Balanced Protein-Energy Supplement in Pregnancy and Early Lactation on Birth Outcomes and Growth in Southern Nepal

Background Undernourished women in many low and middle income countries (LMICs) enter pregnancy with low nutritional reserves, and low maternal body mass index (BMI) and short stature are associated with increased risks of a variety of adverse reproductive outcomes (Kozuki et al., 2015; Rahman et al., 2015). Balanced protein-energy (BE-P) supplementation during pregnancy shows moderate evidence than such dietary supplementation can decrease the risk of stillbirth and small-for-gestational-age (SGA) births (Ota et al., 2015; Stevens et al., 2015). The WHO antenatal care guidelines recommend providing pregnant women in undernourished populations with balanced protein energy dietary supplementation (WHO, 2016). The current evidence base for BP-E supplementation draws from a number of studies where BP-E supplements varied widely in composition and form (Imad \& Bhutta, 2002). Recent studies using fortified blended foods found that although approximately 800 kcal/d was provided, the net energy intake increased by a much smaller amount (200-300 kcal) (Janmohamed et al., 2016; Saville et al., unpublished trial in Nepal).

Evidence regarding dietary practices and preferences among pregnant women in Nepal is generally limited, however, some studies have shown that intra-household food allocation and specific food beliefs/behaviors do not favor adequate energy or micronutrient intake among adolescent girls and adult women (Gittelsohn, 1991; Gittelsohn et al., 1997). Christian et al showed that staple food intake can be reduced during pregnancy; which could be attributed to an aversion to food, lack of appetite, feeling unwell, or concern regarding having a "large" baby and difficulties with labor and delivery (Christian et al., 2006). Low nutrition knowledge may also play a role in suboptimal intake among women (Jones et al., 2005). Dynamic changes in access and use of services offer the opportunity to nest additional nutritional support services within the antenatal care system, including BP-E supplementation. However, information is limited regarding the impact such supplementation will have on the key birth outcomes that affect child health and development. It is this lack of evidence that this project will address.

Methods This project will be done in three phases. Phase one will identify preferable supplement types for use in pregnancy in the South Asian context. Phase two will use the results of phase one to conduct a medium-term feeding trial to assess acceptability and consumption. Phase three will be a large, community-based randomized trial using at least one of these supplement options for use in pregnancy and through the first 6 months post-partum on birth outcomes and early infant growth.

Specific Aim, Phase 1: Identify preferred product types for the provision of fortified BEP supplements among pregnant women in Southern Nepal.

Specific Aim, Phase 2: Assess the 8-week acceptability and consumption of 2 selected supplements with nutrient content consistent with Bill and Melinda Gates Foundation (BMGF) guidance among pregnant women in Southern Nepal.

Specific Aims Phase 3: Community-Based Randomized Field Trial

There is one primary aim and a number of secondary aims for each marginal treatment arm of this trial. Primary aims:

1. To compare the incidence of small-for-gestational-age (SGA) newborn infants born to mothers randomized to receive either a daily balanced protein-energy nutritional supplement with multiple micronutrients or control during pregnancy. SGA will be defined as <10th centile of birthweight for gestational age standard using InterGrowth 21st criteria (Villar et al., 2014).
2. To compare the growth of live-born infants among mothers randomized to the two groups during the first 6 months post-partum. The primary infant growth outcome is Length-for-Age Z (LAZ) score using the WHO Child Growth Study criteria (WHO, 2006).

Secondary aims:

Pregnancy Supplementation

1. To compare the incidence of short-for-gestational-age (ShGA) newborn infants born to mothers randomized to the two groups during pregnancy. ShGA defined as <10th centile of birth length for gestational age standard using InterGrowth 21st criteria.
2. To compare the incidence of small-for-gestational-age (SGA) newborn infants born to mothers randomized to the two groups during pregnancy. SGA defined as <3rd centile of birthweight for gestational age standard using InterGrowth 21st criteria.
3. To compare the incidence of short-for-gestational-age (ShGA) newborn infants born to mothers randomized to the two groups during pregnancy. ShGA defined as <3rd centile of birth length for gestational age standard using InterGrowth 21st criteria.
4. To compare the means and distribution of birth weight, birth length and gestational age (and percent <2500 gm, <37 weeks gestation) in infants born to women randomized to the two groups during pregnancy.
5. To compare the growth through 6 and 12 months of age (LAZ, weight-for-age Z (WAZ), weight-for-length Z (WLZ) scores) of infants born to women randomized to the two groups during pregnancy. 12 months follow-up will be available on approximately 2/3rd of infants.
6. To compare maternal weight gain during pregnancy among women randomized to the two groups during pregnancy.
7. To compare the diversity of the gut microbiome in pregnancy among women randomized to receive either a daily balanced protein-energy nutritional supplement with multiple micronutrients or control during pregnancy.

Post-Partum Supplementation

1. To compare WAZ and WLZ scores through 6 and 12 months of age of infants born to women randomized to receive either a daily balanced protein-energy nutritional supplement or control during the first 6 months post-partum. 12 months follow-up will be available on approximately 2/3rd of infants and women.
2. To compare the weight and BMI of mothers at 6 and 12 months post-partum among women randomized to the two groups during the first 6 months post-partum. 12 months follow-up will be available on approximately 2/3rd of mothers.
3. To compare the diversity of the gut microbiome post-partum among women randomized to receive either a daily balanced protein-energy nutritional supplement with multiple micronutrients or control during post-partum.
4. To compare the diversity of the gut microbiome among infants <6 months of age whose mothers were randomized to receive either a daily balanced protein-energy nutritional supplement or control during the post-partum period.
5. To compare the neurodevelopmental status of infants at 6 months of age whose mothers were randomized to receive either a daily balanced protein-energy nutritional supplement of control during the post-partum period.

Combination of Pregnancy and Post-Partum Supplementation

1. To compare the impact of combined pregnancy and post-partum nutritional supplementation with women who received only pregnancy or post-partum supplementation alone on LAZ, WAZ, and WLZ scores at 6 and 12 months of age among infants born to these mothers. 12 months follow-up will be available on approximately 2/3rd of infants.

Study Design for Phases 1 \& 2: A two-phase, mixed methods approach combining qualitative and quantitative methods for data collection will be used.

Phase 1: 11 product types will be introduced to participants in a single-meal test for their feedback on acceptability. Phase 1 will collect data to describe general preferences across product types and variations.

Phase 2: 2 product types identified in Phase 1 will be tested in an 8-week feeding trial. Each participating woman will experience one selected product type for a total of 8 weeks. Hedonic properties will be assessed at 2 and 8 weeks.

Complementary qualitative data will also be collected using in-depth interviews and focus group discussions during both phases 1 and 2.

Setting for all Phases This project will be conducted in a chronically food insecure study site in southern Nepal (Sarlahi District). This site is representative of a region (South Asia) and meets the criteria of having adequate local security conditions and presence of a strong field research site.

Study population and sample size estimates Phases 1 \& 2:

Pregnant Women (PW): In Phase 1, a convenience sample of 40 PW will be selected to evaluate 11 product types. In Phase 2, women participating in the home feeding trial (n = 40 for each of two product types, total n=80) will additionally participate in in-depth interviews.

Data Collection: Phases 1 \& 2:

Phase 1. Identify preferred product types Direct observation of single meal with hedonic testing: During the single meal observation, each participant will receive 1 serving of each product. Acceptability will be assessed based on Likert-like scoring using hedonic scales of overall liking, appearance, smell, taste and consistency.

Focus group discussion: Focus groups (maximum of 40 women) will be conducted following the single meal direct observation.

Phase 2. Home-feeding trial to assess acceptability and at-home consumption Two product types selected in Phase 1 will be assessed for medium-term acceptability and utilization through an 8-week feeding trial. Two measures of adherence will be assessed at each weekly home distribution; adherence in the last 24 hours using direct questioning and adherence using direct sachet/packet count.

Hedonic testing: Hedonic testing will be done at 8 weeks (de Graaf et al., 2005).

In-depth interviews: For both products, interviews will be conducted with PW and other household members, and health staff.

Focus group discussions: Focus groups among PW receiving both products will be conducted at the end of Phase 2.

Data analysis, Phases 1 \& 2:

Adherence data: Adherence by self-report will be calculated as the percentage of days the supplement was reported to be consumed and mean portion when consumed, and adherence by sachet count will be calculated as total empty packages returned divided by total packages expected to be consumed.

Hedonic test data: Product liking, and left over quantities over time will be summarized by product type using means for continuous variables.

Qualitative data: Analysis will be ongoing throughout data collection, and conducted by the research team using an exploratory, inductive approach that will draw from Grounded Theory and allow for the analysis of emergent themes (Charmaz, 2014; Ridder et al., 2014; Charmaz, 2006). Data will be systematically coded using Dedoose computer software (Consultants, 2016) (Glesne \& Peshkin, 1992).

Phase 3: Randomized Trial Design The design of this study is a community-based, household level-randomized 2x2 factorial trial among women who become pregnant in the study area. Due to interruptions due to the COVID-19 pandemic and funding issues, recruitment was conducted in two non-concurrent cohorts with 50% of recruitment during each cohort.

Study Population The study population for this trial will include all women who become pregnant during two cohorts each over an approximate 6 month period in a set of Nagarpalikas or Gaunpalikas in Sarlahi District, Nepal. Newly pregnant women will be identified through a pregnancy surveillance system.

Intervention

Two interventions strategies will be tested in a factorial design in this trial; nutritional supplementation during pregnancy and nutritional supplementation during the first 6 months post-partum. The exact forms of this supplement will be selected in Phases one and two but its composition will be in conformance with the recommendations from the expert panel convened by the BMGF on nutritional supplementation in pregnancy (Expert Consultation, 2016). Women in households allocated to the control groups will receive a standard package of antenatal services and nutritional counseling by study staff. In all four groups, the following antenatal and post-natal interventions will be offered:

• Women will be encouraged to enroll in routine antenatal care at their local health post/center.
• A clean birthing kit will be provided consisting of a clean blade, string, and plastic disc for cutting the cord, a plastic sheet, a bar of soap, and a tube of chlorhexidine ointment for application to the umbilical stump.
• Women will be encouraged to deliver at a certified birthing facility and participate in the government's incentive scheme.
• Additional nutritional, hygiene, and infant care counseling will be provided by study staff.
• If a woman reports not attending ANC and/or not receiving tetanus toxoid or iron-folic acid supplements, our study staff will provide them.

Masking While it will not be possible to mask the participants or the Ward Data Collectors to the treatment assignment, we will attempt to mask the birth and growth assessment teams to the treatment assignments.

Randomization Randomization will be done at the household level for this study. This means that once the first pregnant woman from a given household is identified and consents to be in the trial, that household will be allocated to either the supplement or non-supplement group (for pregnancy). Any other woman or women from this household who subsequently become pregnant within the 9-month enrollment period, would also be assigned to that group.

We will pre-generate a randomization sequence denoting the two pregnancy allocation possibilities (supplement, no-supplement) using the R software environment for statistical programming (www.r-project.org) using randomly permuted blocks of size 4, 8, and 12, to ensure sequential balance across the groups throughout the enrollment period.

For randomization to the post-delivery exposure (i.e. supplement, no-supplement), a slightly modified procedure will be followed. Here, we will construct two pre-generated allocation sequences, again constructed with randomly permuted blocks of size 4, 8, and 12. These two sequences will be stored in our database, and annotated/labelled to allow for allocating post-delivery exposures stratified by pregnancy exposure.

Substudies A group of substudies will be conducted among a systematic sample of 120 pregnant women per treatment group (total of 240). These substudies include assessments of maternal dietary intake, the gut microbiome using multiplex PCR and metabolomic assays in stool, and infant neurodevelopmental outcomes at 6 months.

Maternal Microbiome SubStudy The objective is to examine whether a BP-E nutritional supplement during pregnancy and breastfeeding will promote a more diverse and normal gut microbiome in pregnant and breastfeeding women.

Methods:

Stool will be collected among all 120 women per group at the following times:

• at the time of baseline enrollment in early pregnancy.
• at the late pregnancy visit (~34 weeks gestation).

Dietary Intake Sub-Study The purpose of this sub-study is to assess whether there is significant dietary diversion occurring in the supplementation groups compared with controls. We will use daily energy and protein intake as the primary indicators of diversion. A quantitative 24-hour recall instrument will be used once in pregnancy (in duplicate) and once in the post-partum period (in duplicate). We will adapt dietary instruments that have been used previously in this population for this sub-study (Sudo et al, 2006; Parajuli et al, 2012; Campbell et al, 2014; Henjum et al, 2015).

Sample Size / Power Calculations These calculations assume the nutritional supplementation in pregnancy and post-partum periods have independent effects on the outcomes of interest. We are not powered to detect specific interaction effects on either primary or secondary outcomes.

Primary Outcomes: Small-for-gestational-age (SGA) is the primary outcome that we have used to drive the sample size calculations. While not a truly individually randomized trial (households will be randomized), the very small likelihood of identifying more than one eligible pregnant women in a household over a 9 month period is quite low (<3.5% of "clusters" will have more than 1 individual outcome). We've estimated the effect of this level of clustering on sample size and found it to be extremely small versus individual randomization. Therefore, all our sample size/power estimates are based on individual randomization.

Over a period of two phases of approximately 6 months, we expect that eligible households in our study area will yield approximately 1600 pregnant women whose infants will be seen within 72 hours of birth for measurement of the primary birth outcome.

Table: Effect Size Detectable for Primary Birth Outcome Given 774 Infants per Group Measured within 72 Hours of Birth Outcome Relative Risk Detectable (Supplemented/Unsupplemented) 80% Power 90% Power Primary SGA (10th centile) 0.825 0.798

We will be able to detect a 17.5% reduction in SGA with 80% power and a 20.2% reduction with 90% power assuming a type 1 error of 5% (two-sided).

The table below presents the detectable relative risks or mean differences for a variety of secondary outcomes with 80% and 90% power.

Table: Detectable Difference in Secondary Outcomes (Marginal Comparisons) Outcome Detectable Relative Risk or Difference Between Groups 80% Power 90% Power Secondary ShGA (10th centile)a 0.778 0.745 SGA (3rd centile)b 0.726 0.686 ShGA (3rd centile)c 0.653 0.604 Mean Birth Weight d 60.7 g 70.2 g Mean Birth Length e 3.0 mm 3.5 mm LAZ at 6 months1 0.16 Z scores 0.18 Z Scores WAZ at 6 months2 0.17 Z scores 0.20 Z scores WLZ at 6 months3 0.18 Z scores 0.21 Z scores Maternal Weight Gain in Pregnancy4 463 g 536 g Maternal Weight at 6 months Post-Partum5 0.89 kg 1.03 kg

a ShGA (10th centile expected rate is 30.1%. b SGA (3rd centile) expected rate is 22.1%. c ShGA (3rd centile) expected rate is 14.4%. d Expected mean birth weight is 2708 g (SD=429). e Expected mean birth length is 47.6 cm (SD=2.2).

1. Expected mean LAZ at 6 months is -1.04.
2. Expected mean WAZ at 6 months is -1.63.
3. Expected mean WLZ at 6 months is -1.23.
4. Expected mean weight gain in pregnancy is 7.55 kg (SD=3.23 kg).
5. Expected mean maternal weight at 6 m PP is 44.07 kg (SD=6.21 kg).

Stool Sub-Study: All substudy women in both groups will have stool collections at each indicated point in time. There are very few guidelines for sample size in microbiome studies. We consider these analyses to be exploratory and may provide important information regarding the mechanisms of how nutritional supplementation can improve birth outcomes and early infant growth.

Dietary Intake Sub-Study: We used two studies conducted among adults in the terai that use quantitative methods to estimate actual intake (Parajuli et al, 2012; Sudo et al, 2006). Both provided estimates of daily intake for a variety of nutrients for adult women and were very similar in their results. We used the average of their estimated intake of energy (kcal) and protein (g) and the associate average standard errors for our calculation. Mean energy intake was 1700 kcal/day (SD=350) and 42.3 g (SD=10.4) of protein. As this sub-study is designed to assess dietary diversion due to the supplement, we used a minimally detectable difference of -12% for energy and protein as an indicator that dietary diversion was occurring. Given these parameters, 120 persons per group will provide over 90% power for energy intake and approximately 85% power for protein intake.

Data Analysis, Phase 3: More detail will be available in the Statistical Analysis Plan document that will be posted to this site.

Data analysis will proceed in two phases, an exploratory data analysis phase followed by a confirmatory analysis phase and focuses on the specific aims of this project. The exploratory phase will involve final data cleaning and editing as well as examination of the distributions for all variables and the application of appropriate transformations. We will populate a CONSORT diagram. Next, comparisons will be made between the four treatment groups on a variety of variables at the household and individual levels to assess their comparability. All eligible birth outcomes will be limited to those measured within 72 hours of delivery.

Pregnancy Intervention Primary Aim: Data analysis will be by intent to treat. The proportion of babies born SGA (10th centile) will be compared in the two treatment groups using generalized estimating equations (GEE). An effect size will be estimated as the proportionate change in the intervention group relative to the control group. A 95% confidence interval will be calculated around this effect size estimate that accounts for the correlation within households. A secondary, post-hoc stratification by level of compliance with supplement use will be done within the active treatment group and the proportion SGA will be compared across groups. A dose-response effect will be estimated by linear regression techniques.

Secondary Aims 1, 2, and 3: A similar analysis to that of the primary aim will be done for the outcome ShGA (10th centile), SGA (3rd centile), and ShGA (3rd centile).

Secondary Aim 4: Mean birthweight and the distribution of birthweights will be compared by treatment group (supplementation during pregnancy versus not). Means and proportions below 2500 grams will be compared using GEE. A comparison of the full distribution of birthweights between treatment groups will also be done as previous research in this study area has indicated that treatment effects may vary by location on the birthweight distribution. (Katz et al, 2006; Dominici et al, 2006). A similar analysis of birth length will be conducted. A similar analysis will be conducted for the mean and distribution of gestational age and the proportion preterm (<37 weeks gestation).

Secondary Aim 5: Mean Length-for-Age, Weight-for-Age, Weight-for-Length, and proportions (<-2 Z scores) at 6 months of age will be compared by treatment group using GEE.

Post-Partum Intervention Primary Aim: Data analysis will be by intent to treat. The mean LAZ score among infants at 6 months of age will be compared in the two treatment groups using GEE. An effect size will be estimated as the proportionate change in the intervention group relative to the control group. A 95% confidence interval will be calculated around this effect size estimate that accounts for the correlation within clusters. A secondary, post-hoc stratification by level of compliance with supplement use during lactation will be done within the active treatment group and the mean LAZ score will be compared across groups. A dose-response effect will be estimated by linear regression techniques.

Secondary Aim 1: A similar analysis to that of the primary aim will be done for the outcome WAZ and WLZ scores.

Secondary Aim 2: Weight (and BMI) of mothers at 6 months post-partum will be compared by treatment group using GEE. An effect size will be estimated as the proportionate change in the intervention group relative to the control group. A 95% confidence interval will be calculated around this effect size estimate that accounts for the correlation within clusters. A secondary, post-hoc stratification by level of compliance with supplement use during lactation will be done within the active treatment group and the mean weight and BMI will be compared across groups. A dose-response effect will be estimated by linear regression techniques.

Combination of Pregnancy and Post-Partum Supplementation

Secondary Aim 1: The combined effect of both pregnancy and post-partum supplementation on 6 and 12 month LAZ, WAZ, and WLZ scores will be estimated two ways. First a stratified analysis will be done comparing mean and 95% CI on Z scores among control in both periods with pregnancy alone, post-partum alone, and both pregnancy and post-partum supplementation. For significance testing we have chosen the following 3 comparisons to apply tests of significance:

• Control in both periods versus post-partum alone.
• Pregnancy alone versus post-partum alone.
• Post-partum alone versus both pregnancy and post-partum. The second approach will be to model the effects of the treatment groups using linear regression models that include an interaction term for the combined treatment arm. A test of interaction will be used but we will primarily rely on interpreting the size of the interaction effect relative to the changes in the period-specific supplementation effects.

A variety of additional analyses will be conducted to explore the consistency of the main effects of these interventions.

Ethical Considerations The project has been approved by Institutional Review Boards (IRBs) at the George Washington University, the Johns Hopkins Bloomberg School of Public Health, and the Nepal Health Research Council. The IRB at the Harvard T.H. Chan School of Public Health has defered to the IRB of record at George Washington University.

Adverse Event Reporting Serious Adverse Events: Serious adverse events (SAE) under GCP guidelines include death, a life-threatening reaction to the supplement or other study procedure, hospitalization, significant or persistent disability or impairment, or a congenital anomaly/birth defect.

Other Adverse Events: These include any untoward medical event that occurs in a participant, any unfavorable sign or symptom of disease temporally associated with the provision of the supplement, or any noxious and unintended response to the supplement. Anticipated adverse events include signs/symptoms of allergic reaction to the supplement, preeclampsia, infection during pregnancy, gestational diabetes. Unanticipated adverse events are either unexpected in terms of nature, severity and frequency. An example would include newborn macrosomia (BW>97th centile according to InterGrowth21 standards).

All adverse events will be reported to the designated local investigator for review within 48 hours. If there is a possibility of the event being related to the study, a report will be made to the independent Safety Officer within 24 hours of receipt by the investigator. The Safety Officer will review the file and make a determination regarding attribution to study procedures or the intervention. If the adverse event is considered by the Safety Officer to be minor in severity, it will be logged and a summary of such events will be reported to the DSMB at their next meeting. If the severity of the adverse event is considered to be moderate or severe by the Safety Officer, he/she will indicate to the Principal Investigator that it should be reported to the DSMB and the cognizant IRBs immediately. The PI will then forward all materials together with the Safety Officer's summary to the DSMB and the IRBs.