Nglazi, Mweete D MPH*†; Lawn, Stephen D MD*‡; Kaplan, Richard MD*; Kranzer, Katharina MRCP(UK), MSc*‡; Orrell, Catherine MBChB, MSc*; Wood, Robin MBBCh, MSc, FCP*; Bekker, Linda-Gail MBChB, FCP, PhD*
Antiretroviral therapy (ART) has become much more widely available in resource-limited countries with a high burden of HIV/AIDS. Four million people were estimated to be receiving ART in low- or middle-income countries by the end of 2008, of whom 2.9 million were in sub-Saharan Africa and 701,000 were in South Africa alone.1 Success in scale-up may be tempered, however, by the challenges associated with rapidly increasing caseloads of patients attending individual clinics. This may potentially undermine the ability to sustain program quality.
Reports of early experiences from ART programs in sub-Saharan Africa were generally favorable, with good immunological, virological, and clinical responses being achieved.2 However, it subsequently emerged that mortality rates within the initial months of ART are disproportionately higher in African programs compared with rates in other regions.3,4 Moreover, meta-analyses of programs that included self-paying patients subsequently highlighted high rates of program attrition after 2 years of follow-up due to a combination of mortality and losses to follow-up.5 Few studies of large cohorts in sub-Saharan Africa have reported on long-term outcomes6-8 and how these outcomes have changed over time as overall cohort size has increased.6 In this study, we reported on outcomes of patients receiving treatment between 2002 and 2009 in a community-based ART cohort in Cape Town, South Africa. We report on trends in these outcomes stratified by cohort year of enrollment, thereby providing an important assessment of temporal trends in early and long-term outcomes in this cohort.
This is a well-characterized ART service in a poor periurban area in Cape Town, South Africa.9-12 Provision of ART at this service commenced in September 2002, and by September 2009, care was provided for more than 3000 patients. The national program criteria for starting ART were a prior AIDS diagnosis [World Health Organization (WHO) stage 4 disease] or a blood CD4 cell count <200 cells per microliter. The first-line ART regimen comprised 2 nucleoside reverse transcriptase inhibitors and a nonnucleoside reverse transcriptase inhibitor. This was provided to patients free of charge. Patients had routine clinical assessment every 2 weeks before ART and after 8 and 16 weeks of treatment and 16-weekly thereafter. CD4 cell count and viral load were monitored before ART and every 16 weeks during ART.
Provision of patient care was supported by peer counselors, most of whom are themselves living with HIV and receiving ART.13 Each new patient enrolling into the clinic was allocated to a peer counselor living in the same area. Through group sessions and individual home visits, patients were educated and provided with counseling support and the need for high levels of treatment adherence were reinforced. From 2006 onward, the number of counselors remained constant at around 30, despite ongoing increases in patient caseload.
Data were obtained from a prospectively maintained ART cohort database of clinical variables, outcomes, treatment regimens, and laboratory data derived from patient notes and laboratory records. ART-naive patients aged ≥15 years who enrolled in this cohort between September 2002 and September 2008 were eligible for this analysis. Follow-up data were censored at September 2009. Staffing levels at this clinic were determined from human resources records. Data collection on this cohort was approved by the Research Ethics Committee of the University of Cape Town, and all patients gave written informed consent to have data collected anonymously for research purposes.
Definitions of Outcomes
Patients were defined as a “loss to follow up” (LTFU) if they had failed to attend the clinic for ≥12 weeks and were not known to have died or to have been transferred out to another facility. The date of LTFU was recorded as the date of last visit to the clinic. “Deaths” referred to all-cause mortality notified from any source. “Transfers-out” referred to patients receiving ART whose care was transferred to another clinic, arising as a result of either increasing numbers of treatment centers in surrounding communities or due to patient home relocation. “Retention in care” refers to patients alive and receiving ART. “Virological failure” was defined as 2 consecutive viral load measurements >1000 copies per milliliter.
Five calendar periods of enrollment were defined as September 2002 to September 2004, followed by 4 calendar periods of 12 months each up to September 2008. The first calendar period combined the first 2 years due to the small numbers of patients enrolled in those years and similarity in patient baseline profile, including CD4 counts and log10 viral loads.
The survival analyses had the following multiple end points: death, loss-to-follow up, and virological failure. Each end point was analyzed separately. Kaplan-Meier estimates were used to describe time-to-event distributions. Log rank tests were used to compare time-to-event after ART initiation across the calendar period of enrollment. Cox proportional hazard models were used to determine the hazard of the event of interest at any time during the follow-up period, adjusting for relevant baseline covariates; calendar period of enrollment was included as a key variable of interest in this model. The proportional hazard assumption has been checked graphically using a log-log plot and the Schoenfeld residuals (tests and graphs).
For analyses involving week 16 viral loads and week 48 CD4 cell counts, data were categorized into binary variables. Bivariate analyses used χ2 tests to compare proportions. Logistic regression models were used to determine the risk of the outcome of interest, adjusting for relevant baseline covariates; calendar period of enrollment was included as a key variable of interest in this model. Log likelihood ratio tests were used to compare relative goodness of fit of nested models. Final models were checked using standard regression diagnostics for logistic regression.
Wald confidence limits were used for all multivariate models. All statistical tests were 2-sided at alpha of 0.05. Stata statistical software, version 10.0 was used for analyses (Stata Corporation, College Station, TX).
Overall, 3162 treatment-naive patients aged 15 years and older had initiated ART by September 2008 (Table 1), with annual enrollment reaching a maximum of 783 patients in the 12-month period 2005-2006. Two thirds of the cohort were women, and most were young adults. Immunodeficiency was typically advanced, as reflected by the WHO clinical stage of disease and CD4 cell counts (Table 1). Median baseline CD4 cell counts differed significantly between successive calendar periods of enrollment, increasing from 87 [interquartile range (IQR), 45-145] cells in 2002-2004 to 121 cells per microliter (IQR, 60-178) in 2007-2008. The most common first-line regimens was stavudine and lamivudine combined with either efavirenz or nevirapine (Table 1). Patients were followed up on ART for a minimum duration of 1 year, a median duration of 2.4 years (IQR, 1.2-3.8), and up to a maximum of 7.0 years. A total of 8061.1 person-years of follow-up accrued during the analysis period. The number of patients per member of clinic staff increased markedly over time, with the patient-to-peer counselor ratio increasing by more than 4-fold during the analysis period and patient-to-doctor ratio approximately doubling (Table 1).
Overall, 334 patients (10.6%) died, 592 patients (18.7%) were LTFU, and 326 patients (10.3%) were transferred to another ART facility. Kaplan-Meier analyses were used to provide probability estimates of long-term losses. The probability of being lost to care through death or LTFU was 37.4% [95% confidence interval (CI), 34.9 to 40.0] at 6 years (Fig. 1A). In addition, some patients were transferred out so that the overall probability of leaving the program for any reason was 52.5% (95% CI, 49.5 to 55.6) after 6 years (Fig. 1A). When examined by calendar period of enrollment, total losses (death, LTFUs, and transfers-out) increased between sequential calendar periods of enrollment (Fig. 1B), although the difference between the latter 2 periods was not statistically significant.
We next examined each of these losses in turn. Probability of death was greatest in the first 1 year of ART with an estimate of 7.9% (95% CI, 7.0% to 8.9%) of patients. However, mortality risk was lower in subsequent years, and the cumulative probability of death after 6 years was just 15.2% (95% CI, 13.1%-17.6%). Mortality did not differ substantially between sequential calendar periods of enrollment with 1-year mortality estimates for the 5 sequential periods of 8.4% (6.0-11.6), 8.3% (6.3-10.7), 8.7% (6.9-11.0), 7.3% (5.6-9.7), and 6.9% (5.1-9.1) (Fig. 1C).
In contrast to deaths, the probability of patients being LTFU was much higher at 29.1% (95% CI, 26.5% to 31.8%) after 6 years, and the proportions LTFU increased significantly between successive calendar periods of enrollment (Fig. 1D). Furthermore, the probability of being transferred-out was 21.6% (95% CI, 18.2% to 25.4%) after 6 years and, similar to LTFU, transfers-out increased between successive calendar periods of enrollment (Fig. 1E).
Having observed that LTFU and transfers-out increased over time with successive calendar periods of enrollment, we next examined the risk of virological failure. By 6 years, the probability of virological failure was 23.1% (95% CI, 20.2 to 26.3) (Fig. 2A), and this proportion also increased with each successive calendar period of enrollment (Fig. 2B).
Risk Factors Associated With Program Losses and Virological Failure
We next used univariate and multivariate analyses to determine factors associated with death, LTFU, transfer-out, and virological failure and included calendar period of enrollment as a key variable of interest. Final fully adjusted models are shown in Table 2.
Higher mortality risk was significantly associated with men, lower baseline CD4 cell count (≤50 cells/μL), and the WHO stage of disease III and IV (Table 2). However, consistent with Figure 1C, mortality was not associated with calendar period of enrollment. Factors significantly associated with LTFU were men, younger age and the WHO stage of disease III and IV. However, greater risk of LTFU was also associated with later calendar periods of enrollment, with adjusted hazard ratios increasing with each successive calendar period (Table 2). Being transferred to another ART facility was not associated with baseline patient characteristics but was strongly associated with cohort year of enrollment (Table 2). Thus, patients enrolled in later years were far more likely to be LTFU or be transferred to another treatment site.
Finally, factors significantly associated with virological failure were younger age (≤25 years), lower baseline CD4 cell count (≤50 cells/μL), and higher baseline viral load (≥5 log10 copies/mL). However, the risk of virological failure also increased significantly for patients in later calendar periods of enrollment (Table 2).
Virological and Immunological Responses to ART
Thus far, data indicate worsening program performance with regard to increasing risk of LTFU and virological failure when comparing patients enrolled in successive calendar periods. To provide a better assessment of patients' initial adherence and response to ART, we next examined viral load suppression rates after 16 weeks of ART and the proportion of patients with a CD4 cell count ≤200 cells per microliter after 48 weeks of ART.
Viral load suppression to <400 copies per milliliter was observed among a very high proportion of patients, exceeding 93% for each successive calendar period of enrollment (Table 3). In multivariate analysis using logistic regression, lack of virological suppression at 16 weeks was significantly associated with age <25 years and high baseline viral load (≥5 log10 copies/mL) but not with cohort year of enrollment (data not shown).
Approximately 90% of patients had a CD4 cell count ≤200 cells per microliter at baseline (Table 1), and this proportion decreased to between 20.8% and 31.1% after 48 weeks of ART comparing different calendar periods of enrollment (Table 3). In fully adjusted analyses, factors significantly associated with CD4 cell count remaining ≤200 cells per microliter at 48 weeks were men, older age (≥41 years), lower baseline CD4 cell counts, and lower baseline viral loads (data not shown). However, similar to virological suppression rates, there was no association with cohort year of enrollment.
To optimize the benefits of ART, a number of overlapping short-term and long-term goals collectively require proficient and supportive delivery of treatment by a team of health care workers and counselors. The overriding goal is to rapidly reduce mortality through diagnosis and treatment of comorbidities, provision of cotrimoxazole prophylaxis, and achievement of optimum virological and immunological responses to ART. Long-term goals include the retention of patients on ART with high rates of treatment adherence and sustained virological suppression.
In this article, 3162 patients were started on ART in a South African treatment service studied over 7 years. Excellent early outcomes (mortality and virological and immunological responses) were sustained in sequential calendar periods of ART initiation. In contrast, however, cumulative rates of longer term adverse outcomes of LTFU and virological failure deteriorated over time as the service enlarged and ratios of patient to health worker staff numbers are adversely affected. As ART services grow in size, patient care and support systems may be overly focused on treatment initiation and the subsequent few months of care and that additional attention and resources are needed for ongoing treatment support in the longer term.
A key outcome of ART programs is combined attrition due to death and LTFU. This was 37.4% after 6 years, which is similar to that observed in another program in Cape Town.6 A previous meta-analysis of cohorts in sub-Saharan Africa reported average losses of approximately 40% after just 2 years.5 The probability of retaining approximately two thirds of our cohort on ART after 6 years represents a considerable success. However, the risk of being lost to the program increased with each successive calendar year of enrollment, suggesting a progressive deterioration in program performance. To explore this further, we examined these different losses separately in addition to early and late virological responses to ART.
Between 8% and 26% of patients die during the first year of ART in treatment programs in sub-Saharan Africa,3 and yet patients in this cohort had a 1-year mortality rate of 7.9%, despite many having advanced immunodeficiency. This represents among the lowest reported mortality rates in the region. Moreover, regardless of escalating patient caseload, 1-year mortality rates did not differ between successive calendar periods of ART initiation. Thus, the overriding goal of maintaining low mortality rates during scale-up of this service was achieved.
Virological suppression is another key early programmatic goal. Excellent suppression rates were observed, with ≥93% of patients having suppression <400 copies per milliliter at 16 weeks, and these rates did not vary significantly between successive years of recruitment. This is indicative of very high rates of initial adherence to effective treatment. We also examined the proportion of patients with a CD4 cell count <200 cell per microliter at 48 weeks because this is the patient subset that remained at high ongoing risk of mortality and of morbidity.12,14 This proportion decreased from approximately 90% at baseline to approximately 20%-30% after 48 weeks of ART and did not significantly differ between sequential calendar periods of ART initiation. Thus, collective first year key outcomes of mortality and virological and immunological responses were sustained over 7 years during scale-up of this service. This suggests that standards of initial care and treatment support provided in the initial months of ART were sustained regardless of overall patient caseload in the service.
After the first year of ART, a key challenge is to retain patients within the cohort and maintain virological suppression. We found that the cumulative probability of being LTFU after 6 years of ART was 29.1%, which is broadly similar to another program in Cape Town.6 However, the proportion LTFU increased significantly between sequential calendar periods of ART initiation as observed elsewhere.6,15,16 Escalating patient caseload may be an important underlying factor as this may be associated with increasing clinic waiting times, shorter consultations, reduced opportunities for counseling, and adherence support and overstretching of human resources for patient tracing following missed appointments. Indeed, over the course of scale-up, the number of new patients enrolled per peer counselor or per doctor increased several fold (Table 1).
The increasing rates of LTFU were also paralleled by increasing risk of virological failure, further underscoring the suggestion that treatment support had diminished over time with increasing clinic size. The probability of virological failure in the whole cohort at 6 years was 23.1% and many patients who develop failure in this cohort have drug resistance mutations.17 Few other data are available from public sector programs in sub-Saharan African due to a lack of availability of routine virological monitoring and directly comparable prospectively collected data are lacking.18 The failure rate we observed is higher than the rate of 14% reported in another program in Cape Town,6 but definitions of failure and frequency of virological monitoring differed between programs. The increasing rate of virological failure has important implications with regard to escalating rates of drug resistance and increased switching to protease-inhibitor-based second line therapy, which is considerably more expensive. Remaining therapeutic options thereafter are few.
The proportion of patients transferred between services in ART programs in sub-Saharan Africa is highly variable, but a large meta-analysis reported a weighted average of just 1.1% after a median of 26 months of ART.19 In contrast, approximately one fifth of our cohort was estimated to transfer care over 6 years of ART. The long-term outcomes of such patients are not known. The probability of being transferred out was much greater for patients initiating ART in more recent calendar periods. There may be several explanations for this. First, expansion in the availability of ART services in surrounding communities and elsewhere in South Africa has increasingly provided access for patients to be treated near their own homes. It is also likely that those enrolled in the clinic in the earlier years compared with later years had more attachment to the clinic, especially as in the earlier years, clinics were small and little was known about likely outcomes. It is also possible that patients who resume work after a period of illness may wish to be transferred to other centers, which are less busy, have shorter waiting times, or are more conveniently located. The South African population is highly mobile, and this provides a challenge to long-term care. Efficient referral and patient tracking systems are needed to facilitate uninterrupted care for such patients.
Strengths of this study include the fact that these data are from a primary care clinic within the South African public sector system. The cohort of more than 3000 patients is very well characterized and has very complete prospectively recorded data on patient outcomes. This cohort has one of the longest durations of follow-up within the context of a public sector ART programs in sub-Saharan Africa. The analysis of outcome data by calendar period of enrollment provides critical insights into the changing treatment outcomes over time.
Limitations include the analysis of multiple end points, which could increase the chance of false-positive findings. To minimize this bias, we analyzed each outcome separately. Some deaths could have remained unascertained and recorded as LTFU. However, low CD4 cell counts were not a risk factor for LTFU (Table 2), and the temporal distribution of deaths and LTFU differed (Figs. 1C, D), suggesting that any such overlap was insubstantial. Although escalating size of the treatment cohort is suggested as a plausible factor underlying the rising risk of LTFU and virological failure over time, a causal association has not been demonstrated and other as yet unidentified factors may be important. The subsequent outcomes of patients transferred out to other services are unknown.
In summary, this analysis of a community-based ART services shows that mortality and immunological and virological responses during the first year were not compromised during scale-up, suggesting that patient preparation for ART and initial medical care were maintained at a high standard. However, the cumulative probability of patients being LTFU or developing virological failure deteriorated over time when comparing successive calendar periods of recruitment. This may reflect a diminishing capacity over time to adequately support patients during long-term therapy as clinic size escalates.
The authors gratefully acknowledge the dedicated staff of the Hannan Crusaid ART clinic and the Desmond Tutu HIV Centre.
1. World Health Organisation. Towards Universal Access: Scaling up Priority HIV/AIDS Interventions in the Health Sector. Progress Report 2009
. Geneva, Switzerland: Wolrd Health Organization; 2010. Available at: http://www.who.int/hiv/pub/tuapr_2009_en.pdf
. Accessed 12 January, 2010.
2. Ivers LC, Kendrick D, Doucette K. Efficacy of antiretroviral therapy programs in resource-poor settings: a meta-analysis of the published literature. Clin Infect Dis. 2005;41:217-224.
3. Lawn SD, Harries AD, Anglaret X, et al. Early mortality among adults accessing antiretroviral treatment programmes in sub-Saharan Africa. AIDS. 2008;22:1897-1908.
4. 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.
5. Rosen S, Fox MP, Gill CJ. Patient retention in antiretroviral therapy programs in sub-Saharan Africa: a systematic review. PLoS Med. 2007;4:e298.
6. Boulle A, Van Cutsem G, Hilderbrand K, et al. Seven-year experience of a primary care antiretroviral treatment programme in Khayelitsha, South Africa. AIDS. 20 2010;24:563-572.
7. Bussmann H, Wester CW, Ndwapi N, et al. Five-year outcomes of initial patients treated in Botswana's National Antiretroviral Treatment Program. AIDS. 2008;22:2303-2311.
8. Sanne IM, Westreich D, Macphail AP, et al. Long term outcomes of antiretroviral therapy in a large HIV/AIDS care clinic in urban South Africa: a prospective cohort study. J Int AIDS Soc. 2009;12:38.
9. Lawn SD, Myer L, Orrell C, et al. Early mortality among adults accessing a community-based antiretroviral programme in South Africa: implications for programme design. AIDS. 2005;19:2141-2148.
10. Bekker LG, Myer L, Orrell C, et al. Rapid scale-up of a community-based HIV treatment service: programme performance over 3 conecutive years in Gugulethu, South Africa. S Afr Med J. 2006;96:315-320.
11. Lawn SD, Myer L, Harling G, et al. Determinants of mortality and nondeath losses from an antiretroviral treatment service in South Africa: implications for programme evaluation. Clin Infect Dis. 2006;43:770-776.
12. 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.
13. Lawn SD, Kaplan R, Wood R, Bekker LG. Promoting retention in care: an effective model in an antiretroviral treatment service in South Africa. Clin Infect Dis. 2007;45:803.
14. Lawn SD, Myer L, Edwards D, et al. Short-term and long-term risk of tuberculosis associated with CD4 cell recovery during antiretroviral therapy in South Africa. AIDS. 2009;23:1717-1725.
15. Boulle A, Bock P, Osler M, et al. Antiretroviral therapy and early mortality in South Africa. Bull World Health Organ. 2008;86:678-687.
16. Brinkhof MWG, Dabis F, Myer L, et al. Early loss of HIv-infected patients on potent antiretroviral therapy programmes in lower-income countries. Bull World Health Organ. 2008;86:559-567.
17. 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.
18. 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.
19. Fox MP, Rosen S. Patient retention in antiretroviral therapy programs up to three years on treatment in sub-Saharan Africa, 2007-2009: systematic review. Trop Med Int Health. 2010;15(Suppl 1):1-15.
© 2011 Lippincott Williams & Wilkins, Inc.