AIDS due to HIV infection is among the leading causes of death in sub-Saharan Africa and the fourth major cause of mortality worldwide.1 More than 60% of the world's HIV-infected people live in sub-Saharan Africa and over half of the HIV-infected adults are women of child-bearing age. The use of antiretroviral (ARV) therapy (ART) for the treatment of HIV-infected persons has had a significant impact on viral load, CD4+ cell count, and HIV-related morbidity and mortality.2–5
Many HIV-infected women in sub-Saharan Africa are diagnosed with HIV during pregnancy and initiate ART for their own health if they meet local eligibility criteria for treatment initiation1,6 or initiate ARV prophylaxis for prevention of mother-to-child transmission (PMTCT). Extension of prophylaxis through the breast-feeding period for women who are not eligible for treatment allows for infant protection and is recommended.6 CD4+ cell counts and WHO staging are used to define HIV progression7,8 and to determine the timing of ART initiation, whereas plasma HIV-1 viral load, if available in resource-limited settings, provides evidence for treatment response to ART.9–11 Current WHO guidelines for ART among adults and adolescents recommend CD4+ cell count testing at baseline to determine treatment eligibility and thereafter to assess treatment response and proper adherence counseling. However, CD4+ cell counts are imperfect for judging need for ARV and also for defining treatment response in the absence of viral load measurements because CD4+ cell count alone has low sensitivity for detecting patients with virologic failure until immunologic sequelae of treatment failure have become apparent.12–16
Good adherence to ARV is essential to achieving clinical success, delaying development of viral resistance and preventing vertical transmission.17–21 However, adherence to medication is a dynamic human behavior that is challenging to quantify within clinical and research settings.22,23 Common methods for estimating adherence levels are self-report, electronic data monitoring, pharmacy refills, drug calendar, and pill counting; each of these measures has its own limitations.24,25 In addition, while adherence ≥95% is necessary for reliable viral suppression, a subset of adherent patients will have virologic suppression without effective immune reconstitution.24–26
High maternal viral load remains a risk factor for HIV vertical transmission; the aim of maternal ARV prophylaxis in PMTCT is to maximally reduce HIV-1 viral load to undetectable levels hence reducing chances of HIV vertical transmission. It is thus important to monitor these women while on ARV prophylaxis to ensure good adherence to medication, immunological improvement, and good virologic outcome. In the Kisumu Breastfeeding Study (KiBS), we sought to evaluate immunological response, virologic suppression, and adherence among women receiving maternal triple ARV prophylaxis consisting of lamivudine/zidovudine and either nevirapine (NVP)-based triple ARV regimen or nelfinavir (NFV)-based triple regimen for PMTCT.27 We also assessed other risk factors that influenced time to achieving viral load suppression.
MATERIALS AND METHODS
The study design has been previously described.27 Briefly, the KiBS was a phase IIb open-label clinical trial designed to evaluate the efficacy of, tolerance to, and adherence to maternal triple ARV prophylaxis, for PMTCT from delivery up to 24 weeks postpartum. Women were recruited at 32–34 weeks of gestation and ARVs initiated at 34 weeks gestation and continued through 24 weeks postpartum. Initially enrolled participants were initiated on NVP/lamivudine/zidovudine (3TC/ZDV) regardless of baseline CD4+ cell counts. However, following an US Food and Drug Administration warning on the risk of hepatotoxicity among women with CD4+ ≥250 cells per microliter,28,29 the regimen was revised. Women with baseline CD4+ counts ≥250 cells per microliter received NFV instead of NVP. Those who met WHO treatment criteria (CD4+ count of <200 cells/mL or WHO stage III or IV) (WHO 2003) at enrollment or before 6 months postpartum remained on ARVs throughout the study.
Participants were enrolled into this substudy, if they had adherence, viral load, and CD4 data in at least 3 time points during the intervention period (32–34 weeks gestation to 24 weeks postpartum) and agreement to exclusively breast-feed up to 24 weeks postpartum. The substudy was approved by the Ethical Review Committee of the Kenyan Medical Research Institute and the Institutional Review Board of the US Centers for Disease Control and Prevention, Atlanta, GA. All participants provided a signed informed consent after being counseled on the benefits and risks of the study.
In this analysis, we focused on intervention visits (baseline to 24 weeks postpartum) only. During study visits, we assessed adherence to medication and infant-feeding recommendations, performed clinical evaluations, and obtained blood for hematologic, biochemical, and virologic monitoring.
Three methods were used to evaluate participant's adherence to study medication. Through pill count, adherence was calculated by trained pharmacy staff over a given period, by subtracting the number of pills returned during every scheduled visit from the number dispensed. Participants were also given a simple user-friendly drug calendar to mark date, day, and time (times in day were described in pictorial forms, ie, sunrise, sunset, etc) when the pills were taken. Participants returned the drug calendars to the pharmacy technician for review during clinic visits. Last, self-report through standard questionnaires administered during routine study visits were used to assess adherence. Participants were asked the number of doses they missed in the past 3 days and within a specified recall period (within the last one month).
Adherence was calculated as the percent of pills dispensed that were actually taken.30,31 Drug calendar and self-report data were used only to further probe adherence issues among participants. We determined adherence levels over 3 periods; from drug initiation through delivery visit, between delivery visit and 14 weeks postpartum, and between 14 weeks and 24 weeks postpartum. We also calculated cumulative (overall) adherence based on the 3 time points. Adherence was dichotomized at a conventionally accepted level of ≥95% or <95%.
Maternal blood specimens were collected in EDTA vacutainer tubes at multiple antenatal and postnatal visits. The blood was aliquoted for CD4+ cell count testing and plasma harvesting. The later was done within 6 hours from the time of collection and plasma stored at −70°C for later HIV-1 viral load testing. Plasma HIV-1 viral load was performed using the Amplicor version 1.5 monitor test standard (Roche Diagnostics Systems, Branchburg, NJ). CD4+ cell counts were enumerated using flow cytometry (FacsCalibur machine, Becton Dickinson San Jose, CA). For HIV testing of infants, dried blood spots were used, which were collected at birth (0–7 days), at 2, 6, 14, and 24 weeks. DNA polymerase chain reaction (PCR) testing was performed in real time on the 14- and 24-week specimens. After any positive test result, PCR testing was performed sequentially on prior untested specimens to identify the first positive specimen.
Data Management and Statistical Analysis
Statistical analysis was conducted using SAS version 9.2 (SAS Institute, Inc, Cary, NC). Analyses included data on participants with at least 3 of the 4 time points; enrollment (baseline), delivery, 14 weeks postpartum, and 24 weeks postpartum. We considered P values < 0.05 to be significant. Any viral load <400 copies per milliliter (<2.6 log) was characterized as undetectable.
Time trends in the proportion of women with CD4+ counts <250 cells per microliter and undetectable viral load was fitted using logistic regression model. Binary outcomes in viral load between CD4+ cell count categories of all women in the cohort were compared using the Pearson χ2 test. The Wilcoxon test was used to compare the duration of ARVs from treatment initiation to the delivery visit for women started on NVP- and NFV-based ARVs. Logistic regression was used to model undetectable viral load at 24 weeks postpartum with the following covariates: treatment combination, duration of ARV use before delivery, and baseline CD4+ and viral load categorical variables. Only women who did not change treatment regimen were included in this model.
Duration to achieving undetectable viral load between women with baseline CD4 counts <250 and CD4 counts ≥250 cells per microliter was compared with Zhao and Sun generalized log-rank test.32 A discrete survival model was used to model the hazard of achieving undetectable viral load and included a number of categorical prognostic baseline variables: CD4+ cell count, viral load, age, parity, type of regimen, WHO stage, and adherence.33 In the analyses of the type of ARV regimen, we considered only women with a baseline CD4 ≥ 250 based on the study design. An interval-censored survival plot was used to describe the time to achieve viral suppression for women on NFV- versus NVP-based ART in a subset analysis of women with detectable viral loads at baseline and CD4+ counts ≥250 cells per microliter.34 Only women with detectable baseline viral loads and who did not change treatment regimen were included in these analyses (n = 288). In these interval-censored survival analyses, observations from women who achieved undetectable viral load were censored in the interval between the dates of their last detectable viral load and their first undetectable viral load measurements. Observations for those women who did not achieve undetectable viral load were censored at their last follow-up.
Between July 2003 and November 2007, 522 women were enrolled into the KiBS; 22 women withdrew from the study between enrollment and the delivery visit and 500 women delivered during the study. Because of various reasons (9 stillbirths, 21 infant deaths, 3 maternal deaths, and 22 withdrawals), only 445 remained in the study. This analysis is restricted to the 434 women who completed the intervention through to 24 weeks postpartum with at least 3 data per specimen collection time points. The baseline characteristics of the study population have previously been described.27 Briefly, age of participants ranged from 15 to 43 years with a median of 24 years [interquartile range (IQR), 21–27], and parity ranged from 0 to 8 with a median of 1 (IQR, 0–1). Of the 434 women, 82% (358) were in WHO stage I, 10% (42) in stage II, 7% (30) in stage III, and none in stage IV (data missing for 4 women). Baseline CD4+ count ranged from 32 to 1165 cells per microliter with a median of 394 cells per microliter (IQR, 255–558), and baseline viral load ranged from <2.6–6.2 log copies per milliliter with a median of 4.5 log copies per milliliter (IQR, 3.9–5.1).
Adherence Over Time
Overall 84% (n = 366) of the participants were adherent during the study based on cumulative adherence data. At various time points, the proportions of participants who were adherent were not dramatically different; 78% (n = 339) between drug initiation through delivery, 83% (n = 360) between the delivery and 14 weeks postpartum, and 84% (n = 366) between 14 weeks and 24 weeks postpartum (Fig. 1). Adherence estimation by pill count, drug calendar, and participant self-report were comparable (data not shown), hence we focused on pill count.
Immunological Response Over Time
The median CD4+ count was 394 cells per microliter at baseline, 463 cells per microliter at delivery, 646 cells per microliter at 14 weeks, and 654 cells per microliter at 24 weeks postpartum. The percentage of women with CD4+ count <250 cells per microliter decreased from 23% (n = 100) at baseline to 5% (n = 22) at 24 weeks postpartum; the percentage of women with CD4+ count >500 cells per microliter increased from 32% (n = 139) at baseline to 69% (n = 295) at 24 weeks postpartum (Fig. 2). Of the 100 participants with a baseline CD4+ count <250 cells per microliter, 22% (n = 22) had a CD4+ count <250 cells per microliter at 24 weeks postpartum, 54% (n = 54) had a CD4+ count from ≥250–≤500 cells per microliter, and 24% (n = 24) had a CD4+ count >500 cells per microliter. There was a significant decrease in the proportion of women with CD4+ counts <250 cells per microliter from baseline to 24 weeks postpartum (P < 0.001).
Viral Load Response Over Time
At baseline, delivery, 14 weeks and 24 weeks postpartum, 6%, 67%, 80%, and 79% (χ2 test P < 0.001) of participants, respectively, had an undetectable viral load (Fig. 3). There was no difference in the proportion of participants who achieved undetectable viral load at 24 weeks postpartum based on CD4+ cell count categories at baseline, <250, ≥250–350, 351–500, and >500 cells per microliter with 80%, 79%, 77%, and 81%, respectively (χ2 for trend P = 0.85). Among the 89 participants who had a detectable viral load at 24 weeks postpartum, 32 never achieved undetectable viral load and 57 achieved an undetectable viral load at one point, but later experienced viral load rebound. In a further analysis, there was no difference in adherence measurements comparing the 57 participants who experienced viral load rebound versus the 32 participants who never achieved undetectable viral load (χ2 test, P = 0.23).
Among those who initiated and remained on NVP-based ARV, 89% (201/226) achieved undetectable viral load compared with 98% (166/170) of those who initiated and remained on NFV-based ARV (unadjusted relative risk = 4.7, P < 0.05). Fewer participants, who initiated on NVP-based ARV, achieved undetectable viral load even after adjusting for maternal CD4+ cell counts and adherence (OR = 0.34, P < 0.0001). In addition, of the 38 participants who had a treatment interruption (mean = 18 days, range, 3–40 days) due to an adverse event or other causes, 92% (35/38) achieved undetectable viral load.
Viral Load at Delivery Among Women With Varying Duration of ART
There was no difference in the median duration on ARV treatment from drug initiation to delivery based on regimen, 40 (IQR, 27–56) and 45 (IQR, 31–55) days for NVP- and NFV-based ARV regimen, respectively (P = 0.1, Wilcoxon test). There was a significant increase in the proportion of women achieving undetectable viral load at delivery, categorized by duration on treatment before delivery; 11 (35%) of the 31 pregnant women who were on therapy for <2 weeks, 44 (54%) of the 81 women who were on therapy for 2–4 weeks, 163 (71%) of the 231 women who were on therapy for 4–6 weeks, and 74 (81%) of the 91 women who were on therapy for >6 weeks (χ2 test for trend, P < 0.0001) before delivery achieved undetectable viral load. Undetectable viral load at delivery was achieved by 58% (130/226) of women on NVP-based maternal triple ARV and by 82% (139/170) of women on NFV-based ARV (χ2 test, P = 0.0001), excluding those who changed regimen (n = 38). This significant difference between the 2 regimens remained when we controlled for baseline CD4+ cell counts, baseline viral load, and duration on treatment [odds ratio (OR) = 2.02; 95% confidence interval (CI): 1.16 to 3.54, P = 0.014]. A similar percentage of women with baseline CD4+ counts <250 cells per microliter 80/100 (80%) versus ≥250 cells per microliter 264/334 (79%) had undetectable viral load by 24 weeks postpartum (χ2 test, P = 0.88). However, the mean duration to achieving undetectable viral load for women with baseline CD4+ counts of <250 cells per microliter (n = 100) was 14 weeks compared with 11 weeks for women with CD4+ counts ≥250 cells per microliter (n = 334) (t test P = 0.005).
Viral Load at 24 Weeks Postpartum Among Adherent and Nonadherent Women
Adherence was significantly associated with achieving undetectable viral load at 24 weeks postpartum. Of the 366 women who were adherent, 82% (n = 300) achieved undetectable viral load compared with 65% (n = 44) among nonadherent women (unadjusted OR = 0.40, 95% CI: 0.23 to 0.70, P = 0.001). Of the 32 participants who did not achieve undetectable viral load, 21 were adherent, whereas 11 were nonadherent.
Viral Load at 24 Weeks Postpartum Among Women Receiving Different ARV Regimen
Two hundred eighty-eight of the 434 women were included in the discrete survival regression model after excluding those with a baseline CD4+ count <250 cells per microliter (n = 100), undetectable baseline viral load (n = 25), or who changed regimen in the course of the intervention (n = 38) (some excluded participants were in more than one group). Of the 288 participants, 140 (49%) and 148 (51%) women were on NFV- and NVP-based regimen, respectively. By 24 weeks postpartum, 221 (92%) of 240 adherent women and 42 (88%) of 48 nonadherent had achieved undetectable viral load (Table 1). Using a discrete-survival modeling in bivariate and multivariable analysis, we did not find a relationship between achieving an undetectable viral load and baseline CD4+ strata ≥250 cells per microliter (250–350, 351–500, and >500), WHO stage, age, and adherence. Compared with women initiated on NVP-based ARV, women initiated on NFV-based ARV were more likely to achieve undetectable viral load in both bivariate and multivariate analyses (Fig. 4). Other variables that were associated with achieving an undetectable viral load in the proportional hazards model included lower baseline viral load, higher parity, and use of NFV-based ARV (Table 1). The regression results for this model were very similar to Cox regression hazard model results except for the age variable.
Viral Load and Mother-to-Child Transmission
Twenty-four infants acquired HIV infection during the 24-week intervention period. Of these infections, 12 (50%) were determined at delivery (0–7 day after delivery), whereas the rest occurred during the breast-feeding period; 8 (33%) between delivery and 14 weeks; and 4 (17%) between 14 weeks and 24 weeks postpartum. Of the 12 women whose infants were infected by the delivery visit, 11 (92%) had a baseline viral load >10,000 copies per milliliter. Of the 11 women whose infants were infected and had a viral load result at the delivery visit, 73% (n = 8) had a detectable viral load. Viral load at delivery was correlated with MTCT at delivery (χ2 test P = 0.0028). This association remained significant after adjusting for maternal CD4+ cell counts and duration of ARV treatment (OR = 0.323, 95% CI: 0.140 to 0.744, P = 0.008). Similarly, of the 12 women whose infants were infected during the breast-feeding period, 11 (92%) had a baseline viral load >10,000 copies per milliliter. Six (75%) of these women who had a viral load result at the time of infant seroconversion had a detectable viral load with 5 (63%) of them having viral load >10,000 copies per milliliter. Viral load was also correlated with MTCT at 14 weeks (P = 0.05) and at 24 weeks (P = 0.01) postpartum.
Significant improvement in immunological status and virological response demonstrated in this population during the 7 months on triple combination therapy for PMTCT provides significant support for viability of ARV intervention initiatives using combination ARV regimens for PMTCT in resource-limited settings. Most study participants had an overall adherence of ≥95% to ART from late in pregnancy through 24 weeks of breast-feeding. The timing of initiation of maternal ARV prophylaxis for PMTCT is important; based on our findings women who had high baseline viral load took longer to achieve undetectable viral load by the delivery visit compared with those who had low baseline viral load. Furthermore, a shorter duration on ARV prophylaxis before delivery was also associated with lower likelihood of achieving undetectable viral load at delivery. Participants who were initiated on NFV-based ARV prophylaxis achieved undetectable viral load faster than those on NVP-based ARV, and more women on NFV-based ARV achieved an undetectable viral load by delivery than on NVP-based ARV. In addition, viral load seemed to be associated with MTCT at the 3 time points.
Most HIV-infected pregnant women who were enrolled in the KiBS achieved good virologic suppression and good immunologic response, with more than 50% of women achieving CD4+ cell count ≥500 by 24 weeks postpartum following initiation of maternal triple ARV prophylaxis for PMTCT of HIV. A large percentage of the participants were ≥95% adherent. The virologic and immunologic outcomes and adherence levels are consistent with those seen in other studies previously conducted in sub-Saharan Africa.3,4 Adherence ≥95% was significantly associated with viral suppression at 24 weeks postpartum in the unadjusted bivariate analysis. When we evaluated a subset of this population, for only women who had baseline CD4+ cell counts ≥250 cells per microliter, detectable baseline viral load, and no change in regimen, using a discrete-survival model, adherence did not remain significant in both bivariate and the multivariate analysis.32–34 In a separate analysis that included all the 434 women, the Wilcoxon test was used to compare time to achieving undetectable viral load by delivery stratified by type of regimen, we found that NFV-based ARV regimen more commonly led to virologic suppression by delivery than NVP-based ARV regimen even when we controlled for duration on ARVs before delivery. Similar findings were observed at 24 weeks postpartum using the discrete survival model. Given that the intervention was initiated in the third trimester, it was not surprising to find that a longer duration on ARV prophylaxis before delivery was associated with increased likelihood of attaining virologic suppression at the time of delivery.
Level of adherence to ART is associated with viral suppression, CD4+ cell counts, HIV-1 drug resistance, and progression to AIDS.35,36 In this study, there was no significant relationship between adherence and viral load suppression in a subset of women with CD4 counts ≥250 cells per microliter. Adherence outcome and virologic outcomes, especially among the adherent women who did not achieve viral suppression in our multivariate analysis, raise questions as to (1) how much adherence is needed to achieve viral suppression to undetectable levels among patients on ARVs, and (2) how sensitive current methods are at assessing adherence in resource-limited settings. Adherence measured by pharmacy refill and pill count is often lower than self-reported adherence, but higher than adherence measured with electronic drug monitoring.22,23 Given that there are measurement errors associated with each method, true adherence lies somewhere between the extremes of these diverse adherence measures. The inherent strengths and weaknesses of each method make it likely that the best method of assessing adherence includes multiple or composite measures.37 However, we found little or no differences in adherence calculation between the 3 methods used in this study (data not shown). One limitation of our measures of adherence in this study is that we have evaluated 434 participants who remained in the study at 24 weeks postpartum and for whom we have relevant data. These participants were possibly the most adherent to study requirements. Thus, it is possible that participants who had withdrawn by 24 weeks may have been less adherent. Second, there is a possibility that some women achieved undetectable viral loads between tests and then rebound within the same interval before the next visit and were not taken into consideration because of the study design.
Most study participants had substantial increases in CD4+ cells counts during the intervention. Although 69% of the participants experienced an increase in CD4+ cell counts to ≥500 cells per microliter by 14 weeks postpartum, there was no difference in the proportion with ≥500 cells per microliter at 24 weeks. This suggests that immunological response to ARVs can be assessed as early as 14 weeks after initiating of treatment.
High maternal viral load is a risk factor for vertical transmission and thus, maternal viral load suppression during pregnancy and after delivery for mothers who exclusively breast-feed their babies can substantially reduce HIV transmission.38–40 We observed good virologic suppression with most (79%) women achieving an undetectable viral load by 24 weeks postpartum, which may have contributed to low postnatal–associated HIV-1 transmission during breast-feeding in this study.27 Women in this study, who had baseline viral load of <5.0 log copies per milliliter, were most likely to achieve undetectable viral load compared with those who had viral load ≥5.0 log copies per milliliter. Early initiation of ARV for PMTCT before delivery may significantly reduce maternal viral load and thus reduce vertical transmission. The use of protease inhibitors in triple combination regimen in PMTCT needs to be evaluated in hopes of improving HIV services to infected pregnant women. We believe that these findings are important and may provide insight into optimization of PMTCT regimens and practices because most HIV transmission in this study population occurred among women with high baseline viremia and low baseline CD4+ counts as shown in our previous findings.27
We also found that participants who initiated maternal NFV-based triple ARV prophylaxis were more likely to achieve viral suppression at delivery and at 24 weeks postpartum than participants initiated on NVP-based ARV. Our findings are in line with the new WHO guidelines for PMTCT and also consistent with findings from other studies.41 The trial was initially designed to study maternal NVP-based triple ARV prophylaxis for PMTCT and was modified to later include NFV in some women, so the finding of improved virologic response with NFV is presented as an unplanned secondary analysis. However, because the trial was not designed to test these 2 drugs in this manner, the results should be considered as preliminary and would require confirmation in another study.
Other studies that have used combination ARV have reported plasma HIV-1 viral load suppression in 60%–90% of their patients10,42 consistent with our study outcome. However, the proportion of participants who did not achieve undetectable viral load in this study is concerning because the population was ARV-naive. Participants received extensive counseling and support from study staff, yet there may have remained some unmet needs and missed opportunities to optimize adherence. Importantly, many participants in this study were enrolled at a time when ARV was not widely available in Kenya (2003) or during the early expansion of ART programs (2004–2006), which may have impacted acceptability of their use. The impact of stigma cannot be underestimated; many participants may not have disclosed their HIV status to their male partners or other family members.43–45 In addition, participants may have taken pills of the bottle but not consumed them, thus leading to overestimation of adherence, and participants may have provided study staff with socially desirable responses. In our study, we did not find any reason why women with parity of greater than 2 were more likely to achieve undetectable viral load compared with those who had less than 2. This finding may require further evaluation to determine the role of parity in virologic outcome.
As observed in this study, viral load remains a major risk factor for MTCT of HIV. As reported by in the KiBS, women with a high baseline viral load were more likely to transmit infection to their infants compared with those with lower levels.27 Early and sustained control of viral load has been associated with a decreasing residual risk of HIV transmission. Perhaps more women would have achieved undetectable viral load by delivery had therapy been initiated earlier in pregnancy, and this would have important implications for HIV prevention of vertical transmission in utero and peripartum.
In summary, to ensure long-term success of PMTCT, which involves extended administration of combination ARV, we must identify feasible and reliable tools to assess adherence and provide real-time interventions to support optimal adherence among these HIV-infected women during the intervention period.
The authors would like to acknowledge Drs Lisa Mills and Michael C. Thigpen for their thorough review of this manuscript. The authors would also like to appreciate vital contributions made by Lazarus Odeny and Richard Ndivo during data analysis, KiBS participating mothers, KiBS staff for their diligent work in the field, and Collins Odhiambo for valuable comments, GlaxoSmithKline for donation of Combivir, Boehringer Ingelheim for donation of Viramune, clinic staffs, New Nyanza Provincial General Hospital, and Kisumu District Hospital for their assistance in recruitment and for caring for study participants, Kenya Medical Research Institute/US Centers for Disease Control and Prevention Kisumu HIV-Research laboratory where all laboratory testing were done. This article is published with the approval of the Director KEMRI.
Authors contributions: M. G. Fowler, P. J. Weidle, and T. K. Thomas contributed to the study design. B. Akoth, J. A. Okonji, R. Masaba, T. K. Thomas, and C. Zeh contributed to study data collection. J. A. Okonji, C. Zeh, P. J. Weidle, J. Williamson, M. G. Fowler, R. Masaba, and T. K. Thomas contributed to the planning and/or conduct of manuscript analyses. T. K. Thomas, P. J. Weidle, C. Zeh, M. G. Fowler, A. Benta, J. A Okonji, and R. Masaba contributed to the manuscript preparation and/or revisions.
2. Mellors JW, Munoz A, Giorgi JV, et al.. Plasma viral load and CD4+ lymphocytes as prognostic markers of HIV-1 infection. Ann Intern Med. 1997;126:946–954.
3. Nkengasong JN, Borget MY, Maurice C, et al.. Distribution of HIV-1 plasma RNA viral load and CD4 + T-cell counts among HIV-infected Africans evaluated for antiretroviral therapy. J Acquir Immune Defic Syndr. 2001;28:99–101.
4. Palella FJ Jr, Delaney KM, Moorman AC, et al.. Declining morbidity and mortality among patients with advanced human immunodeficiency virus infection. HIV Outpatient Study Investigators. N Engl J Med. 1998;338:853–860.
5. Tsague L, Tsiouris FO, Carter RJ, et al.. Comparing two service delivery models for the prevention of mother-to-child transmission (PMTCT) of HIV during transition from single-dose nevirapine to multi-drug antiretroviral regimens. BMC Public Health. 2010;10:753.
6. WHO. Rapid Advice: Use of Antiretroviral Drugs for Treating Pregnant Women and Preventing HIV Infection in Infants Version 2. Geneva, Switzerland: WHO; 2010.
7. Sande MA, Carpenter CC, Cobbs CG, et al.. Antiretroviral therapy for adult HIV-infected patients. Recommendations from a state-of-the-art conference. National Institute of Allergy and Infectious Diseases State-of-the-Art Panel on Anti-Retroviral Therapy for Adult HIV-Infected Patients. JAMA. 1993;270:2583–2589.
8. The UK Collaborative HIV Cohort (CHIC) Study Steering Committee. Rate of AIDS disease or death in HIV-infected antiretroviral therapy-naïve individuals with high CD4 cell counts. AIDS. 2007;21:1717–1721.
9. Semple M, Loveday C, Weller I, et al.. Direct measurement of viraemia in patients infected with HIV-1 and its relationship to disease progression and zidovudine therapy. J Med Virol. 1991;35:38–45.
10. O'Brien WA, Hartigan PM, Daar ES, et al.. Changes in plasma HIV RNA levels and CD4+ lymphocyte counts predict both response to antiretroviral therapy and therapeutic failure. VA Cooperative Study Group on AIDS. Ann Intern Med. 1997;126:939–945.
11. Calmy A, Ford N, Hirschel B, et al.. HIV viral load monitoring in resource-limited regions: optional or necessary? Clin Infect Dis. 2007;44:128–134.
12. Kagaayi J, Makumbi F, Nakigozi G, et al. WHO HIV clinical staging or CD4 cell counts for antiretroviral therapy eligibility assessment? An evaluation in rural Rakai district, Uganda. AIDS. 2007;21:1208–1210.
13. Mee P, Fielding KL, Salome C, et al.. Evaluation of the WHO criteria for antiretroviral treatment failure among adults in South Africa. AIDS. 2008;22:1971–1977.
14. Reynolds SJ, Nakigozi G, Newell K, et al.. Failure of immunologic criteria to appropriately identify antiretroviral treatment failure in Uganda. AIDS. 2009;23:697–700.
15. Jaffar S, Birungi J, Grosskurth H, et al.. Use of WHO clinical stage for assessing patient eligibility to antiretroviral therapy in a routine health service setting in Jinja, Uganda. AIDS Res Ther. 2008;5:4.
16. Carter RJ, Dugan K, El-Sadr WM, et al.. CD4+ cell count testing more effective than HIV disease clinical staging in identifying pregnant and postpartum women eligible for antiretroviral therapy in resource-limited settings. J Acquir Immune Defic Syndr. 2010;55:404–410.
17. Djomand G, Roels T, Ellerbrock T, et al.. Virologic and immunologic outcomes and programmatic challenges of an antiretroviral treatment pilot project in Abidjan, Cote d'Ivoire. AIDS. 2003;17(suppl 3):S5–S15.
18. Mills EJ, Nachega JB, Buchan I, et al.. Adherence to antiretroviral therapy in sub-Saharan Africa and North America: a meta-analysis. JAMA. 2006;296:679–690.
19. Bangsberg DR, Porco TC, Kagay C, et al.. Modeling the HIV protease inhibitor adherence-resistance curve by use of empirically derived estimates. J Infect Dis. 2004;190:162–165.
20. Paterson DL, Potoski B, Capitano B. Measurement of adherence to antiretroviral medications. J Acquir Immune Defic Syndr. 2002;31(suppl 3):S103–S106.
21. Bangsberg DR, Kroetz DL, Deeks SG. Adherence-resistance relationships to combination HIV antiretroviral therapy. Curr HIV/AIDS Rep. 2007;4:65–72.
22. Berg KM, Arnsten JH. Practical and conceptual challenges in measuring antiretroviral adherence. J Acquir Immune Defic Syndr. 2006;43(suppl 1):S79–S87.
23. Berg KM, Demas PA, Howard AA, et al.. Gender differences in factors associated with adherence to antiretroviral therapy. J Gen Intern Med. 2004;19:1111–1117.
24. Tuboi SH, Brinkhof MW, Egger M, et al.. Discordant responses to potent antiretroviral treatment in previously naive HIV-1-infected adults initiating treatment in resource-constrained countries: the antiretroviral therapy in low-income countries (ART-LINC) collaboration. J Acquir Immune Defic Syndr. 2007;45:52–59.
25. Tuboi SH, Pacheco AG, Harrison LH, et al.. Mortality associated with discordant responses to antiretroviral therapy in resource-constrained settings. J Acquir Immune Defic Syndr. 2010;53:70–77.
26. Reynolds S, Nakigozi G, Newell K, et al.. Evaluation of the WHO Immunologic Criteria for ART Failure among Adults in Rakai, Uganda. Presented at: 16th Conference on Retroviruses and Opportunistic Infections (CROI 2009); February 8–11, 2009; Montreal, Canada. Abstract 144.
27. Thomas T, Masaba R, Borkowf CB, et al.. Triple-antireroviral prophylaxis to prevent mother -to-child HIV transmission through breastfeeding—the Kisumu Breastfeeding Study, Kenya: a clinical trial. PLoS Med. 2011;8:e1001015.
29. Boehringer Ingelheim Pharmaceuticals, Inc. Prescribing Information: VIRAMUNE (nevirapine). Ridgefield, CT: Boehringer Ingelheim Pharmaceuticals, Inc; 2008. Investigator's Brochure: Nevirapine (BI-RG-587).
30. Chalker JC, Andualem T, Gitau LN, et al.. Measuring adherence to antiretroviral treatment in resource-poor settings: the feasibility of collecting routine data for key indicators. BMC Health Serv Res. 2010;10:43.
31. Ross-Degnan D, Pierre-Jacques M, Zhang F, et al.. Measuring adherence to antiretroviral treatment in resource-poor settings: the clinical validity of key indicators. BMC Health Serv Res. 2010;10:42.
32. Zhao Q, Sun J. Generalized log-rank test for mixed interval-censored failure time data. Stat Med. 2004;23:1621–1629.
33. Scheike TH, Jensen TK. A discreet survival model with random effects: an application to time to pregnancy. Biometrics. 1997;53:318–329.
34. Wellner JA, Zhan Y. A hybrid algorithm for computation of the nonparametric maximum likelihood estimator from censored data. J Am Stat Assoc. 1997;92:945–959.
35. Bangsberg DR, Charlebois ED, Grant RM, et al.. High levels of adherence do not prevent accumulation of HIV drug resistance mutations. AIDS. 2003;17:1925–1932.
36. Bangsberg DR. Less than 95% adherence to nonnucleoside reverse-transcriptase inhibitor therapy can lead to viral suppression. Clin Infect Dis. 2006;43:939–941.
37. Liu H, Miller LG, Hays RD, et al.. A practical method to calibrate self-reported adherence to antiretroviral therapy. J Acquir Immune Defic Syndr. 2006;43(suppl 1):S104–S112.
38. Jamieson DJ, Sibailly TS, Sadek R, et al.. HIV-1 viral load and other risk factors for mother-to-child transmission of HIV-1 in a breast-feeding population in Cote d'Ivoire. J Acquir Immune Defic Syndr. 2003;34:430–436.
39. Leroy V, Montcho C, Manigart O, et al.. Maternal plasma viral load, zidovudine and mother-to-child transmission of HIV-1 in Africa: DITRAME ANRS 049a trial. AIDS. 2001;15:517–522.
40. O'Donovan D, Ariyoshi K, Milligan P, et al.. A maternal plasma viral RNA levels determines differences in mother-to-child transmission rates of HIV-1 and HIV-2 in the Gambia. AIDS. 2000;14:441–448.
41. Yazdanpanah Y, Sissoko D, Egger M, et al.. Clinical efficacy of antiretroviral combination therapy based on protease inhibitors or non-nucleoside analogue reverse transcriptase inhibitors: indirect comparison of controlled trials. BMJ. 2004;328:249.
42. Hogg RS, Rhone SA, Yip B, et al.. Antiviral effect of double and triple drug combinations amongst HIV-infected adults: lessons from the implementation of viral load-driven antiretroviral therapy. AIDS. 1998;12:279–284.
43. Stirratt MJ, Remien RH, Smith A, et al.. The role of HIV serostatus disclosure in antiretroviral medication adherence. AIDS Behav. 2006;10:483–493.
44. Bikaako-Kajura W, Luyirika E, Purcell DW, et al.. Disclosure of HIV status and adherence to daily drug regimens among HIV-infected children in Uganda. AIDS Behav. 2006;10:S85–S93.
45. Li L, Wu Z, Wu S, et al.. Impacts of HIV/AIDS stigma on family identity and interactions in China. Fam Syst Health. 2008;26:431–442.
Keywords:© 2012 by Lippincott Williams & Wilkins
ARV; ART; antiretroviral naive; prevention of mother-to-child transmission; viral load; CD4+; adherence; viral suppression; Western Kenya