Infant Nutritional Status and Markers of Environmental Enteric Dysfunction are Associated with Midchildhood Anthropometry and Blood Pressure in Tanzania

Objective To assess whether growth and biomarkers of environmental enteric dysfunction in infancy are related to health outcomes in midchildhood in Tanzania. Study design Children who participated in 2 randomized trials of micronutrient supplements in infancy were followed up in midchildhood (4.6-9.8 years of age). Anthropometry was measured at age 6 and 52 weeks in both trials, and blood samples were available from children at 6 weeks and 6 months from 1 trial. Linear regression was used for height-for-age z-score, body mass index-for-age z-score, and weight for age z-score, and blood pressure analyses; log-binomial models were used to estimate risk of overweight, obesity, and stunting in midchildhood. Results One hundred thirteen children were followed-up. Length-for-age z-score at 6 weeks and delta length-for-age z-score from 6 to 52 weeks were associated independently and positively with height-for-age z-score and inversely associated with stunting in midchildhood. Delta weight-for-length and weight-for-age z-score were also positively associated with midchildhood height-for-age z-score. The 6-week and delta weight-for-length z-scores were associated independently and positively with midchildhood body mass index-for-age z-score and overweight, as was the 6-week and delta weight-for-age z-score. Delta length-for-age z-score was also associated with an increased risk of overweight in midchildhood. Body mass index-for-age z-score in midchildhood was associated positively with systolic blood pressure. Serum anti-flagellin IgA concentration at 6 weeks was also associated with increased blood pressure in midchildhood. Conclusions Anthropometry at 6 weeks and growth in infancy independently predict size in midchildhood, while anti-flagellin IgA, a biomarker of environmental enteric dysfunction, in early infancy is associated with increased blood pressure in midchildhood. Interventions in early life should focus on optimizing linear growth while minimizing excess weight gain and environmental enteric dysfunction. Trial registration ClinicalTrials.gov: NCT00197730 and NCT00421668. The study sample included children born in Dar es Salaam, Tanzania, who participated in 1 of 2 randomized controlled trials of multiple micronutrient supplementation to infants. The first trial (ClinicalTrials.gov: {“type”:”clinical-trial”,”attrs”:{“text”:”NCT00197730″,”term_id”:”NCT00197730″}}NCT00197730) randomized 2387 infants born to HIV-infected mothers to either daily administration of multiple micronutrients (vitamins B complex, C, and E) or placebo at 6 weeks of age.18 Randomization of infants occurred between August 2004 and November 2007; follow-up ended in May 2008. The micronutrient supplements did not show an effect on mortality, morbidity, or child growth.18, 19 The second trial (ClinicalTrials.gov: {“type”:”clinical-trial”,”attrs”:{“text”:”NCT00421668″,”term_id”:”NCT00421668″}}NCT00421668) was implemented with a 2 × 2 factorial design assessing the effect of zinc, zinc plus multivitamins (the same combination of vitamins B complex, C, and E as described), multivitamins alone, or placebo among 2400 infants born to HIV-negative women.20 The second trial found that zinc supplementation reduced the risk of acute respiratory and diarrheal infections,20 but that neither supplement alone nor in combination had an effect on rates of stunting, wasting, or underweight.21 The 2 studies were designed to allow for pooled analyses—they were conducted in overlapping clinics with similar staff, they used identical inclusion/exclusion criteria (other than maternal HIV status), and they collected the same sociodemographic and clinical data on all mothers and children. In both trials, infants were randomized at 6 weeks of age, and mothers were asked to bring the children to the clinic for follow-up visits every 4 weeks after randomization. At each monthly follow-up visit, a trained study nurse measured child anthropometry using standard techniques.22 Weight was measured on a digital infant balance scale with 10-g precision (Tanita, Arlington Heights, Illinois) and length with 1-mm precision using a rigid length board with an adjustable foot piece. For the current study, we identified children who participated in the 2 original trials who met the following criteria: children with complete physical descriptions of their home addresses on file, who had anthropometric data at 6 weeks, who had participated in their original trial through 15 months of age, and who were available for contact during the follow-up study recruitment period of June to August 2014. From the 2387 children in the first trial, a list of all children who fulfilled these criteria was generated and simple random sampling was used to select children for follow-up. From the 2400 children in the second trial, we selected from 269 children who had participated in an enteric disease substudy because these children had provided blood specimens at both 6 weeks and 6 months of age.17 Further inclusion criteria in the substudy was that children had length-for-age z-score (LAZ) > −2 at 6 weeks. Microtiter plates were coated with purified Escherichia coli flagellin (100 ng/well) or purified Escherichia coli LPS (2 mg/well). Serum samples from study participants were diluted 1:200 and applied to wells coated with flagellin or LPS. After incubation and washing, the wells were incubated with anti-human IgA (KPL, Milford, Massachusetts) or IgG (GE Healthcare, Little Chalfont, United Kingdom) coupled to a horseradish peroxidase. The quantification of total immunoglobulins was performed with the use of the colorimetric peroxidase substrate tetramethylbenzidine, and absorbance (optical density) was read at 450 nm with the use of an enzyme-linked immunosorbent assay plate reader. Data are reported as optical density-corrected data by subtracting background concentrations, which were determined from the readings in samples that lacked serum. For all children who participated in the follow-up study, trained study nurses measured their weight, height, and blood pressure when the children were between 4 and 9 years of age. Diastolic, systolic, and mean arterial blood pressure in mm Hg were measured with a DINAMAP DPC120X-EN (GE Medical Systems Information Technologies Inc, Milwaukee, Wisconsin). Blood pressure measurements were assessed 5 times by a single observer and then averaged. Body weight was measured with an electronic digital scale accurate to 0.1 kg (Tanita), and standing height was measured with a stadiometer to the nearest 0.1 cm. Ethical approval for both parent trials was granted by the Harvard T.H. Chan School of Public Health Human Subjects Committee, the Muhimbili University of Health and Allied Sciences Committee of Research and Publications, the Tanzanian Institute for Medical Research, and the Tanzanian Food and Drug Authority; the follow-up study was approved by the Harvard T.H. Chan School of Public Health Human Subjects Committee and the Muhimbili University of Health and Allied Sciences Committee of Research and Publications. All mothers provided written informed consent to enroll themselves and their children in the original and follow-up studies. Details on data collection and management from the 2 trials have been published previously.18, 19, 20, 21 In brief, data were double entered by using Microsoft Access software and converted to SAS datasets (version 9.4; SAS Institute, Cary, North Carolina) for analysis. For growth analyses, we calculated age- and sex-specific z-scores for 3 anthropometric indices during infancy: weight-for-length z-score (WLZ), LAZ, and weight-for-age z-score (WAZ) using the 2006 World Health Organization (WHO) growth standards.23 In accordance with WHO recommendations, we set all extreme LAZ (<−6 or >6), WLZ (<−5 or >5), and WAZ (<−6 or >5) values to missing.24 For follow-up anthropometric data, we calculated age- and sex-standardized height-for-age z-score (HAZ), WAZ, and BMIZ using the 2006 WHO growth standards for children younger than 5 years of age and the WHO 2007 Growth Reference for children aged 5 years and older.25 In accordance with WHO guidelines, stunting was defined as HAZ <−2, and overweight as BMIZ >1, and obese as BMIZ >2. Descriptive statistics, including means with standard deviations and frequencies with percentages, were used to summarize sociodemographic information on maternal, child, and household characteristics as well as the growth characteristics of children who participated in the follow-up study. Our outcomes of interest were HAZ, BMIZ, and WAZ; systolic and diastolic blood pressures; and overweight, obesity, and stunting in midchildhood. Our exposures of interest were baseline anthropometric indicators (6-week LAZ, WAZ, and WLZ) and change in each anthropometric indicator from 6 to 52 weeks of age, as well as EED biomarkers (anti-flagellin and anti-LPS IgG and IgA) at 6 weeks and 6 months of age. For each continuous outcome of interest, we first conducted univariate linear regression models with each anthropometric and EED biomarker as predictors of interest. Confounders for multivariate models were selected based on a review of the literature and included which trial the child originally participated in, the trial treatment arm, and the child’s sex for all models; maternal height and maternal education for all models for anthropometric outcomes; and child’s gestational age at birth as well as age at midchildhood follow-up for all blood pressure models. Multivariate models assessing the effect of anthropometry in infancy included each anthropometric indicator at 6 weeks of age as well as the change in the same indicator from 6 to 52 weeks of age to assess relative importance of the 2 time periods for each indicator. The assumption of linearity was assessed using plots of residuals versus predicted values, and normality based on normal quantile plots. For our primary outcomes (anthropometry and blood pressure in midchildhood), data from 113 subjects and 8 covariates in the multivariate linear regression models, provided 80% power to detect a minimal increase of 7% in the R2 value for a predictor of interest at significance level of 0.05. In models where biomarkers of EED were predictors of interest, the number of subjects was reduced to 66 in our analysis, which provided 80% power to detect a minimal increase of 12% in R2 value at a significance level of 0.05. We also conducted univariate and multivariate log-binomial regression models to assess the relationship between each of our predictors of interest and overweight, obesity, and stunting in midchildhood. Given the limited power to assess these outcomes, multivariate models for binary anthropometric outcomes only adjust for which trial the child participated in, the child’s sex, and the corresponding anthropometric indicator at baseline or change from 6 to 52 weeks of age. Multivariate models for stunting in midchildhood also adjust for maternal height. All analyses were conducted in SAS version 9.4 (SAS Institute).