Our data on infant growth among HIV-exposed infants from the SWEN trial in India highlight several important findings. First, we found a high prevalence of baseline malnutrition among both HIV-infected and HIV-exposed, uninfected infants. We observed that differences in stunting and underweight between HIV-infected and uninfected infants emerged early and persisted throughout the first year of life. Secondly, we found that SWEN exposure was associated with lower risk of wasting in HIV-exposed, uninfected infants. Lastly, we confirmed the impact of maternal and infant factors on growth -- low maternal education, maternal advanced disease state as measured by CD4 count and viral load, low infant birth weight, infant HIV infection and morbidity were the key independent factors associated with poor growth outcome. As expected, breastfed infants were found to have better growth outcomes compared to those not breastfed.
The latest national nutritional status data for Indian infants 0–6 months in the general population showed 23% stunting, 32% underweight, and 31% wasting . The nutritional status of infants in the same age group in some of the African countries is comparatively better than those in India, with 9% 19% stunted, 4%-14% underweight, and 3%-16% wasted (Additional file 1: Table S7). Hence, the baseline levels of malnutrition among Indian infants in the community are far worse to begin with than in many other resource-constrained regions.
In our study, there was an overall high prevalence of malnutrition at birth with HIV-exposed infants having approximately 47% stunting, 10% wasting and 26% underweight. In a retrospective study of 162 HIV-exposed infants at a Regional Pediatric Center for HIV in Delhi, prevalence of wasting and stunting was 50.5% and 48.8%, respectively . While our prevalence of stunting was comparable to that in this study, the wasting was lower in our study. In a cohort of antiretroviral-naïve HIV-infected infants in south India, the prevalence of stunting, underweight and wasting was 58%, 63% and 16%, respectively . In contrast to Indian studies, one in Dar es Salaam, Tanzania found less baseline stunting (29%) but similar wasting prevalence (8%) . By 6 months of life, stunting in both HIV-exposed but uninfected (51%) and HIV-infected infants (63%) in our study were significantly worse off compared to infants 0–6 months of age in the general population in India (23%) . In a review of 6 infant growth studies examining HIV exposure and postnatal growth outside of India, only one study in Kenya showed significantly more stunting among HIV-exposed uninfected compared to infants unexposed to HIV. The other studies found a lack of association suggesting that viral exposure without infection is not detrimental to postnatal growth . In our study, HIV-exposed but uninfected infants also had poor growth outcomes, although HIV-infected infants had far worse outcomes. This could partly be explained by the overall high levels of malnutrition at birth in our sample.
Among infected cohort in our study, baseline LAZ, WAZ, and WLZ scores were −1.84, –1.63 and −0.15 respectively as compared to the better LAZ, WAZ and WLZ (−0.62, –0.83, and −1.11 respectively), reported among HIV-infected infants in Democratic Republic of Congo  or Durban, South Africa  (0.13, 0.10, and −0.43) respectively. By 12 months, infected cohort in our study were far worse off especially with respect to LAZ and WAZ scores (−2.98 and −2.08 respectively), as compared to infected infants in the Democratic Republic of Congo study (−1.67 and −1.86, respectively), or Durban, South Africa study (−1.26 and −0.53, respectively).
We confirmed the compounding impact HIV infection has on growth as measured by LAZ and WAZ scores. With onset of HIV infection, the divergence in growth profile in our infants occurred within 4–6 weeks of life and persisted through one year of life. Other studies have noted differences occurring around 3 months of age [13, 16, 17], but most have noted the difference in growth between HIV-infected and uninfected occurring by one year of life [8, 9, 27]. LAZ score among infected infants in our study was lower by 0.35 at 6 months and 0.81 at 12 months. In other studies where negative association was found, HIV-infected infants had lower LAZ scores by 0.23 to 1.55 at 6 months of age, and 0.25 to 0.72 at 12 months . WAZ scores among HIV-infected infants were lower by 0.59 at 6 months and 0.64 at 12 months. The corresponding numbers reported by other studies ranged from 0.20 to 1.72 at 6 months and 0.17 to 0.87 at 12 months. Clearly, infants in our study were worse off both in terms of LAZ and WAZ. Some studies have reported that differences in weight between the two groups were detected around the same time that differences in height  were detected, which is consistent with what we found in our study. As in other studies, we found that the differences in WLZ scores between the two groups were inconsistent and were not statistically significant.
With regard to the timing of infection, not surprisingly those infected at birth have worse WAZ and WLZ during the follow-up period compared to those infected within 6 weeks and those infected after 6 weeks. Critical factors in preventing in utero infection and associated poor growth are getting pregnant women into prenatal care, rapidly testing them for HIV and starting those that are HIV-infected with antiretrovirals as soon as possible. Furthermore protecting infants while being breastfed with regimens such as SWEN are important for preventing HIV and poor growth outcomes overall. We found that infants who remained HIV-uninfected and received SWEN were less likely to be wasted, but SWEN did not significantly reduce the risk of stunting or underweight. Antiretroviral therapy has been shown to improve LAZ and WAZ scores among HIV-infected children [28–38], but this is the first time that early extended exposure to nevirapine in infants has been shown to be associated with improved WLZ scores among HIV-exposed but uninfected infants. The extended dose of nevirapine was well tolerated by the infants in the SWEN arm of our study.
Interestingly, one study in India reported that pre-existing malnutrition impacted negatively on the nutritional response to ART . Furthermore, a recent multicountry trial found that HIV-infected young children treated with a nevirapine-based ART had higher increases in Z scores for height and weight compared to those treated with a ritonavir-boosted lopinavir regimen . We likely did not observe this finding in our HIV-infected cohort due to our small sample size of HIV-infected infants receiving SWEN. The mechanism for this presumed nevirapine effect however is unclear. Nevirapine has been studied in terms of antimicrobial properties but does not appear to have any specific activity against common bacteria impacting childhood illness , but it is not known if it has any effect on pediatric respiratory pathogens. To assess if other characteristics related to growth could explain our finding, we compared the duration of breastfeeding as well as the hospitalization rates between SWEN-exposed vs. SDNVP infants and found no significant differences. Interestingly, the FDA package insert for Nevirapine shows that the drug vehicle includes sorbitol and sucrose, but it is likely that this amount administered is too small to explain it’s effect on growth. While it could be a coincidental result, it is plausible that SWEN protected these infants from acquiring HIV infection, consequently protecting them against compromised nutritional status. Nevirapine may also have an effect on absorption of nutrients and on GI infections. Further understanding of the effect of nevirapine on growth is warranted. Although availability of cART in India has improved, the average age at which HIV-infected infants present for care in India is typically 6 years and many infants with perinatally acquired HIV infection remain undiagnosed and uninitiated on treatment .
We confirmed that importance of several of the maternal and infant factors that have been shown to be associated with poor growth in both HIV-unexposed and HIV-exposed infants in other settings. Maternal hemoglobin level remains a critical predictor of gestational weight gain . It is routine practice to measure maternal hemoglobin levels during antenatal visits, and our data reinforce the need for women with lower hemoglobin levels to be identified during pregnancy should be more closely monitored and managed to prevent adverse outcomes for both maternal and infant health irrespective of the HIV status.
We found that low maternal education significantly increases the risk of stunting as have other studies [42–44]. Maternal education is not only a good measure of their knowledge of health-related issues, prenatal and postnatal infant care, infant feeding practices, and better sanitary habits, but it also impacts health-seeking behavior, income generating capacity, and ability to make autonomous decisions. Ascertaining the level of maternal education at first antenatal clinic contact in conjunction with measuring levels of malnutrition and anemia would be helpful to clinicians in providing adequate information and care to HIV-infected pregnant women and subsequently ensuring positive outcomes for infants. Similar to other studies, we also found that advanced disease status of the mother as measured by her CD4 cell count and HIV viral load at delivery also negatively impacts growth outcomes [20, 45, 46]. It is likely that mothers with low CD4 and/or higher viral load may be unable to provide adequate care for their infants because of the severity of their own illness [47–49]. Several of these factors are associated with maternal health and nutritional status both prior to and during delivery and therefore, emphasis on improving maternal health is critical to addressing adverse growth outcomes in infants, particularly if mother is HIV-infected. Infant factors such as low birth weight, preterm delivery and delivery mode other than normal vaginal delivery were all factors that significantly increased the risk of stunting, underweight, and wasting. Additionally, our results highlight the importance of breastfeeding among HIV-exposed infants.
As with other studies, our study has several strengths and limitations. A major strength of our study is the assessment of timing of infection which is likely to be more accurate due to multiple and frequent testing. While our study had excellent infant HIV diagnosis ascertainment, few children received ART as it was not available in the public sector at the time of our study. We however had prospective follow-up with multiple growth parameter assessments along with excellent retention, therefore our growth data from Indian infants are among the most robust. The limitation of our study was the lack of maternal BMI data, which is an important characteristic impacting infant growth. We also had limited information about weaning practices, especially between 6 and 9 month visits, when solid foods are introduced. Furthermore, our data come from a single site clinical trial where infants received enhanced care so some of our findings may not be generalizable to other settings; however since many children with HIV exposure are managed in urban public sector settings in India, we feel our data contribute relevant observations.