Growth status, inflammation, and enteropathy in young children in Northern Tanzania

Recent evidence suggests that enteropathy of the gut due to environmental conditions (i.e., environmental enteropathy [EE]) in young children is negatively associated with linear growth. Using a case–control study design, we examined the potential determinants of stunting in stunted and non-stunted children 22–28 months of age. Potential determinants included inflammation biomarkers C-reactive protein, alpha-1-acid glycoprotein (AGP), and endotoxin-core antibody (EndoCAb) measured in serum samples; enteropathy markers alpha-1-antitrypsin, neopterin, myeloperoxidase (MPO) measured in stools samples; and demographic, health, feeding, and household characteristics. We also explored the determinants of EE by testing associations of composite EE scores and individual biomarkers with potential risk factors. Fifty-two percent of children (n = 310) were found to be stunted, and mean height-for-age Z scores (HAZ) were -1.22 (standard deviation [SD] ± 0.56) among non-stunted (control) children and -2.82 (SD ± 0.61) among stunted (case) children. Child HAZ was significantly (P < 0.05) and inversely associated with AGP, and child stunting was significantly positively associated (P < 0.05) with low dietary diversity, severe household hunger, and absence of soap in the household. Alpha-1-acid glycoprotein and EndoCAb concentrations were also significantly higher (P < 0.05) among children in households with no soap. Our study documented a seemingly localized cultural practice of young children (25%) being fed their dirty bathwater, which was associated with significantly higher concentrations of MPO (P < 0.05). Alpha-1-acid glycoprotein showed the most consistent associations with child growth and hygiene practices, but fecal EE biomarkers were not associated with child growth. The lack of retrospective data in our study may explain the null findings related to fecal EE biomarkers and child growth. We designed a case–control study to compare inflammation and EE biomarkers against sociodemographic, health and hygiene characteristics between stunted children (cases) and non-stunted children (controls) living in the same communities. The case–control study was nested into the endpoint assessment of the Creating Homestead Agriculture for Nutrition and Gender Equity (CHANGE) impact evaluation—a trial examining the impact of an integrated nutrition intervention (including micronutrient powders [MNPs], homestead food production, and WASH and nutrition education) on child anemia and growth.20 The CHANGE project aimed to improve child growth by supporting caregivers to adopt optimal child-feeding practices and improved WASH practices through a behavior change approach and to adopt improved homestead food production (e.g., vegetable gardens and chicken rearing) through training and service provision. As part of the research design, MNPs were given to all children participating in the program as well as to a control group, with the original goal of reducing anemia levels to understand the ability of the integrated CHANGE package to maintain these reduced anemia levels. For the case–control study, participants were enrolled from January to February 2016 in the control areas of the CHANGE evaluation in which only a limited set of activities were implemented. Specifically, children in the control arm received anemia and malaria diagnosis and treatment during baseline and follow-up surveys of the CHANGE evaluation, which happened 18, 15, and 6 months before recruitment into the case–control study. In addition, they received a 2-month supply of MNPs at the CHANGE baseline and 12-month follow-up surveys (i.e., when children were 6–12 months of age and then 18–24 months of age). No other activities (e.g., support for agriculture and livestock production or behavior change communication on WASH, nutrition, or malaria prevention) were organized as part of the CHANGE program in the control communities. The case–control study recruited all children living in the 10 control wards (i.e., smallest administrative units) of the Sengerema district in Tanzania’s Lake Zone who provided blood samples during the CHANGE end line assessment. These children were also included in previous rounds of the CHANGE study, from whom questionnaire data and blood samples were collected during previous survey rounds. The enrolled children had an age range of 22–28 months during the end line assessment. This design was expected to identify 128 cases and 192 controls (320 children in total), considering an expected stunting rate of 40% and an 80% response rate for stool samples, with a statistical significance level of 0.05 and power of 0.8. Based on the aforementioned assumptions, the target sample size would presumably allow the study to detect an odds ratio of 2 between stunted and non-stunted children with differing endotoxin exposures and other stool biomarkers that were considered clinically relevant a priori. A comprehensive questionnaire was administered as part of the larger study to caregivers of enrolled children. The child’s age was collected as part of the household questionnaire and further validated by confirming the birth date listed on the child’s health card and records from previous rounds of the CHANGE evaluation. The questionnaire also included information on household and individual demographic variables (e.g., age and gender), household food security, household WASH facilities and practices, maternal health and education characteristics, and infant and young child feeding practices. Following the completion of the questionnaire, height and weight were measured from all children using portable wooden stadiometers (Infant/Child ShorrBoard®, Olney, MD) and standing scales (Seca, 874 U, Hamburg, Germany). In preparation for the fieldwork, a height measurement standardization exercise was conducted, and only the best-performing nurses were hired. Capillary blood samples were collected from each child’s middle or ring finger by trained nurses. After the lancet puncture, the first two drops of blood were swiped with dry gauze, and the third and fourth blood droplets were used to measure hemoglobin concentration and malaria status. Following this, approximately 300–400 µL of blood was collected into capillary blood collection tubes with coagulation activator (Microvette® 300 Z, 20.1308; Sarstedt, Nümbrecht, Germany). The blood samples were kept cold (+1 to +4°C) and were centrifuged on the day of blood collection. After centrifugation and separation, the serum was frozen at −20°C. Following blood collection, nurses provided pre-labeled stool sample containers, a clean plastic spoon, and paper to place below a child when defecating, should the child not be wearing a diaper. Each child’s caretaker was instructed to scoop approximately 100 mL of stool directly from the child’s diaper or from clean paper into the pre-labeled stool container directly after a child defecated, securely close the lid, and place the sample in a cool dark spot until it was collected. Three research assistants on motorbikes accompanied the nurses to the field and collected the stool samples the same day, such that the cold chain began 1–4 hours after the stool was passed. After collection, the stool samples were kept in cold boxes (+4 to +8°C) until the samples were delivered to the the National Institute for Medical Research (NIMR) field laboratory in Sengerema, where a slide for microscopy was prepared within 12–24 hours after the stool was passed and a 10–15 mL aliquot of stool was frozen at −20°C. Hemoglobin concentration was measured onsite using a portable hemoglobinometer (Hb201+; HemoCue AB, Ängelholm, Sweden). Current and recent malaria status was assessed using a rapid diagnostic test (Malaria Ag P.f/Pan; Standard Diagnostics, Gyeonggi-do, Republic of Korea) with the ability to detect antigens from Plasmodium faliciparum, Plasmodium vivax, Plasmodium ovale, and Plasmodium malariae. Three markers of systemic inflammation were measured in serum. C-reactive protein (CRP) and alpha-1-acid glycoprotein (AGP) were measured in duplicate at VitMin Lab laboratory (Willstaett, Germany) using the sandwich enzyme-linked immunosorbent assay (ELISA) technique described by Erhardt et al.21 Single IgG endotoxin-core antibody (EndoCAb) concentrations were assessed using commercially available kits (Hycult Biotech, Inc., Uden, The Netherlands) using a 1:400 dilution factor. Approximately, 15% of samples were measured in duplicate, and the results showed high correlation (R2 = 0.91). Microscopy to detect Ascaris lumbricoides (large roundworm) and Strongyloides stercoralis (threadworm) was performed separately by two microscopists with extensive experience detecting soil transmitted helminths burden in stool samples. Analysis of stool biomarkers was conducted at NIMR Mwanza Laboratory (Mwanza, Tanzania). Commercial ELISA kits were used to measure concentrations of three biomarkers in stool samples: neopterin (NEO; GenWay Biotech, San Diego, CA), alpha-1-antitrypsin (AAT; BioVendor, Karasek, Czech Republic), and myeloperoxidase (MPO; Immundiagnostik, Bensheim, Germany). During the extraction process, NEO was diluted with 0.9% saline solution at a dilution factor of 50, and AAT and MPO were diluted with the manufacturer’s dilution buffer at a dilution factor of 63.5. All aliquots were vigorously vortexed for 1 minute, centrifuged according to the manufacture’s specifications, and 1 mL of the resulting supernatant was then removed and frozen at −20°C. Following an initial screening for each biomarker, a second dilution step was conducted to yield concentrations within the standard curves. The final dilutions for NEO, AAT, and MPO were 1:500, 1:12,500, and 1:200, respectively. Of note, the NEO ELISA manufactured by Genway Biotech is only marketed for analysis of serum, plasma, and urine but has been repeatedly used by other researchers to measure NEO in stool.14,22 Positive controls were included as part of each kit for AAT, MPO, and NEO, and controls for all plates were consistently within the ranges provided by the manufacturer. For NEO, the interassay coefficients of variation (CVs) were 21.2% and 14.3% for the low (target concentration: 5.7 nmol/L) and high (target concentration: 18.8 nmol/L) controls, respectively. The interassay CVs for AAT were 4.1% for the low (target concentration: 8.3 ng/mL) and 6.4% high (target concentration: 26.2 ng/mL) controls. For MPO, the interassay CVs were 14.9% for the low (target concentration: 6.9 ng/mL) and 5.9% high (target concentration: 26.4 ng/mL) controls. C-reactive protein and AGP were used in tandem to capture the various stages of inflammatory response as defined by Thurnham et al.23 Incubation was defined as elevated CRP (> 5 mg/L) alone, early convalescence as elevated CRP and AGP (> 1.0 g/L), and late convalescence defined as elevated AGP alone. Incubation, early convalescence, or late convalescence was defined as any inflammation, and no inflammation was defined as CRP ≤ 5 mg/L and AGP ≤ 1.0 g/L. The weight and height measurements were used to calculate gender-specific height-for-age Z scores (HAZ) and weight-for-height Z scores (WHZ) according to the WHO Child Growth Standards.24 Children with HAZ and WHZ scores < −2.0 were classified as stunted and wasted, respectively.25 Hemoglobin concentration < 110 g/L was used to define anemia26 following a subtraction of 2 g/L hemoglobin to adjust concentrations for altitudes 1,000–2,000 m above sea level.26 Using questionnaire data collected within the CHANGE study, minimum dietary diversity for each child was calculated using WHO guidelines,27 although most of the children in our study are outside the age range (i.e., > 24 months of age) recommended by WHO. Despite the age of most children in our study, dietary diversity is still likely a relevant indicator of dietary adequacy as it is often used in other population groups, such as women of reproductive age.28 In addition, caretakers reported recent illnesses of the children in the past 2 weeks according to their own understanding of descriptions used by interviewers. Illnesses examined were watery diarrhea (i.e., “stool like water”) and fever (“hot body”). Caretakers were also asked about when they washed their child’s hands and were able to respond without prompting about all instances where the child’s hands are washed. The responses “after defecation” and “before eating” were the only two responses mentioned of the five critical moments for handwashing.29 With the inclusion of the response that the child’s hands were washed “frequently throughout the day,” a handwashing score was calculated, to identify if the child had zero, one, two, or three times when hands were washed routinely, and then dichotomized to determine if at least one of these key practices was used. Caretakers also reported if the child was fed his/her own bath water 1) anytime when the child was < 1 year of age and 2) anytime in the past month. The practice of feeding children their own bath water had been observed in the region during previous rounds of the CHANGE evaluation. The interviewers observed the presence and type of sanitation facility at the household, using WHO/UNICEF criteria30 and determined if the sanitation facilities were considered “improved” (i.e., flush to piped sewer system/septic tank/pit, ventilated improved pit latrine, pit latrine with a slab, composting toilet) or “un-improved” (i.e., pit latrine without slab, bucket toilet, hanging toilet, no facilities). Household food security in the past month was measured using the standard household food insecurity access scale (HFIAS) classification into food secure, or mildly, moderately, or severely food insecure.31 In addition to the HFIAS, the standard household hunger scale (HHS), which captures more severe hunger-related behaviors,32 was calculated and separated into three categories (little to no hunger, moderate hunger, and severe hunger).33 We constructed a composite “EE score” modifying the approach developed by Kosek et al.14 Using this approach, each child was assigned to one of three categories based on their percentile for NEO, AAT, and MPO concentrations, “where AAT, MPO, and NEO categories were defined as 0 (≤ 25th percentile), 1 (25–75th percentile), or 2 (≥ 75th percentile).”14 In addition, we augmented the “EE score” developed by Kosek et al.14 by incorporating the systemic inflammation markers EndoCAb and AGP into the equation because these markers have been associated with growth or stunting in previous studies.34,35 In this “expanded EE score,” EndoCAb and AGP are categorized using the same approach as developed by Kosek et al., and weighted similar to NEO in the index model (see equation below). To identify factors associated with stunting, inflammation, EE, demographic, health, feeding, and household characteristics were compared independently in stunted and non-stunted children. For these associations, measures of central tendency and variance were calculated for stunted and non-stunted children separately and compared. For normally-distributed variables, means and standard deviations were calculated, and the P-values expressing the difference between non-stunted and stunted children were calculated using the analysis of variance (ANOVA) test. P values less than 0.05 were considered statistically significant. We did not adjust further for multiple comparisons because the goal of the paper was to draw hypotheses on the pathway from inflammation to stunting, rather than prove predefined key hypothesis of these pathways for final decision making.36 For non-normally distributed variables, the medians of each subgroup were compared using the Wilcoxon rank sum test. For dichotomous and categorical variables, odds ratios and their respective confidence intervals (95%) were calculated.37 In addition, adjusted odds ratios were calculated using a logistic regression model to account for child's gender, child’s age in months, child’s anemia status, mother’s anemia status, household ownership of soap, and household hunger. To identify continuous associations between growth, inflammation, and EE, pairwise correlations between HAZ and all markers of inflammation and EE were also conducted. Lastly, to identify risk factors of EE, the mean EE and expanded EE scores were compared for various health, feeding, and hygiene factors. Data analysis was conducted using Stata 14 (StataCorp, College Station, TX). Ethical approval for the case–control study was received from the Lake Zone Ethics Committee (REF: MR/53/100/3) housed at the NIMR Mwanza Center and the International Food Policy Research Institute’s (IFPRI) Institutional Review Board (REF: 2016-7-PHND-M). Ethical clearance for the CHANGE study was obtained from Medical Research Coordinating Committee housed at NIMR headquarters in Dar es Salaam (REF: NIMR/HQ/R.8a/Vol.IX/1721) and IFPRI’s Institutional Review Board (REF: 2014-PHND-12-M). A separate written informed consent statement was used to request consent to collect stool samples for the measurement of EE markers and to measure EndoCAb from serum samples collected for micronutrient biomarkers. This consent statement was read aloud in the local language to the child’s caretaker, who gave consent on the child’s behalf and agreed to collect the stool sample according to the procedures described by the nurse. To meet the requirements of NIMR’s National Ethics Committee, a single dose of albendazole was given to all children who provided stool samples according to procedures described in Tanzania’s public health policy.38 The consent forms were stored in a locked cabinet categorized per NIMR’s standardized filing and storage procedure. Stool sample data contained no identifying information.