More than 2 million children are currently living with HIV-1, of which more than 90% live in Sub-Saharan Africa.1,2 Growth failure, including poor weight gain and linear stunting, is one of the most common conditions in children infected with HIV.3–6 The World Health Organization (WHO) defines growth failure as weight-for-age z-score (WAZ) or height-for-age z-score (HAZ) more than 2 standard deviations (SDs) below the median weight or height, corresponding to the 2.3rd percentile on reference growth charts.7 Previous studies have established that growth failure in children with HIV is a sensitive indicator of disease progression and growth recovery is often observed in response to combination antiretroviral therapy (cART) initiation.8–13
The velocity and degree of growth recovery among children on cART is inconsistent and some never achieve normal growth, with children who start cART at older ages less likely than younger children to achieve growth recovery.9,14–16 However, age does not explain all heterogeneity observed in growth potential and other predictors are not well understood. Children and adults who start cART with advanced disease have been shown to have slower and suboptimal treatment responses.9,17–19 We hypothesized that severity of disease, marked by decreased immune function, at the time of cART initiation impedes growth recovery. Understanding the relationship between baseline immune status and growth recovery could have important implications for cART eligibility criteria for HIV-infected children.
HIV-infected children younger than 15 years who received care at Harriet Shezi Children's Clinic from April 1, 2004 through March 31, 2008 were included in the analysis. Harriet Shezi Children's Clinic is an outpatient pediatric HIV clinic at Chris Hani Baragwanath Hospital in Soweto, South Africa. After initiating cART, children were scheduled for visits after 1 month and 3 months, and then every 3 months thereafter. Clinical information, including weight and length (aged 2 years or younger) or height (aged older than 2 years), was collected at every visit, and laboratory information was collected every 6 months. Additional visits and laboratory tests were scheduled as clinically needed.
Baseline in this cohort study was defined as cART initiation. Eligibility for cART was determined by the South African national guidelines.20 For children younger than 3 years or weighing <10 kg, the first-line regimen included stavudine (d4T), lamivudine (3TC), and ritonavir-boosted lopinavir (LPV/r); the second-line regimen included zidovudine (AZT), didanosine (ddI), and nevirapine (NVP). For children at least 3 years of age and weighing more than 10 kg, the first-line regiment included d4T, 3TC, and efavirenz (EFV); the second-line regimen included AZT, ddI, and LPV/r or EFV. Alternate regimens also included abacavir (ABC), 3TC, and LTV/r or EFV, as well as AZT, 3TC, and LPV/r. To be at risk for growth recovery, children must be experiencing growth failure. Therefore, only children underweight (WAZ ≤ −2) or stunted (HAZ ≤ −2) at baseline were included in follow-up analyses.
The primary analysis assessed the effect of severe baseline immune status on growth recovery. The outcome of interest, growth recovery, was defined as the first occurrence of a WAZ or HAZ > −2 standard deviations of the median value for a given age group and sex. WHO growth charts were used to calculate weight z-scores for children younger than or aged 10 years and height z-scores for all children.21–23 CDC charts were used to calculate weight z-scores for children older than 10 years.24 The primary exposure was degree of immunodeficiency at cART initiation, defined as “severe” (exposed) or “not severe” (unexposed) according to the 2006 WHO CD4 cell-percentage criteria.25 The CD4 thresholds for severe immunodeficiency were <25% for children aged 11 months or younger, <20% for children aged 12–35 months, and <15% for children aged older than 35 months.
Potential confounding factors were assessed via a directed acyclic graph.26 and included the baseline variables: age, severity of growth failure, viral load, and tuberculosis treatment status. In nonstratified models, age, viral load, and z-scores were modeled with Stone and Koo additive splines constrained to be linear in the tails, with knots at the fifth, 35th, 65th, and 95th percentiles.27
To explore if the effect of baseline immunodeficiency on growth recovery varied by other baseline characteristics, we examined effect estimates stratified by age and degree of growth failure at the start of cART. In stratified models, age was categorized as 11 months or younger, 12–35 months, and older than 35 months based on the age cut points for treatment eligibility in the 2006 WHO guidelines. Growth failure at cART initiation was dichotomized as severe or moderate; children with a z-score < −3 were classified as having severe growth failure and children with a z-score between –2 and –3 as having moderate growth failure. For all measures, we carried the last observation forward for missing data.
Descriptive statistics were used to generate the distributions of baseline characteristics (age, sex, degree of immunodeficiency, WHO clinical stage, viral load count, ART regimen, TB treatment status, and growth z-scores) of the study population overall and separately for those who were underweight or stunted. The Kaplan–Meier estimator was used to estimate the overall and stratified survival functions. To quantify the association between baseline immune status and time to growth recovery, hazard ratios (HRs) and corresponding 95% confidence intervals (CIs) were calculated. Two separate analyses were conducted, one for recovery to a normal weight that only included children who were underweight at baseline and one for recovery to a normal length/height that only included children who were stunted at baseline.
We fit Cox proportional hazards models to generate overall and stratified crude and adjusted effect estimates. Only children with complete information on CD4 cell percentage, age, viral load, tuberculosis treatment status, and either weight or length/height were included in follow-up analyses. Follow-up time was censored at time of first growth recovery, loss to follow-up (LTFU), death, or the administrative end of the study (2 years after the start of cART or March 31, 2008), whichever occurred first. We allowed for a 3-month window around the final 2-year visit to account for children who may have arrived late for their scheduled visit. Children who were last seen more than 3 months before the administrative end of the study and who had not yet experienced the event of interest or death were considered LTFU. Those who had no follow-up visit beyond the baseline visit were assigned a survival time of 15 days, the midpoint between the baseline visit and the next scheduled visit 1 month later. This technique allowed children who had no follow-up visit to contribute some information to follow-up analyses rather than being excluded entirely.
The Cox proportional hazards assumption was evaluated with martingale-based residuals28 and a P value was obtained for a Kolmogorov–Smirnov type test. A P value < 0.05 was considered evidence for a violation of the proportional hazards assumption. All models used Efron method for ranking tied event times.29 Analyses were conducted in SAS 9.3 (SAS Institute, Inc, Cary, NC).
In additional multivariate models, we looked at the predictive ability of other study covariates. Specifically, we used Cox regression models to estimate adjusted HRs (and 95% CIs) for the association between covariates and LTFU or death, as well as with growth recovery. We also used logistic regression to estimate adjusted odds ratios (and 95% CIs) for the associations between covariates and severe immunodeficiency (both measured at baseline).
We assessed potential selection bias induced by differential LTFU or death by recalculating the overall unadjusted HR for the association between severe baseline immunodeficiency and growth recovery under a range of assumptions about children who were LTFU or died. The lower extreme scenario, corresponding to an HR that approaches zero, assumed all unexposed (no severe immunosuppression) children who were LTFU or died experienced growth recovery at the time they were censored, whereas all exposed (severe immunosuppression) children contributed 2 years of outcome-free person-time without achieving growth recovery. The upper extreme scenario, corresponding to an HR that approaches infinity, assumed the opposite, that is, all exposed children who were LTFU or who died experienced growth recovery at the time they were censored and all unexposed contributed 2 years of outcome-free person-time. Under more moderate scenarios, we varied the outcome proportion in the exposure groups from 10% to 90%, corresponding to HRs of about 0.05 to 20, and assumed growth recovery occurred at the median follow-up time for children who experienced growth recovery in the full cohort.
Study Population Characteristics
Among the 2399 children younger than 15 years initiating cART during the study period, the median age was 4.6 [interquartile range (IQR): 1.7–7.6] years and approximately half (48.9%) were female (Table 1). The majority (79.4%) had severe immunodeficiency, 71.9% were WHO clinical stage 3 or 4, 59.4% had viral load counts >100,000, and 27.0% were on TB treatment at the time of cART initiation. Overall, 34.1% received a baseline regimen of d4T/3TC/LPV/r, 64.1% received d4T/3TC/EFV, and 1.7% received an alternate regimen (ABC/3TC/LPV/r, ABC/3TC/EFV, AZT/ddI/LPV/r, or AZT/3TC/LPV/r).
Of the 2242 (93.5%) children with a recorded baseline height or weight measurement, 1600 (71.4%) were experiencing growth failure; 986 (61.6%) were underweight and stunted, 128 (8.0%) were only underweight, and 486 (30.4%) were stunted with normal weight. Underweight and stunted children were younger than the overall cohort of children initiating cART, with median baseline ages of 3.6 (IQR: 1.1–7.4) years and 3.8 (IQR: 1.7–7.0) years, respectively. Slightly higher proportions of underweight and stunted children had severe immunodeficiency, were WHO clinical stage 3 or 4, had viral load counts >100,000, received a baseline regimen of d4T/3TC/LPV/r, and were on TB treatment. About half of children with growth failure were severely underweight (56.4% with WAZ < −3) or severely stunted (55.6% with HAZ < −3).
In total, baseline immunodeficiency information was available for 998 underweight children and 1344 stunted children. Of those, 953 underweight children and 1282 stunted children had complete covariate information and were included in follow-up analyses. Underweight children contributed 636.4 person-years of follow-up and stunted children contributed 1270.6 person-years (Table 2). The median follow-up time was 306 days (IQR: 98–621) for stunted children and 168 days (IQR: 52–360) for underweight children. During follow-up, 110 underweight children were LTFU and 86 died before experiencing growth recovery. In the length/height analysis, 161 stunted children were LTFU and 83 died. Among the 36 underweight children who switched regimens before follow-up time was censored, the medium time to switch was 259 (IQR: 84–509) days. In the length/height analysis, 87 children switched regimens, with a medium time to switch of 374 (IQR: 168–563) days.
Over 2 years of follow-up, 81% of underweight children achieved normal weight, with a median time to recovery of 259 days (95% CI: 238 to 279 days). A lower proportion of severely immunodeficient children ultimately achieved recovery to a normal weight (79% vs 87%) (Fig. 1A). Stratified Kaplan–Meier estimates suggest that regardless of baseline immunodeficiency status, older children (aged older than 35 months) and those who were severely underweight at cART initiation were less likely to achieve recovery to a normal weight compared with younger children and those who were only moderately underweight (Figs. 1B, C).
In Cox regression analyses, severe immunodeficiency at baseline was not associated with achieving recovery to a normal weight on cART, with an overall unadjusted HR of 1.05 (95% CI: 0.83 to 1.32) (Table 3). The point estimate was slightly different for children who were severely underweight (HR: 1.42, 95% CI: 0.94 to 2.16) and moderately underweight (HR: 0.99, 95% CI: 0.75 to 1.31) at the start of cART, but overlap of the 95% CIs was substantial. We also did not observe notable modification by baseline age, with HRs (95% CIs) of 1.25 (0.74 to 2.12), 1.02 (0.67 to 1.56), and 1.01 (0.73 to 1.40) for those aged 11 months or younger, 12–35 months, and 35 months or older, respectively. Adjustment for covariates did not substantively alter the overall or stratified effect estimates (Table 3).
Among stunted children, only 64% achieved length/height recovery within the first 2 years of cART, with a median recovery time of 700 days (95% CI: 619 to 787 days). A slightly higher proportion of children with severe immunodeficiency ultimately achieved recovery to a normal length/height compared with those without severe immunodeficiency (68% vs 50%) (Fig. 1D). Similar to the weight analysis, stratified Kaplan–Meier estimates suggest that regardless of baseline immunodeficiency status, older children (aged older than 35 months) and those with severe stunting at cART initiation were less likely to achieve recovery to a normal height than younger children and those who were only moderately stunted (Figs. 1E, F).
Severe immunodeficiency at the start of cART was not strongly associated with length/height recovery, with an overall unadjusted HR of 1.06 (95% CI: 0.83 to 1.34) (Table 3). We observed modest modification by severity of baseline stunting and age. The stratum specific unadjusted HRs among those moderately and severely stunted were 1.27 (95% CI: 0.94 to 1.70) and 0.95 (95% CI: 0.64 to 1.41), respectively. Before adjustment, HRs stratified by baseline age were 0.69 (95% CI: 0.38 to 1.25) among those 11 months or younger, 1.41 (95% CI: 0.86 to 2.32) among those 12–35 months, and 0.95 (95% CI: 0.70 to 1.29) among older than 35 months. However, overlap of the stratified CIs was substantial. Effect estimates adjusted for covariates were mostly similar to unadjusted estimates, although the overall HR increased to 1.29 (95% CI: 1.01 to 1.64).
Association of Baseline Immunodeficiency, Growth Recovery, and LTFU or Death With Other Study Covariates
Older children were less likely to achieve weight and length/height growth recovery, and also less likely to die or to be LTFU (Table 4). In addition to being strongly associated with growth recovery, more severe growth failure at baseline was modestly predictive of severe baseline immunodeficiency and death or LTFU. Children with a history of TB treatment were more likely to have severe immunodeficiency at baseline, but they did not seem to be more or less likely to experience growth recovery compared with children who were not on TB treatment at baseline. Finally, there was a strong positive association between degree of baseline immunodeficiency and viral load.
LTFU and death are common outcomes among HIV-positive clinical cohorts, and it is likely that such children have different characteristics than those who remain in care. Our sensitivity analysis showed that one would have to make extreme improbable assumptions about the association between baseline immunodeficiency and growth recovery among those who were LTFU or who died to alter the conclusion of no association between baseline immunodeficiency and growth recovery. That is, the observed unadjusted effect estimate for the association between severe immunodeficiency and growth recovery among those who were not LTFU and did not die was 1.03 (95% CI: 0.82 to 1.30) in the weight analysis and 1.05 (95% CI: 0.83 to 1.33) in the length/height analysis (see Table, Supplemental Digital Content 1, http://links.lww.com/QAI/A341, which shows the results of the sensitivity analysis). Combining the observed effect estimate from the analysis of weight recovery with a range of unobserved hypothesized effect estimates for those who were LTFU or who died produced weighted posterior estimates that ranged from 0.67 (95% CI: 0.54 to 0.83) under the lower extreme hypothesis that the unobserved HR approached zero to 1.81 (95% CI: 1.44, 2.28) under the other extreme hypothesis that the unobserved HR approached infinity. In the analysis of length/height recovery, the extreme bounds of the HR were 0.57 (95% CI: 0.46 to 0.69) to 1.97 (95% CI: 1.56 to 2.48). The weighted posterior estimates produced under more reasonable hypotheses about the unobserved effect estimate among those who were LTFU or who died ranged from about 0.80 to 1.30 in both analyses (see Table, Supplemental Digital Content 1, http://links.lww.com/QAI/A341, which shows the results of the sensitivity analysis).
In a large cohort of South African children initiating cART, growth failure was common, with more than 70% being either underweight or stunted. Most children (61.6%) were both underweight and stunted, 8.0% were only underweight, and 30.4% were only stunted. A substantial proportion of those underweight and stunted failed to achieve growth recovery within the first 2 years of treatment with cART, with only 64% achieving a normal length/height for age and 81% achieving a normal weight for age. These figures are particularly astonishing, considering that normal growth was defined as reaching a weight or height z-score of -2, corresponding to the 2.3rd percentile on reference growth charts.
Our analysis did not support the hypothesized association between severe baseline immunodeficiency and growth recovery and the effect seemed to be only modestly modified by severity of growth failure or age at the start of cART. Regardless of baseline immunodeficiency, severity of growth failure and age at the start of cART were highly predictive of growth recovery, with older children and those with more severe growth failure notably less likely to achieve growth recovery. In predictive models, baseline TB treatment status was not associated with growth recovery. However, more in-depth analyses to explore the relationship between TB disease and subsequent treatment with growth recovery are warranted by future research.
To date, reports on the effect of baseline immune status and growth recovery due to cART in the published literature are sparse. Similar to our findings, a smaller (N = 225) Thai study reported a crude risk ratio of 1.14 (95% CI: 0.96 to 1.34) for the association between growth failure reversal and each 5% increase in baseline CD4 cell percentage.8 Other studies that examined the association between baseline immunodeficiency and growth on cART included all children regardless of their growth status at baseline and did not provide results stratified by baseline age or growth status,9,30,31 which precludes assessment of whether baseline immunodeficiency effects underweight and stunted children's recovery from growth failure once cART is started.
Furthermore, prior studies of growth among children initiating cART have primarily included children with an older median age at cART initiation. An older median age is suggestive of a cohort of long-term survivors who may be different from HIV-infected children in general. With a relatively young median age and a large proportion of children younger than 2 years of age, the findings of this study are further generalizable to infants and young children, who also experience growth failure and are targeted for cART. Our results suggest that age is predictive of LTFU or death, with a higher rate observed among younger children. This observation is consistent with other studies32,33 and may have implications for the proportion of children reported to achieve growth recovery on cART because younger children are also more likely to achieve growth recovery after treatment is started. However, age was not strongly associated with having severe immunodeficiency at the start of cART and our sensitivity analysis showed that the HR for the effect of severe baseline immunodeficiency on growth recovery is robust unless very extreme assumptions are made about those who were LTFU or who died.
The type of cART regimen that children receive may play a role in their growth recovery. Some studies found that, among children with exposure to single dose NVP for the prevention of mother-to-child transmission who were initially started on cART containing LPV/r, children who were later switched to a regimen containing NVP had a larger percentage increase in WAZ and HAZ than children who remained on a regimen containing LPV/r.34,35 In our study, no children were switched from their baseline regimen to a regimen containing NVP. This may partially explain the poor growth outcomes we observed in our study.
Strengths of our analysis included a large sample size, wide age range, and long follow-up time of 2 years, which allowed for a greater number of children to experience the event of interest, thus increasing the precision of our effect estimates. Using sensitivity analyses, we were also able to determine that the potential bias introduced by LTFU and death was likely minimal. An important limitation of the study lies in the lack of data on socioeconomic status, food security, and nutritional support provided in addition to cART. Complex bidirectional relationships exist between immunodeficiency status and the covariates included in this analysis, including severity of growth failure, viral load, and TB treatment status. We made the assumption that all potential confounding was time fixed.
In conclusion, growth failure clearly remains an important problem among HIV-infected children and, despite the availability of cART, many fail to recover to a normal weight or height. Poor growth recovery after cART initiation is concerning, as it may reflect an inadequate treatment response.12,13,36 Degree of baseline immunosuppression does not seem to substantially influence growth recovery on cART. Our results suggest that, in children initiating ART at time of severe growth failure, the potential for growth recovery is limited. The biological or social phenomena underlying this observation are unclear and merit further investigation. Our Kaplan–Meier curves suggest that the proportion of children that have the potential to achieve weight recovery begins to plateau before 2 years of treatment with cART, while children may continue to achieve length/height recovery beyond 2 years. Baseline age may also be an important factor that influences growth recovery, and additional research is needed to design appropriate interventions for affected children. Our finding that a substantial proportion of children do not achieve normal growth despite 2 years of cART highlights the need for early treatment initiation, before the presence of growth failure and independent of the level of immunodeficiency.
1. UNAIDS. Global Report: UNAIDS Report on the Global AIDS Epidemic 2010. Geneva, Switzerland: WHO; 2010.
2. World Health Organization. WHO Recommendations on the Diagnosis of HIV Infection in Infants and Children. Geneva, Switzerland: WHO; 2010.
3. Hirschfeld S. Dysregulation of growth and development in HIV-infected children. J Nutr. 1996;126(10 suppl):2641S–2650S.
4. Arpadi SM. Growth failure in children with HIV infection. J Acquir Immune Defic Syndr. 2000;25(suppl 1):S37–S42.
5. Miller TL, Evans SJ, Orav EJ, et al.. Growth and body composition in children infected with the human immunodeficiency virus-1. Am J Clin Nutr. 1993;57:588–592.
6. McKinney RE Jr, Robertson JW. Effect of human immunodeficiency virus infection on the growth of young children. Duke Pediatric AIDS Clinical Trials Unit. J Pediatr. 1993;123:579–582.
7. Kessler DB, Baker SS, Silverman LA. Growth assessment and growth failure. Consens Pediatr. 2004;1:3.
8. Aurpibul L, Puthanakit T, Taecharoenkul S, et al.. Reversal of growth failure in HIV-infected Thai children treated with non-nucleoside reverse transcriptase inhibitor-based antiretroviral therapy. AIDS Patient Care STDS. 2009;23:1067–1071.
9. Musoke PM, Mudiope P, Barlow-Mosha LN, et al.. Growth, immune and viral responses in HIV infected African children receiving highly active antiretroviral therapy: a prospective cohort study. BMC Pediatr. 2010;10:56.
10. Kabue MM, Kekitiinwa A, Maganda A, et al.. Growth in HIV-infected children receiving antiretroviral therapy at a pediatric infectious diseases clinic in Uganda. AIDS Patient Care STDS. 2008;22:245–251.
11. Nachman SA, Lindsey JC, Moye J, et al.. Growth of human immunodeficiency virus-infected children receiving highly active antiretroviral therapy. Pediatr Infect Dis J. 2005;24:352–357.
12. Verweel G, van Rossum AM, Hartwig NG, et al.. Treatment with highly active antiretroviral therapy in human immunodeficiency virus type 1-infected children is associated with a sustained effect on growth. Pediatrics. 2002;109:E25.
13. Yotebieng M, Van Rie A, Moultrie H, et al.. Six-month gain in weight, height, and CD4 predict subsequent antiretroviral treatment responses in HIV-infected South African children. AIDS. 2010;24:139–146.
14. Bolton-Moore C. Clinical outcomes and CD4 cell response in children receiving antiretroviral therapy at primary health care facilities in Zambia. JAMA. 2007;298:1888–1899.
15. McGrath CJ, Chung MH, Richardson BA, et al.. Younger age at HAART initiation is associated with more rapid growth reconstitution. AIDS. 2011;25:345–355.
16. Sutcliffe CG, van Dijk JH, Munsanje B, et al.. Weight and height z-scores improve after initiating ART among HIV-infected children in rural Zambia: a cohort study. BMC Infect Dis. 2011;11:54.
17. Hammer SM, Eron JJ Jr, Reiss P, et al.. Antiretroviral treatment of adult HIV infection: 2008 recommendations of the International AIDS Society-USA panel. JAMA. 2008;300:555–570.
18. Meyers TM, Yotebieng M, Kuhn L, et al.. Antiretroviral therapy responses among children attending a large public clinic in Soweto, South Africa. Pediatr Infect Dis J. 2011;30:974–979.
19. Palombi L, Marazzi MC, Guidotti G, et al.. Incidence and predictors of death, retention, and switch to second-line regimens in antiretroviral- treated patients in sub-Saharan African Sites with comprehensive monitoring availability. Clin Infect Dis. 2009;48:115–122.
20. Meyers T, Eley B, Leoning W. Guidelines for the Management of HIV-Infected Children. Johannesburg, South Africa: Jacana Media; 2005.
21. WHO Multicentre Growth Reference Study Group. WHO Child Growth Standards based on length/height, weight and age. Acta Paediatr Suppl. 2006;450:76–85.
22. de Onis M, Onyango AW, Borghi E, et al.. Development of a WHO growth reference for school-aged children and adolescents. Bull World Health Organ. 2007;85:660–667.
23. WHO Multicentre Growth Reference Study Group. WHO Child Growth Standards: Methods and Development: Length/Height-for-Age, Weight-for-Age, Weight-for-Length, Weight-for-Height and Body Mass Index-for-Age. Geneva, Switzerland: World Health Organization; 2006.
24. National Center for Health Statistics. 2000 CDC Growth Charts for the United States: Methods and Development. Hyattsville, MD: Centers for Disease Control and Prevention; 2002.
25. Technical Reference Group on Paediatric HIV Care and Treatment. Antiretroviral Therapy of HIV Infection in Infants and Children in Resource-Limited Settings: Towards Universal Access. Recommendations for a Public Health Approach, Geneva, Switzerland: WHO; 2006.
26. Greenland S, Pearl J, Robins JM. Causal diagrams for epidemiologic research. Epidemiology. 1999;10:37–48.
27. Stone C, Koo C. Additive splines in statistics. Proc Stat Comp Sect Am Statist Assoc. 1985;27:45–48.
28. Lin D, Wei L, Ying Z. Checking the Cox model with cumulative sums of martingale-based residuals. Biometrika. 1993;80:557–572.
29. Efron B. The efficiency of Cox's likelihood function for censored data. J Am Stat Assoc. 1977;72:557–565.
30. Kekitiinwa A, Lee KJ, Walker AS, et al.. Differences in factors associated with initial growth, CD4, and viral load responses to ART in HIV-infected children in Kampala, Uganda, and the United Kingdom/Ireland. J Acquir Immune Defic Syndr. 2008;49:384–392.
31. Weigel R, Phiri S, Chiputula F, et al.. Growth response to antiretroviral treatment in HIV-infected children: a cohort study from Lilongwe, Malawi. Trop Med Int Health. 2010;15:934–944.
32. Low risk of death, but substantial program attrition, in pediatric HIV treatment cohorts in Sub-Saharan Africa. J Acquir Immune Defic Syndr. 2008;49:523–531.
33. Fetzer BC, Hosseinipour MC, Kamthuzi P, et al.. Predictors for mortality and loss to follow-up among children receiving anti-retroviral therapy in Lilongwe, Malawi. Trop Med Int Health. 2009;14:862–869.
34. Coovadia A, Abrams EJ, Stehlau R, et al.. Reuse of nevirapine in exposed HIV-infected children after protease inhibitor-based viral suppression: a randomized controlled trial. JAMA. 2010;304:1082–1090.
35. Palumbo P, Lindsey JC, Hughes MD, et al.. Antiretroviral treatment for children with peripartum nevirapine exposure. N Engl J Med. 2010;363:1510–1520.
36. Arpadi SM. Growth Failure in HIV-Infected Children. Durban, South Africa: World Health Organization, Department of Nutrition for Health and Development; 2005.