ReMAPP is a prospective cohort study in five countries. It is embedded within the Pregnancy Risk, Infant Surveillance, and Measurement Alliance (PRISMA) Maternal and Newborn Health study: a longitudinal, open cohort study that seeks to evaluate pregnancy risk and health and development for women and infants [32]. The population for ReMAPP is nested in the PRISMA study cohort. As illustrated in Fig 1, further selection will be done from the primary cohort of enrolled ReMAPP participants to identify a clinically ‘healthy’ analytical subcohort and cross-sectional sub-samples for etiology assessment. Data from the primary cohort, analytical subcohort, and cross-sectional sample will be used to achieve aims one, two, and three, respectively. Enrollment for ReMAPP was initiated on a rolling basis by site and began between September 2022 and December 2023 (22 September 2022 in Pakistan, 15 December 2022 in Zambia, 28 December 2022 in Ghana, 14 April 2023 in Kenya, 20 June 2023 in India (Vellore), and 12 December 2023 in India (Hodal). Enrollment is expected to be completed by April 2025, with all follow up and data collection concluding across sites by December 2025. Results will be disseminated by study aim; it is anticipated that all results will be shared by early 2026.

This study focuses on countries in South Asia and sub-Saharan Africa with high anemia prevalence [33]. There are six research sites: Kintampo, Ghana; Kisumu and Siaya, Kenya; Lusaka, Zambia; Karachi, Pakistan; Vellore, India; and Hodal, India. Each site has identified and mapped geographic catchment areas, which have at least 1500 deliveries per year.

Participants eligible for the ReMAPP study are pregnant women who live within the catchment area, are <45 years and meet country-specific minimum age requirements (Ghana: 15 years of age; Kenya: 18 years of age or emancipated minors; Pakistan: 15 years of age or emancipated minors; Zambia: 15 years of age; India: 18 years of age), have a viable intrauterine pregnancy, are less than 20 weeks gestation as verified via ultrasound, intend to stay in the study area until six weeks post-delivery, and who provide informed consent.

All participants enrolled in the PRISMA study will be invited to participate in the ReMAPP study if they meet eligibility criteria. Enrollment will continue until the target sample size (n = 1650 to 2000 per site) is achieved. These participants will form the ReMAPP primary cohort, and will undergo serial hemoglobin testing by complete blood count and be followed up at regular study visits through one year postpartum.

Upon enrollment, all ReMAPP participants will undergo screening based on predetermined criteria to obtain sociodemographic information, obstetric history, vital signs, anthropometric measurements, and laboratory results. If a participant meets the 21 screening criteria (Box 1), they will be included in the clinically ‘healthy’ analytical subcohort. Selected participants may also be excluded from the final ‘healthy’ analytical subcohort if any of the following occur during follow-up: multiple pregnancies previously not identified, severe conditions not initially evident including cancer, HIV, tuberculosis, or malaria, or severe pregnancy-related conditions requiring hospital admission including severe preeclampsia/eclampsia.

Among the participants screened, a cross-sectional sample of up to 300 participants (100 per trimester, both anemic and nonanemic) per site will be selected to undergo biomarker-intensive testing. Any method of population-representative sampling may be used within each trimester stratum, provided that selection occurs over the same time period to eliminate time bias.

Data will be collected using a harmonized set of forms developed for the PRISMA study [32]. Beyond the routine PRISMA study visit procedures, ReMAPP participants will provide biological specimens to undergo additional laboratory testing (Table 2). For hemoglobin, the primary measure of interest, participants will be serially assessed at less than 20 weeks, at 20 weeks, 28 weeks, and 36 weeks gestation and at six weeks postpartum by complete blood count using a venous blood sample and a five-part differential automated hematology analyzer. Additional testing will be done at enrollment, as a part of the screening laboratory tests for the clinically ‘healthy’ subcohort. Biomarker-intensive testing will be done for the sample of 300 participants (100 per trimester), either during their enrollment visit (gestational age less than 14 weeks), 20-week, or 32-week ANC visit. Methods and timing of laboratory assessments will be standardized across sites to ensure comparability and quality.

Each site laboratory will adhere to the Laboratory Quality Assurance and Quality Control Standard Operating Procedures manual developed specifically for this study. For full blood count measurement, all participating labs will be enrolled onto the United Kingdom National External Quality Assurance Scheme for Hematology for monthly quality assurance [35]. For full blood count and all other analytes, the College of American Pathologists’ External Quality Assessment Program is required [36].

Each site will enroll 1,650 to 2,000 participants in the ReMAPP study, to achieve an overall study sample of 12,000 participants across the six sites. The sample size was selected assuming that about 30% (n = 600) will meet the eligibility criteria for the ‘healthy’ subcohort. This sample size will also capture sufficient cases of the primary maternal and newborn outcomes of interest, based on previous regional prevalence estimates, in order to conduct a pooled decision limits analysis with at least 90% power to estimate association between anemia and each outcome.

Our power speculation is based on the distribution of hemoglobin values in pregnancy by gestational age presented in the INTERGROWTH-21st study [18]. For the ‘healthy’ analytical subcohort, a sample size of 600 participants per site allows for single-site estimation of hemoglobin thresholds for any maternal anemia (i.e., 5th percentile) and severe maternal anemia (i.e., 2.5th percentile). As illustrated in Fig 2, there will be sufficient precision to distinguish between these two percentiles without overlapping 95% confidence intervals. If sites are unable to identify 600 participants for the healthy cohort, a minimum sample size of 300 participants will be sufficient for single-site hemoglobin threshold estimation.

For the cross-sectional sample, n = 300 was determined to minimize the burden of additional specimen collection while allowing for exploratory pooled analyses of anemia etiology. The calculation is based on the following assumptions: one third of the participants have a certain contributing factor (e.g., the prevalence of vitamin A insufficiency); 30% of participants without the factor have anemia; and 50% with the factor have anemia. At the two-sided significance level of 95% and 80% statistical power, we can detect a prevalence ratio of 1.7 with a sample size of between 240 and 270 [37]. We inflated the sample to 300 to account for the fact that some contributing factors are less prevalent in certain study sites.

Participants will be collaboratively involved at multiple phases of the study. During initial protocol workshops, investigators carefully evaluated the additional burden of study participation; for this reason, the size of the biomarker-intensive sample was kept to a minimum and invasive blood draws limited to select visit timepoints. As a part of the translation and validation process for questionnaires to measure fatigue, patients’ experiences were centered and informed scale validation procedures. Participants’ health priorities and experiences will also be explored in corresponding qualitative research about maternal morbidity. Upon completion of the study, we intend to share the main findings with participants and community members via appropriate local dissemination methods. Finally, we will ensure that participants’ contributions to this research are acknowledged in any subsequent reports, presentations, and publications.

This protocol, the informed consent documents and any subsequent modifications will be reviewed and approved by the relevant institutional review board (IRB) and ethics review committee (ERC) responsible for oversight of the study at each site. IRB approvals for this research were received from the following ERCs in each country: Ghana (Ghana Health Service, GHC-ERC 019/09/20 and the Kintampo Health Research Center Institutional Ethics Committee, KHRCIEC/2020-17), Kenya (KEMRI Scientific and Ethics Review Unit, 04-10-358-4166), Zambia (University of Zambia Biomedical Research Ethics Committee, IRB00001131 of IORG0000774; University of North Carolina at Chapel Hill Biomedical IRB, Study #14-2113), India (Office of Research, Christian Medical College, Vellore, India Ethics Committee, IRB No. 14553; Ethics Review Committee, Society for Applied Studies, New Delhi, SAS/ERC/ReMAPP Study/2022, Department of Health Research, EC/NEW/INST/2022/DL/0140), Pakistan (National Institutes of Health—Health Research Institute, National Bioethics Committee Ref: No.4-87/NBC-962/23/593 and Aga Khan University Ethics Review Committee, 2022-7197-21350), and the United States (Columbia University IRB-AAAU7504; The George Washington University IRB NCR224396; Harvard University IRB23-1093; The University of Alabama at Birmingham IRB-300013081). The rights and safety of all study participants will be protected. Written informed consent, assent and parental consent will be sought from all study participants as appropriate prior to enrollment. Site-specific unique identification numbers will be issued to participants instead of names to protect their identity. A limited data set will be stored in a secure cloud server where only trained study staff will have credentials for access. Study participants with abnormal biochemical results will be referred for further clinical assessment and management according to study-specific standard operating procedures.

Aim 1 will establish trimester-specific and postpartum hemoglobin clinical decision limits for any anemia based on associations with adverse maternal, fetal, and infant outcomes (Table 3). For this analysis, hemoglobin and outcome data points from all participants will be pooled. Nonparametric relationships between risks of adverse events and hemoglobin levels will be plotted using real data [38]. A series of candidate thresholds in hemoglobin values will be considered and disease risks before and after each threshold will be compared. Thresholds with clinically meaningful differences in the disease risks and statistically sound significance levels will be selected. Clinical decision limits for overall adverse events will be reported, as well as disease-specific clinical decision limits.

Aim 2 will establish gestational-age-specific hemoglobin reference limits for any, mild, moderate, or severe anemia in a reference population (i.e., clinically ‘healthy’ subcohort) during pregnancy and within 42 days postpartum [41,42]. This analysis will include hemoglobin data points from all participants eligible for the clinically ‘healthy’ subcohort. Fractional polynomial regression (FPR) will be used to model and visualize the hemoglobin changing curves of different percentiles against gestational age [18]. Sensitivity analysis of the percentile curves predicted by FPR models will be performed excluding one site at a time. If the hemoglobin distributions are similar across sites and between-site between is less than within-site variance, we will pool the data; if not, analyses will remain stratified by site.

Hemoglobin levels below the 5th or 2.5th percentiles at the corresponding time points during pregnancy are typically considered abnormally low and likely indicative of anemia. Therefore, the reference limits are defined as tail percentiles of hemoglobin levels in the reference population. Tail percentiles, such as 5th or 2.5th percentiles, will be modeled as a smooth polynomial function of gestational weeks using hemoglobin levels. With the final validated model, we will estimate the 2.5th, 5th, 95th, and 97.5th percentiles predicted from the FPR curves at specific gestational timepoints from 14 to 40 weeks and postpartum.

Aim 3 will describe the underlying contributing factors of anemia in pregnancy among a sample of up to 1,800 participants (up to 300 per site), with approximately equal numbers in each trimester. Analyses will be performed on pooled, regional, and site-specific data to identify shared and unique risk factors for anemia. We will quantify the association between contributing factors and anemia (defined using the WHO cutoffs and study-established reference and decision limit thresholds) by calculating relative risks from a generalized linear log-binomial model with robust standard errors. If these models fail to converge, we will use modified Poisson estimates, which produce valid but less efficient estimates of the log-binomial model. We will run separate models for proximal, medial, and distal risk factors (Fig 3). Distal risk factors will be included as potential confounders to proximal and medial models. We will estimate the population attributable risk fraction using prevalence of the exposure/risk factor and the relative risk of anemia in the exposed versus unexposed group. Partial population attributable risk will be calculated by fitting multivariate models with log link for anemia adjusting for multiple risk factors. For contributing factors significantly associated with risks of anemia, we will further report the change in continuous hemoglobin concentrations associated with the factor.

The data generated in the course of the study will be reviewed on a regular basis for quality per our established Data Quality Assurance and Quality Control Standard Operating Procedures monthly. Following the end of the study, meetings will be organized at each study site to share results with community members, local authorities, and study participants. Findings will also be shared with local and international stakeholders, including the Bill & Melinda Gates Foundation, WHO, CDC, United States Agency for International Development, and Ministries of Health in each participating country. Abstracts will be developed for dissemination through both local and international scientific conferences and publication in peer-reviewed journals.