Patients with antiretroviral treatment (ART) failure are increasingly encountered in resource-limited settings.1–3 Rates of switching to second-line ART are low, and concerns have been raised that patients may be experiencing long periods with virologic failure.4,5
In settings without access to resistance tests, evidence is needed to refine the monitoring strategy for detecting treatment failure and switching regimens.6 Laboratory monitoring [CD4 count ± viral load (VL)] improves immune recovery7 and reduces morbidity and mortality;8,9 however, the optimal use of these tests is unclear.9,10 Virological monitoring may conserve first-line regimens by detecting nonadherence before viral resistance develops,11 limiting resistance accumulation through earlier detection of virological failure12,13 and reducing transmission of resistant virus.14 Although late switching, with complex resistance, seems to have minimal effect on early second-line outcomes, the role of virological monitoring in determining longer-term outcomes is yet to be determined.15 Resistance, however, may not be the primary etiology in all patients with virological failure and switching early, without adequately excluding nonadherence, could rapidly exhaust available treatment options and drive up program costs.16 The optimal adherence support, monitoring frequency, acceptable duration of viremia, and virological threshold at which to switch regimens, whether from a clinicoimmunological, resistance, or cost-effectiveness perspective, are ill defined.12,14,17–22
The World Health Organization (WHO) provides pragmatic recommendations; in countries without resistance tests but with routine virological monitoring, adherence interventions should be instigated after a first raised VL. Patients considered adherent with confirmed VL >5000 copies per milliliter should switch regimens.23 Few studies have described programmatic practice of this approach.5,24,25 In South Africa, the country with the world's largest antiretroviral treatment program, 6-monthly virological monitoring is standard of care, routine resistance testing is not available and current national guidelines (2010) advocate switching at a VL threshold of 1000 copies per milliliter.26 In a multisite treatment program, we aimed to describe rates of virological failure, subsequent outcomes, and factors associated with a decision to switch to second-line ART during the episodes of viraemia.
Study Design and Setting
This observational retrospective cohort study used prospectively collected clinical data from the Aurum Institute ART program, South Africa. The workplace component comprised 56 clinics serving employees of large predominantly mining companies. The community program had 81 urban and periurban private general practitioner and nongovernment organization clinics serving patients with limited resources. In both programs, HIV-related treatment was provided free of charge.
This study was approved by the Research Ethics Committees of the University of KwaZulu Natal, South Africa, and the London School of Hygiene and Tropical Medicine, United Kingdom. Access to the program database was granted by Aurum Institute. This database contains data collected as part of routine clinical care for the purposes of monitoring and evaluation. The workplace employers also provide additional data on reasons for leaving the program through employers’ records and hospital death registers, and dates of death are confirmed through program links with the National Death Register. The Research Ethics Committees approval of this study waived the need for patient consent as data had been collected as part of routine practice, and all data received were both retrospective and anonymous.
In the workplace, patients were eligible for first-line ART (efavirenz or nevirapine with zidovudine/lamivudine until 2008, then tenofovir/emtricitabine thereafter) if CD4 ≤250 cells per cubic millimeter, WHO stage IV, or CD4 ≤350 cells per cubic millimeter plus WHO stage III. In the community program, eligibility criteria for first-line ART (nevirapine or efavirenz with stavudine/lamivudine) were WHO stage IV or CD4 ≤200 cells per cubic millimeter. Treatment monitoring involved 3 monthly visits, with CD4 count and VL monitored at baseline, 6 weeks and 6 monthly intervals thereafter. Program guidelines recommended intensifying adherence counseling after a first raised VL with health care workers trained in identifying appropriate patient-centred interventions, for example, treatment partners and medication alarms. If a subsequent VL remained high, a switch to second-line ART was recommended; the threshold triggering a switch changed from 5000 to 1000 copies per milliliter over the course of the program. Second-line ART in the workplace program comprised abacavir, didanosine and boosted lopinavir, and in the community programme zidovudine, didanosine, and boosted lopinavir. All community clinics were doctor led; however, some workplace clinics were nurse led with doctors consulted concerning complications, including virological failure.
Patients ≥15 years old commencing first-line ART between January 1, 2003, and December 31, 2008, and with at least 6 months of follow-up were eligible for inclusion. Clinics with <50 patients on first-line ART were excluded.
At each visit, health care workers used standardized data collection forms to record clinical and program data, for example, WHO staging diagnoses, adverse events, prescriptions and reported reasons for leaving the program. Self-reported adherence was recorded as part of routine data collection; however, no data was collected on adherence interventions implemented as part of the adherence support package. Data were entered into a central database with laboratory data transferred electronically. Where civil identification numbers were available, deaths were verified through the National death register; and in the workplace, through employment records and hospital death registers.
Virological failure on first-line ART was defined as the second of 2 consecutive (≤9 months apart) measurements, both >1000 copies per milliliter and occurring ≥6 months after commencing therapy. Patients were defined as having viral resuppression if a subsequent VL measurement on first-line ART was <400 copies per milliliter and to have switched to second-line ART if a protease inhibitor–based regimen was prescribed. If the patient died within 3 months of the last visit, death was determined to be the primary reason for loss to the program. Patients with no clinic contact for ≥6 months, and with no known reason for leaving the program, were defined as lost to follow-up 3 months after their last visit.
Rates of virological failure, overall and by clinic, were described for patients commencing first-line ART. Analysis was restricted to patients with ≥1 VL measurement after 6 months of first-line ART with follow-up right censored at earliest of first episode of virological failure, death, loss to program, or end of follow-up (January 31, 2009, defined as administrative censoring). Analyses of viral resuppression and switching to second-line ART after first virological failure were restricted to patients with at least 1 episode of virological failure before June 30, 2009, allowing at least 6 months potential follow-up for events to occur. Patients entered this analysis on date of second consecutive VL measurement >1000 copies per milliliter with follow-up right-censored at earliest of event of interest, death, loss to program, or end of follow-up (December 31, 2009). In determining rates of switching, follow-up was right censored on date of resuppression (and vice versa). Taking account of their competing risks, the cumulative incidence at 12 months of switching regimens and viral resuppression and death on first-line ART were described.27
For patients developing virological failure prior to 31/12/2009, factors associated with switching to second-line ART at the next, and subsequent, clinic visits were assessed by constructing a visit dataset. Patients with ≥1 visit following first or subsequent episodes of virological failure were eligible for inclusion. Visits of interest were those at which the health care worker believed the patient to be viremic and made a decision to continue first-line ART or switch to second-line ART (Fig. 1); thus we included visits occurring between >3 days after the date of virological failure to ≤3 days after the date of viral resuppression on first-line ART. Visits after date of switching to second-line ART were excluded. Our rationale was based on discussions with health care workers, who reported that a minimum of 3 days were required for samples to get to the laboratory, be processed, and for results to be made available. Therefore, in the 3 days after venesection, health care workers would be unaware that the patient had fulfilled a definition of virological failure or achieved viral resuppression.
A switch to second-line ART was defined as the first visit a protease inhibitor regimen was prescribed. Covariates were updated using information available to the clinician at the time of the visit. “VL and CD4 count” were defined as the closest available result from a sample taken 7 months to 3 days before the visit. “Change in VL and CD4 count” as percentage change in absolute values between the last 2 available results (within 2–28 weeks of each other and >3 days before the visit). “Duration of viremia” as time between second VL >1000 copies per milliliter and date of visit. “Change in weight” between last and current appointment, where appointments were greater than 1 week apart was measured in kilogram per month. Patients with no clinic contact in preceding 4 months were defined as having “delayed clinic attendance”. For each clinic, the “clinics' rate of virological failure,” as described above, was used as a proxy measure of their experience in managing virological failure. The individual clinic rate was categorized as “high” or “low” if it was above or below the median for the overall rate of virological failure by clinic.
Logistic regression, using random effects to control for individual-level and clinic-level clustering, was used to explore predictors of switching to second-line ART at any given visit. A backward step-wise approach was taken with covariates included and retained in the multivariable model if P < 0.2 [likelihood ratio test (LRT)]. Colinearity was assessed by comparing standard errors between successive models. To explore clinic-level clustering, we looked to see if clinic factors, although not associated with switching in the univariable analysis, would account for clinic-level clustering in the final model.
Sensitivity analyses were performed with the final model restricted to patients known to be ART naive on commencing ART, and to those with virological suppression before failure; and also stratified by time between the visit and diagnosis of virological failure (<12 vs. ≥12 months) and program. Statistical analyses were performed using Stata version 11 (STATA Corporation, College Station, TX)
In total, 10,402 out of 13,537 patients starting first-line ART had at least 1 VL measurement after 6 months of first-line ART [median 5; interquartile range (IQR): 4–7]. Of these, 1867 (17.9%) patients [1146 (31.1%) from 12 workplace clinics and 721 (10.7%) from 51 community clinics] experienced virological failure on first-line ART over a median of 2.6 years (IQR: 1.69–3.51) from initiating ART to leaving program or administrative censoring (Fig. 2). The first episode occurred a median of 1.4 years (range: 1.1–2.0) after commencing ART.
Rates of virological failure varied between clinics [median clinic rate 4.8 per 100 person-years (IQR: 3.3–7.7)], with the median clinic rate in the larger clinics (>500 patients initiated first-line ART, n = 7) twice that seen in smaller clinics (<500 patients, n = 56); 10.6 per 100 person years and 4.6 per 100 person years, respectively. Accounting for clinic-level clustering by the use of robust standard errors, the overall rate of virological failure on first-line ART was 7.4 per 100 person-years [95% confidence interval (CI): 5.8 to 9.5]; 6.8 per 100 person-years (95% CI: 5.1 to 9.1) between 6 and 12 months; 14.1 per 100 person-years (95% CI: 10.7 to 18.6) between 12 and 24 months; and 6.2 per 100 person-years (95% CI: 4.8 to 8.1) thereafter. The median time between the first and second raised VL was 4.9 months (IQR: 2.8–6.1).
The characteristics of patients with virological failure and clinics are presented in Table 1 and Supplemental Digital Content 1 (see Table, http://links.lww.com/QAI/A345), respectively. The majority of patients were male (73.1%) with a median age at virological failure of 41 years and median CD4 count 184 cells per cubic millimeter. Twelve workplace clinics (4 hospital outpatients and 8 occupational health clinics) and 51 community clinics (49 general practitioners, 3 nongovernment organization clinics, and 1 specialist clinic) were included in the study. Fifty-nine clinics started enrolling patients on first-line ART between 2003 and 2006 and 4 clinics between 2007 and 2008. On conducting the study, 7 clinics had enrolled >500 patients on first-line ART, 7 clinics 250–499 patients, and 49 clinics 50–249 patients. Seven clinics had switched >7.5% of patients to second-line ART; 8 clinics 5%–7.5%; 16 clinics 2.5%–4.9%; 16 clinics <2.5%; and in 11 clinics, no patients had switched to second-line ART.
Outcomes in Patients With First-Line Virological Failure
One thousand six hundred sixty-eight (89.3%) patients had at least 6 months' follow-up after a second VL >1000 copies per milliliter [median 1.6 years (IQR: 0.8–2.9)]. Over 1921.8 person-years of first-line virological failure, 296 (17.7%) patients achieved viral resuppression, 361 (21.6%) switched to second-line ART, 107 (6.4%) died, 312 (18.7%) were lost to follow-up, 49 (2.9%) transferred out, 67 (4.0%) left program for other reasons, and 476 (28.5%) were administratively censored after ≥6 months of viremia (Fig. 3). Taking account of competing risks, at 12 months the cumulative incidence of switching to second-line ART, viral resuppression and death was 16.9%, 13.2%, and 4.6%, respectively.
After failure, the rate of viral resuppression on first-line ART was 15.4 per 100 person-years (95% CI: 11.7 to 20.2); highest in the first year at 16.7 per 100 person-years (95% CI: 12.4 to 22.4), falling to 14.1 per 100 person-years (95% CI: 11.0 to 18.2) in the second year, and 11.8 per 100 person-years (95% CI: 7.4 to 18.8) thereafter. Sixty patients had at least 1 further episode of first-line virological failure. The rate of switching after failure was 18.8 per 100 person-years (95% CI: 16.9 to 20.9); highest in the first year at 21.3 per 100 person-years (95% CI: 13.5 to 33.4), falling to 14.8 per 100 person-years (95% CI: 8.1 to 26.7) in the second year, and 15.0 per 100 person-years (95% CI: 6.3 to 35.8) thereafter. Overall rates of resuppression and rates of switching varied by clinic as follows: 10.6 per 100 person-years (IQR: 0–25.0) and 16.9 per 100 person-years (IQR: 2.8–49.7), respectively.
Predictors of Switching to Second-Line ART
Three hundred forty-five (18.5%) patients had no recorded visits after virological failure (156 administrative censoring, 117 lost to follow-up, 31 died, 27 other, 14 transferred out). Eight thousand eight hundred sixty visits were recorded on the remaining 1522 patients from 63 clinics. Two thousand eight hundred six visits occurred during periods of resuppression, or after switch to second-line ART, resulting in 6054 visits on 1495 patients from 63 clinics in the visit dataset (Fig. 1). Visits occurred either during 1 episode (n = 1441 patients), 2 episodes (n = 52 patients), or 3 episodes (n = 2 patients) of virological failure, with a median of 3 visits/episode (IQR: 2–5). Two patients had no visits recorded during the first episode of virological failure; however, visits were recorded during subsequent episodes. 88.4% (1320 of 1493) of patients visited the clinic within 6 months of date of first virological failure (Table 1). During the viremic episodes, the median time-interval between visits was 2.8 months (IQR: 1.4–3.5).
There was strong evidence of within-clinic clustering for switching outcome (LRT, P < 0.01), with weak evidence of within-individual clustering (LRT, P = 0.08). Random effect analyses, which accounted for both clinic-level and individual-level clustering did not differ from analyses accounting for clinic-level clustering alone; therefore, the simpler models are presented. On univariable analysis, switching was associated with ART exposure pre-programme, calendar year of visit, visit number during viremic episode, delayed clinic attendance, current CD4 and VL result, and CD4 and VL trend preceding the visit (Table 2).
Multivariable analysis (Table 2) showed that independent determinants of switching included the clinic visit being the third or subsequent visit during the viremic period, the visit occurring in 2008 versus earlier years, the patient being ART experienced pre-programme, or the most recent VL result being log10 ≥4 or CD4 count <100 cells per cubic millimeter. Patients with no clinic contact in the 4 months preceding the visit, or declining VL (trend), were less likely to switch regimens. In the final model, there was strong evidence of clinic-level clustering (LRT, P <0.001), which remained after accounting for program, clinic size, and clinic rate of virological failure.
When the final model was stratified by duration of virological failure or program; or restricted to those known to be ART naive or with prior viral suppression, the direction of associations remained the same (see Tables, Supplemental Digital Content 2–5, http://links.lww.com/QAI/A345).
In this large multicentre ART program in South Africa, despite 6 monthly virological monitoring and guidelines recommending a switch after confirmed virological failure, rates of switching to second-line ART were low. Many patients experienced extended periods of viremia after virological failure, in part, due to relatively infrequent VL monitoring and clinic visits. However, other factors including delayed clinic attendance, VL (magnitude and trend), and patients’ immune status seemed to influence health care workers’ decisions regarding when to switch regimens.
Although current program guidelines advise intensifying adherence interventions and switching regimens if a second VL remains high,23,26 this rarely happened in practice. Instead, after a second raised VL, many patients experienced prolonged periods of viremia with the rate of viral resuppression on first-line ART approaching the rate of switching to second-line ART. By the end of follow-up, nearly a quarter of patients remaining on first-line ART were virologically suppressed. Others report high levels of viral resuppression after a single raised VL,25,28–31 however, these results highlight that even after a second raised viral load, viral resuppression remains possible due to, for example, a change in patients adherence behavior or removal of a drug interaction. Excluding nonadherence as a cause of viremia is challenging; current adherence measurement tools are inadequate,32 and strategies to improve adherence can take time and are often context specific requiring multidimensional interventions.33 These difficulties may be compounded by weaknesses in the health care system, for example, out-of-date patient contact details lack of an appointment system, or outreach workers may create delays in detecting and tracing patients with raised viral loads or missed appointments.34
When deciding to switch to second-line ART, health care workers seem to use the magnitude and trend in VLs, prior ART experience, and clinic attendance to evaluate the probability of resistance as opposed to nonadherence. As reported elsewhere, patients with high VLs were more likely to switch regimens35–37; although there is a suggestion in our study that health care workers were less likely to switch patients with a VL >5 log10 as compared with 4–5 log10, perhaps recognizing that these patients have a lower likelihood of resistance and may be nonadherent.38–40 Patients whose VL had fallen to 400–1000 copies per milliliter or whose VL had declined by >50%, suggesting improved adherence, and patients with delayed clinic attendance who were likely to have run out of medication were less likely to be switched. A decision not to switch patients in whom adherence interventions are working or where clinic appointments have been missed seems rational.
Patients with a low CD4 count were more likely to switch to second-line ART, confirming findings from other studies.5,35–37,41 A low CD4 count may reflect immunological progression occurring as a result of viral replication due to treatment interruption, nonadherence, or resistance.42,43 There is no consistent relationship between current CD4 count and odds of detecting resistance.38,40 This may in part relate to the mutations studied; M184V, a mutation which emerges early, has been associated with higher concurrent CD4 count,44,45 whereas some46,47 but not all studies44,48 have demonstrated an association between low CD4 count and thymidine analogue mutations. Regardless of the reason, patients with a low CD4 count are at high risk of short-term clinical progression,19,49 and it may be this factor in itself which prompts a decision to switch to second-line ART.
The frequency of clinic visits contributes to prolonged viremia, with results from 1 visit only available to influence decision-making at the subsequent visit. One-fifth of patients had no record of clinic attendance after confirmed virological failure, therefore, health care workers had no opportunity to assess and instigate appropriate interventions. Interestingly, duration of viremia at the time of visit was not associated with switching regimens, but number of visits was. Regardless of duration of viremia, health care workers may require repeated visits to exclude alternative causes of viremia and ensure that the patient is engaged in care.
Rates of switching and resuppression varied markedly between clinics, supporting observations by others.4,36,50 This heterogeneity was not accounted for by the individual-level or clinic-level variables measured. Others have demonstrated the role of contextual and program factors on the timing of ART initiation,51 however, their influence on the timing of the switch decision is not fully understood.52 Factors such as variations in workload, available adherence support measures within the clinic, procedures for tracing of patients who have missed appointments, and the influence of contextual factors on individual health care workers and patients decision-making regarding virological failure need explored.
Although other studies in resource-limited settings have reported on rates and predictors of switching, these are largely from specialist centers and describe rates of switching after initiation of first-line ART, evaluating predictors either from this time point or at 6 months.4,53–55 In contrast, our analysis describes rates of switching in patients with virological failure and explores determinants of switching at the time decisions are made, that is, at the clinic visit. By taking this approach, rather than a time-to-event analysis, we have avoided the potential bias introduced by competing risks of viral resuppression and interval censoring due to frequency of visits. Although some visit records may be missing, we do not believe this will introduce bias but rather weaken associations.
This study was conducted using routine program data, and although reasons for stopping treatment including switching regimens were available, reasons for continuing a failing regimen were not. We were unable to fully explore the role nonadherence might have on the timing of the switch decision as our only available measure of nonadherence was patients’ self-report. This is recognized to be a suboptimal measure32 and was often missing. In addition other factors, not captured in routine reporting, may affect decision-making, for example, patient’s readiness to switch regimens, health care workers’ concerns regarding limited future regimens, and experience in prescribing second-line ART, or health system factors.52–54,56–58 Finally, some patients may have switched to a protease inhibitor for reasons other than treatment failure, for example, toxicity. In a study using the same cohort, the majority of patients switched to second-line ART whilst viremic did so for treatment failure;59 we therefore believe there is minimal misclassification in our outcome measurement.
Given the large numbers of people living with HIV in sub-Saharan Africa, the need for life-long treatment, the high cost of second-line and third-line drugs, and limited resources, it is vital that the currently available ART regimens are used effectively. These results highlight that even in programs with access to virological monitoring, health care workers encounter difficulties in determining which patients are likely to have resistance and should switch regimens and which need further adherence support. Current tools for measuring nonadherence are inadequate32 and alternative approaches, including targeted resistance testing, warrants further investigation.60 For some patients, attempts to improve nonadherence will be unsuccessful; patients will remain viremic and at risk of resistance accumulation. Switching such patients to a boosted protease inhibitor-based regimen, which is more tolerant of nonadherence, is one strategy; however, the cost-effectiveness in resource-limited settings needs explored. Evidence to support decisions, together with clear algorithms for the management of virological failure, particularly where there is ongoing suboptimal adherence are needed; this becomes particularly pertinent in light of increasing task-shifting of HIV care to nurses and field workers.
The authors would like to thank the patients and health care teams at the participating sites and the staff at Aurum Institute for their assistance with this study.
2. Barth RE, van der Loeff MF, Schuurman R, et al.. Virological follow-up of adult patients in antiretroviral treatment programmes in sub-Saharan Africa: a systematic review. Lancet Infect Dis. 2010;10:155–166.
3. Renaud-Thery F, Duncome C, Kerr S, et al.. Systematic review of first-line treatment failure and attrition rates [abstract 827]. Paper presented at: 17th Conference on Retroviruses and Opportunistic Infections; February 16–19, 2010; San Francisco, CA.
4. Keiser O, Tweya H, Boulle A, et al.. Switching to second-line antiretroviral therapy in resource-limited settings: comparison of programmes with and without viral load monitoring. AIDS. 2009;23:1867–1874.
5. Zhou J, Li PC, Kumarasamy N, et al.. Deferred modification of antiretroviral regimen following documented treatment failure in Asia: results from the TREAT Asia HIV Observational Database (TAHOD). HIV Med. 2010;11:31–39.
6. Chang LW, Harris J, Humphreys E. Optimal monitoring strategies for guiding when to switch first-line antiretroviral therapy regimens for treatment failure in adults and adolescents living with HIV in low-resource settings. Cochrane Database Syst Rev. 2010;4:CD008494.
7. Laurent C, Kouanfack C, Laborde-Balen G, et al.. Monitoring of HIV viral loads, CD4 cell counts, and clinical assessments versus clinical monitoring alone for antiretroviral therapy in rural district hospitals in Cameroon (Stratall ANRS 12110/ESTHER): a randomised non-inferiority trial. Lancet Infect Dis. 2011;11:825–833.
8. Mugyenyi P, Walker AS, Hakim J, et al.. Routine versus clinically driven laboratory monitoring of HIV antiretroviral therapy in Africa (DART): a randomised non-inferiority trial. Lancet. 2010;375:123–131.
9. Mermin J, Ekwaru JP, Were W, et al.. Utility of routine viral load, CD4 cell count, and clinical monitoring among adults with HIV receiving antiretroviral therapy in Uganda: randomised trial. BMJ. 2011;343:d6792.
10. Jourdain G, Ngo-Giang-Huong N, Le Coeur S, et al.. PHPT-3: a randomized clinical trial comparing CD4 vs viral load ART monitoring/switching strategies in Thailand [abstract 44]. Paper presented at: 18th Conference on Retroviruses and Opportunistic Infections; February 27-March 2, 2011; Boston, MA.
11. Wilson D, Keiluhu AK, Kogrum S, et al.. HIV-1 viral load monitoring: an opportunity to reinforce treatment adherence in a resource-limited setting in Thailand. Trans R Soc Trop Med Hyg. 2009;103:601–606.
12. 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.
13. Cozzi-Lepri A, Phillips AN, Martinez-Picado J, et al.. Rate of accumulation of thymidine analogue mutations in patients continuing to receive virologically failing regimens containing zidovudine or stavudine: implications for antiretroviral therapy programs in resource-limited settings. J Infect Dis. 2009;200:687–697.
14. Phillips AN, Pillay D, Garnett G, et al.. Effect on transmission of HIV-1 resistance of timing of implementation of viral load monitoring to determine switches from first to second-line antiretroviral regimens in resource-limited settings. AIDS. 2011;25:843–850.
15. Ajose O, Mookerjee S, Mills EJ, et al.. Treatment outcomes of patients on second-line antiretroviral therapy in resource-limited settings: a systematic review and meta-analysis. AIDS. 2012;26:929–938.
16. Long L, Fox M, Sanne I, et al.. The high cost of second-line antiretroviral therapy for HIV/AIDS in South Africa. AIDS. 2010;24:915–919.
17. Murri R, Lepri AC, Cicconi P, et al.. Is moderate HIV viremia associated with a higher risk of clinical progression in HIV-infected people treated with highly active antiretroviral therapy: evidence from the Italian cohort of antiretroviral-naive patients study. J Acquir Immune Defic Syndr. 2006;41:23–30.
19. Ledergerber B, Lundgren JD, Walker AS, et al.. 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.
20. Kousignian I, Abgrall S, Duval X, et al.. Modeling the time course of CD4 T-lymphocyte counts according to the level of virologic rebound in HIV-1-infected patients on highly active antiretroviral therapy. J Acquir Immune Defic Syndr. 2003;34:50–57.
21. 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.
22. Walensky RP, Ciaranello AL, Park JE, et al.. Cost-effectiveness of laboratory monitoring in sub-Saharan Africa: a review of the current literature. Clin Infect Dis. 2010;51:85–92.
23. World Health Organization. Antiretroviral therapy for HIV infection in adults and adolescents: recommendations for a public health approach. 2010. Available at: http://www.who.int/hiv/pub/arv/adult2010/en/
. Accessed September 15, 2011.
24. Keiser O, Tweya H, Braitstein P, et al.. Mortality after failure of antiretroviral therapy in sub-Saharan Africa. Trop Med Int Health. 2010;15:251–258.
25. Orrell C, Kaplan R, Wood R, et al.. Virological breakthrough: a risk factor for loss to followup in a large community-based cohort on antiretroviral therapy. AIDS Res Treat. 2011;2011:469127.
26. National Department of Health, South Africa. Clinical Guidelines for the Management of HIV & AIDS in Adults and Adolescents. 2010.
27. Coviello V, Boggess M. Cumulative incidence estimation in the presence of competing risks. Stata J. 2004;4:103–112.
28. Hoffmann CJ, Charalambous S, Sim J, et al.. Viremia, resuppression, and time to resistance in human immunodeficiency virus (HIV) subtype C during first-line antiretroviral therapy in South Africa. Clin Infect Dis. 2009;49:1928–1935.
29. Castelnuovo B, Sempa J, Agnes KN, et al.. Evaluation of WHO criteria for viral failure in patients on antiretroviral treatment in resource-limited settings. AIDS Res Treat. 2011;2011:736938.
30. Brinkhof MW, Boulle A, Weigel R, et al.. Mortality of HIV-infected patients starting antiretroviral therapy in sub-Saharan Africa: comparison with HIV-unrelated mortality. PLoS Med. 2009;6:e1000066.
31. Messou E, Chaix ML, Gabillard D, et al.. Association between medication possession ratio, virologic failure and drug resistance in HIV-1-infected adults on antiretroviral therapy in Cote d'Ivoire. J Acquir Immune Defic Syndr. 2011;56:356–364.
32. Gill CJ, Hamer DH, Simon JL, et al.. No room for complacency about adherence to antiretroviral therapy in sub-Saharan Africa. AIDS. 2005;19:1243–1249.
33. Simoni JM, Amico KR, Smith L, et al.. Antiretroviral adherence interventions: translating research findings to the real world clinic. Curr HIV/AIDS Rep. 2010;7:44–51.
34. Braitstein P, Daggy J, Easterbrook P, et al.. Retaining adults in HIV care: impact of key program characteristics on patient loss to follow-up (LTFU) in the IeDEA East Africa Consortium [abstract 1015]. Presented at 18th Conference on Retroviruses and Opportunistic Infections; February 27-March 2, 2011; Boston, MA.
35. Lee KJ, Dunn D, Porter K, et al.. Treatment switches after viral rebound in HIV-infected adults starting antiretroviral therapy: multicentre cohort study. AIDS. 2008;22:1943–1950.
36. Davies MA, Moultrie H, Eley B, et al.. Virologic failure and second-line antiretroviral therapy in children in South Africa–the IeDEA Southern Africa collaboration. J Acquir Immune Defic Syndr. 2011;56:270–278.
37. Petersen ML, van der Laan MJ, Napravnik S, et al.. Long-term consequences of the delay between virologic failure of highly active antiretroviral therapy and regimen modification. AIDS. 2008;22:2097–2106.
38. Marconi VC, Sunpath H, Lu Z, et al.. Prevalence of HIV-1 drug resistance after failure of a first highly active antiretroviral therapy regimen in KwaZulu Natal, South Africa. Clin Infect Dis. 2008;46:1589–1597.
39. Bangsberg DR, Moss AR, Deeks SG. Paradoxes of adherence and drug resistance to HIV antiretroviral therapy. J Antimicrob Chemother. 2004;53:696–699.
40. Prosperi MC, Mackie N, Di Giambenedetto S, et al.. Detection of drug resistance mutations at low plasma HIV-1 RNA load in a European multicentre cohort study. J Antimicrob Chemother. 2011;66:1886–1896.
41. Danel C, Gabillard D, Inwoley A, et al.. Medium-term probability of success of antiretroviral treatment after early warning signs of treatment failure in West African adults. AIDS Res Hum Retroviruses. 2009;25:783–793.
42. Kranzer K, Ford N. Unstructured treatment interruption of antiretroviral therapy in clinical practice: a systematic review. Trop Med Int Health. 2011;16:1297–1313.
43. Seyler C, Adje-Toure C, Messou E, et al.. Impact of genotypic drug resistance mutations on clinical and immunological outcomes in HIV-infected adults on HAART in West Africa. AIDS. 2007;21:1157–1164.
44. van Zyl GU, van der Merwe L, Claassen M, et al.. Antiretroviral resistance patterns and factors associated with resistance in adult patients failing NNRTI-based regimens in the western cape, South Africa. J Med Virol. 2011;83:1764–1769.
45. Doualla-Bell F, Gaseitsiwe S, Ndungu T, et al.. Mutations and polymorphisms associated with antiretroviral drugs in HIV-1C-infected African patients. Antivir Chem Chemother. 2004;15:189–200.
46. Orrell C, Walensky RP, Losina E, et al.. HIV type-1 clade C resistance genotypes in treatment-naive patients and after first virological failure in a large community antiretroviral therapy programme. Antivir Ther. 2009;14:523–531.
47. Hosseinipour MC, van Oosterhout JJ, Weigel R, et al.. The public health approach to identify antiretroviral therapy failure: high-level nucleoside reverse transcriptase inhibitor resistance among Malawians failing first-line antiretroviral therapy. AIDS. 2009;23:1127–1134.
48. Sigaloff KC, Hamers RL, Wallis CL, et al.. Unnecessary antiretroviral treatment switches and accumulation of HIV resistance mutations; two arguments for viral load monitoring in Africa. J Acquir Immune Defic Syndr. 2011;58:23–31.
49. Srasuebkul P, Lim PL, Lee MP, et al.. Short-term clinical disease progression in HIV-infected patients receiving combination antiretroviral therapy: results from the TREAT Asia HIV observational database. Clin Infect Dis. 2009;48:940–950.
50. Lee K, Lyall H, Walker A, et al.. Wide disparity in switch to second-line therapy in HIV infected children in CHIPS. Paper presented at Eighth International Congress on Drug Therapy in HIV Infection; November 12–16, 2006; Glasgow, United Kingdom.
51. Nash D, Wu Y, Elul B, et al.. Program-level and contextual-level determinants of low-median CD4+ cell count in cohorts of persons initiating ART in eight sub-Saharan African countries. AIDS. 2011;25:1523–1533.
52. Srasuebkul P, Calmy A, Zhou J, et al.. Impact of drug classes and treatment availability on the rate of antiretroviral treatment change in the TREAT Asia HIV Observational Database (TAHOD). AIDS Res Ther. 2007;4:18.
53. Landier J, Akonde A, Pizzocolo C, et al.. Switch to second-line ART in West African routine care: incidence and reasons for switching. AIDS Care. 2011;23:75–78.
54. Orrell C, Harling G, Lawn SD, et al.. Conservation of first-line antiretroviral treatment regimen where therapeutic options are limited. Antivir Ther. 2007;12:83–88.
55. Luca D, Marazzi MC, Mancinelli S, et al.. Prognostic value of virological and immunological responses after 6 months of antiretroviral treatment in adults with HIV-1 infection in Sub-Saharan Africa. J Acquir Immune Defic Syndr. 2011;59:236–244.
56. Hawkins C, Murphy RL. Management of antiretroviral failure and resistance in developing countries. Curr Opin HIV AIDS. 2009;4:538–544.
57. Harries AD, Zachariah R, van Oosterhout JJ, et al.. Diagnosis and management of antiretroviral-therapy failure in resource-limited settings in sub-Saharan Africa: challenges and perspectives. Lancet Infect Dis. 2010;10:60–65.
58. Pitts M, JG, Koelmeyer R. If it ain't broke, don’t fix it: the impact of patient and doctor concerns on commencing and changing antiretroviral treatment. Presented at: 6th IAS Conference on HIV Pathogenesis, Treatment and Prevention; July 17–20, 2011; Rome, Italy. Abstract: CDB364.
59. Johnston V, Fielding K, Charalambous S, et al.. Second-line antiretroviral therapy in a workplace and community-based treatment programme in South Africa: determinants of virological outcome. PLoS ONE. 2012;7(5):e36997.
60. Rosen S, Long L, Sanne I, et al.. The net cost of incorporating resistance testing into HIV/AIDS treatment in South Africa: a Markov model with primary data. J Int AIDS Soc. 2011;14:24.