In Sub-Saharan Africa, AIDS-related morbidity and mortality are the highest in the world, mainly due to the spread of the HIV-1 epidemic and the limited access to diagnosis and treatment.1 Observations on the efficacy of antiretroviral treatment (ART) in low-income to middle-income countries (LMIC) show encouraging results in the short and medium term.2,3 There is not yet consensus on how ART efficacy should be monitored in these settings. In most ART programs, diagnosis of treatment failure is solely based on clinical criteria or CD4 cell counts: monitoring of plasma HIV RNA, a standard of care in high-income countries, is rarely employed. Although the addition of CD4 counts to clinical monitoring was associated with better survival in ART programs,4,5 the monitoring of plasma HIV RNA has not been associated with a significant additional benefit on survival and was not found to be reasonably cost-effective.4,6,7 Nonetheless, ART program monitoring strategies without HIV RNA are associated with considerable drug resistance accumulation.8,9 Moreover, by a rational use of viral load monitoring, unnecessary switches to expensive second-line regimens may be prevented.2,10–16
Observational cohort studies from Europe and North America showed that the initial response to treatment is strongly predictive of clinical progression. In particular, both the HIV RNA level and the CD4 count at 6 months after treatment initiation were independently associated with subsequent AIDS events and death either in treatment-naive or experienced patients.17–19 To date, no such analysis has been attempted in patients from ART programs in LMIC.
In this study, we show the results of the association of viral load responses at 6 months after starting a first-line ART with patients survival, loss to follow-up, and initiation of a second-line regimen in a cohort of patients from sub-Saharan African ART sites. We also show the estimated survival probability according to the 6-month CD4 and viral load categories, for up to 2 years thereafter.
PATIENTS AND METHODS
Study Sites and Dream Program
Patients were selected from 5 sites from the public sector all of which were in partnership with the Dream program. Two sites are in Mozambique (the Machava center in Matola city, Maputo province, and the Manga Chingoussura center, Sofala province), 2 urban sites in Malawi (in Blantyre and in Lilongwe), and 1 in Guinea (in the city of Conakry). Details of the characteristics of the Dream program have been described elsewhere.2 Briefly, in this free-of-charge program, patients starting ART undergo clinical monitoring after 2 and 4 weeks, at month 3, and at 3-months intervals thereafter, HIV RNA measurements are performed every 6 months and CD4 cell counts every 3 months. Pharmacy appointments for medication pick-up are scheduled at day 1 of ART initiation, every 2 weeks during the first month, and on a monthly basis thereafter. Computer-based verification of patients appointment-keeping (for clinical visits and pharmacy drug pick-ups) takes place at each of the selected sites.
Criteria for patients inclusion in the analysis were to be ≥15 years old, nonpregnant, naive to ART, having started a first-line ART regimen [containing 2 nucleoside reverse transcriptase inhibitor (NRTI) and a non-NRTI or 3 NRTI] with available pretherapy laboratory values, and having a 6-month visit, defined as an available CD4 count or HIV RNA measure or a recorded clinic visit attendance around 6 months (in a time window of 12–36 weeks), plus a subsequent follow-up after the 6-month visit. With more than 1 visit in the 6-month time window available, the closest to week 24 was chosen.
This study was approved by the Ministries of Health from Mozambique and Guinea Conakry and by the National Health Sciences Research Committee of the Ministry of Health in Malawi.
Variables Collected and Outcomes Definitions
Allowing for a maximum of 180 days before ART initiation, the last pretreatment available, hemoglobin, body mass index (BMI), CD4 count and HIV RNA (the same method over the study period and throughout the sites: detection limit 50 copies/mL, bDNA assay, Versant 3.0, Siemens Diagnostics), and type of first-line regimen employed were retrieved for analysis. Clinical stage was electronically recorded by the sites only from December 2005, so it is available only for a subset of the patients. Medication adherence was indirectly inferred by counting the missed scheduled clinical visits and the missed pharmacy appointments, allowing for 7 days of delay. Scheduled and actual dates of clinical visits and pharmacy pick-ups were electronically recorded, so that missed appointments could be easily calculated comparing dates of actual and scheduled appointment. The 7 days delay was allowed given that during the program, antiretrovirals were always delivered in excess so that the patients escort could last up to 7 days after the following scheduled pharmacy appointment, to account for unforeseen events or transportation difficulties. The proportion of missed appointments was dichotomized using thresholds at 80% and 100% (for the first 6 months of follow-up) and at 95% (for the whole follow-up). The date of last follow-up, the reasons for follow-up interruption (death, loss to follow-up, transfer-out), and the complete treatment history were also retrieved, along with the date of the switch to a second-line ART, defined by the initiation of a protease inhibitor in the regimen.
Deaths were ascertained using information from hospitals, families, and by actively searching for patients not showing up using phone calls and visits by the home care teams of the sites. The date of loss to follow-up was defined as the date the patients were last seen before being lost for >3 months, unless they returned later.
HIV RNA values were log10 transformed. The predictors of the 6-month treatment virologic response (an HIV RNA value of <1000 copies/mL) and immunologic response (an increase from baseline CD4 counts of ≥100 cells/μL) were analyzed by univariable and multivariable logistic regression. For time-to-event analyses, patients follow-up started from the date of the 6-month visit (using, in order of priority, the date of the HIV RNA measurement, of the CD4 count, and of the clinical visit). Separate analyses were employed for each of the following outcomes: death from all causes, loss to follow-up, and initiation of a second-line regimen. For patients not reaching a specific outcome, analysis time ended at the earliest of the following: the date of the most recent clinic or pharmacy attendance; date of transfer-out; June 30, 2007. Kaplan–Meier estimates of patient survival were employed using different combinations of 6-month HIV RNA and CD4 categories, assessing differences with a log-rank test. Univariable and multivariable Cox proportional hazards models were used to identify hazard ratios of the association of the tested variables with the outcome events, after stratifying by clinical site. All analyses were carried out using the SPSS v. 18 software package (SPSS Inc, Chicago, IL).
Patients Characteristics, ART Regimens, and 6-Month Virological and Immunological Outcomes
Patient selection started from 3660 individuals initiating ART with at least 1 follow-up visit, included in a previous analysis2 as follows: of these, 268 died, 29 were lost to follow-up, 52 were transferred-out to another site before reaching the 6-month window. Another 555 patients did not reach the 6-month window for insufficient follow-up. Of the 2756 remaining patients, 2539 had a subsequent follow-up and were included in this analysis (see Flow Chart, Supplemental Digital Content 1, http://links.lww.com/QAI/A278). Females were 1585 (62.4%). Median calendar month of ART initiation was March 2006 (IQR: October 2004 to September 2006). At that time, patients median age was 34 years (IQR: 28–41), 479 of 1569 (30.5%) were on WHO stage 3 or 4 (970 unknown), their median CD4 was 215 cells per microliter (IQR: 108–328), with 23.0% having baseline CD4 <100 and 46.1% <200 cells per microliter. Patients median pre-ART HIV RNA was 4.61 log10 copies per milliliter (IQR: 3.86–5.09), median hemoglobin concentration 10.9 g/dL (IQR: 9.3–12.3), and median BMI 20.58 kg/m2 (IQR: 18.65–23.01). The employed first-line regimens were stavudine + lamivudine + nevirapine in 1545 (60.9%); zidovudine + lamivudine + nevirapine in 883 (34.8%), zidovudine or stavudine + lamivudine + efavirenz in 41 (1.6%); zidovudine or stavudine + lamivudine + abacavir in 34 (1.3%); other in 28 (1.1%). During follow-up, 479 of 2525 patients (19.0%) missed 5% or more drug pick-up appointments (14 had missing data), although 688 of 2539 patients (27.1%) missed 5% or more clinical visit appointments. At 6 months after ART initiation (median 181 days, IQR: 171–185), the median HIV RNA levels (n = 1913; 626 had missing data) were 1.69 log10 copies/mL (IQR: 1.69–2.13) and the median change from baseline HIV RNA was −2.60 log10 copies/mL (IQR: −0.78 to −3.21;); 85.0% achieved <1000 copies per milliliter and 60.4% <50 copies per milliliter. The 6-month median CD4 cell count was 334 cells per microliter (IQR: 215–466) and the median change from baseline CD4 cell count was +112 cells per microliter (IQR: +35 to +194).
Predictors of 6-Month Virological and Immunological Outcomes
The analyses of the pre-ART factors associated with 6-month viral load response and higher 6-month CD4 cell count recovery are summarized in Table 1. At multivariable analysis, a 6-month HIV RNA <1000 copies per milliliter was independently predicted by older age, lower baseline HIV RNA, higher baseline CD4 counts, and 100% on time attendance to pharmacy appointments during the first 6 months. A better 6-month CD4 cell count recovery was independently predicted by higher pre-ART CD4 counts and viral load and by a less advanced clinical stage at baseline; a higher visit attendance during the first 6 months showed a trend toward a better CD4 recovery.
Death and Concomitant Conditions After the 6-month Measures and Predictors of Death
After the 6-month visit, during a cumulative follow-up of 3112 person-years, 91 patients died from any cause (mortality rate 2.9 per 100 person-years follow-up). The 91 deaths occurred during a hospitalization in 17 cases, at home in 41 cases and in a nonspecified location in the remaining 33. The causes of death could not be specifically investigated in this cohort. We attempted to search all potentially life-threatening conditions included in the patient's electronic medical records during the 6 months (in case of chronic conditions) or 2 months (in case of acute events) before the recorded date of death. In 42 cases (45%), no potentially life-threatening morbid condition was recorded, in 7 cases (8%), a WHO stage 3/4 was recorded without further specifications; in 20 (22%), there was a diagnosis of tuberculosis (in 3 cases with concomitant cancer of different types, in 1 case with severe diarrhea, and 1 case with liver failure); pneumonia was recorded in 4 cases; there was a record of Kaposi sarcoma, of deaths due to violent causes severe diarrhea/wasting each in 3 cases; malaria, cardiovascular disorders, unspecified cancers, unspecified severe infection, and liver failure were recorded each in 2 cases; renal disorder, non-Hodgkin lymphoma, surgery complications, and unspecified allergic reaction each in 1 case.
At univariable analysis, WHO stage 4 as compared with 1 or 2, a 6-month viral load ≥10,000 as compared with <50 copies per milliliter and 6-month CD4 strata <200 versus >500 cells per microliter were associated with higher risk of death, whereas female gender, higher baseline CD4, hemoglobin and BMI levels, and attendance to >95% of pharmacy appointments were associated with a lower risk of death. Baseline HIV RNA, analyzed as a continuous variable did not show any association with death. Table 2 shows the adjusted hazard ratios for death from any cause from fitting a multivariable model. Factors independently associated with death from all causes were the 6 months HIV RNA levels, along with the gender, the pharmacy appointments attendance, the pre-ART WHO AIDS stage, and hemoglobin level. Patients with 6-month missing CD4 counts also showed an independent association with higher mortality, whereas CD4 values <50 cells per microliter in that time window showed a trend toward a higher hazard of death. Of note, lower 6 month CD4 cell counts and higher HIV RNA levels were both independently predictive of all-cause mortality in a model excluding the respective baseline values (not shown).
The prognostic value of the 6-month HIV RNA and CD4 measures was further analyzed by estimating the probability of survival at several time points after the 6-month HIV RNA measurement, according to different viral load and CD4 strata (Fig. 1). For this analysis, only patients with an available 6-month HIV RNA and CD4 count were used. In the stratum with 6 month CD4 count <200 cells per microliter, a 6-month viral load >1000 copies per milliliter was already predictive of a decreased estimated survival thereafter (Fig. 1). The estimated probabilities of surviving 24 months after the 6-month HIV RNA ranged from 0.69 (95% confidence interval: 0.68 to 0.70) in the stratum with 6-month CD4 <200 cells per microliter and HIV RNA ≥10,000 copies per milliliter to 0.95 (95% confidence interval: 0.94 to 0.96) in the stratum with 6-month CD4 ≥200 cells per microliter and HIV RNA <1000 copies/mL.
Patients Drop-Out Rate After the 6-Month Measures and Its Predictors
After the 6-month visit, during a cumulative follow-up of 3112 person-years, 71 patients were lost to follow-up (2.3 per 100 person-years follow-up). The analysis of the association of variables with the risk of loss to follow-up is summarized in Table 3. At univariable analysis, 6-month viral load ≥10,000 copies per milliliter as compared with <50 copies per milliliter, lower attendance to pharmacy drug pick-up appointments, and younger age were associated with higher hazards for being subsequently lost to follow-up. In the multivariable analysis age, attendance to visits and drug pick-up appointments, and 6-month viral load were independently associated with the subsequent patients retention in the program.
Start of a Second-Line Regimen After the 6-month Measures and Its Predictors
After the 6-month visit, 206 patients were switched to a second-line regimen during 2785 person-years follow-up (7.4 per 100 person-years follow-up). Analysis of the predictors of the time to switch to a second-line regimen is summarized in Table 4. At both univariable and multivariable analysis, switch to second-line ART was inversely associated with more recent calendar year of ART initiation and predicted by the 6-month CD4 count<200 cells per microliter and HIV RNA >50 copies per milliliter.
In this cohort of patients treated with a standard first-line regimen in sub-Saharan Africa, we first showed that predictors of 6-month virological and immunological responses were the respective baseline values of HIV RNA and CD4 counts; virological response was also predicted by the attendance to pharmacy appointments, an indirect measure of patient's adherence to the ART program. This was previously observed on this cohort,20 and it underscores the importance of adherence monitoring and implementation in ART sites in LMIC.
More importantly, this study shows that an HIV RNA level ≥10,000 copies per milliliter at 6 months of combination ART was an independent predictor of subsequent mortality. Concomitantly, the 6-month CD4 cell count strata were also independently predictive of patient's survival. Nonetheless, when both baseline CD4 and HIV RNA data were included in the prediction model, only 6-month viral load remained independently predictive of subsequent survival. This confirms previous similar analyses conducted in developed countries.17–19 New in this analysis, indirect measures of patient's adherence were also analyzed. Among these, lower attendance to pharmacy appointments was also independently predictive of subsequent death from any cause. It is important to notice that the 6-month viral load predicted progression to death independently from this adherence measure, suggesting that the viral load monitoring adds clinically relevant prognostic information in patients on ART. To our knowledge, this is the first study showing the prognostic value of follow-up viral load after ART initiation in LMIC. A single study performed in South Africa identified that the time-updated CD4 count after ART initiation is strongly predictive of mortality,21 but viral load responses were not analyzed.
The viral load threshold for predicting mortality identified in this cohort (HIV RNA ≥10,000 copies/mL) corresponds to an initial WHO-defined criterion for virological failure.22,23 Our finding reassures that this cut-off, initially identified in patients failing protease inhibitors-based ART in high-income countries,24 has a prognostic equivalent also in patients undergoing a standard WHO-recommended first-line regimen in LMIC, an evidence that was still missing.23 More recently, WHO redefined its virological failure threshold at 5000 copies per milliliter.25 It should be noticed that, in specific subgroups exposed to a higher risk, such as those whose CD4 counts at 6 months remained <200 cells per microliter, even an HIV RNA between 1000 and 9999 copies per milliliter was independently predictive of all-cause death. Moreover, patients with a 6-month viral load of ≥1000 copies per milliliter with a concomitant CD4 count <200 cells per microliter showed the worst prognosis in this analysis. Interestingly, in this lower 6-month immunological stratum, the estimated proportion surviving was comparably lower whether the patients had and HIV RNA 1000–9999 or ≥10,000 copies per milliliter. Given the prognostic relevance of this observation, a virological failure could be reconsidered at a lower threshold in patients with concomitant CD4 counts lower than 200 cells per microliter. This finding argues in favor of the use of viral load for monitoring in ART programs in LMIC. To optimize the costs of such a monitoring, we propose to evaluate in prospective studies the use of a standard measurement at a fixed time point of around 6 months. In addition, cheaper semiquantitative viral load assays calibrated to detect relevant HIV-1 RNA thresholds could be developed and validated26 along with alternative pooling strategies.27 Moreover, if viral load suppression is obtained and maintained, the risk of virological rebound will decrease over time, maintaining a sufficient adherence.28 As a consequence, ART program monitoring strategies could include viral load determination in the initial period, followed by a less expensive adherence monitoring once virologic control is sustained for a sufficient period.
The 6-month viral load >10,000 copies per milliliter was also independently predictive of subsequent loss to follow-up. Other independent predictors of patient's loss to follow-up after month 6 were younger age and less regular attendance to pharmacy and clinic appointments. Lack of patient retention into care is a major limitation to the success of ART programs in LMIC.29 Active investigations on this population of lost to follow-up identified that a relevant proportion of these patients had died early after the initiation of ART.30 At the clinical sites of the patients used in this study, an active search of patients not showing up at the scheduled clinical appointments takes place through phone calls and home visits. Therefore, the relative proportion of patients dropped from the program was low. In this study population, surviving enough to receive a viral load test 6 months after starting ART, the predictors of being subsequently lost to follow-up were not variables typically associated with mortality (such as CD4 counts and clinical stage), but rather indirect indicators of lower medication adherence, such as measures of appointment keeping, viral load, and younger age.31 We interpret this finding with the fact that the patients dropped from the program after the sixth month typically represent a less adherent population. Specific investigations for the reasons leading to reduced adherence and patients retention in this population are advisable.32 Interestingly, the viral load measure offers an added predictive value of patient's retention, independently from some classic program adherence measures.
The switch to second-line therapy was associated with 6 months CD4 counts <200 cells per microliter, detectable HIV RNA, and earlier calendar year of ART initiation. These findings indicate that classic immunological but also virological criteria were used for patients switch, which is compatible with the comprehensive monitoring of both markers in this cohort. An increasingly prudent attitude over calendar years toward switching to second-line regimens may be due to the higher propensity to adopt a drug-sparing policy and a decreasing confidence toward the efficacy and tolerability of some early second-line regimes. It should be noticed that second-line drugs were steadily available during the development of the program in the study sites, therefore, access limitations should not account for the observed phenomenon. However, in the period analyzed, available second-line regimens were almost exclusively based on unboosted protease inhibitors (not shown), whose efficacy in this context is certainly limited. Moreover, the more conservative attitude toward switching to second-line ART may reflect a wider application of the WHO-suggested criteria of first implementing adherence counseling and repeat viral load testing to avoid unnecessary switches.25 Further investigation is required to assess whether the availability of ritonavir-boosted protease inhibitors at reduced prices33,34 will increase the propensity toward the prescription of second-line drugs in these settings, for patients with any type of failure, including those with exclusive virological failure.
In conclusion, this observational study shows how a single viral load measure at 6 months after standard first-line therapy initiation in a sub-Saharan African setting is a prognostic marker for disease prognosis and subsequent patient retention into care. The detection of a viral load above 1000 or 10,000 copies per milliliter around this period, according to the population subset analyzed, could be employed as a trigger for specific interventions at individual or ART program level.
The authors express their gratitude to Pierre Kondiano, clinical coordinator of the DREAM center in Conakry, and to Elard Alumando, DREAM coordinator in Malawi, for cohort coordination support and to the clinical staff and data managers in the DREAM centers of Machava, Manga Chingoussoura, Lilongwe, Blantyre, and Conakry, for invaluable dedication. Many other public and private organizations support DREAM and private citizens in many European countries. The authors express their gratitude to all of them.
2. Palombi L, Marazzi MC, Guidotti G, et al.. Incidence and predictors of death, retention, and switch to second-line regimens in antiretroviral-treated patients in Sub-Saharan African sites with comprehensive monitoring availability. Clin Infect Dis. 2009;48:115–122.
3. Braitstein P, Brinkhof MW, Dabis F, et al.. Mortality of HIV-1-infected patients in the first year of antiretroviral therapy: comparison between low-income and high-income countries. Lancet. 2006;367:817–824.
4. Coutinho A, Mermin J, Ekwaru J, et al.. Utility of routine viral load, CD4 cell count, and clinical monitoring among HIV-infected adults in Uganda: a randomized trial. Presented at: 15th Conference on Retroviruses and Opportunistic Infections; February 3–6, 2008; Boston, MA. Abstract 125.
5. Dart trial team. Routine versus clinically driven laboratory monitoring of HIV antiretroviral therapy in Africa (DART): a randomized non-inferiority trial. Lancet. 2009;375:123–131.
6. Phillips AN, Pillay D, Miners AH, et al.. Outcomes from monitoring of patients on antiretroviral therapy in resource-limited settings with viral load, CD4 cell count, or clinical observation alone: a computer simulation model. Lancet. 2008;371:1443–1451.
7. Bendavid E, Young SD, Katzenstein DA, et al.. Cost-effectiveness of HIV monitoring strategies in resource-limited settings: A southern African analysis. Arch Intern Med. 2008;168:1910.
8. Ndembi N, Goodall RL, Dunn DT, et al.. Development of Antiretroviral Treatment in Africa Virology Group and Trial Team. Viral rebound and emergence of drug resistance in the absence of viral load testing: a randomized comparison between zidovudine-lamivudine plus nevirapine and zidovudine-lamivudine plus abacavir. J Infect Dis. 2010;201:106–113.
9. Gupta RK, Hill A, Sawyer AW, et al.. Virological monitoring and resistance to first-line highly active antiretroviral therapy in adults infected with HIV-1 treated under WHO guidelines: a systematic review and meta-analysis. Lancet Infect Dis. 2009;9:409–417.
10. Kantor R, Diero L, DeLong A, et al.. Misclassification of first-line antiretroviral treatment failure based on immunological monitoring of HIV infection in resource-limited settings. Clin Infect Dis. 2009;49:454–462.
11. Mee P, Fielding KL, Charalambous S, et al.. Evaluation of the WHO criteria for antiretroviral treatment failure among adults in South Africa. AIDS. 2008;22:1971–1977.
12. Reynolds SJ, Nakigozi G, Newell K, et al.. Failure of immunologic criteria to appropriately identify antiretroviral treatment failure in Uganda. AIDS. 2009;23:697–700.
13. Calmy A, Ford N, Hirschel B, et al.. HIV viral load monitoring in resource-limited regions: optional or necessary? Clin Infect Dis. 2007;44:128–134.
14. Moore DM, Mermin J. Monitoring antiretroviral failure in resource-poor settings. Lancet. 2008;371:1396–1397.
15. Lawn SD, Bekker LG, Calmy A, et al.. Monitoring of antiretroviral therapy in low-resource settings. Lancet. 2008;372:287–288.
16. Elliott JH, Lynen L, Calmy A, et al.. Rational use of antiretroviral therapy in low-income and middle-income countries: optimizing regimen sequencing and switching. AIDS. 2008;22:2053–2067.
17. Chêne G, Sterne JA, May M, et al.. Antiretroviral Therapy Cohort Collaboration. Prognostic importance of initial response in HIV-1 infected patients starting potent antiretroviral therapy: analysis of prospective studies. Lancet. 2003;362:679–686.
18. Grabar S, Le Moing V, Goujard C, et al.. Response to highly active antiretroviral therapy at 6 months and long-term disease progression in HIV-1 infection. J Acquir Immune Defic Syndr. 2005;39:284–292.
19. Lanoy E, May M, Mocroft A, et al.. Prognosis of patients treated with cART from 36 months after initiation, according to current and previous CD4 cell count and plasma HIV-1 RNA measurements. AIDS. 2009;23:2199–2208.
20. San Lio MM, Carbini R, Germano P, et al.. Evaluating adherence to highly active antiretroviral therapy with use of pill counts and viral load measurement in the drug resources enhancement against AIDS and malnutrition program in Mozambique. Clin Infect Dis. 2008;46:1609–1616.
21. Lawn SD, Little F, Bekker LG, et al.. Changing mortality risk associated with CD4 cell response to antiretroviral therapy in South Africa. AIDS. 2009;23:335–342.
22. Gilks CF, Crowley S, Ekpini R, et al.. The WHO public-health approach to antiretroviral treatment against HIV in resource-limited settings. Lancet. 2006;368:505–510.
23. World Health Organization. Antiretroviral Therapy for HIV Infection in Adults and Adolescents: Recommendation for a Public Health Approach. Geneva, Switzerland: World Health Organization; 2006.
24. The PLATO Collaboration. Predictors of trend in CD4-positive T-cell count and mortality among HIV-1-infected individuals with virological failure to all three antiretroviral-drug classes. Lancet. 2004;364:51–62.
25. World Health Organization. Rapid Advice. Antiretroviral Therapy for HIV Infection in Adults and Adolescents. Geneva, Switzerland: World Health Organization. 2009. Available at: http://www.who.int/hiv/pub/arv/rapid_advice_art.pdf
Accessed December 16, 2011.
26. Fiscus SA, Cheng B, Crowe SM, et al.. HIV-1 viral load assays for resource-limited settings. PLoS Med. 2006;3:e417.
27. van Zyl GU, Preiser W, Potschka S, et al.. Pooling strategies to reduce the cost of HIV-1 RNA load monitoring in a resource-limited setting. Clin Infect Dis. 2011;52:264–270.
28. Rosenblum M, Deeks S, Van Der Laan M, et al.. The risk of virologic failure decreases with duration of HIV suppression at greater than 50% adherence to antiretroviral therapy. PLoS ONE. 2009;4:e7196.
29. Rosen S, Fox MP, Gill CJ. Patient retention in antiretroviral therapy programs in sub-Saharan Africa: a systematic review. PLoS Med. 2007;4:e298.
30. Brinkhof WG, Pujades-Rodriguez M, Egger M. Mortality of patients lost to follow-up in antiretroviral treatment programmes in resource-limited settings: systematic review and meta-analysis. PLoS ONE. 2009;4:e5790.
31. Nachega JB, Hislop M, Nguyen H, et al.. Antiretroviral therapy adherence, virologic and immunologic outcomes in adolescents compared with adults in southern Africa. J Acquir Immune Defic Syndr. 2009;51:65–71.
32. Losina E, Toure` T, Uhler LM, et al.. Cost-effectiveness of preventing loss to follow-up in HIV treatment programs: a Cote d'Ivoire appraisal. PLoS Med. 2009;6:e1000173.
34. World Health Organizations. Transaction Prices for Antiretroviral Medicines and HIV Diagnostics From 2008 to October 2009. A Summary Report From the Global Price Reporting Mechanism. Geneva, Switzerland: WHO: 2009. Available at:http://www.who.int/hiv/amds/GPRMsummaryReportNov2009.pdf
Accessed December 16, 2011.