Enhanced Nutrition Package (ENP) Intervention

The ENP intervention consisted of (1) nutritional counselling, (2) monthly supply of high quality adequately iodised salt, (3) iron-folate (IFA) tablets and (4) women with an upper-middle arm circumference (MUAC) of <23 cm received a fortified balanced energy-protein (BEP) supplement monthly. The BEP supplement was a locally produced, micronutrient fortified, corn soya blend (World Food Programme Super Cereal). The nutrition routine care arm consisted of IFA provision.

Enhanced Infection Management Package (EIMP) Intervention

The EIMP intervention consisted of screening and treating for urinary tract infections (UTIs), sexually transmitted infections (STIs) and reproductive tract infections (RTIs) and enhanced deworming. UTIs were treated based on sensitivity patterns. Symptomatic RTIs (eg, trichomonas, bacterial vaginosis) were diagnosed with point-of-care tests and treated accordingly with their partners. For intestinal parasites, presumptive deworming was conducted at the second trimester visit with mebendazole (500 mg). At the next antenatal care visit, approximately 1 month later, stool was screened with wet-mount microscopy and targeted treatment was provided for identified parasitic infections. The infection routine care arm involved syndromic management of maternal infections according to Ethiopian Ministry of Health recommendations.

Table 2

ENAT study visits and biospecimen collection

Maternal urine

Participants were provided with detailed instructions on how to collect the proper sample. Mid-stream clean catch urine samples were obtained from the participants at the study health centre using sterile urine collection cups. The collected samples were then transferred into 10 mL boric acid vacutainers for transport until urine culture was processed at APHI for the main study. The samples were aliquoted into 5 mL sterile cryotubes for storage at APHI for future analysis. Additionally, UTI testing was done with urine culture, and antibiotic susceptibility testing was conducted using Vitek cards.

Maternal vaginal swab

Participants were instructed by the health centre laboratory technicians and study nurses on how to self-collect mid-vaginal swab samples. Formative work was conducted prior to the study regarding the self-collection of swabs, and it was both highly acceptable (96%) and high-quality. Based on this formative work, sample collection instructions were revised to improve quality. Following handwashing with soap, women used nylon swabs to swab their vagina and placed them on a specimen collection tray. Two swabs were collected. STIs were detected using self-collected vaginal swabs tested for gonorrhoea and chlamydia with the GeneXpert platform. The laboratory technicians then transferred the remaining swabs into the DNA/RNA shield collection tubes, securely capped them and gently inverted the tubes several times (ensuring no foaming). Samples were transported to the APHI and aliquoted into sterile cryotubes and stored at −80°C.

Maternal blood

Maternal stool

Participants were instructed by the health centre laboratory techs and study nurses on how to collect stool samples. Women were instructed to urinate into the toilet before collecting stool in the study-provided container while avoiding urine, soil, detergent, fragrance or water contaminating the sample. The laboratory tech evaluated the stool sample, completed the Bristol Stool Chart and measured the pH using nitrazine pH paper. Samples were then homogenised by using a plastic spoon and stirring the sample. Stool samples were aliquoted for Genotek analysis using an OMNIgene. GUT collection kit for TaqMan analysis was collected using a DNA/RNA shield tube. All stool samples were refrigerated at the local health centre after collection and transported to APHI. At APHI, Genotek samples were aliquoted into two cryovials and were stored at −80°C. TaqMan samples were first stored in the DNA/RNA shield tube, and additional stool was aliquoted into three cryovials and stored at −80°C.

Infant blood

Infant stool

Mothers were instructed to wipe down the infant’s groin and public area with wipes provided by the study. Without applying any lotions or ointments, mothers lay the infant on the diaper or a cloth and stimulated the anal region using a wet cotton bud. Once the infant had a bowel movement, mothers collected the stool sample following the same instructions as their own samples (described earlier). Samples were processed, transported and stored using the same procedures as maternal stool samples.

A copy of the samples was shipped from Ethiopia to Boston, MA in the USA on dry ice and with temperature monitors for temporary storage and testing/analysis as previously described. Any samples with a single copy were retained in Ethiopia in compliance with Ethiopian regulatory requirements.

Table 3

Summary of analyses to be performed on biospecimens

Inflammation

Maternal diet and infections during pregnancy both may influence inflammatory processes and impact the development and growth of the fetus and infant.20 21 Using infant and maternal DBS, we will examine the effects of the ENAT interventions on maternal and newborn inflammation biomarkers including C-reactive protein (CRP), α1-acid glycoprotein (AGP), interleukin-6 and 6R (IL-6 and IL-6R) and investigate the association of maternal and newborn inflammation on long-term neurodevelopment in infants.

Micronutrients

Metabolomics

Metabolomics is the systematic analysis of human biofluids, such as blood, to describe the composition and metabolic changes within the human body at a specific time point. It is essential to understand the biological processes and pathways influenced by interventions involving biological elements like supplements and antibiotics. Using maternal and infant whole blood samples collected across multiple time points via VAMS and DBS, we will conduct untargeted metabolomics analysis using rapid liquid chromatography-mass spectroscopy (rLC-MS) at Sapient Analytics (California, USA).25 Internal control samples will be used to standardise testing and counteract potential batch effects. This discovery work will help us to identify metabolites both in pregnant women and infants that are associated with the ENAT interventions, as well as adverse birth outcomes such as preterm birth, small-for-gestational-age birth and LBW.

Microbiome profiling

Multiple studies have established the importance of the maternal microbiome on infant birth and growth outcomes; however, there is limited data on the effects of intervention during pregnancy.26 Using stool samples collected via OMNIgene.GUT collection kits, we will profile the maternal and infant gut microbiomes through shotgun metagenomic sequencing at the University of Wisconsin (Madison, WI, USA). This approach will accurately measure the richness and abundance of microbes and the differences between those receiving the ENAT nutrition intervention, infection intervention, both or neither. Additionally, we will also investigate how the maternal microbiome during pregnancy may impact infant birth and neurodevelopmental outcomes. This analysis will be conducted in collaboration with the Sonnenberg Laboratory at Stanford University (Stanford, CA, USA).

Gut enteropathogens

It has been hypothesised that enteropathogen infection may cause intestinal inflammation and contribute to stunting and micronutrient deficiencies in infants.13 Stool samples collected in DNA/RNA shield will be sent to the University of Virginia for testing using custom-designed TaqMan array cards (TAC). We will determine the pathogenic burden within pregnant women. We will also examine the impact of the ENAT deworming interventions on stool pathogens.

Telomeres

Leucocyte telomere length will be determined from maternal and cord blood DBS DNA. DNA will be extracted from the samples using the Qiagen Qiamp Blood Mini Kit protocol at the Drury Laboratory at Boston Children’s Hospital (Boston, MA, USA). Telomere length will be assayed following a standardised protocol.27

Intervention effects

In our primary analysis, we will examine differences in biomarkers or ’omics profiles across the randomised intervention study arms. We will use intention-to-treat as the primary analyses where individuals are analysed according to their originally assigned study arm, and in secondary analysis, we will also explore effects of individual interventions (ie, BEP) and adherence/dose. Outcomes will be compared between the primary factors (ENP or EIMP) to determine their marginal effects (ENP vs not-ENP, and EIMP vs not-EIMP). We will estimate differences in biomarkers, using linear mixed regression models. Potential confounding bias should be minimised due to the randomised study design; however, we will assess group differences in demographic and clinical characteristics at both cluster- and individual randomisation levels, as recommended by Consolidated Standards of Reporting Trials.28 Baseline covariate imbalances associated with both the outcome and exposure status (p<0.10) will be included in adjusted models as potential confounders. Multiple births will be excluded from analyses using birth outcomes due to the known increased rate of preterm birth and growth restriction among multiple deliveries. When large volumes of tests are conducted for metabolomics or genomics analyses, we will use the Benjamini-Hochberg method to correct for multiple testing, as appropriate.

Adverse pregnancy/child development outcomes

In our secondary analyses, we will examine the associations of biomarkers or ’omics profiles and adverse birth outcomes (eg, preterm birth, stillbirth and growth restriction) or child development outcomes (eg, EEG outcomes, MRI outcomes, eye tracking, etc). In most of these analyses, we will use a prospective cohort study design and estimate differences among birth or child development outcomes using linear or logistic regression models (based on the outcome), and adjust for covariates as relevant for specific outcomes. Details for each individual relationship examined will be included in the publications for each individual analysis.