An estimated 2.1 million children younger than 15 years were living with HIV infection worldwide by the end of 2007, 90% in sub-Saharan Africa.1,2 An estimated 780,000 children were eligible for antiretroviral treatment (ART), that is, presenting with a World Health Organization (WHO) clinical stage 4 or stage 3, irrespective of the absolute CD4 count or percentage, or WHO stage 1 or stage 2 and severe immunodeficiency.3 However, only an estimated 1 in 10 eligible children have received treatment so far and fewer than 5% of the total population on ART are children.4 This is despite initial reports from pilot treatment cohorts and programs in the African region, indicating that children respond well to treatment with improved survival and low rates of toxicity.5-12 However, these reports are restricted to 1 cohort, 1 country, or 1 program. The Kids' Antiretroviral Treatment in Lower-Incomes Countries (KIDS-ART-LINC) Collaboration, a multicountry multiprogram network of cohorts of HIV-infected children throughout sub-Saharan Africa, pools data from children on ART followed in participating pediatric HIV clinics to document their prognosis and other key program outcomes.13 The objectives of the analysis presented here were to estimate the 2-year probabilities of death and loss to follow-up (LTFU) of HIV-infected children initiating ART in sub-Saharan Africa and to quantify the effect of individual patient variables at ART initiation and clinic/hospital characteristics on these outcomes.
PATIENTS AND METHODS
We conducted a pooled analysis of individual data from cohorts participating in the KIDS-ART-LINC Collaboration.14 Participating pediatric HIV clinics formally agreed to be included in this collaborative network, had local institutional review board's exemption or approval, and had a computerized clinical data collection system. The present analysis included individual data available by May 2007. These data were collected at each site according to local procedures. Selected variables (see below) were extracted, standardized, transferred, and pooled into a central database using standard operating procedures. Cohorts that provided data on at least 30 children were included in the present analysis. Eligibility criteria for inclusion of records of children in the analysis were as follows: ART initiation between birth and 15 years of age included, at least 1 day of follow-up, sex recorded, initiation of ART with at least 3 drugs irrespective of drug regimen, and start of ART at least 6 months before the closing date of the local database.
In most of the sites, biological assessments in children not yet on ART were performed every 6 months. If they were not available on the date of ART initiation, we included the closest measurements within the last 200 days before the start of ART date up till 30 days after this date as baseline values. A child was considered LTFU if the last contact recorded was at least 6 months before the closure date of the local database, and the child was not known to have died or moved to another treatment program or to another place. In addition, children aged 16 years or more at last contact were assumed transferred to adult care and not considered LTFU. Fifteen years was the median age cutoff for defining a child in the sites participating in the KIDS-ART-LINC Collaboration.14 All patients with follow-up less than 2 years were censored at their last visit date.
Age at ART initiation was categorized in 4 groups: younger than 12 months, 12-35 months, 36-59 months, and 5 years and older. First-line ART regimens included nonnucleoside reverse transcriptase inhibitor (NNRTI)-based combinations and other regimens (protease inhibitor-based [PI] or nucleoside reverse transcriptase inhibitor-based), either with zidovudine or stavudine (d4T). Severe immunodeficiency was defined according to WHO criteria3: %CD4 <25% or CD4 count <1500 cells per cubic millimeter for children younger than 12 months; % CD4 <20% or CD4 count <750 cells per cubic millimeter for children between 12 and 35 months; % CD4 <15% or CD4 count <350 cells per cubic millimeter for children between 36 and 59 months; and % CD4 <15% or CD4 count <200 cells per cubic millimeter for children 5 years and older. Severe anemia was defined according to US guidelines15: hemoglobin (Hb) <10 g/dL for children younger than 21 days; Hb <8 g/dL for children between 22 and 35 days; Hb <7 g/dL for children between 36 and 56 days; and Hb <7.5 g/dL for children 57 days and older. Because of the high proportion of missing data with regard to weight and WHO clinical stage, a combined variable was created, referred as severe clinical status and defined as weight-for-age z score <−3 (severe underweight according to United Nations Children's Fund http://www.unicef.org/infobycountry/stats_popup2.html) or WHO clinical stage ≥3. Site characteristics included the type (public clinic vs. others) and location according to 3 regions (western, eastern, and southern Africa) and, more precisely, location in a malaria area or not, to assess the potential influence of such coinfection (http://www.who.int/ith/maps/malaria2005_en.gif).
We applied an intent-to-continue treatment approach, ignoring changes to treatment, treatment interruptions, and terminations. Time was measured from the start of ART and ended at the earliest of the date of death, the date of the last follow-up visit, the date of last contact, or month 24 after starting ART. Probabilities of death and LTFU and their respective 95% confidence intervals (CIs) were estimated using the Kaplan-Meier method.16 Fisher exact tests and Kruskal-Wallis tests were applied to compare percentages and medians, respectively, when relevant. The statistical significance threshold was 5%, and all reported P values were 2 sided.
We studied the association of baseline individual and site characteristics with the end points, death or LTFU, within the subset of children for whom all baseline variables were available, adjusting for sex, age, and ART regimen. Furthermore, between-cohort heterogeneity was accounted for by including a random effect (or shared frailty) on cohort. We selected a Weibull model17,18 but verified that the use of a Cox proportional hazard model instead gave similar results (data not shown).
In the context of sensitivity analyses, we used death combined with LTFU as a composite end point to evaluate a worst case scenario assuming all LTFUs were unreported deaths. Sensitivity analyses were also performed to evaluate the influence of the availability of baseline variables on the different estimates. Missing data were considered as a third category to the initially dichotomous variables. All statistical analyses were conducted with STATA 9.2 software. Results for death or LTFU were presented as probability estimates, rates, and hazard ratios (HRs) with 95% CIs.
Individual data were received from 9 programs plus 8 additional sites of the mother-to-child transmission (MTCT) Plus International Initiative (www.mtctplus.org). General characteristics of the sites have been reported elsewhere,14 with 3644 children overall of whom 2666 were reported to have received ART. One site had less than 30 children on ART (n = 23) and was therefore excluded from the analysis.
This analysis is based on data from 5 clinics in western Africa, 4 in eastern Africa, and 7 in southern Africa. All sites were urban facilities. Twelve were public clinics, and all clinics but the 5 South African ones were located in an endemic malaria zone (Table 1). In total, 2405 children in 16 cohorts met the eligibility criteria and were included in the analyses; 52.4% were male; the median age at ART initiation was 4.9 years (interquartile range [IQR]: 2.1-8.4), varying substantially between cohorts (Table 1). Reasons for non inclusion of 238 children were as follows: age at ART initiation younger than 1 day or older than 15 years (n = 29), follow-up <1 day (n = 44), sex unknown (n = 2), mono/dual ART recorded (n = 26), and ART initiation <6 months before the closing date of the local database (n = 137) (Fig. 1).
At baseline, 69.5% (n = 1211) of children presented with severe immunodeficiency, 6.3% (n = 99) with severe anemia, and 52.5% (n = 797) with a severe clinical status; relevant baseline individual data were not available for all children (Table 2).
The main first-line ART regimens were NNRTI based. However, children younger than 12 months more often received a PI-based first-line regimen (66.5% as compared with 37.1% in older children, P < 0.001). Infants were also more likely than older children to have severe immunodeficiency (83.6% vs. 67.1%, P < 0.001) and severe clinical status (67.6% vs. 51.4%, P = 0.001). They were more likely to have missing data for Hb (46.6% vs. 33.3%, P < 0.001) and clinical status (65.5% vs. 32.9%, P < 0.001) but had fewer missing CD4 values (15.5% vs. 29.2%, P < 0.001).
During a median follow-up of 20.3 months (IQR: 11.7-27.9), 153 deaths were registered, 88 (57.5%) occurred in the first 3 months of treatment, 26 (17.0%) in the following 3 months, 25 (16.3%) in the following 6 months, and 14 (9.2%) in the second year. The number of children LTFU (n = 187) was 34 (18.2%), 26 (13.9%), 49 (26.2%), and 78 (41.7%), respectively, for the same periods. Transfer to another treatment program or a move to another location was reported in 7.7% of children (n = 185) and decreased with time (102 during the first year and 83 during the second one).
The overall estimated cumulative probabilities of death at 6, 12, and 24 months were 4.8% (CI: 4.0 to 5.8), 6.0% (CI: 5.0 to 7.1), and 6.9% (CI: 5.9 to 8.1), respectively. The overall cumulative probabilities of LTFU at 6, 12, and 24 months were 2.6% (CI: 2.0 to 3.4), 5.0% (CI: 4.1 to 6.1), and 10.3% (CI: 8.9 to 11.9), respectively.
Univariate and multivariate analyses were restricted to 1058 children from 8 cohorts who had complete data on clinical status, immunodeficiency, and anemia available (Fig. 1). The details on laboratory procedures used in the sites enrolled in the analyses are reported in Table 1. In this subset (Table 2), 68 deaths and 44 LTFUs occurred in the first year on treatment, with 8 additional deaths and 22 LTFUs in the second year. The median follow-up was 17.7 months (IQR: 11.0-25.8). Probabilities of death were 5.4% (CI: 4.1 to 7.0) at 6 months, 6.7% at 1 year (CI: 5.3 to 8.4), and 8.0% (CI: 6.4 to 10.1) at 2 years. Baseline severe immunodeficiency, severe anemia, and severe clinical status were independently associated with the risk of death. Probabilities of LTFU were 2.8% (CI: 1.9 to 4.1) at 6 months, 4.6% (CI: 3.4 to 6.2) at 1 year, and 8.4% (CI: 6.5 to 10.7) at 2 years, but only severe clinical status was associated with LTFU through the end of the second year (Table 3). As no LTFU was observed in the 2 not-for-profit participating clinics, the type of facility could not be explored as a prognostic factor in univariate and multivariate analyses. The risk of death was associated neither with age at ART initiation nor with the type of first-line regimen (NNRTI vs. PI-based or d4T vs. zidovudine), and there was no interaction between age at ART initiation and ART regimen, immunodeficiency, anemia, or clinical status.
In sensitivity analyses combining death and LTFU as an end point, baseline severe immunodeficiency, severe anemia, and severe clinical status remained independent predictive factors of the combined risk (data not shown). The subset used for the prognosis analyses had a lower proportion of children younger than 12 months (7.6% vs. 16.0%, P < 0.001), had a higher proportion of west African children (68.8% vs. 30.4%, P < 0.001), and was more likely to be from public facilities (94.2% vs. 81.8%, P < 0.001) compared with the subset where baseline data were not available. Children with all biological data available were at higher risk of death but at the same risk of LTFU at 2 years than those without these data (data not shown). When missing data were included as a separate category for baseline clinical and biological variables and the overall study sample was included, the above findings were confirmed (Table 4). Furthermore, children presenting without severe clinical status were at lower risk of death than those without these data available. Although no interaction between age at ART initiation and the 3-category variable for immunodeficiency was found, the influence of missing data was different among children younger than 12 months or older as shown in Figure 2. Indeed, children younger than 12 months with missing data were younger (median age at ART initiation: 6.4 months; IQR: 0.5-11.9) and were more likely to receive a PI (86%) and d4T (58%) than those with available data (median age at ART initiation: 8.4 months, IQR: 2.2-12.0, PI = 59%; d4t = 40%, with P values of 0.0001, 0.0001, and 0.009, respectively). In addition, when performing univariate analysis on the overall sample (N = 2405) for age at ART initiation, children younger than 12 months were at higher risk of death (HR = 3.1; CI: 2.0 to 4.9) but not at higher risk of LTFU (HR = 1.4; CI: 0.8 to 2.3) compared with older children.
Overall, as illustrated by Figure 2, the instantaneous hazard of LTFU was fairly constant over follow-up time, but the hazard of death leveled off over time.
This pooled analysis of routinely collected individual data on about 2400 children on ART followed in health care facilities in sub-Saharan Africa up to mid-2007s indicates a relatively low probability of death within the first 2 years of treatment (7%), with 75% of the deaths occurring in the first 6 months, as observed in other ART programs of much smaller scale such as in Port-au-Prince, Haïti (n = 236),19 and in Mzuzu, Malawi (n = 439),20 or larger ones such as in Lusaka, Zambia (n = 2398).12 However, the probability of LTFUs was higher (10%) than that of mortality, with 60% of the LTFUs occurring in the first year of treatment. Infants younger than 12 months of age at ART initiation were at increased risk of death compared with older ART-treated children, although, in the subset of children for whom all covariables were available, this risk difference was no longer significant. This suggested that the restriction of the sample resulted in a preadjustment on age and biological and clinical conditions. The young children enrolled presented more advanced HIV disease, highlighting the limited availability of early HIV diagnostic testing and weak links between prevention of mother-to-child transmission (PMTCT) and HIV-exposed infant follow-up services21,22 or early treatment whose benefit has been shown recently in South Africa.23
Our data confirm the relatively low mortality of children enrolled in ART programs in lower incomes countries5-8,11,19,24,25 and provide reliable and precise estimates, thanks to the large sample size. We highlight also the challenge of LTFU and attrition from ART programs. However, these data should be interpreted with some caution as it is not known how much LTFUs were due to unreported death and to which extent the true mortality rate was therefore underestimated. On the other hand, the probabilities of death and of LTFU were associated with different risk factors and were differently distributed over time after ART initiation, which would suggest limited bias of misclassification. Probabilities of death after 2 years on ART were available in only 3 reports and ranged from about 9% in Zambia12 and South Africa26 to 12% in Côte d'Ivoire.6,9 With data on 1184 children younger than 13 years receiving a fixed-dose combination of d4T, lamivudine, and nevirapine within the Médecins Sans Frontières (MSF) network in Africa and Asia, the estimated 12-month probability of death on ART was 5% and 13%, with the end point combining death and LTFU.8 This last end point was 18% at 2 years27 in a subset of children younger than 5 years (median age of about 3 years). Our findings confirmed those previously reported by MSF.
Previous evaluations of pediatric HIV treatment programs could only investigate a limited and selected number of prognostic factors with regard to survival. First-line antiretroviral regimens with PIs and low baseline CD4 cell counts were5,11,12,19,20 or were not8,26,27 found to be associated with poorer survival. Other determinants included high baseline HIV plasma viral load,5 baseline WHO clinical stage 4, age younger than 12 or 16 months,5,8,11,12,19,20 low weight-for-age Z score,11,19,20 and anemia.12 Our findings confirmed the importance of the overall baseline clinical status on survival after ART initiation, including not only the HIV staging but also the nutritional status28 and baseline anemia.
In a study conducted in Lusaka, 158 children had withdrawn from care and 382 were at least 30 days late for follow-up, representing 5.4% and 13.0% of the 2928 children who started ART, but no hypothesis was made regarding the reasons for these observations.12 In our study, only a severe clinical status was significantly associated with LTFU. In addition, the instantaneous hazard of LTFU did not seem to be affected by time on treatment, suggesting that factors independent from treatment effect, such as the type of facility, the size of the cohort, or unmeasured socioeconomic factors,21 are indeed the underlying cause of LTFU. In contrast, the leveling off of the instantaneous hazard of death was likely to be explained by the good response to treatment with time. In addition, there is no published report on the transition of adolescent to adult care in sub-Saharan Africa, and our assumption that children older than 15 years should not be considered as LTFU while fitting the definition may have led to underestimation of this outcome. Reasons for program attrition are incompletely understood, even for adults,29-31 and no pediatric data are available so far. However, factors such as unreported death, unreported move, death or disease of the caregiver, lack of search procedures, inability to pay for transport, drugs, laboratory tests, drug shortages, fear of discrimination, and bad experience with the care staff could all contribute to overall LTFU. International studies with systematic data collection of such parameters are needed to better understand this serious program pitfall.
This study has some limitations. First, 40% of the children received a PI-based first-line regimen, which is not in accordance with 2006 WHO guidelines.3 This was mostly due to the greater availability of pediatric formulations of lopinavir/ritonavir and nelfinavir for children weighing <10 kg in some sites. As there has been no international pediatric ART program evaluation outside the MSF experience so far, it is unclear how this distribution of first-line ART regimens reflected the current real-life antiretroviral management in pediatric programs in sub-Saharan Africa in the first years of the rollout. In general, the extent to which the results of this analysis are representative of the whole of sub-Saharan Africa in the same period cannot be fully ascertained. However, it provides data on various treatment programs in different countries, which may be closer to the reality than studies within 1 program8 or 1 country.12 Second, the present study was based on data collected routinely in health care facilities with their own procedures. The proportion of missing data was substantial, potentially leading to biased subgroup analyses and to the use of atypical summary measure with regard to the clinical status in lieu of standard predictors (WHO clinical stage, z score). We thoroughly evaluated the extent of these biases by performing sensitivity analyses. Other data were not available at all in some cohorts because they were not routinely recorded or not entered electronically. For instance, data on adherence, previous PMTCT drug exposure, plasma HIV RNA viral load, co-trimoxazole intake, or tuberculosis history were not available for this analysis, although expected to have implications in the response to treatment.32-35 Of note, most of these variables have also not been investigated in other pooled data sets as they are difficult to collect and validate outside research settings.8,27 Finally, causes of death were not available, this variable being poorly recorded in the context of routine follow-up8,12 like in most of the sites of the KIDS-ART-LINC Collaboration. The lack of standardization in the recording of the data could have also introduced a bias, but this is likely to be minimal given that the data extracted were those commonly used and with an agreed and simple definition and thus not requiring further standardization. Finally, results of laboratory tests were obtained using similar techniques among the different sites.
In conclusion, 1- and 2-year estimated risks of death in African children starting ART are now reported in the KIDS-ART-LINC Collaboration with reliable and precise estimates in line with individual study reports on smaller cohorts. ART is still mainly initiated at advanced disease stages in African children. Age at ART initiation is strongly associated with the risk of dying, and the younger the children, presenting with more advanced disease, the worse the prognosis. The gradually increasing attrition rate over time of African children on ART is also worrisome. The early and comprehensive institution of treatment plans for children is urgently needed to reverse the current pediatric survival patterns in sub-Saharan Africa and maintain as many children as possible in long-term care.
We acknowledge all children whose data were used in this study. We also would like to thank all persons who contributed to record, enter, prepare, and send the data to the KIDS-ART-LINC Collaboration. Special thanks to Jack Whitescarter, Paolo Miotti (National Institutes of Health/OAR), Brigitte Bazin, and Séverine Blesson (ANRS) for encouraging and supporting this international database project. Writing committee: Elise Arrivé and Benoit Marquis, Inserm U897, Institut de Santé Publique et de Développement (ISPED), Bordeaux, France; Nathan Tumwesigye, African Network for the Care of Children Affected by AIDS (ANECCA), Kampala, Uganda; Martin W. G. Brinkhof, Department of Social and Preventive Medicine, University of Bern, Switzerland; Patricia Fassinou, Centre Hospitalier Universitaire de Yopougon, Abidjan, Côte d'Ivoire; Mark Cotton, Tygerberg Children's Hospital, Cape Town, South Africa; Louise Wemin, Centre de Prise en Charge, de Recherche et de Formation (CEPREF) Enfants, ACONDA, Abidjan, Côte d'Ivoire; Andrew Boulle, Khayelitsha Hospital, Cape Town, South Africa; Margaret Holland, Connaught Clinic, Harare, Zimbabwe; Lorna Renner, Korle Bu Hospital, Accra, Ghana; Pierre Kariyo, Association Nationale de Soutien aux Séropositifs (ANSS), Bujumbura, Burundi; Akum Aveika, Medical Research Council, Fajara, The Gambia; Alain Azondekon, UPEIV, Hôpital d'Intruction des Armées, Cotonou, Benin; Rosalind Carter, MTCT Plus Network, Columbia University, New York, NY; Valériane Leroy, Inserm U897, ISPED, Bordeaux, France; Mary-Pat Kieffer, USAID East Africa, Nairobi, Kenya; Leticia Namale, ANECCA, Kampala, Uganda; Marie-Louise Newell, Africa Center, Hlabisa, KwaZulu Natal, South Africa; Dorothy Mbori-Ngacha, University of Nairobi, Nairobi, Kenya; François Dabis, Inserm U897, ISPED, Bordeaux, France. The KIDS-ART-LINC Collaboration is organized as follows: Principal investigators: Dorothy Mbori-Ngacha and François Dabis. Central coordinating team: Elise Arrive, Daniel Kyabayinze (2006), Benoit Marquis, Leticia Namale (2007), Susan Olet (2007), and Nathan Tumwesigye. Steering committee (in alphabetic order): Alain Azondékon, UPEIV/Hôpital d'Instruction des Armées, Cotonou, Bénin; Alice Zougrana Kaboré, Hôpital Charles de Gaulle, Ouagadougou, Burkina Faso; Andrew Boulle, David Coetzee, Khayelitsha Hospital, Cape Town, South Africa; Avi Violari, Rikash Jokhan, Perinatal HIV Research Unit, Chris Hani Baragwanath Hospital, Soweto, South Africa; Christine Ondoa-Onama, Arua Regional Referral Hospital, Arua, Uganda; Elain Abrams, Rosalid Carter, 12 MTCT Plus sites (http://www.mtctplus.org); Frida Sunjoh, Paediatric service, Provincial Hospital, Bamenda, Cameroon; Haby Signaté Sy, Fatou Ly Ndiaye, Hôpital Albert Royer, Dakar, Sénégal; Higgins Massawe, Augustin Massawe, Muhimbili National Hospital, Dar Es Salaam, Tanzania; Julius Kiwanuka, Mbarara University, Mbarabra, Uganda; Kankasa Chipepo, Mwiya Mwiya, University Teaching Hospital, Lusaka, Zambia; Lorna Renner, Bamenla Goka, Korle Bu Hospital, Accra, Ghana; Louise Wemin, CEPREF Enfants, Abidjan, Côte d'Ivoire; Margaret Pascoe, Connaught Clinic, Harare, Zimbabwe; Maria Nanyonga, Nsambya Hospital, Kampala, Uganda; Mariam Sylla, Fatoumata Dicko-Traoré, Hôpital Gabriel Touré, Bamako, Mali; Marie-Claude Uwurukundo, CHU Butare, Butare, Rwanda; Mark Cotton, Helena Rabie, Hans Prozesky, Clair Edson, Heinrich Weber, and Ayanda Madide, Tygerberg Children's Hospital, Cape Town, South Africa; Mark Paterson, Provincial Hospital of Mutare, Mutare, Zimbabwe; Narcisse Muganga, CHU Kigali, Kigali, Rwanda; Patricia Fassinou, Centre Hospitalier Universitaire (CHU) de Yopougon, Abidjan, Côte d'Ivoire; Pelagie Nimbona, Association Nationale de Soutien aux Séropositifs (ANSS), Bujumbura, Burundi; Pierre Kariyo, University of Kamenge Central Hospital, Bujumbura, Burundi; Ralf Weigel, Lighthouse Clinic, Lilongwe, Malawi; Sarah Rowland Jones, Akum Aveika, Medical Research Council, Fajara, The Gambia; Tim Meade, CorpMed Clinic, Lusaka, Zambia. Advisory scientific committee (in alphabetical order): Chewe Luo (United Nations Children's Fund), Elaine Abrams (MTCT Plus, Columbia University), Marie-Louise Newell (Ghent Group and European Collaborative Study, Africa Center, Hlabisa, South Africa), Mary-Pat Kieffer (USAID), Matthias Egger (ART-LINC of IeDEA, University of Bern), Paolo Miotti (National Institutes of Health/OAR), Ruth Nduati (ANECCA), and Valériane Leroy (Ghent Group and ISPED Bordeaux).
1. WHO. Towards Universal Access. Scaling up priority HIV/AIDS interventions in the health sector. Available at: http://www.who.int/hiv/pub/towards_universal_access_report_2008.pdf
. Accessed July 7, 2008.
2. UNAIDS. 2007 AIDS epidemic update. Available at: http://data.unaids.org/pub/EPISlides/2007/2007_epiupdate_en.pdf
. Accessed July 7, 2008.
3. WHO. Antiretroviral therapy of HIV infection in infants and children in resource-limited settings: towards universal access. Available at: http://www.who.int/hiv/pub/guidelines/art/en/index.html
. Accessed July 7, 2008.
4. WHO. Children and AIDS, a stocktaking report. Available at: http://www.who.int/hiv/paediatric/stocktakingreport.pdf
. Accessed July 7, 2008.
5. Eley B, Davies MA, Apolles P, et al. Antiretroviral treatment for children. S Afr Med J
6. Fassinou P, Elenga N, Rouet F, et al. Highly active antiretroviral therapies among HIV-1-infected children in Abidjan, Cote d'Ivoire. AIDS
7. Nyandiko WM, Ayaya S, Nabakwe E, et al. Outcomes of HIV-infected orphaned and non-orphaned children on antiretroviral therapy in western Kenya. J Acquir Immune Defic Syndr
8. O'Brien DP, Sauvageot D, Zachariah R, et al. In resource-limited settings good early outcomes can be achieved in children using adult fixed-dose combination antiretroviral therapy. AIDS
9. Rouet F, Fassinou P, Inwoley A, et al. Long-term survival and immuno-virological response of African HIV-1-infected children to highly active antiretroviral therapy regimens. AIDS
10. Kariyo P, Mbuzenakamwe M, Baramperanye E, et al. Antiretroviral therapy in a hundred Burundian children with AIDS. Paper presented at: XVI International AIDS Conference; August 13-18, 2006; Toronto, Ontario. Abstract MOPE0208.
11. Mbewe M, Bolton C, Levy J, et al. Children enrolled in a public HIV care and treatment program in Lusaka, Zambia: rapid scale-up and first-year clinical outcomes. Paper presented at: XVI International AIDS Conference; August 13-18, 2006; Toronto, Ontario. Abstract MOAB0201.
12. Bolton-Moore C, Mubiana-Mbewe M, Cantrell RA, et al. Clinical outcomes and CD4 cell response in children receiving antiretroviral therapy at primary health care facilities in Zambia. JAMA
13. Mbori-Ngacha D, Kyabayinze D, Arrivé E, et al. Paediatric ART in sub-Saharan Africa
: program characteristics in the multi-center international KIDS-ART-LINC collaboration. Paper presented at: XVI International AIDS Conference; August 13-18, 2006; Toronto, Ontario. Abstract MOPE0257.
14. Arrivé E, Kyabayinze D, Marquis B, et al. Cohort profile: the paediatric antiretroviral treatment programmes in lower-income countries (KIDS-ART-LINC) collaboration. Int J Epidemiol
15. DAIDS. Table for grading the severity of adult and pediatric adverse events. Available at: http://www3.niaid.nih.gov/research/resources/DAIDSClinRsrch/PDF/Safety/DAIDSAEGradingTable.pdf
. Accessed July 7, 2008.
16. Kaplan E, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc
17. Nardi A, Schemper M. Comparing Cox and parametric models in clinical studies. Stat Med
18. Braitstein P, Brinkhof MW, Dabis F, et al. Mortality of HIV-1-infected patients in the first year of antiretroviral therapy: comparison between low-income and high-income countries. Lancet
19. George E, Noel F, Bois G, et al. Antiretroviral therapy for HIV-1-infected children in Haiti. J Infect Dis
20. Bong CN, Yu JK, Chiang HC, et al. Risk factors for early mortality in children on adult fixed-dose combination antiretroviral treatment in a central hospital in Malawi. AIDS
21. Jones SA, Sherman GG, Varga CA. Exploring socio-economic conditions and poor follow-up rates of HIV-exposed infants in Johannesburg, South Africa. AIDS Care
22. Manzi M, Zachariah R, Teck R, et al. High acceptability of voluntary counselling and HIV-testing but unacceptable loss to follow up in a prevention of mother-to-child HIV transmission programme in rural Malawi: scaling-up requires a different way of acting. Trop Med Int Health
23. Violari A, Cotton M, Gibb D, et al. Antiretroviral therapy initiated before 12 weeks of age reduces early mortality in young HIV-infected infants: evidence from the Children with HIV Early Antiretroviral Therapy (CHER) Study. Paper presented at: 4th IAS Conference on HIV Pathogenesis, Treatment and Prevention; July 22-25, 2007; Sydney, Australia. 2007. Abstract WESS103.
24. Diack MA, Signate Sy H, Diagne Gueye NR, et al. Epidemiological and clinical aspects of paediatric HIV infections in Albert-Royer Paediatric Hospital (Dakar, Senegal) [in French]. Arch Pediatr
25. Puthanakit T, Oberdorfer A, Akarathum N, et al. Efficacy of highly active antiretroviral therapy in HIV-infected children participating in Thailands National Access to Antiretroviral Program. Clin Infect Dis
26. Reddi A, Leeper SC, Grobler AC, et al. Preliminary outcomes of a paediatric highly active antiretroviral therapy cohort from KwaZulu-Natal, South Africa. BMC Pediatr
27. O'Brien DP, Sauvageot D, Olson D, et al. Treatment outcomes stratified by baseline immunological status among young children receiving nonnucleoside reverse-transcriptase inhibitor-based antiretroviral therapy in resource-limited settings. Clin Infect Dis
28. van Kooten Niekerk NK, Knies MM, Howard J, et al. The first 5 years of the family clinic for HIV at Tygerberg Hospital: family demographics, survival of children and early impact of antiretroviral therapy. J Trop Pediatr
29. Giordano TP, Gifford AL, White AC Jr, et al. Retention in care: a challenge to survival with HIV infection. Clin Infect Dis
30. Anglaret X, Toure S, Gourvellec G, et al. Impact of vital status investigation procedures on estimates of survival in cohorts of HIV-infected patients from sub-Saharan Africa
. J Acquir Immune Defic Syndr
31. Brinkhof MWG, Dabis F, Myer L, et al. Early loss to program in HIV-infected patients starting potent antiretroviral therapy in lower-income countries. Bull World Health Organ
32. Chintu C, Bhat GJ, Walker AS, et al. Co-trimoxazole as prophylaxis against opportunistic infections in HIV-infected Zambian children (CHAP): a double-blind randomised placebo-controlled trial. Lancet
33. Hesseling AC, Westra AE, Werschkull H, et al. Outcome of HIV infected children with culture confirmed tuberculosis. Arch Dis Child
34. Lockman S, Shapiro RL, Smeaton LM, et al. Response to antiretroviral therapy after a single, peripartum dose of nevirapine. N Engl J Med
35. Sethi AK, Celentano DD, Gange SJ, et al. Association between adherence to antiretroviral therapy and human immunodeficiency virus drug resistance. Clin Infect Dis