Study setting

We will conduct this randomized effectiveness study in Akaki and Nifas Silk-Lafto, two sub-cities of Addis Ababa, Ethiopia. These sub-cities have significant burdens of undernutrition among women of reproductive age. Between 2018 and 2020, among women of reproductive age with a child aged 6 to 59 months, 14.0% in Akaki and 10% in Nifas Silk-Lafto had a BMI less than 18.5 kg/m2; 8.6% in Akaki and 11.7% in Nifas Silk-Lafto had a MUAC less than 23 cm; and 17.9% in Akaki and 13.8% in Nifas Silk-Lafto had either a low BMI or a low MUAC (unpublished data).

Randomization design

We will conduct the study in nine public antenatal health facilities, five in Akaki and four in Nifas Silk-Lafto. This study will include randomization at both cluster level and individual level (Fig. 1). We first conducted cluster randomization among the nine health facilities to allocate three of them to universal BEP (Arm 4). Then, within the remaining six facilities, we will conduct individual randomization within each facility to allocate individual pregnant women to one of three arms: control (Arm 1), targeted BEP based on baseline nutritional status (Arm 2), or targeted BEP based on baseline nutritional status and monthly GWG monitoring (Arm 3). Therefore, all participants in the three antenatal facilities assigned to Arm 4 will receive the same universal BEP intervention. Participants within the other six facilities will receive different interventions based on individual randomization.

Fig. 1

Cluster- and individual-level randomization of a randomized effectiveness study to compare effectiveness, cost-effectiveness, and implementation across different targeting strategies of antenatal balanced energy and protein (BEP) supplementation in Addis Ababa, Ethiopia

To ensure comparability of the three health facilities for Arm 4 versus the six facilities for Arms 1 to 3, we matched the nine antenatal facilities into three triplets before the cluster randomization. Each triplet includes three facilities similar in the annual number of deliveries and the proportion of low birthweight newborns out of all live births. We used coarsened exact matching and the cem command [22] in STATA 16 (StataCorp, College Station, Texas) for the initial matching, followed by manual fine-tuning to arrive at two matches, with three facilities per match. Then, we randomly assigned one facility out of each triplet to Arm 4. A statistician not directly involved in study implementation conducted the cluster and individual randomization. For the individual randomization, a separate randomization list was generated in permutated blocks for each facility. The study nurses responsible for recruiting and assigning interventions are not aware of the blocking information. Due to the adaptive nature of interventions in this study, blinding the participants and study teams to the assigned interventions will not be possible.

Screening and enrollment

The inclusion criteria include (1) pregnant women aged 18 to 49; (2) attending antenatal visits in one of the study health facilities; (3) with a gestational age of 24 weeks or less; (4) no known allergies to peanuts or soy; (5) having resided in the current location for at least 12 months; (6) intending to continue antenatal follow-up in the health facility; (7) intending to give birth and remain in the study area until 6 weeks after delivery; and (8) willing to take the BEP supplements for the entire duration of the pregnancy if eligible. We will exclude those above 24 weeks of gestation due to the limited time the participants may receive the intervention and the difficulty of estimating GWG. Criteria 5 to 8 are enacted to restrict to pregnant women most likely to adhere to the allocated intervention and minimize loss to follow-up.

Upon enrollment, trained study nurses will collect information on demographic characteristics, socioeconomic status, household food insecurity, health conditions, and reproductive history using interviewer-administered questionnaires. Study nurses will measure height, weight, MUAC, and blood pressure; assess fingerstick hemoglobin concentration; and conduct a dipstick proteinuria test.

Gestational age assessment and quality assurance

At the enrollment visit, women will provide a self-report of their last menstrual period (LMP). Those with an LMP between 8 and 24 weeks are eligible for the study and will receive an ultrasound assessment for pregnancy confirmation and gestational age dating. If the ultrasound shows less than 8 weeks gestation, the participants will be offered another ultrasound in 4 weeks. For women who cannot recall their LMP despite support from the study nurses, they will be provided with the ultrasound assessment at enrollment, and the ultrasound-based gestational age estimate will be used for eligibility confirmation. During data analysis, we will triangulate the gestational age estimates based on self-report and ultrasound assessment to determine the best obstetric estimate for the relevant outcomes (e.g., preterm birth and SGA). Two study nurses were trained to perform the ultrasound scans at each health facility.

An external sonographer will review a random sample of the ultrasound scan images weekly. Each image will be scored based on a pre-specified scoring spreadsheet with five domains: crown-rump length, biparietal diameter, head circumference, abdominal circumference, and femur length. Each domain has a score ranging from 0 to 7, with the total possible score ranging between 0 and 35. The external sonographer will also provide written comments for each image to offer specific and individualized feedback. In addition to the external quality assurance, we will provide quarterly reinforcement training and standardization for all study nurses responsible for ultrasound scans.

Pregnancy follow-up

We will assess each participant every 4 weeks during their visits to the antenatal facilities until delivery, death, or fetal loss, whichever occurs first. Data collected during the monthly follow-up visits will include pregnancy status, anthropometry, and clinical assessments. The clinical assessments will include a physical examination, blood pressure measurement, collection of morbidity symptoms, and dipstick proteinuria test. During 28–32 weeks of gestation, we will conduct another fingerstick assessment of hemoglobin concentration. We will conduct a food frequency questionnaire during 28–32 weeks of gestation. We will conduct 24-h dietary recalls once in the second trimester during 14–28 weeks of gestation and once in the third trimester after 32 weeks of gestation.

Delivery assessment

We will assess delivery outcomes within 3 days (< 72 h) of birth, regardless of whether a participant delivers in a health facility or at home. An outcome assessment team will contact participants weekly after 36 weeks of gestation to maintain good rapport with the participant and ensure that the delivery information will be recorded promptly. We will also give participants the phone number of a study-specific ‘birth hotline’ so they can report their deliveries. When the outcome assessment team is alerted of a potential birth in advance, they will record delivery details, including vital status and anthropometry. For deliveries at home, the outcome assessment team will visit the home at the earliest possible time and no later than 72 h after birth to assess the vital status and measure the newborn’s anthropometry, including weight, length, and head and chest circumference.

Postnatal follow-up

The outcome assessment team will visit all study participants at home or in the health facility at 6 weeks postpartum to assess maternal and infant vital and health status and perform maternal and infant anthropometric measurements.

Interventions

The participants will be randomized to four study arms (Fig. 2). Participants in all four arms will receive the standard of antenatal care per the WHO antenatal guidelines [12] and the Ethiopian antenatal guidelines, including iron and folic acid supplementation. Arm 1 will be the control arm and will only receive this standard of care.

Fig. 2

Study design of comparing different targeting approaches of antenatal balanced energy and protein supplementation in Addis Ababa, Ethiopia. BEP, balanced energy and protein; BMI, body mass index; GWG, gestational weight gain; MUAC, mid-upper arm circumference

Participants in Arm 2 (targeted BEP based on baseline nutritional status), in addition to the standard of care, will receive BEP supplements if their baseline BMI is less than 18.5 kg/m2 or their baseline MUAC is less than 23 cm. The cutoff of 18.5 kg/m2 for BMI corresponds to the WHO definition of underweight [23]. The cutoff of 23 cm for MUAC is selected based on prior literature [24, 25] and analyses of the predictive power of adverse pregnancy outcomes based on internal unpublished data, prioritizing specificity over sensitivity to reduce false positives (i.e., so fewer participants would receive the supplements unnecessarily). MUAC is relatively easy to measure; hence, it has the potential of complementing BMI for screening high-risk pregnant women in need of targeted interventions [26]. Weight gain during the first trimester of pregnancy is minimal, and most of the GWG occurs during the second and third trimesters [14]. Therefore, for participants enrolled before the end of the first trimester (136/7 weeks), we will use the weight measured at enrollment as the proxy for pre-pregnancy weight. For participants enrolled during the second trimester (but no later than 24 weeks of gestation), we will impute early-pregnancy gestational weight, as further described below. We will use the baseline MUAC measurement regardless of gestational age at enrollment. Participants who meet either the low BMI or the low MUAC criterion will receive the BEP supplements until the end of their pregnancy.

Participants in Arm 3 (targeted BEP based on baseline nutritional status and monthly GWG monitoring), in addition to the standard of care, will receive BEP supplements if their baseline BMI is less than 18.5 kg/m2 or their baseline MUAC is less than 23 cm, in the same way as described for Arm 2. Additionally, we will use monthly measurements of GWG to inform the initiation of additional BEP supplements for those with inadequate GWG. If the participant is already on antenatal BEP supplements initiated based on baseline undernutrition, she will receive additional BEP supplements on top of the ration she is already receiving. If the participant does not have undernutrition at baseline and thus is not already on the supplements, she will start receiving BEP supplements from that visit in which inadequate GWG was identified. The supplementation initiated based on inadequate GWG will be provided until the end of pregnancy regardless of gestational weight gain assessments during subsequent visits.

All participants in Arm 4 (universal BEP) will receive BEP supplements until the end of pregnancy, regardless of baseline nutritional status or GWG. This arm mimics the universal BEP supplementation as currently recommended by the WHO. For participants with overweight or obesity (i.e., BMI \(\ge\) 25 kg/m2) at enrollment during the first trimester, or based on imputation for those enrolled in the second trimester, we will stop the provision of supplements if they start to gain excessive GWG. Definitions of inadequate and excessive GWG are provided below.

Balanced energy and protein supplements

We will use Plumpy’Sup as the BEP supplement. The product is a smooth, homogeneous, thick, dark-brown-colored paste containing peanuts, sugar, oil, milk powder, and vitamin and mineral premix. It is ready to use and can be squeezed out of the sachet and eaten directly without cooking, mixing, or diluting. Each 100-g sachet of Plumpy’Sup includes 540 kcal of energy, 12 g of protein, and an array of micronutrients. The nutrient composition of the supplement aligns with the recommended specifications for the nutritional composition of BEP based on the expert consultation held by the Bill & Melinda Gates Foundation (Table 1) [27].

Table 1 Nutrition composition of Plumpy’Sup and expert recommendation from the Bill & Melinda Gates FoundationDistribution of nutritional supplements and assessment of compliance

The BEP supplements will be distributed by study nurses at baseline visits and every month when participants come to the antenatal facility for follow-up visits. In addition to the amount determined by the intervention assignment, we will also provide a modest quantity of extra supplements to compensate for the practice of intra-household sharing. We determined the amount to compensate for intra-household sharing through a formative period. We will assess compliance with the assigned intervention by requesting participants to bring back unused sachets and packages of used sachets during their monthly follow-up visits. We will also directly ask the participants about the use of the supplements in the previous week and since the last visit.

Early-pregnancy weight imputation and gestational weight gain monitoring

A significant proportion of pregnant women in practical settings do not present themselves at antenatal facilities until later in pregnancy, typically during the second trimester. Therefore, to approach a real-world setting to the greatest extent possible, we will also include participants enrolled during the second trimester of pregnancy. However, gestational weight beyond the first trimester does not reflect the pre-pregnancy or early-pregnancy nutritional status. It thus cannot be used as a proxy for baseline nutritional status or as an anchor point for the calculation of GWG. Therefore, for participants enrolled during the second trimester of pregnancy (but no later than 24 weeks of gestation), we will impute their early-pregnancy gestational weight.

As part of the preparatory activities, we developed and validated an imputation equation that estimates a pregnant woman’s weight during early pregnancy, using individual characteristics and gestational weights collected later during pregnancy. We used individual characteristics, including current gestational weight, current gestational age, and height, as potential predictors. We developed the equation using the GWG Pooling Project data, which combined data from 55 studies and approximately 150,000 pregnant women in LMICs [28, 29]. The GWG Pooling Project has accumulated a wealth of data on longitudinal gestational weights, including pre-pregnancy weights from some studies and early-pregnancy weights from many studies [28, 29]. We conducted external validation of the tool using internal data from Ethiopia [30].

We incorporated the imputation algorithm into the electronic tool for data collection. The imputation equation estimates a pregnant woman’s early-pregnancy weight, an essential quantity for evaluating baseline nutritional status and a critical anchor point for assessing GWG adequacy. The electronic tool also has the built-in capacity of monthly GWG adequacy monitoring based on the Institute of Medicine (IOM) recommendations and the (measured or imputed) early-pregnancy weight.

Quantifying gestational weight gain adequacy

Inadequate weight gain at each visit will be quantified using the GWG percent adequacy ratio, calculated as the ratio of the observed GWG to the weight gain recommended by the IOM based on pre-pregnancy BMI [14]. First, we will calculate the observed GWG as the difference between each weight measurement and the early-pregnancy weight (measured for those enrolled during the first trimester or imputed for those enrolled during the second trimester). Second, we will estimate the IOM-recommended GWG at the time of the weight measurement. Finally, we will calculate the percent adequacy by dividing the observed by the recommended GWG. The formulas are shown below.

$$\mathrm(\mathrm) =\frac}}\times 100\%$$

$$\mathrm=\text-\mathrm$$

$$}=\left(\frac}\;1\;}}}}\right)\left(13.86\;}-}\right)+\left(}-13.86}\right)\times }\;2\;\& \ }\;3$$

The expected first-trimester GWG will be 2, 1, and 0.5 kg for pregnant women with underweight/normal weight, overweight, and obesity, respectively. The recommended weekly rate of GWG in the second and third trimesters will be 0.51 kg/week for underweight, 0.42 kg/week for normal weight, 0.28 kg/week for overweight, and 0.22 kg/week for obesity [14].

GWG adequacy ratio accounts for the gestational age at each weight measurement and takes advantage of the well-established IOM recommendations. To account for the allowable range specified in the IOM standards and in accordance with previous studies [28, 29, 31,32,33], we will define inadequate GWG at each visit as a percent adequacy ratio less than 90%, adequate GWG as a percent adequacy ratio between 90 and 125%, and excessive GWG as a percent adequacy ratio greater than 125%.

Formative research

During the preparatory phase of the study, we conducted formative research for a preliminary evaluation of the potential acceptability of the BEP intervention. We established an initial understanding of potential facilitators and barriers to the continued use of BEP supplements. Forty-five participants were selected purposively from the nine health centers and provided with BEP sufficient for 4 weeks with clear instructions. After this period, we conducted in-depth interviews (IDIs) with 39 participants. We used a structured framework during the IDIs to elicit contextual data relevant to pregnant women’s general perspectives on dietary preferences and culinary practice. The framework included questions about factors that may influence the acceptability and continued use of the BEP supplements, such as flavor preferences, food-sharing dynamics, and local food environments.

Cost data collection

We will focus on three major cost categories for empirical cost data collection: (a) direct costs required to provide each intervention; (b) costs or cost-savings resulting from changes in utilization for health services that would plausibly be impacted by the intervention, such as medical care for participants with pregnancy complications; and (c) costs and cost-saving borne by patients, as could results from seeking and receiving related medical care, as well as productivity losses due to pregnancy-related morbidity and accessing healthcare.

For intervention costs, we will use an ingredients approach [34] to identify and cost all resources required for the targeted and universal approaches of BEP delivery. The costs will be broadly composed of four categories, including personnel time (e.g., nurses, sonographers); BEP supplements (cost of products and shipping); medical equipment and supplies (e.g., ultrasound machines and anthropometric measurement tools); and infrastructure and logistical support (e.g., storage space for BEP supplements). For each category, we will create data collection instruments to record resource utilization and use these to quantify the resources required throughout pregnancy. We will review financial records and conduct key informant interviews at each facility to estimate resource requirements if the program is to be scaled up.

For costs due to changes in utilization of related health services, we will conduct interviews to identify health services plausibly affected by the interventions (i.e., by changing the probability of different pregnancy outcomes). Based on these interviews, we will identify the relevant services and collate information on the utilization of these services for each study arm. We will collect cost data to calculate unit costs for each service and multiply them by recorded service utilization to estimate total costs per patient for related health services, for each study arm.

For patient-incurred costs, we will conduct interviews with a sub-sample of study participants to understand the direct medical and non-medical costs incurred to participate in the interventions and receive related services. In the same survey, we will collect information on time spent participating in interventions (beyond the routine antenatal process), as well as time spent unable to work or contribute to household activities due to pregnancy-related morbidity, to estimate productivity losses.

Qualitative evaluation

Toward the end of the study, we will use focus group discussions (FGDs) and IDIs to generate qualitative implementational evidence regarding the feasibility and acceptability of different modalities of antenatal BEP supplementation and the facilitating and inhibitory factors of uptake and continued use of the supplements.

The qualitative evaluation will be conducted from the perspectives of pregnant women and healthcare providers. We will conduct FGDs among pregnant women participating in the targeted or universal BEP intervention. We will select eight participants from each of the nine antenatal health centers to form the focus group for that facility. Therefore, the total number of participants for the FGDs will be 72 pregnant women (eight participants per facility from nine facilities). For the six facilities with a mixture of participants of Arms 2 and 3, we will ensure that participants in both targeted arms will be represented in the focus group and will only include participants who received the supplements under the corresponding targeting criterion. We will also invite one study nurse from each antenatal facility to form a separate focus group with nine nurses. The study nurses for the FGDs will be selected from those closely involved with the day-to-day study activities, including the distribution of the BEP supplements.

We will develop semi-structured FGD questions and guides for pregnant women and healthcare providers. The FGDs with pregnant women will cover a range of thematic topics, including general questions on household food practices and beliefs, diet during pregnancy and lactation, availability and access to nutritious foods, antenatal care-seeking behaviors, and supplement-specific questions, including experience with product taste, product preparation, levels and barriers of adherence and continued use, dynamics of intra-household sharing, and interface of the BEP products with local food environments. The FGDs with nurses will cover the burdens, enablers, and barriers of targeted or universal distribution for healthcare providers and their perspectives on sustainability, buy-in, and scale-up. All FGD questions and guides will be pretested and modified prior to implementation.

The FGDs will be held at the antenatal facilities and last approximately 2 to 2.5 h. We will use private rooms perceived as neutral as possible. Participants will be encouraged to speak in the language in which they are most comfortable and confident. The FGDs will be audio-recorded and transcribed from the local language into English. The focus groups will be audio-recorded and conducted with one moderator (a trained research assistant) and one notetaker who will support with additional observational notes that may not be captured through the audio recording.

In addition to FGDs with pregnant women and nurses, we will conduct one-on-one, IDIs with the head administrator of each of the nine antenatal facility. The interviews with the administrators will elicit additional information on the healthcare providers’ experience with various modalities of supplement delivery and understand their perspectives on sustainability, buy-in, and scale-up of targeted or universal supplementation. The IDIs with the head administrators will last approximately 60 min and be conducted by a trained research assistant supported by an additional notetaker in their antenatal facilities.

The expected participant timeline is illustrated in Table 2. The study protocol was designed and developed in accordance with the Standard Protocol Items: Recommendations for Interventional Trials (SPIRIT) 2013 checklist (Additional file 1, SPIRIT Checklist).

Table 2 Schedule of enrollment, intervention, and assessments based on SPIRIT guidelines1Sample size calculation

We used SGA as the primary outcome to calculate the required sample size, as evidence suggests that one of the primary effects of antenatal BEP supplements is to reduce the risk of SGA [9]. We used a two-sample test of proportion to conduct sample size calculation, comparing Arm 2 to Arm 1, and comparing Arm 3 to Arm 1, individually. We used the Power and Sample Size Calculation Program by Dupont and Plummer to conduct the calculation [35]. The following assumptions were used:

(1)

SGA incidence of 30% in the control group based on unpublished data from the Akaki or Nifas Silk-Lafto sub-cities;

(2)

An effect size (risk ratio) of 0.80 based on the Cochrane systematic review by Ota et al. (2015) [9];

(3)

A 14% rate of loss to follow-up;

(4)

A two-sided significance level of 0.05 and a power of 90%.

Under these assumptions, the target sample size per arm is 1337 pregnant women, and the total number of participants for Arms 1 to 4 would be 5348 participants (i.e., 1337 × 4), or 5400 after rounding to the next hundred. Therefore, we will enroll 5400 pregnant women in this study. Additional file 2 shows the detectable RRs given alternative assumptions of SGA incidence and statistical power.

Data management and monitoring

The Addis Continental Institute of Public Health (ACIPH) will be responsible for data collection and management. Data will be electronically collected using pre-programmed questionnaires and tools on research tablets. All data collected as part of the study will be uploaded and stored in a secure server at ACIPH, with access restricted to authorized members of the study team only. Data will be stored in coded format, stripped of any identifiers and with the linkage key stored separately. Any paper forms will be stored in locked cabinets that are accessible only to authorized members of the study team. De-identified data will be transferred to Harvard T.H. Chan School of Public Health for joint monitoring, including the examination of data values for outliers and potential errors.

Data analysisEffectiveness analysis

We will evaluate the effectiveness of the individual-based targeting strategies by comparing the incidences of adverse pregnancy outcomes across arms, using Arm 1 (control group) as the reference. We will use SGA as the primary outcome. The secondary outcomes will include inadequate and excessive GWG close to delivery, stillbirths, preterm births, LBW, macrosomia, large-for-gestational-age births, anemia during the third trimester, pregnancy-related death, neonatal death, and perinatal death, as defined in Table 3. We will use log-binomial models to calculate RRs and the 95% CIs for the following comparisons: (1) Arm 2 versus Arm 1; (2) Arm 3 versus Arm 1; (3) Arm 4 versus Arm 1; and (4) Arm 2 versus Arm 3. We will use modified Poisson models with robust variance estimates to achieve model convergence whenever necessary [36]. We will conduct the primary analyses using the intention-to-treat approach. The outcomes will be compared across arms among all participants in each arm, regardless of whether they meet the criteria to receive BEP, so we can quantify the impacts of the targeting strategies at the population level, not only merely among those who received the supplementation.

Table 3 Definition of primary and secondary outcomes

We will investigate the modification of the effect of the targeted intervention on primary and secondary outcomes by potential effect modifiers at baseline, including maternal age, dietary intake, anemia status, and reproductive health history. We will explore effect modification by including interaction terms in regression models and assess statistical significance using likelihood ratio tests. We will conduct the analyses for effectiveness using SAS 9.4 (SAS Institute Inc., Cary, North Carolina) at a two-sided \(\alpha\) level of 0.05.

Cost-effectiveness analysis

We will use the cost data and effectiveness results to conduct an incremental cost-effectiveness analysis comparing the four intervention approaches [37, 38]. We will develop a simple decision-analytic model to extrapolate long-term health outcomes, including changes in survival, quality-of-life, and disability-adjusted life years (DALYs), based on the empirical endpoints of the study. We will use this model to estimate the health losses associated with each study outcome. We will apply these values to the distribution of outcomes obtained for each study arm. We will summarize these differences as life years gained and DALYs averted by the study arms relative to each other.

We will assess cost-effectiveness from a range of perspectives, including provider, the health system, and societal. The first primary cost-effectiveness endpoint will be the incremental cost per instance of SGA averted, matching the primary endpoint of the effectiveness evaluation. The second primary cost-effectiveness endpoint will be the incremental cost per DALY averted, based on the results of the decision-analytic modeling. While this outcome will require additional assumptions beyond the data collected in the study, it is used for many cost-effectiveness analyses of health interventions, allowing us to compare cost-effectiveness results to other health interventions and established cost-effectiveness benchmarks [39]. Secondary cost-effectiveness endpoints will be the incremental cost per adverse pregnancy outcome averted (the union of adverse outcomes in Table 3) and the incremental cost per life year saved (based on the results of the decision-analytic modeling). We will use established methods to describe uncertainty in cost-effectiveness results [40].

Qualitative analysis

Qualitative data collected from the FGDs and IDIs will be analyzed using thematic analysis. The thematic analytical approach allows for the systematic identification and analysis of patterns and themes within a qualitative dataset [41]. This inductive process begins as soon as data (i.e., interview transcripts and field notes from the FGDs and IDIs) become available. We will begin with an “open coding” process where the qualitative data are explored without making prior assumptions about what might be discovered. Dominant, recurring themes will be identified through the review of transcripts and field notes, and we will develop a thematic framework iteratively [42]. Salient concepts will then be coded, and all transcripts will be imported for analysis by labeling phrases within the transcripts. A random selection of the transcripts will be independently reviewed by two researchers in the study team to cross-check codes, ensure consistency, and maintain quality assurance.

The emerging trends from the qualitative coding will be critically analyzed to ensure that the emerging themes are relevant to the research objectives [43], namely: household food practices and beliefs, diet during pregnancy and lactation, availability and access to nutritious foods, antenatal care-seeking behaviors, and supplement-related themes including experience with product taste, product preparation, levels and barriers of adherence and continued use, dynamics of intra-household sharing, and interface of the BEP products with local food environments, and the healthcare providers’ perspectives on sustainability, buy-in, and scale-up. We will conduct the coding and analyses of the qualitative data using the Dedoose analytic software (Version 8.2.32; SocioCultural Research Consultants, 2016).

Training, standardization, and monitoring

Before study implementation, all study staff will be trained in the study objectives, intervention design, and good clinical practice. Study staff will also undergo intensive training in study processes such as consenting, anthropometry measurements (including the monthly evaluation of GWG), BEP supplements provision, and pregnancy outcomes assessment. Inter- and intra-observer standardization exercises for all anthropometry measures will be conducted and repeated every 6 months during the study implementation period. Weighing scales, length board, MUAC tape, and digital blood pressure monitor will be routinely calibrated using standard measurement equipment.

The adverse events (AEs) and serious adverse events (SAEs) considered for this study will include miscarriage, stillbirth, maternal deaths, infant deaths, maternal hospitalization (for any reason except hospitalization for childbirth), and infant hospitalization (for any reason). The occurrence of these AEs/SAEs will be ascertained through the routinely collected data. The AEs and SAEs will be reported to IRB as part of the progress reports annually. The study principal investigators are responsible for reporting these AEs/SAEs to the institutional review boards in the United States and Ethiopia at the time of annual review. In the unlikely event of harm resulting from participation in this trial, we will review these incidents to determine if they are related to trial participation. In cases where harm is determined to be directly attributed to trial participation, we will facilitate the necessary medical care for the participants.

We formulated a technical advisory group (TAG) consisting of four senior experts on epidemiology and nutrition. Prior to the start of the study, we consulted with the TAG for their technical input on various study design features including the definitions of inadequate and excessive GWG. A clinical trial expert external to the study will be invited to conduct external trial auditing on a regular basis.

Ethical considerations

This study was approved by the Institutional Review Board at Harvard T.H. Chan School of Public Health (Protocol #: IRB22-1245) and the Institutional Ethical Review Board of Addis Continental Institute of Public Health (Ref: ACIPH/IRB/008/2022). The study is registered at ClinicalTrials.gov (registration number: NCT06125860). Important protocol modifications (e.g., changes to eligibility criteria, outcomes, analyses) will be communicated to these Institutional Review Boards and the clinical trial registry at ClinicalTrials.gov.

Written individual informed consent in the local language of Amharic will be obtained by study nurses from eligible pregnant women before their inclusion in the study. The form will be read aloud to participants who cannot read. For those who cannot sign, a thumb imprint will be taken and witnessed by an impartial literate witness.