Secondary Logo

Journal Logo

Implementation Science

Extending Visit Intervals for Clinically Stable Patients on Antiretroviral Therapy: Multicohort Analysis of HIV Programs in Southern Africa

Haas, Andreas D. PhDa; Johnson, Leigh F. PhDb; Grimsrud, Anna PhDc; Ford, Nathan PhDd; Mugglin, Catarina MDa; Fox, Matthew P. DSce,f,g; Euvrard, Jonathan MPHb; van Lettow, Monique PhDh,i; Prozesky, Hans MMedj; Sikazwe, Izukanji MDk; Chimbetete, Cleophas MPHl; Hobbins, Michael PhDm; Kunzekwenyika, Cordelia MPHn; Egger, Matthias MDa,b; for IeDEA Southern Africa

Author Information
JAIDS Journal of Acquired Immune Deficiency Syndromes: August 1, 2019 - Volume 81 - Issue 4 - p 439-447
doi: 10.1097/QAI.0000000000002060

Abstract

INTRODUCTION

Since 2015, the World Health Organization (WHO) has recommended treating all people living with HIV, and today, more than 20 million people are on antiretroviral therapy (ART) with the goal to expand ART to reach about 30 million by 2020.1–4 WHO's 2016 HIV treatment guidelines recommend a differentiated care approach to meet the needs of a rapidly growing and increasingly diverse cohort of people on ART.2 Differentiated care is a public health approach to adapting service delivery to patient needs.2,5,6 Differentiated care has been widely adopted by countries with support from major donors including U. S. President's Emergency Plan For AIDS Relief, the Global Fund, and the Bill & Melinda Gates Foundation.7–9

Differentiated appointment spacing, that is, extending visit intervals for clinically stable patients to 3–6 months, is one of the key recommendations for differentiated care.2 National guidelines of several Southern African countries introduced extended visit intervals for patients stable on ART long before this recommendation was adopted by the WHO in 2016. For example, the 2004 South African National Antiretroviral Treatment Guidelines recommended that stable patients should be seen every 3 months.10 In Malawi, 2-monthly follow-up visits for stable patients were introduced in 2006.11

HIV programs need simple and reliable criteria for identifying clinically stable patients to implement differentiated care models safely and effectively. Early versions of national treatment guidelines did not define criteria for clinical stability.10,11 The WHO 2017 criteria for defining clinically stable patients include receiving ART for at least 1 year, no adverse drug reactions, no current illness, a good understanding of lifelong adherence, and evidence of treatment success. Programs with routine viral-load monitoring are advised to use virological criteria as a marker of treatment success (ie, 2 consecutive viral-load measurements below 1000 copies/mL), and programs with CD4 monitoring should use immunological criteria (ie, rising CD4 cell counts or a CD4 cell count >200 cells/µL).5,12 Programs without access to viral load or CD4 monitoring rely on clinical criteria to determine treatment success.13,14

We examined time trends in visit frequency in ART programs in Southern Africa from 2004 to 2017 to describe the scale-up of extended visit intervals. We studied associations between visit frequency and clinical, immunological, and virological criteria for clinical stability to explore whether health care workers used available monitoring tools to identify patients who could be seen less frequently.

METHODS

Antiretroviral Therapy Programs

The International epidemiology Databases to Evaluate AIDS (IeDEA) is a global collaboration of ART programs. We included 10 ART programs that participate in the Southern African region of IeDEA (IeDEA-SA).15 Data were collected at ART initiation (baseline) and each follow-up visit using standardized instruments. ART programs regularly transfer data sets to data centers at the Universities of Cape Town, South Africa, and Bern, Switzerland, for data curation and statistical analysis. ART programs vary in size: some programs like Tygerberg or Lighthouse operate at 1 or 2 large urban clinics, whereas other programs like CIDRZ support over 300 urban and rural clinics in several districts. Four cohorts from South Africa performed routine viral-load monitoring with viral-load testing and CD4 cell counts measured annually. In 2 programs from Zambia and Zimbabwe, patient monitoring was based mainly on annual CD4 cell counts, and viral loads were not routinely monitored. In 4 programs from Lesotho, Malawi, Mozambique, and Zimbabwe, monitoring was mainly based on clinical criteria.16

Outcomes and Definitions

Our primary outcome was the rate of clinic visits. A clinic visit was defined as attendance at a health facility for clinical assessment or pharmacy refill. Gaps between visits longer than 1 year were regarded as unscheduled treatment interruptions. Programs were classified as viral-load monitoring programs if most of their patients had a viral load measured during the first year on ART. Programs that measured the CD4 cell count for most of their patients during the first time on ART were classified as using CD4 monitoring. All other programs were deemed to be using clinical monitoring.

Clinical stability was defined according to WHO's 2017 criteria.5,6 We classified patients as clinically stable if they received ART for at least 1 year and met a criterion for clinical stability. We used a different criterion for clinical stability for each monitoring strategy: in programs with viral-load monitoring, we classified patients with a viral load of <1000 copies/mL as clinically stable; in programs with CD4 monitoring, patients with a CD4 cell count of >200 cells/µL were deemed clinically stable; and in programs using clinical monitoring, patients without current tuberculosis were classified as stable. Our definition of clinical stability in programs using clinical monitoring relied solely on diagnosis of tuberculosis because we had no data on symptoms of other HIV-related diseases or side effects. Patients on ART for at least 1 year who did not meet the criterion for clinical stability were classified as receiving a failing ART regimen. Clinical stability was defined as a time-varying covariate carried forward until a next laboratory measurement or OI start or end date was recorded. We categorized visit dates into 4 periods: calendar years 2004–2007, 2008–2011, 2012–2015, and 2016–2017. We used the WHO criteria to define the clinical stage.17 We defined the CD4 count at ART initiation as the value closest to the date of ART initiation within 3 months before and 1 month after initiation. CD4 cell counts were grouped into <200, 200–349, 350–500, and >500 cells/µL. Age at ART initiation was grouped into 16–24, 25–34, 35–49, and 50 years or older. Viral load at ART initiation was defined as the value closest to the date of ART initiation within 6 months before and 7 days after ART initiation. Time on ART was categorized in years: 1, 2, 3, 4, and 5–10.

Participants

HIV-1–infected patients aged ≥16 years who initiated ART between January 2004 and September 2017 and had at least 1 visit after the first year of ART were eligible for analysis. We excluded patients with insufficient data to define clinical stability by 12 months on ART (ie, patients from programs with viral-load monitoring without a viral-load measurement between 4 and 12 months after ART initiation and patients from programs with CD4 monitoring without a CD4 cell count measurement between 3 and 12 months after ART initiation) (Fig. 1).

FIGURE 1.
FIGURE 1.:
Flow of eligibility of patients.

Statistical Analysis

We used summary statistics to describe characteristics of patients at ART. We estimated incidence rate ratios (IRRs) and 95% confidence intervals (CIs) for predictors of visit frequency using univariable and multivariable Poisson regression models. In Poisson regression, each visit represents an event and the time between 2 consecutive visits is the exposure time. We used robust standard errors to adjust for clustering of visits at the patient level. Patients were followed-up from 1 year after initiation of ART for up to 10 years. We examined the following predictors: calendar year, time on ART, CD4 cell count, WHO clinical stage and age at ART initiation, and treatment program. Calendar year and time on ART were assessed at each visit and modeled as time-varying covariates. We added a continuous predictor for the proportion of patients receiving efavirenz-based ART at a program to assess whether the use of less-toxic regimen explained the relationship between calendar year and visit frequency.

We modeled associations between criteria for clinical stability and visit frequency in 2 steps to account for the heterogeneity between ART programs. We first ran multivariable Poisson regression models and estimated adjusted IRRs and 95% CIs for the effect of clinical stability on visit frequency in each ART program. Models adjusted for time on ART, CD4 cell count, WHO clinical stage and age at ART initiation, and calendar year. We then pooled IRRs by criterion for clinical stability using random effects meta-analysis. We present adjusted IRRs for the difference in the rate of visits between stable patients and patients failing ART in each treatment program in a forest plot.

Finally, we used multivariable flexible parametric survival analysis to examine time-dependent effects of clinical stability on the visit rate (ie, varying effects of clinical stability over time on ART) and to show absolute differences in the rate of visits between stable patients and patients receiving failing ART. In this survival analysis, each visit after ART initiation represents a failure event and multiple visits per patient are treated as multiple failures. Patients remain at risk of experiencing subsequent events after failure. Each visit date after ART initiation marks the end of the previous time interval and the beginning of the next interval (ie, gap between visits). We used Royston–Parmar models with restricted cubic splines and interaction terms between restricted cubic splines and predictors for clinical stability to model time-dependent effects.18–20 We used robust standard errors to adjust for clustering of multiple failure events within patients. We ran separate models for each criterion for clinical stability and predicted and plotted adjusted hazard rates and 95% CIs for specific levels of covariates. Models included predictors for clinical stability, calendar year, CD4 cell count at ART initiation, age, gender, and treatment program.

Missing WHO clinical stage and CD4 cell count at ART initiation were included as a separate category in all models. Unscheduled treatment interruptions (ie, 1 year gaps) were excluded from analyses. In sensitivity analysis, we set the duration of these intervals to the median visit interval for the treatment program and time on ART. Statistical analysis was performed in Stata (Version 15; Stata Corporation, College Station, TX).

Ethical Considerations

Local review boards and ethics committees for each treatment program that provided data approved the use of the data included in this study. The Cantonal Ethics Committee of the Canton of Bern, Switzerland, approved data merging and the collaborative analyses. Local review boards and the Cantonal Ethics Committee of the Canton of Bern waived the requirement to obtain informed consent.

RESULTS

Characteristics of Patients and ART Programs

As shown in Table 1, we included 180,837 patients from 10 ART programs in 7 countries: 38,045 (21.0%) patients came from 4 programs with viral-load monitoring; 85,555 (47.3%) from 2 programs with CD4 monitoring; and 57,237 (31.7%) from 6 programs with clinical monitoring. In South African ART programs, efavirenz (EFV)-based ART was used for at least half of patients since 2005. The proportion of patients receiving EFV-based ART increased further with the phase out of stavudine (d4T) in 2010–2011. Most other programs widely introduced EFV-based ART as the standard first-line regimen in 2013–2014 (see Figure S1, Supplemental Digital Content, http://links.lww.com/QAI/B314). We followed-up patients from January 1, 2004, up to August 31, 2017, for 876,801 person-years. The median follow-up time was 4.36 years [interquartile range (IQR) 2.55–6.88]. The median age at ART initiation was 35 years (IQR 30–42), and 117,240 (64.8%) of included patients were women. The median CD4 cell count at ART initiation was 171 cells/µL (IQR 91–264). Almost half of the patients with a known WHO clinical stage (46.3%, 68,901 of 148,809 patients) initiated ART in the stage 3 or 4.

TABLE 1.
TABLE 1.:
Patient Characteristics at Initiation of ART

Predictors of Visit Frequency

Table 2 shows unadjusted and adjusted IRRs for predictors of visit frequency in programs using viral-load monitoring, CD4 monitoring, and clinical monitoring. There was a trend toward fewer visits in more recent years across all monitoring strategies. Rates of visits declined especially after 2011. Adjusted IRRs for the years 2012–2015 compared with years 2004–2007 were 0.75 (95% CI: 0.75 to 0.76) in programs using viral-load monitoring, 0.54 (CI: 0.54 to 0.55) in programs using CD4 monitoring, and 0.77 (CI: 0.76 to 0.78) in programs using clinical monitoring. Time on ART and treatment program was also associated with visit frequency. Age, gender, and WHO clinical stage at ART initiation were weakly associated with the visit rate. In viral load and CD4 monitoring programs, a higher CD4 cell counts at the start of ART were associated with a slightly lower visit frequency. Adjusted IRRs for the change in visit frequency per 100% increase in the proportion of patients receiving EFV-based ART were 0.64 (CI: 0.62 to 0.66) in programs using viral-load monitoring, 0.56 (CI: 0.55 to 0.57) in programs using CD4 monitoring, and 0.88 (CI: 0.88 to 0.89) in programs using clinical monitoring. The proportion of patients receiving EFV-based ART partially explained the relationship between calendar year and visit frequency. After controlling for the proportion of patients receiving EFV-based ART, adjusted IRRs for the years 2016–2017 compared with years 2004–2007 changed from 0.68 (CI: 0.68 to 0.69) to 0.74 (CI: 0.73 to 0.74) in programs using viral-load monitoring, from 0.61 (CI: 0.60 to 0.61) to 0.88 (CI: 0.87 to 0.89) in programs using CD4 monitoring, and from 0.69 (CI: 0.69 to 0.70) to 0.77 (CI: 0.76 to 0.78) in programs using clinical monitoring (see Table S1, Supplemental Digital Content, http://links.lww.com/QAI/B314).

TABLE 2.
TABLE 2.:
Predictors of Visit Frequency

Clinical Stability and Visit Frequency

Clinically stable patients were seen less frequently than patients receiving failing regimens. Pooled adjusted IRRs comparing visit rates for clinically stable patients with patients on failing ART were 0.82 (95% CI: 0.73 to 0.90) for patients classified based on the virological criterion, 0.81 (0.69 to 0.93) for patients classified based on the clinical criterion, and 0.90 (0.85 to 0.96) for patients classified based on the immunological criterion for clinical stability. Stable patients were seen less frequently than patients on failing regimens in all treatment programs, but as shown in Figure 2, the strength of the association varied considerably.

FIGURE 2.
FIGURE 2.:
Forest plot of IRRs comparing clinically stable patients and patients receiving failing regimens. Forest plot of adjusted IRRs comparing visit rates between clinically stable patients and patients receiving failing ART regimens in 10 treatment programs in Southern Africa. Patients from programs using viral-load monitoring were classified based on a virological criterion for clinical stability; patients from programs using CD4 monitoring were classified based on an immunological criterion for clinical stability; and patients from programs using clinical monitoring were classified based on a clinical criterion for clinical stability. IRRs are adjusted for calendar year, time on ART, CD4 cell count, age, WHO clinical stage at ART initiation, and gender. Separate models were fitted for each treatment program, and estimates were pooled by criterion for clinical stability using random effects meta-analysis. N, number of patients in each treatment program at the beginning of follow-up (ie, 1 year after initiation of ART).

Figure 3 shows the adjusted hazard rate of visits by criteria for clinical stability. Stable patients were seen less frequently than individuals on a failing ART regimen in programs using virological, immunological, and clinical monitoring. However, absolute differences varied: differences were much larger for patients classified based on virological and clinical criteria than for patients classified based on the immunological criterion (about 0.5 visits compared with 2 visits per year). The same pattern was seen when we predicted the visit rate for other calendar years, treatment programs, or baseline CD4 cell counts.

FIGURE 3.
FIGURE 3.:
Rate of visits by clinical stability. Adjusted hazard rates of visits per year (solid lines) and 95% CIs (dashed lines) for clinically stable patients and patients receiving failing ART regimens. Patients from programs using viral-load monitoring were classified based on a virological criterion for clinical stability; patients from programs using CD4 monitoring were classified based on a immunological criterion for clinical stability; and patients from programs using clinical monitoring were classified based on a clinical criterion for clinical stability. Models adjusted for clinical stability, calendar year, CD4 cell count at ART initiation, age, gender, and treatment program. Hazard rates were predicted for years 2016–2017 for 25–34 years old female patients starting ART with a CD4 cell count of 200–349 cells/µL at the largest treatment program using the respective monitoring strategy.

Our results were not sensitive to the strategies for handling unscheduled treatment interruptions (ie, exclusion or replacement with medians).

DISCUSSION

The WHO has recently recommended longer visit intervals for clinically stable patients on ART,2 and countries have widely adopted differentiated ART delivery7–9 to ensure effectiveness and sustainability of HIV programs and to reduce the burden of care for patients and clinics.21–25 The frequency of visits is an important measure of the effectiveness of ART delivery, and associations between criteria for clinical stability and visit frequency indicate the level of differentiation of visit intervals.26 In this study, we described trends in the frequency of visits between 2004 and 2017 in 10 HIV programs in Southern Africa and examined associations between virological, immunological, and clinical criteria for clinical stability and the rate of clinic visits.

We observed a clear trend toward fewer visits in recent years, which was independent of the patients' clinical and immunological stage at the start of ART. This trend was partially explained by the proportion of patients receiving EFV-based ART. The virological and clinical criteria for clinical stability were moderately associated with the rate of visits, but the immunological criterion for clinical stability and other patient characteristics including WHO stage at ART initiation, gender, and age was only weakly associated. The association of fewer visits among stable patients was consistently found across all treatment programs, but the strength of the association varied considerably.

One possible explanation for time trends in visit frequency is that patients' better health at the start of ART in recent years allowed for programs to extend visit intervals. The WHO progressively increased the CD4 cell count thresholds for ART eligibility and the median CD4 cell count of patients at the start of ART increased substantially over the past decade.17,27–31 However, we adjusted our analysis for WHO clinical stage and CD4 cell count at ART initiation, and trends in visit frequency remained after controlling for these factors. Our data do not therefore support the hypothesis that trends in appointment spacing are driven by changes in ART eligibility criteria. We believe that other causes such as the introduction of less-toxic first-line regimens after 2010 and the widespread introduction of multimonth prescriptions were the drivers of the reduction in visit frequency.14,17,32 The finding that the trend toward fewer visits in recent years was partially explained by the proportion of patients receiving EFV-based ART supports our hypothesis that the introduction of less-toxic first-line regimens was one of the reasons for the reduction in visit frequency.

To implement differentiated ART delivery safely and effectively, programs need simple and reliable criteria for identifying clinically stable patients. Despite limited access to viral-load testing in several countries in this analysis, we see little evidence that programs were willing to use immunological criteria to differentiate visit intervals. By contrast, there was stronger evidence for differentiated appointment spacing based on virological and clinical criteria. A recent study from Zambia confirms our finding of little differentiation of appointment intervals based on immunological criteria. For every 50 cells/µL increase in the CD4 count, time between appointments increased by only 1 day.32 Sensitivity and positive predictive values of immunological criteria for identifying individuals with virological treatment failure are low,33 and caregivers may be reluctant to extend visit intervals based on these criteria because they may fear that virologically failing patients may not get the care they need.

The need for simple criteria for clinical stability and continued on-site supervision and mentoring to strengthen adherence to guidelines is further reinforced by data from Malawi which suggest that differentiating appointment spacing based on a complex set of criteria may not be feasible.14 According to Malawi's national guidelines, patients are eligible for 3-month ART refills if they meet the following criteria: 18 years or older, on first-line ART, no adverse drug reaction, no current illness or opportunistic infection, good adherence, and if viral-load testing was performed, then viral load <1000 copies/mL.13 A recent evaluation of differentiated ART delivery in Malawi shows that these criteria were not widely applied. A large proportion of eligible individuals did not receive 3-month ART refills, but many patients who were ineligible had been switched to extended refill intervals. In a large number of facilities, an equally large proportion of eligible and ineligible patients received 3-month refills.14

Our data suggest that caregivers are comfortably extending visit intervals for virologically suppressed patients. Viral load is a direct measure of treatment adherence and treatment efficacy that is easy to interpret and well trusted by health care workers. The study from Zambia showed that clinical stability is a highly transient state. Regimen switching, severe immunodeficiency, and new WHO Stage III/IV disease were common among patients who had reached clinical stability.32 Viral load enables early detection of treatment failure and early intervention.16,34 Despite additional costs for viral-load tests, differentiated care based on viral monitoring is a cost-effective strategy in low-resource settings.22 However, scale-up of viral-load testing is complex, and routine viral-load testing is not yet widely available in many resource-limited settings,35,36 and even where viral-load testing has been implemented, the results may not improve clinical management if not delivered in a timely manner or acted on when making clinical decisions.16,37,38

In the past decade, visit frequency has decreased substantially. Visits are anticipated to further decrease to 6-month intervals becausse countries fully implement the WHO 2016 guidelines for differentiated care.2 Widespread implementation of the most recent guidelines only began in mid-2016, and our study had insufficient follow-up time to explore the impact of the latest guideline change. Full implementation of the WHO 2016 guidelines has the potential to reduce the burden of treatment for patients, improve retention in care, decongest clinics, and reduce costs.22,23,39 The clinical and public health impact of differentiated care is an important area for future evaluation.

Our study included data from a broad range of treatment programs with different monitoring strategies, and we could examine associations between virological, immunological, and clinical criteria for clinical stability and visit schedules. With long-term follow-ups, we could examine patterns in visit spacing from the beginning of the scale-up of ART in Africa until 2017. Among the study's limitations, multimonth ART prescribing for stable patients on ART is a complementary approach to differentiated service delivery that is being increasingly adopted in sub-Saharan Africa.14,23 We also had insufficient data to distinguish between visits with clinical consultation and pharmacy refill visits without clinical consultation, and therefore, could not evaluate the potential benefit of fewer clinical follow-up visits. We had insufficient data on adverse drug reactions and adherence and could not consider these criteria in our definition of clinical stability. Although our study included data from a large number of urban and rural primary, secondary, and tertiary facilities, the programs participating in IeDEA may not be representative of the national ART program in the different countries. Finally, we had little data on opportunistic infections after start of ART, and our definition of clinical stability in clinical monitoring programs relied solely on diagnoses of current tuberculosis.

ART programs in Southern Africa reduced visit frequency over the past decade. Differences in visit rates between stable patients and patients failing ART were variable and modest overall. Because larger differences were seen particularly in programs using virological rather than immunological criteria for stability, we conclude that greater access to routine viral-load monitoring may increase scale-up of differentiated ART delivery.

ACKNOWLEDGMENTS

The authors thank all patients, doctors, and study nurses in the participating facilities. The authors thank Christopher Ritter (Institute of Social and Preventive Medicine, University of Bern, Switzerland) for editorial assistance.

REFERENCES

1. World Health Organization. Guideline on When to Start Antiretroviral Therapy and on Pre-exposure Prophylaxis for HIV: World Health Organization; 2015.
2. World Health Organization. Consolidated Guidelines on the Use of Antiretroviral Drugs for Treating and Preventing HIV Infection Recommendations for a Public Health Approach. 2nd ed. Geneva, Switzerland: WHO; 2016.
3. Joint United Nations Programme on HIV/AIDS (UNAIDS). AIDSInfo Online Database: World Health Organization. Available at: http://aidsinfo.unaids.org/. Accessed May 2, 2019.
4. Joint United Nations Programme on HIV/AIDS (UNAIDS). 90-90-90 an Ambitious Treatment Target to Help End the AIDS Epidemic: UNAIDS; 2014.
5. World Heatlh Organization (WHO). Key Considerations for Differentiated Antiretroviral Therapy Delivery for Special Populations: Children, Adolescents, Pregnant and Breastfeeding Women and Key Populations. Geneva, Switzerland: WHO; 2017.
6. World Heatlh Organization (WHO). Guidelines for Managing Advanced HIV Disease and Rapid Initiation of Antiretroviral Therapy. Geneva, Switzerland: WHO; 2017.
7. US Presient's Emergency Plan for AIDS Relief (PEPFAR). PEPFAR Technical Considerations for COP/ROP 2016: US Department of State; 2016.
8. The Global Fund. Differentiated Care For HIV and Tuberculosis: A Toolkit for Health Facilities. Geneva, Switzerland: The Global Fund; 2015.
9. Duncombe C, Rosenblum S, Hellmann N, et al. Reframing HIV care: putting people at the centre of antiretroviral delivery. Trop Med Int Health. 2015;20:430–447.
10. National Department of Health South Africa. National Antiretroviral Treatment Guidelines: National Department of Health; 2004.
11. Ministry of Health Malawi. Guidelines for the Use of Antiretroviral Therapy in Malawi: Ministry of Health Malawi; 2006.
12. Waldrop G, Doherty M, Vitoria M, et al. Stable patients and patients with advanced disease: consensus definitions to support sustained scale up of antiretroviral therapy. Trop Med Int Heal. 2016;21:1124–1130.
13. Ministry of Health Malawi. Clinical Management of HIV in Children and Adults: Ministry of Health Malawi; 2011.
14. Prust ML, Banda CK, Nyirenda R, et al. Multi-month prescriptions, fast-track refills, and community ART groups: results from a process evaluation in Malawi on using differentiated models of care to achieve national HIV treatment goals. J Int AIDS Soc. 2017;20(suppl 4):41–50.
15. Egger M, Ekouevi DK, Williams C, et al. Cohort Profile: the international epidemiological databases to evaluate AIDS (IeDEA) in sub-Saharan Africa. Int J Epidemiol. 2012;41:1256–1264.
16. Haas AD, Keiser O, Balestre E, et al. Monitoring and switching of first-line antiretroviral therapy in adult treatment cohorts in sub-Saharan Africa: collaborative analysis. lancet HIV. 2015;2:e271–8.
17. World Health Organization. Antiretroviral Therapy for HIV Infection in Adults and Adolescents Recommendations for a Public Health Approach: World Health Organization; 2010.
18. Royston P, Parmar MKB. Flexible parametric proportional-hazards and proportional-odds models for censored survival data, with application to prognostic modelling and estimation of treatment effects. Stat Med. 2002;21:2175–2197.
19. Royston P, Lambert PC. Flexible Parametric Survival Analysis Using Stata: Beyond the Cox Model; 2011.
20. Lambert PC, Royston P. Further development of flexible parametric models for survival analysis. Stata J. 2009;9:265–290.
21. Barker C, Dutta A, Klein K. Can differentiated care models solve the crisis in HIV treatment financing? Analysis of prospects for 38 countries in sub-Saharan Africa. J Int AIDS Soc. 2017;20(suppl 4):21648.
22. Working Group on Modelling of Antiretroviral Therapy Monitoring Strategies in Sub-Saharan Africa. Phillips A, Shroufi A, Vojnov L, et al. Sustainable HIV treatment in Africa through viral-load-informed differentiated care. Nature. 2015;528:S68–S76.
23. Mutasa-Apollo T, Ford N, Wiens M, et al. Effect of frequency of clinic visits and medication pick-up on antiretroviral treatment outcomes: a systematic literature review and meta-analysis. J Int AIDS Soc. 2017;20(suppl 4):21647.
24. Grimsrud A, Bygrave H, Wilkinson L. The case for family-centered differentiated service delivery for HIV. J Acquir Immune Defic Syndr. 2018;78(suppl 2):S124–S127.
25. Grimsrud A, Lesosky M, Kalombo C, et al. Implementation and operational research: community-based adherence clubs for the management of stable antiretroviral therapy patients in Cape Town, South Africa: a cohort study. J Acquir Immune Defic Syndr. 2016;71:e16–e23.
26. Ehrenkranz PD, Calleja JM, El-Sadr W, et al. A pragmatic approach to monitor and evaluate implementation and impact of differentiated ART delivery for global and national stakeholders. J Int AIDS Soc. 2018;21. doi:10.1002/jia2.25080.
27. World Health Organization. Scaling up Antiretroviral Therapy in Resource Limited Settings: Guidelines for a Public Health Approach. Geneva, Switzerland: WHO; 2002.doi:10.1097/00019048-200203000-00012.
28. World Health Organization. Antiretroviral therapy for HIV Infection in adults and adolescents in resource-limited settings: towards universal access. Recommendations for a Public Health Approach, 2006: World Health Organization.
29. World Health Organization. Consolidated Guidelines on the Use of Antiretroviral Drugs for Treating and Preventing HIV infection. Recommendations for a Public Health Approach: World Health Organization; 2013.
30. Chan AK, Ford D, Namata H, et al. The Lablite project: a cross-sectional mapping survey of decentralized HIV service provision in Malawi, Uganda and Zimbabwe. BMC Health Serv Res. 2014;14:352.
31. IeDEA and COHERE Cohort Collaborations. Global trends in CD4 cell count at the start of antiretroviral therapy: collaborative study of treatment programs. Clin Infect Dis. 2018;66:893–903.
32. Roy M, Holmes C, Sikazwe I, et al. Application of a multi-state model to evaluate visit burden and patient stability to improve sustainability of HIV treatment in Zambia. Clin Infect Dis. 2018;8:11–83.
33. Rutherford GW, Anglemyer A, Easterbrook PJ, et al. Predicting treatment failure in adults and children on antiretroviral therapy: a systematic review of the performance characteristics of the 2010 WHO immunologic and clinical criteria for virologic failure. AIDS. 2014;28 (suppl 2):S161–S169.
34. Jobanputra K, Parker LA, Azih C, et al. Impact and programmatic implications of routine viral load monitoring in Swaziland. J Acquir Immune Defic Syndr. 2014;67:45–51.
35. Médecins Sans Frontières Access Campaign. Achieving undetectable: what questions remain in scaling-up HIV virologic treatment monitoring? Available at: https://msfaccess.org/sites/default/files/MSF_IssueBrief_undetectable6_10.pdf. Accessed May 2, 2019.
36. Roberts T, Cohn J, Bonner K, et al. Scale-up of routine viral load testing in resource-poor settings: current and future implementation challenges. Clin Infect Dis. 2016;62:1043–1048.
37. Peter T, Zeh C, Katz Z, et al. Scaling up HIV viral load—lessons from the large-scale implementation of HIV early infant diagnosis and CD4 testing. J Int AIDS Soc. 2017;20–e25008.
38. Awungafac G, Amin ET, Fualefac A, et al. Viral load testing and the use of test results for clinical decision making for HIV treatment in Cameroon: an insight into the clinic-laboratory interface. PLoS One. 2018;13:e0198686.
39. Mody A, Roy M, Sikombe K, et al. Improved retention with 6-month clinic return intervals for stable human immunodeficiency virus-infected patients in Zambia. Clin Infect Dis. 2018;66:237–243.
Keywords:

HIV; differentiated antiretroviral therapy delivery; differentiated service delivery; differentiated care; stable patients; Africa

Supplemental Digital Content

Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved.