Over the last decade, scale-up of antiretroviral therapy (ART) to reach more than 12 million persons living with HIV (PLHIV) in low- and middle-income countries (LMIC) has contributed to population-level reductions in mortality1 and declines in HIV incidence.2 In Mozambique, ranked 178th out of 187 countries on the Human Development Index, with <0.3 doctors per 10,000 population,3 scale-up of ART to treat more than 800,000 PLHIV by 2016 is a highlight of global health achievements.2 However, with >75% of HIV program funding coming from international donors4 and plateaued international funding since 2009, Mozambique's ART program operates in a very resource-constrained environment, where monitoring of service quality and outcomes is especially important.5,6 For example, an evaluation of Mozambique's national ART program during 2004–2007 reported concerning increases in rates of loss to follow-up (LTFU) and recommended future investigation of this trend.7 In addition, Mozambique faces a new challenge in 2016 as it plans to phase in test-and-treat guidelines recommended by the World Health Organization (WHO),8 which will further increase strain on health facilities as more patients become ART eligible.
Therefore, to inform future ART scale-up in this challenging environment, the Ministry of Health (MOH) and partners initiated a program evaluation to answer 3 primary questions: (1) what trends can be observed in ART patient characteristics and outcomes over the first decade of scale-up, (2) what lessons can be learned from Mozambique's experience implementing universal treatment for pregnant and breastfeeding women (Option B+) that might inform phased rollout of test-and-treat guidelines starting in 2016, and (3) can ART service delivery to stable patients on ART >6 months through Community ART Support Groups (CASGs) help reduce LTFU rates from the national ART program? Although incidence of LTFU among CASG participants was reported to be low in a small single-province pilot,9,10 outcomes of the nation-wide pilot and the effect of CASG participation on LTFU in a comparative analysis have not yet been reported.
ART Eligibility and Monitoring
Mozambique's ART guidelines changed over time (see Table, Supplemental Digital Content 1, http://links.lww.com/QAI/A862). There have been 3 guideline phases: 2004–2009, 2010–2012, and 2013–2016. During these phases, adults with WHO stage I/II were eligible if CD4 was ≤200, ≤250, and ≤350 cells per microliter, respectively.11 In addition, since 2013, all HIV-infected pregnant women have been eligible for life-long ART through prevention of mother-to-child transmission Option B+ (referred to as Option B+ in this manuscript).12
At ART initiation, monthly for 4 months, at 6 months, and then semi-annually, patients attend clinician check-ups and standardized MOH-recommended records are completed. Most patients collect medications monthly from clinic pharmacies. For all patients late for monthly ART pick-up appointments, text messaging, followed by telephonic tracing, and if necessary, home visits, are recommended, but resource-limitations constrain tracing activities.
Community ART Support Groups
CASGs were first piloted in Tete province in 2008 to address increasing LTFU and alleviate patient-load on overburdened clinics.10 CASGs are groups of usually 6 patients, who take turns at monthly medication collections for the entire group, meaning that individual patients only attend the health facility semi-annually instead of monthly. To be CASG-eligible, adult ART patients must be stable, have taken ART for >6 months, have a CD4 >200 cells per microliter, and no active WHO stage III/IV conditions. A small pilot in Tete showed low LTFU among CASG participants,9,10 and MOH expanded the pilot to 69 ART facilities nation-wide during 2010–2013.
Study Design and Population
This was an observational cohort study analyzing patient-level data prospectively entered into EPTS (a Microsoft Access database) by trained data clerks during 2004–2013. By December 2013, 170 of 288 adult ART facilities (60%) nationally were using EPTS covering 67% of all ART patients. Built-in consistency and legal range checks help maintain EPTS data quality. Semi-annually, data are transferred to a central data warehouse, co-managed by MOH and other stakeholders for further quality checks, data set concatenation, and data management.
For this analysis, only adults ≥15 years old at ART initiation, who started ART during 2004–2013, were eligible. Facility-level databases were closed in April 2014.
Primary ART outcomes were mortality and LTFU, whereas attrition (documented death or LTFU) was a secondary outcome.13 Per Mozambique standards, patients were considered LTFU if ≥60 days late for their next scheduled medication pick-up appointment. Mortality ascertainment occurred through passive reporting, and through MOH-recommended tracing activities.
Exposure variables are listed in Tables 1 and 2. Sex was coded as a 3-level variable (male, nonpregnant female, and pregnant female).14 Rate of site expansion was coded as patient rank divided by site-specific duration of ART services (eg, for a site's 100th ART enrollee, enrolled 4 months after ART service initiation, the expansion rate is 25).6 For each site, size was defined by calendar year as the number of current ART enrollees by calendar year end.6
Data were analyzed using STATA 13 (StataCorp, 2009, Stata Statistical Software, Release 13, College Station, TX).
Complete data were available for time-to-event analysis and for key exposure variables, including ART initiation year, CASG participation, age, sex, site expansion, and site size. For certain other variables (Tables 1 and 2), some data were missing.
To best manage missing covariate data, multiple imputation is preferred to complete case analysis,15,16 because complete case analysis can yield biased parameter estimates.16 First, patterns of missing data were explored using the “mi misstable patterns” command to assess plausibility of the missing at random assumption. Although the missing at random assumption cannot be proven, it was considered plausible for this analysis. Therefore, the mi17 procedure in STATA was used to create 20 imputed data sets for each outcome (death and LTFU) separately.7 Missing covariate data were only imputed if <35% of data were missing. For all analyses using imputed data, estimates were combined across imputed data sets according to Rubin's rules.18 Intrafacility correlation was accounted for using the generalized Huber/White/sandwich estimator of robust standard errors.19,20
To evaluate trends in patient characteristics over time, associations between baseline characteristics and ART initiation year were assessed using regression models appropriate for variable type. To evaluate outcome trends over time, competing risk models were used to estimate mortality and LTFU for each annual cohort of ART enrollees during 2004–2013.13,21–23 Because a nationally representative study to ascertain vital status of patients LTFU has not yet been implemented, a published nomogram, based on a meta-regression analysis of 15 LTFU tracing studies from sub-Saharan Africa, was used to estimate true 1-year mortality incidence for each annual cohort.24
To evaluate outcome determinants, proportional hazards regression models were used to estimate adjusted hazard ratios (AHRs).13,22 Multivariable models were adjusted for covariates considered a priori risk factors: year of ART initiation, age, sex, marital status, number of children, education, employment, referral source, WHO stage, baseline weight, baseline CD4, patient's CASG participation, site-level availability of CASG, rate of site expansion, site size, and ART regimen. CASG participation after ART initiation was coded as time-varying in multivariable models to avoid survivor bias, per published precedent.25 The proportional hazards assumption was assessed using visual methods and the Grambsch and Therneau test.
Use of routine, anonymized data for this study was approved by the Mozambican Ethics Committee and the Center for Global Health Science Office at the Centers for Disease Control and Prevention.
Trends in Patient Characteristics at ART Initiation
During 2004–2013, 306,335 adults initiated ART at facilities with EPTS. Annual adult ART enrollment rates increased 37-fold from 1784/year in 2004 at 9 facilities to 65,442/year in 2013 at 170 facilities. During 2004–2013 at ART enrollment, the percentage who were male declined from 45% to 27% and the percentage who were nonpregnant females from 55% to 51%, whereas the percentage who were pregnant females increased from 0.3% to 22% (P < 0.001), with the steepest increase occurring during 2012–2013 (from 10% to 22%) (Table 1). In 2013, following initiation of Option B+, pregnant females with CD4 count >350 cells per microliter and WHO stage I/II represented 8% of all 2013 enrollees, 11% of all females, and 35% of all pregnant females (up from 0% in 2004). The female-to-male ART enrollee ratio increased from 1.2 to 2.7 during 2004–2013.
Median age at ART enrollment declined from 35.9 in 2004 to 31.1 years in 2013 (P < 0.001). In each calendar year, median age of pregnant female ART enrollees was lower than that of nonpregnant females and males (Table 1). During 2004–2013, the percentage referred after HIV testing and counseling (HTC) at health facilities decreased from 68% to 34%, whereas the proportion referred from antenatal care (ANC) increased from 0% to 24% (P < 0.001) with the steepest annual increase occurring during 2012–2013 (from 12% to 24%).
Median time from HIV diagnosis to ART initiation declined substantially from 4.27 to 0.46 years during 2004–2013 (P < 0.001) with declines similar for males, nonpregnant females, and pregnant females (Table 2). WHO stage IV prevalence at ART initiation declined from 28% to 4% during 2004–2013 (P < 0.001). Median CD4 at ART initiation increased from 139 to 235 per microliter during 2004–2013 (P < 0.001), with increases most pronounced for pregnant females (81–309 per microliter) compared with males (129–187 per microliter) and nonpregnant females (150–231 per microliter) (Table 2).
During 2004–2009, stavudine, lamivudine, and nevirapine were the most common regimens, accounting for 73%–88% of regimens (see Figure and Table, Supplemental Digital Content 2, http://links.lww.com/QAI/A921). However, during 2010–2012, zidovudine, lamivudine, and nevirapine became more common, accounting for 58%–81% of regimens. During 2012–2013, increased tenofovir, lamivudine, and efavirenz use was noted for males (from 0% to 12%), nonpregnant females (from 0% to 16%), and pregnant females (from 0% to 40%). During 2004–2013, stavudine use declined from 92% to 7%, whereas zidovudine and tenofovir use increased from 5% to 71% and 0%–20%, respectively. Nevirapine use declined from 91% to 63%, whereas efavirenz use increased from 6% to 35%.
During 2010–2013, 6766 patients joined a CASG at 69 ART facilities. Compared with non-CASG participants, CASG participants were more likely to be female (67% vs. 76%, P < 0.001), be older at ART enrollment (median age 33.0 vs. 36.2 years, P < 0.001), have no education (13% vs. 21%, P < 0.001), be unemployed (68% vs. 77%, P = 0.033), and have had a longer time from diagnosis to ART initiation (1.48 vs. 4.34 years, P < 0.001) (Table 3). However, at ART enrollment, WHO stage IV disease prevalence (10% vs. 8%, P = 0.315) and median CD4 count (182 vs. 195 per microliter, P = 0.463) were similar between participants and nonparticipants in CASG. Twenty percent of CASG enrollees joined CASG in months 0–12, 23% in months 12–24, and 58% after 24 months of ART.
Trends in Outcomes
Over 517,608 years of ART follow-up, 26,910 (9%) patients transferred out, 113,420 (37%) were LTFU, and 16,486 (5%) died. Overall, documented mortality was 4% and LTFU 27% at 1 year of follow-up. After correcting for estimated rates of undocumented mortality among those LTFU, corrected 1-year mortality was about 13% and corrected LTFU 18% (see Table, Supplemental Digital Content 3, http://links.lww.com/QAI/A922).
During 2004–2013, observed 6-month mortality declined from 7% to 2% but LTFU increased from 24% to 30% (Table 4). Similarly, during 2004–2012, documented 1-year mortality declined from 9% to 3%, but 1-year LTFU increased from 26% to 31%. Corrected 1-year mortality declined from about 17% among 2004 enrollees to about 13% among 2012 enrollees, whereas corrected 1-year LTFU increased from about 18% to 21% (see Table, Supplemental Digital Content 3, http://links.lww.com/QAI/A922).
After 9 years of ART for 2004 enrollees, 13% had died and 51% were LTFU, giving overall retention of 37% (Table 4).
In bivariate regression, more recent year of ART initiation was predictive of lower documented mortality rates, (See Table, Supplemental Digital Content 4, http://links.lww.com/QAI/A923). However, in multivariable analysis, no statistically significant association between year of ART initiation and mortality was observed (Table 5). In contrast, later calendar year of ART initiation was associated with higher LTFU in both crude and multivariable analyses (Table 5).
Among CASG participants, incidence of death after ART initiation was 0.3% at 2 years and 1.4% at 4 years, whereas LTFU incidence was 2.9% at 2 years and 10.1% at 4 years. After controlling for confounders (all variables listed in Table 5), and introducing CASG participation as a time-varying covariate to avoid survivor bias, compared with nonparticipation in CASG, CASG participation was associated with 35% lower LTFU rates [AHR 0.65, 95% confidence interval (CI), 0.46–0.91], but similar mortality (Table 5). In sensitivity analyses, restricting the cohort to the 128,364 enrollees starting ART during 2010–2013 at only those 69 clinics offering CASG programs, CASG participation was associated with 55% reduced LTFU rates (AHR 0.45, 95% CI: 0.32 to 0.64), but similar mortality (See Table, Supplemental Digital Content 5, http://links.lww.com/QAI/A924).
Option B+ Outcomes
In 2013, 6-month mortality was lowest among the Option B+ group of pregnant women enrolled with CD4 >350 cells per microliter and WHO stage I/II (0.1%), compared with other pregnant ART enrollees (1%), nonpregnant females (3%), and males (5%); however, 6-month LTFU was highest for the Option B+ group (38%), compared with other pregnant ART enrollees (26%), nonpregnant females (18%), and males (23%).
Other Outcome Predictors
In multivariable analysis, compared with nonpregnant females, males had higher LTFU and mortality, and pregnant females had higher LTFU but lower mortality (Table 5).
In multivariable analysis, compared to patients with CD4 <50/μL at ART initiation, patients with CD4 counts between 50/μL and 500/μL had 7%–15% lower LTFU (Table 5). However, LTFU rates were similar between those with CD4 <50/μL (23.9/100 person-years) and CD4 >500/μL (32.0/100 person-years). In contrast, compared to patients with CD4 <50 μL, all patients with higher CD4 counts had 36%–54% lower mortality (Table 5). Other markers of advanced disease (ie, advanced WHO stage and low weight at ART initiation) were predictive of both LTFU and mortality (Table 5).
Compared with zidovudine-containing regimens, stavudine-containing regimens had 48% higher mortality, but similar LTFU. Compared with nevirapine-containing regimens, efavirenz-containing regimens had 9% higher LTFU but similar mortality, whereas protease inhibitor-containing regimens had 29% higher LTFU but similar mortality. AHRs derived using the multiple imputation approach were similar to AHRs derived using a complete case analysis approach (See Table, Supplemental Digital Content 6, http://links.lww.com/QAI/A925).
With a cohort size of 306,335, this is the largest single-country cohort of adult ART enrollees described to date.26 The analysis is timely in that it exemplifies a key success (ie, declines in observed ART mortality rates), a key challenge for the future (ie, increases in observed LTFU rates), and a potential partial solution to address the problem of increasing LTFU (ie, ART distribution through CASG). Other notable findings include the high 6-month LTFU among Option B+ enrollees with CD4 >350 cells per microliter and WHO stage I/II at ART initiation and the high LTFU rates among patients starting ART with CD4 >500 cells per microliter. Both these findings have implications for Mozambique as it begins to pilot test-and-treat in 2016.
Similar to other studies in LMIC,5,6,26 observed ART mortality rates were lower among enrollees in more recent calendar years compared with enrollees in earlier years. Since in crude analysis, later year of ART initiation was associated with lower observed mortality rates, but in multivariable analysis, this association disappeared, changes in patient characteristics at ART initiation over calendar time (eg, declining percentage of enrollees who were male, declining prevalence of CD4 <50/μL and declining stavudine use), explain, at least partly, the declining ART mortality in our analysis. Since 20%–60% of patients LTFU from ART are likely to have died,27 we used a published nomogram to explore whether a trend of declining 1-year ART mortality was still present after correcting for estimated undocumented death. Point estimates of corrected 12-month mortality declined from 17% in 2004 to 13% in 2012, but nomogram-generated 95% CIs for these estimates overlapped. Although observed data suggest that ART mortality has declined during 2004–2013, a tracing study to ascertain outcomes of patients LTFU is needed, and is planned, to confirm these findings.
Given the association between patient characteristics at ART initiation and outcomes, understanding patient characteristics and changes over time is important for program managers. The increasing female-to-male ART enrollee ratio from 1.2 to 2.7 during 2004–2013 was probably due to scale-up of HTC at ANC and Option B+ initiation in 2013 and is not explained by slight increases in population-level female-to-male ratios among PLHIV (from 1.43 to 1.49).28,29 As discussed in depth separately,29 increased ART enrollment among men is needed to reduce disproportionately high HIV-related morbidity and mortality among men30 and reduce HIV incidence among their sexual partners.29 Increases in median CD4 count at ART initiation, and declining prevalence of WHO stage IV, similar to other reports,5,31–33 probably reflect expanded HTC, ART access, and ART eligibility. Expanding ART access and ART eligibility are also reflected in the remarkable declines in time from HIV diagnosis to ART initiation during 2004–2013. However, more rapid average annual increases in median baseline CD4 for nonpregnant females (9.1 cells·μL−1·yr−1) and pregnant females (25.3 cells·μL−1·yr−1) compared with males (6.4 cells·μL−1·yr−1) suggest ART access increased more rapidly for females than males. This highlights the importance of ANC HTC and Option B+ as gateways to early ART for females, and the need to identify gateways that work for men.34
Due to stavudine-induced severe adverse events, WHO recommended stavudine phase-out in 2009.35 Mozambique responded rapidly with significant declines in first-line stavudine prescription during 2009–2010, and by 2013, only 7% of adults started stavudine-containing first-line regimens. This study contributes new evidence that, compared with zidovudine, first-line stavudine-containing regimens carry mortality risk.26,35 Complete stavudine phase-out is required, especially since costs of tenofovir and zidovudine have declined substantially.36
Similar to prior Mozambique analyses7 and other studies from LMIC,6,37,38 observed ART LTFU rates increased over time, being nearly 3-fold higher among 2013 enrollees than 2004 enrollees. Unlike the mortality analysis, later calendar year of ART initiation was associated with higher LTFU in both crude and adjusted regression, suggesting that unmeasured factors associated with calendar time explain increasing LTFU. With only 3 pharmacy staff per 100,000 people, the increasing patient-to-pharmacist ratio has been cited as a cause for increasingly long clinic wait times for ART patients,39 resulting in patient dissatisfaction and LTFU.39–41 Per precedent,42 we used rate of site expansion and site size as proxy variables for patient-to-provider ratios, but our proxy variables were not important explanatory variables in the LTFU regression. One explanation, as reported previously,39 is that patient-to-provider ratios do not correlate with pharmacist-to-patient ratios, which may be the underlying site-level factor determining LTFU rates.39 Alternately, other factors might explain increasing LTFU such as increasing undocumented transfer between health facilities or declines in recordkeeping quality.43,44 In addition, a recent model-based analysis reported that treatment interruptions are more likely to be misclassified as LTFU in recent cohorts compared with early cohorts, and this might explain some of the observed increases in LTFU rates over time.44 This highlights the need for an LTFU tracing study, which could help to identify drivers of LTFU and help determine to what extent observed LTFU is true LTFU.45
By including CASG participation as a time-varying covariate and controlling for confounders, this is the first attempt to quantify effect of CASG participation on LTFU rates and the first nation-wide report of CASG outcomes.9,10 This analysis complements a separately described propensity score-matched cohort analysis; both analytic approaches observed a statistically significant LTFU reduction associated with CASG participation. For both analytic approaches, a limitation is that agreeing to join a CASG might be correlated with better health-seeking behavior, which might result in better retention regardless of CASG participation.9 For example, CASG participants had a longer median time from HIV diagnosis to ART initiation, which might suggest they sought HIV testing earlier than nonparticipants, a possible indicator of health-seeking behavior. However, considering that CASG scale-up carries other benefits for the ART program, including decongestion of ART facilities and reduced burden on providers,9,46 and appears to be highly preferred by most CASG participants,46 these findings warrant continued CASG scale-up. Since these data were analyzed, CASG scale-up has continued and by April 2016, about 64,932 (8%) of all 802,659 patients were enrolled in CASGs nation-wide.
Notably, CASG participants were more commonly unemployed and uneducated than nonparticipants, which might indicate that CASG participation is more attractive for patients with fewer financial resources.9,10 This makes intuitive sense since key patient benefits of CASG participation, as reported by CASG participants, are the cost- and time-savings associated with biannual rather than monthly clinic attendance.46 Low male uptake of CASG warrants further research. Offering male-only CASGs might improve male participation in CASG.47 Alternately, other service delivery models that emphasize privacy might be preferred by some stable male patients.48 Possible reasons for lower LTFU following CASG participation include reduced patient transport costs, reduced patient time at the clinic, increased patient accountability, and improved social support.10 Given the shortage of healthcare workers, task-shifting to patients might be a cornerstone of future ART expansion and sustained ART coverage in Mozambique and similar LMIC.10,49–53
Similar to recent program data from South Africa54 and Canada,55 but in contrast to research cohort data from LMIC,42 ART initiation at CD4 >500/μL was associated with LTFU.42 In our programmatic setting, patients who feel healthy at ART initiation and experience dissatisfaction with monthly ART refills at crowded facilities might be at higher LTFU risk than those who feel sicker.56–58 Further research to understand LTFU causes among patients with CD4 >500/μL at ART initiation is needed to inform test-and-treat rollout in a way that maximizes patient health and HIV prevention benefits.8,56
Similar to most reports,59–61 pregnancy at ART initiation was associated with higher LTFU but lower mortality. In contrast to pregnant females, who were mostly referred from routine HTC at ANC, nonpregnant women were largely referred from clinics and VCT centers where they were seeking healthcare at the time of diagnosis. Therefore, inferior motivation for life-long ART might explain higher LTFU among pregnant versus nonpregnant enrollees.62 Other barriers to ART retention among pregnant women might include declining motivation to stay on ART post-partum, new financial constraints, new childcare responsibilities, post-partum depression, and navigating complex referral pathways.62
Our 6-month LTFU among Option B+ enrollees with CD4 >350 and WHO stage I/II at ART initiation (38%) is much higher than reports from Malawi (17%).63 However, similar to Malawi, the Option B+ group had higher LTFU than pregnant woman starting ART per national standards at CD4 ≤350 or WHO stage III/IV.63 Our analysis suggests that high CD4 count at ART initiation, which may be associated with “feeling healthy,” might partly explain higher LTFU in the prevention of mother-to-child transmission B+ group compared with other pregnant ART enrollees, but further research is needed.64 Models of Option B+ that facilitate retention are urgently needed.
Strengths of the analysis include the large cohort size, number of clinics included (170), duration of observation (10 years), and proportion of Mozambique's adult ART population captured (67%). As with all observational studies, some associations might be confounded by unmeasured confounders. For example, the association between efavirenz use and LTFU might be confounded by higher likelihood of prescribing efavirenz-based regimens to TB co-infected enrollees.26 Other limitations include the fact that findings are limited to clinics with EPTS and the possibility that missing covariate data introduced nondifferential measurement error. In addition, despite MOH-recommended tracing efforts, reported LTFU rates are probably overestimates and observed mortality rates underestimates because of incomplete death documentation.27
Mozambique's rapid expansion of ART access is a significant national and global health achievement. This analysis suggests that initiation of better ART regimens at earlier disease stages in later calendar years at least partly explains declining ART mortality. However, LTFU rates increased over time and interventions to reverse this trend are needed. Further scale-up of CASG is one intervention to help reverse trends of increasing LTFU. In addition, targeted LTFU prevention strategies to address high LTFU among pregnant women overall, Option B+ enrollees specifically, and all patients starting ART with CD4 >500/μL are needed.
1. Reniers G, Slaymaker E, Nakiyingi-Miiro J, et al. Mortality trends
in the era of antiretroviral therapy
: evidence from the network for analysing longitudinal population based HIV/AIDS data on Africa (ALPHA). AIDS. 2014;28(suppl 4):S533–S542.
2. World Health Organization. Global Update on the Health Sector Response to HIV; 2014. Available at: http://www.who.int/hiv/pub/progressreports/update2014/en/
. Accessed April 10, 2015.
3. World Health Organization. Global Health Observatory Data Repository. Available at: http://apps.who.int/gho/data/node.country.country-MOZ?lang=en
. Accessed April 10, 2015.
4. Joint United Nations Programme on HIV/AIDS. Global Report: UNAIDS Report on the Global AIDS Epidemic 2013. Available at: http://www.unaids.org/en/media/unaids/contentassets/documents/epidemiology/2013/gr2013/UNAIDS_Global_Report_2013_en.pdf
. Accessed February 17, 2014.
5. Auld AF, Ekra KA, Shiraishi RW, et al. Temporal trends
in treatment outcomes for HIV-1 and HIV-2-infected adults enrolled in cote d'Ivoire's national antiretroviral therapy
program. PLoS One. 2014;9:e98183.
6. Grimsrud A, Balkan S, Casas EC, et al. Outcomes of antiretroviral therapy
over a 10-year period of expansion: a multicohort analysis of African and Asian HIV programs. J Acquir Immune Defic Syndr. 2014;67:e55–e66.
7. Auld AF, Mbofana F, Shiraishi RW, et al. Four-year treatment outcomes of adult patients enrolled in Mozambique
's rapidly expanding antiretroviral therapy
program. PLoS One. 2011;6:e18453.
8. World Health Organization. Guideline on when to start antiretroviral therapy
and on pre-exposure prophylaxis for HIV. Available at: http://www.who.int/hiv/pub/guidelines/earlyrelease-arv/en/
. Accessed October 19, 2015.
9. Decroo T, Koole O, Remartinez D, et al. Four-year retention and risk factors for attrition among members of community ART groups in Tete, Mozambique
. Trop Med Int Health. 2014;19:514–521.
10. Decroo T, Telfer B, Biot M, et al. Distribution of antiretroviral treatment through self-forming groups of patients in Tete Province, Mozambique
. J Acquir Immune Defic Syndr. 2011;56:e39–e44.
11. World Health Organization. Antiretroviral therapy
for HIV infection in adults and adolescents: recommendations for a public health approach—2010 revision. Available at: http://whqlibdoc.who.int/publications/2010/9789241599764_eng.pdf
. Accessed August 4, 2015.
12. World Health Organization. Consolidated guidelines on the use of antiretroviral drugs for treating and preventing HIV infection: recommendations for a public health approac. Available at: http://apps.who.int/iris/bitstream/10665/85321/1/9789241505727_eng.pdf
. Accessed February 17, 2014.
13. Auld AF, Harrison K, Azih C, et al. Evaluation of Swaziland's hub-and-spoke model for decentralizing access to antiretroviral therapy
services. JAIDS. 2015;69:e1–e12.
14. Pati R, Lahuerta M, Elul B, et al. Factors associated with loss to clinic among HIV patients not yet known to be eligible for antiretroviral therapy
(ART) in Mozambique
. J Int AIDS Soc. 2013;16:18490.
15. Schafer JL, Graham JW. Missing data: our view of the state of the art. Psychol Methods. 2002;7:147–177.
16. Graham JW. Missing data analysis: making it work in the real world. Annu Rev Psychol. 2009;60:549–576.
17. White IR, Royston P. Imputing missing covariate values for the cox model. Stat Med. 2009;28:1982–1998.
18. Rubin DB. Multiple Imputation for Nonresponse in Surveys. New York, NY: J. Wiley & Sons; 1987.
19. Rogers WH. Regression standard errors in clustered samples. Stata Tech Bull Reprints. 1993;3:88–94.
20. Williams RL. A note on robust variance estimation for cluster-correlated data. Biometrics. 2000;56:645–646.
21. Wandeler G, Keiser O, Pfeiffer K, et al. Outcomes of antiretroviral treatment programs in rural Southern Africa. J Acquir Immune Defic Syndr. 2012;59:e9–e16.
22. Schoni-Affolter F, Keiser O, Mwango A, et al. Estimating loss to follow-up in HIV-infected patients on antiretroviral therapy
: the effect of the competing risk of death in Zambia and Switzerland. PLoS One. 2011;6:e27919.
23. Coviello V, Boggess M. Cumulative incidence in the presence of competing risks. Stata J. 2004;4:103–112.
24. Egger M, Spycher BD, Sidle J, et al. Correcting mortality for loss to follow-up: a nomogram applied to antiretroviral treatment programmes in sub-Saharan Africa. PLoS Med. 2011;8:e1000390.
25. Cornell M, Lessells R, Fox MP, et al. Mortality among adults transferred and lost to follow-up from antiretroviral therapy
programmes in South Africa: a multicenter cohort study. J Acquir Immune Defic Syndr. 2014;67:e67–e75.
26. Farahani M, Price N, El-Halabi S, et al. Trends
and determinants of survival for over 200,000 patients on antiretroviral treatment in the Botswana National Program: 2002–2013. AIDS. 2016;30:477–485.
27. Brinkhof MW, Pujades-Rodriguez M, Egger M. Mortality of patients lost to follow-up in antiretroviral treatment programmes in resource-limited settings: systematic review and meta-analysis. PLoS One. 2009;4:e5790.
28. Muula AS, Ngulube TJ, Siziya S, et al. Gender distribution of adult patients on highly active antiretroviral therapy
(HAART) in Southern Africa: a systematic review. BMC Public Health. 2007;7:63.
29. Auld AF, Shiraishi RW, Mbofana F, et al. Lower levels of antiretroviral therapy
enrollment among men with HIV compared with women—12 countries, 2002—2013. MMWR Morb Mortal Wkly Rep. 2015;64:1281–1286.
30. Hawkes S, Buse K. Gender and global health: evidence, policy, and inconvenient truths. Lancet. 2013;381:1783–1787.
31. Lahuerta M, Wu Y, Hoffman S, et al. Advanced HIV disease at entry into HIV care and initiation of antiretroviral therapy
during 2006–2011: findings from four sub-saharan African countries. Clin Infect Dis. 2014;58:432–441.
32. Farahani M, Vable A, Lebelonyane R, et al. Outcomes of the Botswana national HIV/AIDS treatment programme from 2002 to 2010: a longitudinal analysis. Lancet Glob Health. 2014;2:e44–e50.
33. Siedner MJ, Ng CK, Bassett IV, et al. Trends
in CD4 count at presentation to care and treatment initiation in sub-Saharan Africa, 2002–2013: a meta-analysis. Clin Infect Dis. 2015;60:1120–1127.
34. Centers for Disease C. Prevention. Differences between HIV-infected men and women in antiretroviral therapy
outcomes—six African countries, 2004–2012. MMWR Morb Mortal Wkly Rep. 2013;62:946–952.
35. Velen K, Lewis JJ, Charalambous S, et al. Comparison of tenofovir, zidovudine, or stavudine as part of first-line antiretroviral therapy
in a resource-limited-setting: a cohort study. PLoS One. 2013;8:e64459.
36. Medicines Sans Frontieres. Untangling the Web of Antiretroviral Price Reductions. 17th ed; 2014. Available at: http://www.msfaccess.org/sites/default/files/MSF_UTW_17th_Edition_4_b.pdf
. Accessed April 3, 2015.
37. Cornell M, Grimsrud A, Fairall L, et al. Temporal changes in programme outcomes among adult patients initiating antiretroviral therapy
across South Africa, 2002–2007. AIDS. 2010;24:2263–2270.
38. Nglazi MD, Lawn SD, Kaplan R, et al. Changes in programmatic outcomes during 7 years of scale-up at a community-based antiretroviral treatment service in South Africa. J Acquir Immune Defic Syndr. 2011;56:e1–e8.
39. Lambdin BH, Micek MA, Koepsell TD, et al. Patient volume, human resource levels, and attrition from HIV treatment programs in central Mozambique
. J Acquir Immune Defic Syndr. 2011;57:e33–e39.
40. Lara J, Lambdin B. Exploratory Analysis of the Facility-Level Factors Associated With ART Abandonment in Central Mozambique
; November 7–11, 2009. Proceedings of the 137th Meeting of the American Public Health Association, Philadelphia, PA, 2009.
41. Wouters E, Heunis C, van Rensburg D, et al. Patient satisfaction with antiretroviral services at primary health-care facilities in the Free State, South Africa–a two-year study using four waves of cross-sectional data. BMC Health Serv Res. 2008;8:210.
42. Gabillard D, Lewden C, Ndoye I, et al. Mortality, AIDS-morbidity, and loss to follow-up by current CD4 cell count among HIV-1-infected adults receiving antiretroviral therapy
in Africa and Asia: data from the ANRS 12222 collaboration. J Acquir Immune Defic Syndr. 2013;62:555–561.
43. Decroo T, Panunzi I, das Dores C, et al. Lessons learned during down referral of antiretroviral treatment in Tete, Mozambique
. J Int AIDS Soc. 2009;12:6.
44. Johnson LF, Estill J, Keiser O, et al. Do increasing rates of loss to follow-up in antiretroviral treatment programs imply deteriorating patient retention? Am J Epidemiol. 2014;180:1208–1212.
45. Geng EH, Odeny TA, Lyamuya RE, et al. Estimation of mortality among HIV-infected people on antiretroviral treatment in east Africa: a sampling based approach in an observational, multisite, cohort study. Lancet HIV. 2015;2:e107–e116.
46. Rasschaert F, Decroo T, Remartinez D, et al. Adapting a community-based ART delivery model to the patients' needs: a mixed methods research in Tete, Mozambique
. BMC Public Health. 2014;14:364.
47. Cornell M, Myer L. Moving beyond gender stereotypes. Lancet. 2013;382:506.
48. Cornell M, McIntyre J, Myer L. Men and antiretroviral therapy
in Africa: our blind spot. Trop Med Int Health. 2011;16:828–829.
49. Joint United Nations Programme on HIV/AIDS. Community-Based Antiretroviral Therapy Delivery
. Available at: http://www.unaids.org/sites/default/files/media_asset/20150420_MSF_UNAIDS_JC2707.pdf
. Accessed January 22, 2016.
50. Philips M, Zachariah R, Venis S. Task shifting for antiretroviral treatment delivery in sub-Saharan Africa: not a panacea. Lancet. 2008;371:682–684.
51. Zachariah R, Ford N, Philips M, et al. Task shifting in HIV/AIDS: opportunities, challenges and proposed actions for sub-Saharan Africa. Trans R Soc Trop Med Hyg. 2009;103:549–558.
52. Kredo T, Ford N, Adeniyi FB, et al. Decentralising HIV treatment in lower- and middle-income countries. Cochrane Database Syst Rev. 2013:CD009987.
53. Green AHIV. In Mozambique
: starting, and staying on, treatment. Lancet. 2016;387:420–421.
54. Grimsrud A, Cornell M, Schomaker M, et al. CD4 count at antiretroviral therapy
initiation and the risk of loss to follow-up: results from a multicentre cohort study. J Epidemiol Community Health. 2016;70:549–555.
55. Lima VD, Reuter A, Harrigan PR, et al. Initiation of antiretroviral therapy
at high CD4+ cell counts is associated with positive treatment outcomes. AIDS. 2015;29:1871–1882.
56. Anglaret X, Scott CA, Walensky RP, et al. Could early antiretroviral therapy
entail more risks than benefits in sub-Saharan African HIV-infected adults? A model-based analysis. Antivir Ther. 2013;18:45–55.
57. Nachega JB, Uthman OA, del Rio C, et al. Addressing the Achilles' heel in the HIV care continuum for the success of a test-and-treat strategy to achieve an AIDS-free generation. Clin Infect Dis. 2014;59(suppl 1):S21–S27.
58. Katz IT, Essien T, Marinda ET, et al. Antiretroviral therapy
refusal among newly diagnosed HIV-infected adults. AIDS. 2011;25:2177–2181.
59. Toro PL, Katyal M, Carter RJ, et al. Initiation of antiretroviral therapy
among pregnant women in resource-limited countries: CD4+ cell count response and program retention. AIDS. 2010;24:515–524.
60. Kaplan R, Orrell C, Zwane E, et al. Loss to follow-up and mortality among pregnant women referred to a community clinic for antiretroviral treatment. AIDS. 2008;22:1679–1681.
61. Boyles TH, Wilkinson LS, Leisegang R, et al. Factors influencing retention in care after starting antiretroviral therapy
in a rural South African programme. PLoS One. 2011;6:e19201.
62. Nachega JB, Uthman OA, Anderson J, et al. Adherence to antiretroviral therapy
during and after pregnancy in low-income, middle-income, and high-income countries: a systematic review and meta-analysis. AIDS. 2012;26:2039–2052.
63. Tenthani L, Haas AD, Tweya H, et al. Retention in care under universal antiretroviral therapy
for HIV-infected pregnant and breastfeeding women (“Option B
+”) in Malawi. AIDS. 2014;28:589–598.
64. Clouse K, Schwartz S, Van Rie A, et al. “What they wanted was to give birth; nothing else”: barriers to retention in option B
+ HIV care among postpartum women in South Africa. J Acquir Immune Defic Syndr. 2014;67:e12–e18.