoret
while four are located 60–120 km away. HIV care is pro-
vided by physicians and clinical officers trained and men-
tored within the AMPATH model .
Detailed algorithms consistent with World Health
Organization(WHO)guidelines forcare ofHIV-infected
patients were developed locally and followed throughout
the study period . Patients receiving antiretroviral
drugs were seen 2 weeks after initiation of therapy and
every month thereafter. During these visits, patients
underwent clinical and adherence assessments and were
dispensed antiretroviral drugs. For patients found to be
adherent with their medication and clinically stable, visit
intervals were sometimes extended to every 2 months.
Laboratory testing was based on local protocols and
clinical necessity. Per protocol, a complete blood count
with white cell differential count, CD4 cell count and
alanine aminotransferase assay were performed at baseline
and every 6 months. No viral load testing was performed
because of funding limitations. During the early part of
the project, when funds were limited, self-pay patients
without resources did not have all laboratory testing done
at the specified intervals. Patients meeting WHO criteria
for treatment started antiretroviral drugs at the sole
discretion of their clinician . Because of funding
limitations during the first 30 months of the HIV clinic
operation, first priority for treatment was given to
severely ill patients. During this period, some patients
received free antiretroviral drugs and testing through the
Maternal to Child Transmission Plus (MTCT-Plus)
program, philanthropic donations, and foundation grants,
while other patients paid for their medications. By the
latter part of this period, significant funding for HIV/
AIDS care was available through the President’s Emer-
gency Plan for AIDS Relief (PEPFAR; http://www.
and antiretroviral therapy was given free of charge to all
patients meeting treatment criteria. Education about
the nurse during the clinic session when the drugs were
initiated. The standard antiretroviral treatment regimen
consisted of three drugs: stavudine, lamivudine, and ne-
virapine. To avoid drug interactions, efavirenz was
substituted for nevirapine in patients receiving induction
therapy for tuberculosis. No patient experienced treat-
ment interruptions owing to stock shortages. A 25%
decline in CD4 cell count from the maximum attained
after initiation of antiretroviral drugs was considered a
regimen failure and triggered a change in patients’
antiretroviral regimen. This level of decline was chosen
42 AIDS 2006, Vol 20 No 1
as the definition of antiretroviral therapy failure beca-
use Hughes and colleagues  documented an average
within-subject coefficient of variation for CD4 cell
count of 25% among HIV-infected patients in the United
Data collection and management
Clinicians completed standard initial and return encoun-
ter forms at all AMPATH clinic visits (http://amrs.iu-
kenya.org/download/forms). The initial encounter form
included standard demographic, historical, psychosocial,
physical, and laboratory data as well as the medications
provided (antiretroviral drugs and opportunistic infection
prophylaxis). Follow-up data were collected on inter-
current symptoms, medication adherence, new diagno-
ses, laboratory data, and modifications of drug regimens.
Dedicated data entry clerks entered this information
into the AMPATH Medical Records System, which uses
an MS-Access database (Microsoft Corp., Redmond,
Washington State, USA) .
Astandard adherence assessment was addedto the follow-
up clinic visit form in June 2003, allowing assessment of
adherence at every visit. It consisted of one question
pills did the patient take?’ The available responses were:
‘none’, ‘few’, ‘half’, ‘most’, and ‘all’. In the analyses,
adherence was considered as perfect (every response at
every visit is ‘all’) or imperfect (any response other than
‘all’ reported at even a single visit). Mortality data were
captured using a passive surveillance system.
Frequency tables were produced for all categorical
variables and were compared via the Fisher’s exact test.
For continuous measures, the median and 95% trimmed
ranges (the 2.5th and 97.5th percentiles) were generated
and compared between groups using the Kruskal–Wallis
test. Time-to-event analyses were performed using the
Kaplan–Meier method. The events of interest were the
time until a greater than 10% decline in weight and a
greater than 25% decline in CD4 cell count compared
with the maximum level attained after initiation of
antiretroviral drugs. The survival distributions were
compared by the log rank test. CD4 cell count and
weight changes over time were analyzed via linear and
non-linear mixed-effects models . A change-point
model was fitted to the square root of CD4 cell counts to
account for the expected rapid increase immediately after
initiation of therapy followed by a less-steep increase
thereafter [19,20]. The location of the change point in
time was allowed to vary and the one best reflecting the
data was determined through a search among candidate
points. The baseline square root CD4 cell count and the
slope before and after the change point were allowed to
the model) while the effect of other explanatory factors
such as gender and adherence was considered fixed. An
exponential growth model was fitted to weight data with
three components reflecting initial weight, final weight,
and rate of weight increase. Interaction of all three with
other model predictors was explored. A patient-level
random initial weight, final weight, and rate of increase
were included. The association between CD4 cell count
a predictor the most recent CD4 cell count prior to the
10% or 20% decline in weight or the last CD4 cell count
for those patients who did not experience an event (these
were censored at their last visit). Statistical significance in
the Cox model was assessed by Wald-type tests. Analyses
were performed by the SAS system version 9.1 (SAS
Institute, Cary, North Carolina) and S plus version 6.2
(Insightful, Seattle, Washington State, USA).
As of September 2004, 2059 non-pregnant adult patients
initial clinic visit to initiation of antiretroviral therapy was
6.9 weeks [95% confidence interval (CI), 6.0–7.1].
Patient characteristics at the first clinic visit are given in
Table 1. Median age at enrollment was 37 years [trimmed
range (TR): 24.4–57] and approximately 60% of the
participants were women. Among patients with available
sociodemographic data, approximately two-thirds were
married, one third were employed outside the home, and
> 90% had attended school. The median duration of
school attendance was 9 years (TR, 0–16). Sixty-three
percent of individuals had disclosed their HIV status to
someone, with 25% having disclosed to their sexual
partner. The Moi Hospital HIV Clinic (clinic 1)
accounted for roughly two-thirds of study patients; the
majority of the remaining patients were cared for at the
Mosoriot Rural Health Center (clinic 2), the first rural
As noted in Table 2, one third of patients presented with
asymptomatic disease; however the median CD4 cell
count was only 86 cells/ml (TR, 2–395). Women had
significantly lower WHO stage and a higher CD4 cell
count than men (P < 0.001). The median weight for
both men and women at enrollment was < 60 kg.
As of February 2005, the median duration of follow-up
after initiation of antiretroviral therapy was 40 weeks
within 3 months prior to antiretroviral drugs initiation)
were available for 1639 patients, with a median CD4 cell
count of 82 cells/ml (TR, 2–378). Approximately one
quarter (27.2%) of this cohort received treatment for
tuberculosis during the study period. There were 111
(5.4%) documented deaths and 505 (24.5%) patients were
lost to follow-up (no visit for at least 3 months). The
Antiretroviral therapy in sub-Saharan Africa Wools-Kaloustian et al.43
1- and 2-year estimated loss-to-follow-up rates were
22.4% and 29.7%, respectively. These differed signifi-
cantly between clinics, approaching 30% at 1 year in
clinics 1 and 3, while the rates in clinics 2 and 4 were less
than half of that (P < 0.001). The more distant rural
clinics were not included in this analysis as they had been
open for a much shorter period and were responsible for
only 6.2% of cohort patients.
By February 2005, 1766 patients with adherence data had
been enrolled. More than one adherence assessment was
available for 92% of these patients, and 78% reported
perfect adherence at all assessments. Of the 22% who
reported taking less than all of their medications
(imperfect adherence) at oneor morevisits,85% reported
problems at only one visit. Adherence rates increased as
the duration of follow-up increased, with reports of
perfect adherence from approximately 93% of patients
followed for at least 1 month and nearly 100% of patients
followed for 38 months.
Longitudinal analysis of patient weight and CD4
Using a non-linear growth model, weight increased
significantlyafter initiating antiretroviraltherapyandthen
tapered over time (Fig. 1). The average increase in weight
over the 3 years of observation was 4.42 kg (P < 0.001).
Weight loss of > 10% from maximal weight was
experienced by 27% and of > 20% by 5% of all patients
within 1 year after antiretroviral drug initiation. The
median time to a 10% weight loss was 88 weeks (95% CI,
The median time to the initial CD4 cell count
measurement after starting antiretroviral drugs was 34
weeks. CD4 cell counts rose rapidly for the first 6 weeks
after therapy initiation (prior to the change point),
resulting in an overall mean CD4 cell increase of
109 cells/ml (Fig. 2). CD4 cell count increased more
slowly thereafter, resulting in estimated increases of 160,
225 and 297 cells/ml at 12, 24, and 36 months,
respectively. Based on the Kaplan–Meier analysis of 3
years of observation, < 50% of patients experienced a
presumed antiretroviral failure (greater than 25% decline
from maximal CD4 cell count after initiation of
antiretroviral drugs; Fig. 3).
44 AIDS 2006, Vol 20 No 1
Table 1. Sociodemographics at enrollment.
Median age [years (trimmed range)] (n ¼ 1878)
Female [No. (%)]
Ethnic groups [No. (%)] (n ¼ 2019)
Married [No. (%)] (n ¼ 2035)
Employed [No. (%)] (n ¼ 1405)
Attended school [No. (%)] (n ¼ 1430)
No. years [median (trimmed range)] (n ¼ 1393)
Disclosure of HIV status [No. (%)] (n ¼ 1379)
Other family member
Other household member
Clinic site [No. (%)] (n ¼ 2059)
Clinic 1 (Urban)
Trimmed range, 2.5th and 97.5th percentile of available observation.
Table 2. Clinical characteristics at enrollment by gender.
Clinical characteristicTotalMale FemaleP valuea
WHO stage [No. (%)]b
CD4 cell count
Median [cells/ml (trimmed range)]
Median [kg (trimmed range)]
Body mass index
Median (trimmed range)
86 (2–395)77 (1–366) 91 (2–430)
2035 804 1231
54 (36.0–80.5)56.6 (41.0–80.2)51 (35.0–80.5)
19.2 (13.7–28.4)18.9 (14.0–26.7)19.5 (13.5–29.5)
Trimmed range, 2.5thand 97.5thpercentile of available observations.
aP value from Kruskal–Wallis test except for WHO stage, which was from Fisher’s exact test.
bCalculated from intake data when clinician assessment was unavailable.
CD4 cell count was a statistically significant predictor of
subsequent10%and 20%weight loss(P < 0.001 inboth).
Among two individuals with a 100 cell/ml count
difference, the individual with the lower CD4 cell count
had one-third higher risk of a 10% weight decline and a
52% greater risk for a 20% weight decline. The risk of
death for a patient with a CD4 cell count at the time of
treatment initiation < 100 cells/ml was over three times
higher than that for a patient with CD4 cell count
> 100 cells/ml (log rank test P < 0.001).
A few patients (4.9%) initiated an antiretroviral regimen
other than the three-drug standard but were subsequently
converted to the standard regimen. A regimen change
fromthestandardwasmadein 242(11.9%)patients aftera
median of 150 weeks. The time to initiation of second-
line therapy was slightly shorter for patients who initiated
P ¼ 0.073).
The effect of adherence on patient outcomes
Patients with imperfect adherence experienced some-
what lower average weight gains (1.04 kg less) than those
reporting perfect adherence (P ¼ 0.059; Fig. 1). Perfectly
adherent patients were significantly less likely to experi-
ence a 10% decline in weight compared with those with
imperfect adherence (P ¼ 0.011).
Antiretroviral therapy in sub-Saharan Africa Wools-Kaloustian et al.45
Fig. 1. Non-linear growth model for weight in men and
women. The estimated mean weight is given for perfect
adherence (–—) and imperfect adherence(- - -).
0 20 4060 80100 120140160 180
CD4 cell count (cells/µl)
Time after starting therapy (weeks)
Fig. 2. Linear change-point model of CD4 cell counts by
adherence over time. The raw means over time are given as
points connected into a curve; the predicted value from the
change-point model is given as a single line. The two sets of
data are given for perfect adherence (–—) and imperfect
adherence (- - -).
0 2550 75100125150 175
Patients without treatment failure (%)
Time after starting therapy (weeks)
Fig. 3. Kaplan–Meier analysis of the time from antiretroviral therapy initiation to a greater than 25% decline in CD4 cell count
(considered as treatment failure) among patients reporting perfect and imperfect adherence.
Adherence was a strong predictor of overall CD4 cell
response. During the first 6 weeks of therapy, a mean
increase of 112 cells/ml was seen among perfectly
adherent patients compared with 106 cells/ml among
those reporting imperfect adherence (P ¼ 0.032; Fig. 2).
Thereafter, the rate of CD4 cell increase in this latter
group was roughly half that seen in the perfectly adherent
group, giving rise to a CD4 cell increase at 1 year that was
43 cells/ml lower than the increase in the perfectly
adherent group (P < 0.001; Fig. 2). Imperfect adherence
was also associated with shorter time until a presumed
antiretroviral therapy failure (P < 0.001; Fig. 3).
The effect of gender on patient outcomes
Males were significantly more likely to be lost to follow-
up than were females (P ¼ 0.004). Among individuals
continuing care, gender was not associated with level of
adherence. Men were heavier by an average of 5.35 kg
and had a significantly faster rate of weight gain than
did women (P < 0.001; Fig. 1). CD4 cell count increases
were greater in women than men during the initial 6
weeks of antiretroviral therapy (P ¼ 0.007). After the
initial 6 weeks, women experienced CD4 cell count
increases nearly double those of men (P ¼ 0.003). Time
to antiretroviral therapy failure was also slightly longer in
women than in men (P ¼ 0.082).
We have established that it is possible to administer
antiretroviral therapy and to achieve substantial clinical
responses within a public sector setting in western Kenya.
This cohort has demonstrated clinical benefit in terms of
both CD4 cell count and weight increases well into the
third yearof follow-up. In the absence of viral load data, a
direct comparison with previous studies is difficult.
However, CD4 cell increases attained on antiretroviral
therapyareconsistent with those seen elsewhere [21–23].
Our findings of a rapid rise in CD4 cell counts during the
initial few weeks of therapy, followed by a slower rise, are
consistent with earlier reports on CD4 cell kinetics
[19,20]. In addition, the median CD4 cell increase seen at
52 weeks after antiretroviral therapy initiation is similar to
the response seen in studies conducted in the United
States and Europe [21–23]. One limitation of this analysis
is that regimen changes were not taken into account. As
such, the mean CD4 cell increases seen at later time
periods reflect responses to both primary and secondary
regimens (as 11.9% of our patients switched to second-
Previous clinical trials assessing durability of response to
regimens based on non-nucleoside reverse transcriptase
inhibitors (NNRTI) have used either detectable HIV
RNA or the combination of detectable HIV RNA and
regimen change as the outcome measure defining treat-
ment failure [7,24,25]. The 3-year treatment failure rates
documented in these studies range from 20 to 48%.
used to define treatment failure in our cohort. This
outcome likely follows virological failure by several
months, which may indicate that our population is
experiencing treatment failure (as defined by detectable
HIV RNA) at a slightly earlier point than seen in prior
studies. This is not unexpected given that the patients
described here are part of a clinic cohort rather than
participants in a clinical trial. As such, the median time to
failure of an NNRTI-based first-line therapy within a
public sector clinic in sub-Saharan Africa appears to be
approximately 3 years.
Because of the cross resistance between the non-
nucleoside reverse trancriptase inhibitors, second-line
regimens will need to be protease inhibitors based. In
reverse transcriptase inhibitors with minimal cross resist-
ance to those used in the initial regimen will also be
required. Use of a protease inhibitor in the second-line
regimen has significant cost and monitor implications that
in the developing world. We believe that viral load data
would be invaluable for the early identification of patients
experiencing antiretroviraltherapy failure and wouldassist
in preventing the development of multiple drug resistance
mutations. However, at current cost, this technology is
such as ours. Therefore, we strongly support development
of low-cost technologies for viral load determination as
part ofthepolicyfor providingantiretroviraldrugs in sub-
Since measurements of CD4 cell counts are not routinely
available in resource-poor settings, and because weight is
an independent predictorof HIV progression, weight was
assessed as one of the clinical outcomes [26–28]. We
found that mean weight gains mirrored CD4 cell
increases immediately following antiretroviral therapy
initiation. However, weight was found to have a within-
subject coefficient of variation of 20%. This may have
contributed to the discrepancy seen between our
finding of consistent mean weight increases throughout
the study period and the Kaplan–Meier model, which
demonstrated that a significant percentage of the study
cohort experienced a greater than 10% decline in weight.
We initially attributed the substantial variability in weight
to patients’ inability to access food consistently. However,
further analysis showed that weight declines correlated
with CD4 cell count, indicating that the at least some of
the variability in weight reflects variations in clinical
status. Because of the discrepancies between the two
models, noted above, we believe that, in the absence of
a reliable food supply, weight should be used with
caution in assessing individual responses to antiretroviral
46 AIDS 2006, Vol 20 No 1
Studies conducted in the United States have shown that
adherence to an antiretroviral regimen is a significant
predictor of viral suppression and clinical response
[29,30]. More than three-quarters of our patients claimed
single item to assess patient adherence was predictive of
immunological and clinical responses in this cohort. This
finding warrants further exploration, with particular
attention to the sensitivity of brief adherence assessments
in a busy clinic setting.
We acknowledge that the 24.5% loss to follow-up rate for
our study population might have had a significant impact
on our findings. In the presence of a passive reporting
system for deaths, we believe that much of our loss to
follow-up is from unreported deaths. This assumption
is based on the low death rate (5.4%) for this population
in which 55% of individuals have WHO stage 3 or
4 disease. Other likely reasons for loss to follow-up in-
clude patients moving outside the catchment area and
problems with transportation access. Deaths likely re-
present poor responses to therapy, which, if captured,
would negatively impact both weight and CD4 cell
trends. This could result in the introduction of a poten-
tially serious bias and possible overestimation of both
CD4 cell count and weight response to therapy. The
up for reasons other than death is difficult to predict.
Because the median duration of follow-up for this cohort
is approximately 1 year, all models discussed above are
later time periods.
Despite the limitations of this study, we have documented
significant clinical and immunological responses in this
population that appear to persist for at least 3 years
after initiation of antiretroviral therapy. These findings
provide significant support for the viability of large-scale
HIV treatment initiatives in resource-limited settings.
However, in order to ensure the long-term success of
treatment programs, there remains a significant need to
identify feasible and reliable tools to assess adherence,
methods to identify and locate patients who are lost to
follow-up, low-cost monitoring technologies, and ratio-
nal second-line drug regimens.
We thank the staff and patients of the AMPATH HIV
Clinics and our collaborators from Indiana University,
Brown University, Columbia University, and the MTCT-
Sponsorship: The Purpleville Foundation, MTCT-plus,
the United States Agency for International Develop-
ment (USAID) and the President’s Emergency Plan for
AIDS Relief (PEPFAR) funded patient care activities
including the purchase of antiretroviral drugs. The
Rockefeller Foundation and the Fogarty International
Center (grant 1-D43-TW01082) contributed funding to
the development of the Electronic Medical Records
1. UNAIDS. Report on the Global HIV/AIDS Epidemic: Fourth
Global Report. Geneva: World Health Organization; 2004.
Steinbrook R. The AIDS epidemic in 2004. N Engl J Med 2004;
AVANTI 2. Randomized, double-blind trial to evaluate the
efficacy and safety of zidovudine plus lamivudine versus
zidovudine plus lamivudine plus indinavir in HIV-infected
antiretroviral-naive patients. AIDS 2000; 14:367–374.
Hammer SM, SquiresKE, Hughes MD, Grimes JM, DemeterLM,
Currier JS, et al. A controlled trial of two nucleoside analogues
plus indinavir in persons with human immunodeficiency virus
infection and CD4 cell counts of 200 per cubic millimeter or
less. AIDS Clinical Trials Group 320 Study Team. N Engl J Med
Montaner JS, Reiss P, Cooper D, Vella S, Harris M, Conway B,
et al. A randomized, double-blind trial comparing combina-
tions of nevirapine, didanosine, and zidovudine for HIV-
infected patients: the INCAS Trial. Italy, the Netherlands,
Canada and Australia Study. JAMA 1998; 279:930–937.
Coetzee D, Hildebrand K, Boulle A, Maartens G, Louis F,
Labatala V, et al. Outcomes after two years of providing
antiretroviral treatment in Khayelitsha, South Africa. AIDS
Laurent C, Ngom Gueye NF, Ndour CT, Gueye PM, Diouf M,
Diakhate N, et al. Long-term benefits of highly active antire-
troviral therapy in Senegalese HIV-1-infected adults. J Acquir
Immune Defic Syndr 2005; 38:14–17.
Weidle PJ, Malamba S, Mwebaze R, Sozi C, Rukundo G,
Downing R, et al. Assessment of a pilot antiretroviral drug
therapy programme in Uganda: patients’ response, survival,
and drug resistance. Lancet 2002; 360:34–40.
DesclauxA, Ciss M,TaverneB, Sow PS, EgrotM, FayeMA, et al.
Access to antiretroviral drugs and AIDS management in
Senegal. AIDS 2003; 17 (Suppl 3):S95–S101.
Harries AD, Nyangulu DS, Hargreaves NJ, Kaluwa O,
Salaniponi FM. Preventing antiretroviral anarchy in sub-
Saharan Africa. Lancet 2001; 358:410–414.
Einterz RM, Kelley CR, Mamlin JJ, van Reken DE. Partner-
ships in international health. The Indiana University-Moi
University experience. Infect Dis Clin North Am 1995; 9:
Mamlin JJ, Kimaiyo S, Nyandiko W, Tierney WM (eds).
Academic institutions linking access to treatment and preven-
tion: Case study. Geneva: World Health Organization; 2004.
Voelker R. Conquering HIV and stigma in Kenya. JAMA 2004;
Siika A, Rotich J, Simiyu C, Kigotho E, Smith F, J. S, et al.
An electronic medical record system for ambulatory care of
HIV-infected patients in Kenya. Int J Med Informat 2005;
Cohen J, Kimaiyo S, Nyandiko W, Siika A, Sidle J, Wools-
Kaloustian K, et al. Addressing the educational void during
the antiretroviral therapy rollout. AIDS 2004; 18:2105–2106.
World Health Organization. Chronic HIV Care with ARV
Therapy: Integrated Management of Adolescent and Adult
Illness. Geneva: World Health Organization; 2004.
Hughes MD, Stein DS, Gundacker HM, Valentine FT, Phair JP,
Volberding PA. Within-subject variation in CD4 lymphocyte
count in asymptomatic human immunodeficiency virus infec-
tion: implications for patient monitoring. J Infect Dis 1994;
Pinheiro J, Bates D. Mixed Effects Models in S and S plus. New
York: Springer; 2000.
Lange N, Carlin BP, Gelfand AE. Hierarchical Bayes models for
the progression of HIV infection using longitudinal CD4 T-cell
numbers. J Am Statist Assoc 1992; 87:615–626.
Antiretroviral therapy in sub-Saharan Africa Wools-Kaloustian et al.47
20. Abrams DI, Goldman AI, Launer C, Korvick JA, Neaton JD, Download full-text
CraneLR,etal.A comparative trialof didanosineor zalcitabine
after treatment with zidovudine in patients with human
immunodeficiencyvirus infection. TheTerry BeirnCommunity
Programs for Clinical Research on AIDS. N Engl J Med 1994;
Staszewski S, Morales-Ramirez J, Tashima KT, Rachlis A, Skiest
D, Stanford J, et al. Efavirenz plus zidovudine and lamivudine,
efavirenz plus indinavir, and indinavir plus zidovudine and
lamivudine in the treatment of HIV-1 infection in adults.
Study 006 Team. N Engl J Med 1999; 341:1865–1873.
Smith CJ, Sabin CA, Youle MS, Kinloch-de Loes S, Lampe FC,
Madge S, et al. Factors influencing increases in CD4 cell counts
of HIV-positive persons receiving long-term highly active anti-
retroviral therapy. J Infect Dis 2004; 190:1860–1868.
van Leth F, Phanuphak P, Ruxrungtham K, Baraldi E, Miller S,
Gazzard B, et al. Comparisonof first-line antiretroviral therapy
with regimens including nevirapine, efavirenz, or both drugs,
plus stavudine and lamivudine: a randomised open-label trial,
the 2NN Study. Lancet 2004; 363:1253–1263.
Shafer RW, Smeaton LM, Robbins GK, de Gruttola V, Snyder
SW, D’Aquila RT, et al. Comparison of four-drug regimens and
pairs of sequential three-drug regimens as initial therapy for
HIV-1 infection. N Engl J Med 2003; 349:2304–2315.
25. Tashima K, Staszewski S, Nelson M, Rachlis A, Skiest D, Stryker
R, et al. Durable viral suppression on EFV-based HAART: 168
weeks of follow-up. XV International Conference on AIDS.
Bangkok, Thailand 2004 [abstract TuPeB4547].
Palenicek JP, Graham NM, He YD, Hoover DA, Oishi JS,
Kingsley L, et al. Weight loss prior to clinical AIDS as a
predictor of survival. Multicenter AIDS Cohort Study Investi-
gators. J Acquir Immune Defic Syndr Hum Retrovirol 1995;
Wheeler DA. Weight loss and disease progression in HIV
infection. AIDS Read 1999; 9:347–353.
Mwamburi DM, Wilson IB, Jacobson DL, Spiegelman D,
Gorbach SL, Knox TA, et al. Understanding the role of HIV
load in determining weight change in the era of highly active
antiretroviral therapy. Clin Infect Dis 2005; 40:167–173.
Mannheimer S, Friedland G, Matts J, Child C, Chesney M. The
consistency of adherence to antiretroviral therapy predicts
persons in clinical trials. Clin Infect Dis 2002; 34:1115–1121.
Paterson DL, Swindells S, Mohr J, Brester M, Vergis EN, Squier
C, et al. Adherence to protease inhibitor therapy and outcomes
in patients with HIV infection. Ann Intern Med 2000; 133:
48 AIDS 2006, Vol 20 No 1