Portal de Periódicos da CAPES

Growth of infants born to HIV-infected women in South Africa according to maternal and infant characteristics

2010; Wiley; Volume: 15; Issue: 11 Linguagem: Inglês

10.1111/j.1365-3156.2010.02634.x

1365-3156

Kartik K. Venkatesh, Mark N. Lurie, Elizabeth W. Triche, Guy de Bruyn, Joseph I. Harwell, Stephen T. McGarvey, Glenda Gray,

Gestational Diabetes Research and Management

Objective To evaluate growth parameters assessed by weight and length in HIV-infected and HIV-uninfected infants born to HIV-infected mothers in South Africa from birth to 6 months of age. Methods We calculated z-scores for weight-for-age (WAZ), length-for-age (LAZ) and weight-for-length (WLZ) among a cohort of 840 mother–infant dyads. Multivariable Cox proportional hazards models with time-varying covariates were used to estimate the risk of falling <−2 z-scores for WAZ, LAZ, and WLZ as a function of infant and maternal characteristics. Results By 6 months after birth, a fifth of infants had WAZ <−2, 19% had an LAZ <−2, and 29% had a WLZ <−2. WLZ and WAZ were significantly lower in HIV-infected infants than in uninfected infants by 3 months of age and LAZ by 6 months of age (P < 0.001). The risk of WAZ falling <−2 was associated with decreasing maternal CD4 cell count (adj. HR for CD4 cell count <200 cells/μl: 1.64; 95% CI: 1.10–2.43), premature birth (adj. HR: 2.82; 95% CI: 2.06–3.86) and formula feeding (adj. HR: 3.35; 95% CI: 1.64–6.85). The risk of LAZ falling <−2 was associated with increasingly lower maternal age (adj. HR for<20 years: 0.54; 95% CI: 0.31–0.96), lower maternal CD4 cell count (adj. HR for CD4 cell count <200 cells/μl: 1.72; 95% CI: 1.14–2.59), premature birth (adj. HR: 2.37; 95% CI: 1.70–3.30) and formula feeding (adj. HR: 4.22; 95% CI: 1.85–9.62). The risk of WLZ falling <−2 was significantly associated with infant HIV infection (adj. HR: 1.64; 95% CI: 1.16–2.32) and formula feeding (adj. HR: 1.78; 95% CI: 1.11–2.83). The risk of WAZ and LAZ falling <−2 was more than two times greater for HIV-infected infants than for uninfected infants with gastrointestinal infections. Conclusions HIV-infected infants were more likely to be stunted and wasted than uninfected infants, which often occurred within 3 months after birth. Infants who were born to mothers with advanced HIV disease, formula-fed and co-infected with HIV and gastrointestinal infections were at greater risk for growth disturbances. Further interventions are needed to promptly initiate both HIV-infected mothers and infants on appropriate antiretroviral therapy and nutritional supplementation. Croissance des nourrissons nés de mères infectées par le VIH en Afrique du Sud selon les caractéristiques maternelles et infantiles Objectif: Investiguer les paramètres de croissance évaluée par le poids et la taille chez les nourrissons infectés ou non par le VIH nés de mères infectées par le VIH en Afrique du Sud, de la naissance à six mois d'âge. Méthodes: Nous avons calculé les scores z du poids-pour-l'âge (WAZ), de la taille-pour-l'âge (LAZ) et du poids-pour-la taille (WLZ) dans une cohorte de 840 dyades mère-enfant. Des modèles multivariés de risques proportionnels de Cox avec des covariables variant dans le temps ont été utilisées pour estimer le risque pour un scores z chutant à <-2 pour WAZ, LAZ et WLZ selon les caractéristiques des nourrissons et des mères. Résultats: À 6 mois après la naissance, un cinquième des nourrissons avaient un WAZ <-2, 19% avaient un LAZ <-2 et 29% avaient un WLZ <-2. WLZ et WAZ étaient significativement plus faibles chez les nourrissons infectés par le VIH que chez ceux non infectés à l'âge de 3 mois et LAZ était plus faible à l'âge de 6 mois (p <0,001). Le risque pour un WAZ chutant à <-2 a été associéà une diminution des CD4 de la mère (HR ajusté pour le nombre de cellules CD4 <200 cellules/μl: 1,64; IC95%: 1,10 - 2,43), à la naissance prématurée (HR ajusté: 2,82; IC95%: 2,06 - 3,86) et à l'alimentation au biberon (HR ajusté: 3,35; IC95%: 1,64 - 6,85). Le risque pour LAZ de chuter à <-2 a été associé avec l'âge maternel de plus en plus bas (RH ajusté pour moins de 20 ans: 0,54; IC95%: 0,31 – 0,96), à un plus faible nombre de cellules CD4 de la mère (RH ajusté pour le nombre de cellules CD4 <200 cellules/μl: 1,72; IC95%: 1,14 – 2,59), à la naissance prématurée (HR ajusté: 2,37; IC95%: 1,70 à 3,30) et à l'alimentation au biberon (HR ajusté: 4,22; IC95%: 1,85 - 9.62). Le risque du WLZ de chuter à <-2 était significativement associéà l'infection du nourrisson par le VIH (HR ajusté: 1,64; IC95%: 1,16 - 2,32) et à l'alimentation par le biberon (HR ajusté: 1,78; IC 95%: 1,11 - 2,83). Le risque pour WAZ et LAZ chutant à <-2 était plus de deux fois plus élevé pour les nourrissons infectés que pour ceux non infectés par le VIH avec des infections gastro-intestinales. Conclusions: les nourrissons infectés par le VIH étaient plus susceptibles de croissance altérée et d'amaigrissement que les enfants non infectés, ce qui survient souvent dans les 3 mois après la naissance. Les nourrissons nés de mères infectées par le VIH à un stade avancé, recevant du lait maternisé et coinfectés par le VIH et des infections gastro-intestinales étaient plus à risque pour des altérations de la croissance. Des interventions supplémentaires sont nécessaires pour initier sans tarder à la fois les mères et les nourrissons infectés par le VIH, au traitement antirétroviral approprié et à des suppléments nutritionnels. Crecimiento de los niños nacidos de mujeres infectadas con VIH en Sudáfrica de acuerdo con las características maternas e infantiles. Objetivo: Evaluar los parámetros de crecimiento mediante el peso y la longitud en niños infectados y sin infectar con VIH, nacidos de mujeres infectadas con VIH en Sudáfrica, desde el nacimiento y hasta los seis meses de edad. Métodos: Hemos calculado los z-scores para peso-por-edad (PEZ), longitud-por-edad (LEZ), y peso-por-longitud (PLZ) en una cohorte de 840 madres e hijos. Se utilizaron modelos de riesgos proporcionales de Cox con datos multifactoriales y covariables de tiempo variable para estimar el riesgo de caer <-2 z-scores para PEZ, LEZ, y PLZ como una función de las características infantiles y maternas. Resultados: Para cuando alcanzaban los 6 meses de edad, una quinta parte de los bebés tenían un PEZ<-2, 19% tenían un LEZ<-2, y 29% tenían un PLZ<-2. PLZ y PEZ eran significativamente más bajos en bebés infectados con VIH que en los bebés no infectados a los 3 meses de edad, y PEZ a los 6 meses de edad (p<0.001). El riesgo de que el PEZ cayese <-2 estaba asociado con un descenso en el conteo de CD4 materno (adj. HR para conteo de células CD4 <200 células/μl: 1.64; 95%IC: 1.10-2.43), parto prematuro (adj. HR: 2.82; 95%IC: 2.06-3.86), y lactancia artificial (adj. HR: 3.35;95%IC: 1.64-6.85). El riesgo de que LEZ cayese <-2 estaba asociado con una menor edad materna (adj. HR for<20 años: 0.54; 95%IC: 0.31-0.96), menor conteo de CD4 materno (adj. HR para conteo de células CD4 <200 células/μl: 1.72;95% IC: 1.14-2.59), parto prematuro (adj. HR: 2.37; 95%IC: 1.70-3.30), y lactancia artificial (adj. HR: 4.22; 95%IC: 1.85-9.62). El riesgo de que PEZ cayese <-2 estaba significativamente asociado con la infección infantil por VIH (adj. HR: 1.64; 95%IC: 1.16-2.32), y la lactancia artificial (adj. HR: 1.78; 95%IC: 1.11-2.83). El riesgo de que PEZ y LEZ cayesen 90%) live in sub-Saharan Africa (United Nations 2008). Paediatric HIV infection is associated with growth retardation and is likely a significant contributor to infant malnutrition and morbidity (Bakaki et al. 2001). Poor growth can be one of the first clinically recognized manifestations of HIV infection in children and has a significant impact on short-term survival (Berhane et al. 1997; Arpadi 2006). Disturbances in growth are likely to occur well before the onset of opportunistic infections or other HIV-related disease manifestations (Oleske et al. 1983). It has been suggested that more than half of all HIV-infected children experience abnormal growth patterns (Hirschfeld 1996; Arpadi 2006). HIV-infected children suffer stunting and wasting within the first few months after birth (Moye et al. 1996; Fausto et al. 2009). Despite expanding access to antiretroviral therapy (ART) in resource-limited settings, it is estimated that less than half of those in need are currently receiving treatment (Bartlett & Shao 2009). Major gaps remain in adequate coverage of mother-to-child transmission (MTCT) services as well as ART access in sub-Saharan Africa for women and their infants (Becquet et al. 2009), highlighting the need for continued studies to examine the impact of HIV infection on infant health outcomes. The degree to which maternal characteristics, vertically acquired HIV infection, or an adverse environment place children at increased risk for growth retardation remains to be fully elucidated (Bailey et al. 1999; Webb et al. 2008; Patel et al. 2010). Growth may be one of the most sensitive indicators of disease progression in HIV-infected infants (Chearskul et al. 2002). Studies conducted among infants born to HIV-infected mothers can provide insight into the nutritional implications of paediatric HIV infection. This prospective cohort study examined growth patterns during the first 6 months of life among both HIV-infected and HIV-uninfected South African infants born to HIV-infected mothers. We estimated the risk of child growth retardation according to both child and maternal characteristics. The study was conducted at three hospitals in South Africa: Chris Hani Baragwanath Hospital (Soweto), Coronation Hospital (Johannesburg) and Mombray Hospital (Cape Town) between October 2000 and September 2002. The primary study was a randomized, open-label, multicentre clinical trial (RCT) to compare the use of single-dose nevirapine (NVP) to 6 weeks of zidovudine (AZT) commenced within 24 h of delivery to prevent MTCT in infants of antiretroviral (ARV)-naïve HIV-infected mothers; more details on the methods can be found elsewhere (Gray et al. 2005; Venkatesh et al. 2010). Clean public drinking water was available in this urban setting (Statistics South Africa 2008). The study was undertaken before ART was readily available in South Africa. This study was approved by the Gauteng Department of Health Provincial Review Committee, the University of the Witwatersrand Committee for Research on Human Subjects and the Mowbray Maternity Hospital Research Committee in Cape Town. Women delivering without prior knowledge of their HIV status were offered post-partum voluntary counselling and rapid onsite testing within 24 h of delivery. Eligible women testing HIV positive were offered enrolment. Infants were excluded if they were preterm weighing <1200 g, required ventilation, were unable to take oral medication or had congenital abnormalities. Infants found to be HIV-infected received co-trimoxazole prophylaxis from 6 weeks of age. Enrolled women were counselled on infant feeding practices per current South African infant feeding guidelines. Subsidized formula was provided to women who chose not to breastfeed. Women who decided to breastfeed were encouraged to breastfeed exclusively for 3–6 months (National Department of Health 2001). Mother–infant pairs were followed for 6 months and for mothers who chose to breastfeed, until one month after cessation of breastfeeding. The probability of infant HIV infection was 12.8% at 6 weeks and 16.3% at 12 weeks (Gray et al. 2005). Among the 1051 mother–child dyads originally enrolled in this study, 205 patients completed the first visit but then did not attend any follow-up visits, and an additional six patients were missing length and weight data. These patients were excluded because of lack of adequate longitudinal growth data. The current analyses have been conducted on the remaining 840 mother–child dyads. The final cohort of patients (N = 840) and those patients excluded from the final cohort (N = 211) were of similar maternal age, maternal plasma viral load (PVL), maternal CD4 cell count, infant gender, infant birth weight and infant ARV prophylaxis status. Standardized procedures for enrolment and data collection were used at all study sites. Baseline socio-demographic data as well as medical and pregnancy history were collected from all enrolled mothers. Follow-up paediatric visits at 10, 45 days, 3 and 6 months after birth included a clinical examination, blood sample for HIV-1 diagnosis and measurement of child weight in kilograms and length in centimetres. All study staff were trained to measure weight and length, and an independent clinical research associate monitored measurement methods to ensure that practices were comparable across sites. Length was measured by two members of staff using a stationary headboard with a sliding vertical foot piece. Weight was measured on an electronic infant scale that was regularly calibrated. Because this was a secondary analysis of a RCT to decrease MTCT, the timing of the growth assessments was based on the primary therapeutic aim of the RCT. This report addresses an a priori secondary aim of the RCT designed to estimate predicted changes in infant growth and its determinants. At each visit, feeding of the infant and breastfeeding practices were ascertained via interviewer-administered structured standardized questionnaires. Child hospitalization diagnoses were abstracted from patient records by two authors (KKV and GEG). Maternal blood samples were tested with Determine HIV-1/2 tests (Abbot Laboratories, Abbot Park, IL, USA), and, if reactive, a second confirmatory test using the Uni-Gold HIV test (Trinity Biotech, Wicklow, Ireland) was performed. Women who tested negative with the initial Determine test were considered uninfected. Maternal blood was sent for HIV-1 RNA quantitative HIV testing (viral load) and CD4 cell counts. Infant blood samples were collected for HIV-1 DNA polymerase chain reaction (PCR) using the Roche Amplicor Monitor version 1.5 qualitative PCR assay (Roche Diagnostics, Basel, Switzerland). Confirmed infant HIV-1 infection was defined as two consecutive blood samples testing positive for HIV-1 DNA by PCR. Infants who had a single documented positive result and were then lost to follow-up were considered infected. An infant was considered to be HIV-uninfected when PCR at 45 days or later was negative in the absence of ever breastfeeding. In breastfed infants, retesting occurred 1 month after breastfeeding cessation. These infants were then considered HIV-uninfected if this sample was negative. Premature births were defined as infants born before 36 weeks of gestation. WHO International Classification of Disease (ICD-10) criteria were used to classify gastrointestinal and respiratory hospitalizations (WHO 2007). At each visit, infant feeding was categorized as formula feeding or breast feeding. In the univariate analysis, mother–infant dyads were stratified by infant HIV status (infected vs. uninfected). Standard normal deviates (z-scores) for length-for-age (LAZ), weight-for-age (WAZ) and weight-for-length (WLZ) were calculated using National Center for Health Statistics reference using EPIINFO (Centers for Disease Control and Prevention, Atlanta, GA). A −2.00 z-score was used as the cut-off for low values of these anthropometric indices. Because infant feeding status could change between study visits, infant feeding method (formula vs. breast) at each visit was assessed as a time-varying covariate, so that feeding status could change over time (Cleves et al. 2008). Confounding variables were included in the multivariable model based on a priori confounders identified from the literature and variables that were significant at a 0.20 level in the univariate analysis. Cox proportional hazards models were used to estimate the risk of reaching <−2.00 for WAZ, LAZ and WLZ, which have been used in earlier studies (Bailey et al. 1999). Time-to-event was measured from enrolment to the study visit at which the outcome was observed. To assess the effects of infant- and maternal-specific independent variables on growth retardation, separate multivariable models were run containing only infant and maternal characteristics (Models I and II, respectively), and a final multivariable model was then run that included both infant and maternal characteristics (Model III). To assess whether the influence of covariates on growth outcomes varied by infant HIV status and maternal disease (Isanaka et al. 2009), we also examined effect modification through stratified multivariable analyses by infant HIV status (infected vs. uninfected) and maternal PVL (PVL < 100 000 vs. PVL > 100 000). All data analyses were conducted using STATA (STATACORP, version 10.0; College Station, TX, USA) software. Table 1 presents demographic and clinical characteristics of the 840 mother–infant dyads overall and stratified by infant HIV status. Maternal median PVL was significantly higher among HIV-infected than uninfected infants (68 650 copies/μl vs. 13 150 copies/μl; P < 0.0001), and maternal median CD4 cell count was significantly lower (357 cells/μl vs. 459 cells/μl; P = 0.0003). Mothers of HIV-infected infants were more likely to have died during study follow-up than mothers of uninfected infants (4.3%vs. 1.6%; P = 0.0485). The proportion of male infants was higher in the HIV-infected group than in the uninfected group (59.3%vs. 46.5%; P = 0.01). Respiratory and gastrointestinal infections were more common among HIV-infected than uninfected infants (P < 0.0001). Maternal age, mode of delivery, gestational age, weight at birth, premature birth, feeding method and ARV prophylaxis status were similar in HIV-infected and HIV-uninfected infants. Table 2 presents the growth parameters of median weight, median length, WAZ, LAZ and WLZ at 10, 45 days, 3 and 6 months after birth stratified by infant HIV status. By the end of the study, 20% of infants had WAZ <−2, 19% had an LAZ <−2, and 29% had a WLZ <−2. LAZ or WAZ <−2 occurred at the same time soon after the birth (WAZ: median of 14 days; and LAZ: median of 15 days). However, WLZ <−2 occurred at a later age (median of 42 days). By 3 months, HIV-infected infants had lower mean WAZ (−0.11 vs. 0.89; P < 0.0001) and mean WLZ (−0.91 vs.−0.04; P < 0.0001) than uninfected infants. By 6 months, HIV-infected infants had a lower mean LAZ than uninfected infants (0.54 vs. 1.33; P = 0.0003). HIV-infected infants had lower weights than uninfected infants by 3 months (4800 g vs. 5440 g; P < 0.0001). Table 3 presents the Cox proportional hazards regression models for predictors of WAZ falling <−2 based on only maternal correlates (Model I), only infant correlates (Model II) and both maternal and infant correlates (Model III). In the maternal model, the risk of WAZ falling <−2 was significantly associated with maternal CD4 cell count 100 000 copies/ml (HR: 1.57; 95% CI: 1.10–2.24). In the infant model, the risk of WAZ falling <−2 was significantly associated with infant gastrointestinal infection (HR: 1.99; 95% CI: 1.08–3.70), premature birth (HR: 2.89; 95% CI: 2.13–3.92) and formula feeding (HR: 3.50; 95% CI: 1.71–7.15). In the combined maternal and infant model, the risk of WAZ falling <−2 was significantly associated with maternal CD4 cell count <200 cells/μl (HR: 1.64; 95% CI: 1.10–2.43), maternal CD4 cell count between 200 and 350 cells/μl (HR: 1.62; 95% CI: 1.09–2.41), premature birth (HR: 2.82; 95% CI: 2.06–3.86) and formula feeding (HR: 3.35; 95% CI: 1.64–6.85). Table 4 presents the Cox proportional hazards regression models for predictors of LAZ falling <−2 based on only maternal correlates (Model I), only infant correlates (Model II) and both maternal and infant correlates (Model III). In the maternal model, the risk of LAZ falling <−2 was significantly associated with maternal age <20 years (HR: 0.54; 95% CI: 0.30–0.95) and maternal CD4 cell count <200 cells/μl (HR: 1.93; 95% CI: 1.29–2.88). In the infant model, the risk of LAZ falling <−2 was significantly associated with infant HIV infection (HR: 1.62; 95% CI: 1.07–2.49), premature birth (HR: 2.47; 95% CI: 1.79–3.41) and formula feeding (HR: 4.47; 95% CI: 1.96–10.17). In the combined maternal and infant model, the risk of LAZ falling <−2 was significantly associated maternal age <20 years (HR: 0.54; 95% CI: 0.31–0.96), maternal age 20–25 years (HR: 0.65; 95% CI: 0.43–0.98), maternal age 25–30 years (HR: 0.61; 95% CI: 0.39–0.96), maternal CD4 cell count <200 cells/μl (HR: 1.72; 95% CI: 1.14–2.59), premature birth (HR: 2.37; 95% CI: 1.70–3.30) and formula feeding (HR: 4.22; 95% CI: 1.85–9.62). In the maternal model (Model I), the risk of WLZ falling <−2 was not significantly associated with any of the maternal correlates. In the infant model (Model II), the risk of WLZ falling <−2 was significantly associated with infant HIV infection (HR: 1.75; 95S% CI: 1.25–2.46), respiratory infection (HR: 0.38; 95% CI: 0.18–0.78) and formula feeding (HR: 1.86; 95% CI: 1.17–2.96). In the combined maternal and infant model (Model III), the risk of WLZ falling <−2 was significantly associated with infant HIV infection (HR: 1.64; 95% CI: 1.16–2.32), respiratory infection (HR: 0.39; 95% CI: 0.18–0.80) and formula feeding (HR: 1.78; 95% CI: 1.11–2.83). Table 5 presents two stratified Cox proportional hazards models by infant HIV status (infected vs. uninfected) and maternal PVL (PVL > 100 000 copies/ml vs. <100 000 copies/ml) for the risk of WAZ and LAZ falling <−2, adjusting for all other covariates. The risk of WAZ falling <−2 was almost three times greater for HIV-infected infants than uninfected infants with gastrointestinal infections. Similarly, the risk of LAZ falling <−2 was four times greater for HIV-infected infants than uninfected infants with gastrointestinal infections. A similar pattern was observed for WAZ falling <−2 with mothers having PVL > 100 000 copies/ml compared to mothers with PVL < 100 000 copies/ml. We found that HIV infection is an independent predictor of infant stunting (LAZ) and wasting (WLZ), but not being underweight (WAZ). HIV-infected infants experienced significantly greater growth retardation within 3 months after birth compared with HIV-uninfected infants. WAZ z-scores were significantly lower in HIV-infected infants than in uninfected infants by 0.31 at 45 days, 1.00 at 3 months and 1.33 at 6 months. Similarly, WLZ z-scores were lower by 0.87 at 3 months and 0.76 at 6 months, and LAZ z-scores were lower by 0.79 at 6 months. At birth, infants had similar weight regardless of HIV status. Formula-fed infants were at consistently greater risk for growth retardation than breastfed infants. In accordance with earlier studies from both the developed world and resource-limited settings, these findings suggest that the difference in growth between HIV-infected and uninfected infants occurs soon after birth and increases with age (Moye et al. 1996; Bailey et al. 1999; Webb et al. 2008). A cohort study of infants from Durban, South Africa, found that HIV-infected infants had early (i.e. by 3 months) and sustained low mean z-scores for LAZ and WAZ but not for WLZ compared with uninfected infants (Bobat et al. 2001). An earlier infant cohort in Rwanda also found that stunting but not wasting was more common among HIV-infected infants (Lepage et al. 1996). In our study, infant HIV infection was associated with stunting and wasting. It has been suggested that height velocity measured as LAZ is more strongly associated with HIV clinical progression, immune reconstitution and decline in viral replication among ART-experienced US children than weight velocity measured as WAZ (Benjamin et al. 2003). This study supports earlier findings that wasting is characteristic of HIV-infected children and that HIV infection is likely a causative factor in initiating and maintaining the process of wasting (Bailey et al. 1999). It is possible that HIV-infected infants may experience proportional declines in both height and weight, such that normal height-for-weight is maintained (Lepage et al. 1996; Bobat et al. 2001), or wasting may become apparent only later after the development of more advanced immunodeficiency (Bailey et al. 1999). The current study is notable for identifying wasting in HIV-infected infants within 3 months of birth. Lower WAZ and LAZ scores among HIV-infected infants were detected at the same time and before lower WLZ scores (in the case of WAZ), which is in accordance with earlier studies (Newell et al. 2003; Isanaka et al. 2009) (Agostoni et al. 1998). While in protein-energy malnutrition it would be anticipated that decreases in weight would occur before height, the pattern of concurrent decreases in height and weight may indicate that other mechanisms may underlie HIV-related growth failure, such as secondary infections associated with HIV. Further longitudinal studies are needed to examine the sequence of occurrence of different indices of anthropometric undernutrition. Maternal CD4 cell count was an independent predictor of being underweight and stunted. These findings are consistent with recent data from a Tanzanian cohort (Webb et al. 2008), but not an earlier study from the Congo (Bailey et al. 1999). Infant disease progression by 6 months of age has been shown to be associated with maternal immune suppression (Rich et al. 2000; Abrams et al. 2003), and poor maternal health during pregnancy may adversely impact the infant immune system (Chougnet et al. 2000; Nielsen et al. 2001). Women with advanced immunodeficiency may be too weak to provide adequate care and nurturing to their children after delivery. In our study, maternal PVL at birth did not significantly predict impaired infant growth (though maternal PVL did predict being underweight in the infant model). The risk of rapid progression in infants during the first 6 months of life may be more a product of a relatively ineffective immune response rather than of viral replication. Breastfeeding was consistently protective against being underweight, stunted and wasted. Among HIV-infected mothers, growth faltering has been associated with cessation of breastfeeding (Arpadi et al. 2009). Among non-HIV-infected mothers, sustained breastfeeding has been shown to be associated with a lower risk of gastrointestinal and respiratory illness in infants from multiple regional settings (Raisler et al. 1999; Arifeen et al. 2001; Quigley et al. 2007). Decreasing breast milk viral load through ART is likely to reduce the risk of infection (Kumwenda et al. 2008; Kilewo et al. 2009) and also provide the infant with the beneficial aspects of breastfeeding if the mother chooses to breastfeed. These findings suggest that with expanded access to maternal ART in resource-limited settings and owing to the increased risk of infections associated with formula feeding, breastfeeding should be encouraged (Patel et al. 2010). Although gastrointestinal infections did not independently predict growth failure (though it did predict being underweight in the infant model), when infants were stratified by HIV status and maternal PVL, HIV-infected infants and infants with mothers with PVL > 100 000 copies/ml who also had a gastrointestinal infection were at greater risk of experiencing <−2 WAZ and LAZ. A study among Tanzanian HIV-infected women and their infants found that episodes of diarrhoea or respiratory infections were related to significantly lower WLZ but not LAZ scores, but interactions between HIV status and diarrhoea or respiratory illness were not considerable (Webb et al. 2008). Diarrhoea is strongly associated with decreased growth and increased mortality (Villamor et al. 2002). In HIV-infected infants, initiation of ART leads to a decrease in the incidence of opportunistic infections (Gona et al. 2006). Further interventions aimed to prevention, early detection and management of gastrointestinal infections among HIV-infected infants is warranted. Preterm infants were at significantly increased risk of being underweight and stunted, which is in accordance with earlier data (Webb et al. 2008). HIV-infected women may be at higher risk for preterm labour and intrauterine growth retardation (IUGR) (Halsey et al. 1990; Taha et al. 1995). Also, IUGR may depend on the degree of maternal immunodeficiency (Braddick et al. 1990). The most severe preterm infants weighing <1200 g did not meet study inclusion criteria, limiting the generalizability of these findings. It is also possible that increased losses to follow-up attributable to mortality could have reduced the contribution of later anthropometric measurements among preterm infants. To our knowledge, this is one of the largest cohorts to report growth outcomes among infants born to HIV-infected mothers. It is difficult to compare our results with other studies examining growth outcomes among HIV-exposed infants because of differences in populations studied, varying prenatal growth patterns, the availability of ART, food supplementation or socioeconomic conditions, viral subtypes, prevalence of maternal sexually transmitted infections, varying sample size and statistical power, and other intercurrent nutritional deficiencies. A strength of the current study was that we controlled for important maternal characteristics, which many earlier studies have not (Isanaka et al. 2009). A limitation of the current study is the lack of a control group of infants born to HIV-uninfected women (Makasa et al. 2007). We do not have reasons for losses to follow-up, though staff attempted to locate women who did not attend study visits. In the survival model employed for analysis, participants were censored at their last attended study visit (Leung et al. 1997). Bias owing to incomplete follow-up data could lead to misclassification as infants lost to follow-up may be at greater risk for growth retardation. However, the current study documented significant growth retardation within 3 months of birth. We were only able to follow infant growth for up to six months, and hence these findings may not apply to long-term infant growth; however, the measurement of infant growth at close intervals allowed for ascertaining changes in anthropometry between HIV-infected and HIV-uninfected infants during a time of rapid growth. Because of the exclusion criteria of the original clinical trial, these findings may be generalizable only to full-term infants. Because of the close association between infant growth with immune function and HIV disease progression, understanding growth patterns of HIV-infected infants is important for developing appropriate treatment and clinical programmes. HIV-infected children should be promptly initiated on ART, as treatment has a positive sustained influence on growth as well as decreased PVL and increased CD4 cell count (Verweel et al. 2002; Guillén et al. 2007). Children with greater PVL are shorter and lighter than children with lower PVL (Arpadi et al. 2000; Nachman et al. 2005). Promptly identifying HIV-infected infants and assessing their growth is important to determine disease stage and prognosis as well as the effectiveness and toxicities of ART. Early nutritional management among HIV-exposed infants is warranted and should be a priority of comprehensive MTCT clinical programmes.