Prenatal fortified balanced energy-Protein supplementation and birth outcomes in rural Burkina Faso: A randomized controlled efficacy trial

Background APUrov: iPdilnegasbecaolannficremdtheantaelrlgheya-dpirnogtleeivnel(sBaErePr)epsruepspenletemdeconrtrseicstlay:promising intervention to improve birth outcomes in low- and middle-income countries (LMICs); however, evidence is limited. We aimed to assess the efficacy of fortified BEP supplementation during pregnancy to improve birth outcomes, as compared to iron-folic acid (IFA) tablets, the standard of care. Methods and findings We conducted an individually randomized controlled efficacy trial (MIcronutriments pour la SAntéde la Mère et de l’Enfant [MISAME]-III) in 6 health center catchment areas in rural Burkina Faso. Pregnant women, aged 15 to 40 years with gestational age (GA) <21 completed weeks, were randomly assigned to receive either fortified BEP supplements and IFA (intervention) or IFA (control). Supplements were provided during home visits, and intake was supervised on a daily basis by trained village-based project workers. The primary outcome was prevalence of small-for-gestational age (SGA) and secondary outcomes included large-for-gestational age (LGA), low birth weight (LBW), preterm birth (PTB), gestational duration, birth weight, birth length, Rohrer's ponderal index, head circumference, thoracic circumference, arm circumference, fetal loss, and stillbirth. Statistical analyses followed the intention-to-treat (ITT) principle. From October 2019 to December 2020, 1,897 pregnant women were randomized (960 control and 937 intervention). The last child was born in August 2021, and birth anthropometry was analyzed from 1,708 pregnancies (872 control and 836 intervention). A total of 22 women were lost to follow-up in the control group and 27 women in the intervention group. BEP supplementation led to a mean 3.1 percentage points (pp) reduction in SGA with a 95% confidence interval (CI) of -7.39 to 1.16 (P = 0.151), indicating a wide range of plausible true treatment efficacy. Adjusting for prognostic factors of SGA, and conducting complete cases (1,659/1,708, 97%) and per-protocol analysis among women with an observed BEP adherence ≥75% (1,481/1,708, 87%), did not change the results. The intervention significantly improved the duration of gestation (+0.20 weeks, 95% CI 0.05 to 0.36, P = 0.010), birth weight (50.1 g, 8.11 to 92.0, P = 0.019), birth length (0.20 cm, 0.01 to 0.40, P = 0.044), thoracic circumference (0.20 cm, 0.04 to 0.37, P = 0.016), arm circumference (0.86 mm, 0.11 to 1.62, P = 0.025), and decreased LBW prevalence (-3.95 pp, -6.83 to -1.06, P = 0.007) as secondary outcomes measures. No differences in serious adverse events [SAEs; fetal loss (21 control and 26 intervention) and stillbirth (16 control and 17 intervention)] between the study groups were found. Key limitations are the nonblinded administration of supplements and the lack of information on other prognostic factors (e.g., infection, inflammation, stress, and physical activity) to determine to which extent these might have influenced the effect on nutrient availability and birth outcomes. Conclusions The MISAME-III trial did not provide evidence that fortified BEP supplementation is efficacious in reducing SGA prevalence. However, the intervention had a small positive effect on other birth outcomes. Additional maternal and biochemical outcomes need to be investigated to provide further evidence on the overall clinical relevance of BEP supplementation. Our research was reported using the Consolidated Standards of Reporting Trials (CONSORT) 2010 checklist (S1 CONSORT checklist) [17]. The MISAME-III protocol was published previously [18]. In brief, the study was a community-based, nonblinded individually randomized 2 × 2 factorial RCT, with directly observed daily supplement intake, conducted in the Houndé health district situated in the Hauts-Bassins region of Burkina Faso. The present manuscript details the primary and secondary birth outcomes only. The maternal and postnatal study outcomes will be reported separately. The study protocol was approved by the ethics committee of Ghent University Hospital in Belgium (B670201734334) and Centre Muraz in Burkina Faso (N°2018–22/MS/SG/CM/CEI). Women aged between 15 and 40 years and living in the study catchment villages were identified through a census in the study area (n = 10,165). A network of 142 trained village-based project workers visited all eligible women at their homes every 5 weeks to identify pregnancy early, by screening for self-reported amenorrhea. Potential cases were referred to the health center for a urinary pregnancy test. Once gestation was confirmed, the MISAME-III study purpose and procedures were explained in the local languages Mooré, Dioula, or Bwamu. Study eligibility criteria were (i) pregnancy confirmed by a urinary pregnancy test and ultrasound examination; and (ii) written informed consent. Exclusion criteria were (i) gestational age (GA) ≥21 completed weeks; (ii) women who planned to leave the area during their pregnancy or deliver outside the study area; and (iii) women allergic to peanuts. Study inclusion ran from October 30, 2019 to December 12, 2020, and the final child was born on August 7, 2021. The climate of the study setting is Sudano-Sahelian, with one dry season conventionally running from September to October to April. Malaria transmission is perennial, with seasonal variations. Regional health statistics from the 6 healthcare centers showed that 1.8% of adults suffered from hookworm or another parasitic infection and 0.9% from a sexually transmitted disease in 2021. The prevalence of pregnant women that suffered from a HIV infection was estimated to be 0.7% [19]. We randomly allocated women to the prenatal control or intervention group. A stratified permuted block randomization schedule was used to allocate women to the study groups. These blocks were generated per health center in blocks of 8 (4 control and 4 intervention) before the start of the study using Stata V.15.1 (Stata, College Station, Texas, United States of America) by a research analyst who was not involved in the study (FB). The allocation was coded with the letters A for the prenatal control and B for the prenatal intervention group and concealed in sequentially numbered sealed opaque envelopes by study employees, not in direct contact with participants. The study midwives, who enrolled the participants, assigned the women to the study groups by drawing a next sealed envelope with the letter code. Postrandomization, we excluded women without a confirmed pregnancy using the ultrasound examination, women with GA ≥21 completed weeks, and multifetal pregnancies [20]. It was not possible to blind the supplement allocation from study participants and trained village-based project workers because the products are readily identifiable. Outcome assessors (study physician, midwives, and field supervisors) were different from study collaborators (trained village-based project workers) who distributed the study supplements. However, given the nonblinded nature of the study, outcome assessors could have been aware of the study group allocation by asking the mother. Researchers who analyzed the data were not blinded. Women in the intervention group received a daily BEP supplement and IFA tablet for the duration of their pregnancy. In a formative study, the most preferred and suitable fortified BEP supplement was selected for administration in the MISAME-III efficacy trial [21,22]. The BEP supplement is an LNS in the form of an energy-dense peanut paste fortified with MMNs. The product is ready to consume, does not require a cold chain, and is highly stable with a long shelf life. On average, the 72g fortified BEP provided 393 kcal and consisted of 36% lipids, 20% protein, and 32% carbohydrates. Protein came from soy (61%), milk (25%), and peanut (15%). Furthermore, the MMN content covered at least the daily estimated average requirements of micronutrients for pregnant women, except for calcium, phosphorous, and magnesium, which were lower [23]. The complete nutritional composition of the fortified BEP is provided in Table 1 [24]. Women in the control group received daily only an IFA tablet (65 mg iron [form: FeH2O5S] and 400 μg folic acid [form: C19H19N7O6]; Sidhaant Life Sciences, Delhi, India), in accordance with the standard of care in Burkina Faso. aIngredients: vegetable oils (rapeseed, palm, and soy in varying proportions), defatted soy flour, skimmed milk powder, peanuts, sugar, maltodextrin, soy protein isolate, vitamin and mineral complex, and stabilizer (fully hydrogenated vegetable fat and mono- and diglycerides). b1 μg vitamin A RE = 3.333 IU vitamin A. c1 μg cholecalciferol = 40 IU vitamin D. d1 mg α-tocopherol = 2.22 IU vitamin E. BEP, balanced energy–protein; IU, international unit, RE, retinol equivalent. Both supplements were delivered on a daily basis and, to the extent possible, consumed under supervision by our trained village-based project workers during home visits. When women had a short and scheduled absence of home, supplements were given to the women in advance, and intake was considered nonobserved for the respective days. The trained village-based project workers also encouraged pregnant women to attend at least 4 ANC consultations. Study participants were designated as lost to follow-up if they moved from the study area, withdrew their participation, or if they could not be reached for more than 3 months. At enrollment (i.e., first ANC visit), pregnancy antecedents were collected and maternal height, weight, mid-upper arm circumference (MUAC), and hemoglobin (Hb) concentration were measured. Maternal height was measured to the nearest 1 cm with a ShorrBoard Infant/Child/Adult (Weigh and Measure, Olney, Maryland, USA) and weight to the nearest 100 g with a Seca 876 scale (Seca, Hanover, Maryland, USA); the accuracy of the scales was verified on a weekly basis. Maternal MUAC was measured to the nearest 1 mm with a Seca 212 measuring tape. Hb concentration was assessed again between 30 and 34 weeks of gestation (i.e., third ANC visit) using a HemoCue Hb 201+ (HemoCue, Ängelholm, Sweden); a calibration check was done weekly. Furthermore, a comprehensive socioeconomic and demographic questionnaire was administered at enrollment [18]. During each subsequent ANC visit, the study midwives measured all anthropometrics and screened for potential adverse events by checking blood pressure, urine protein, body temperature, edema, and fetal activity. Following Burkinabè guidelines, enrolled women received preventative malaria prophylaxis (3 oral doses of sulfadoxine–pyrimethamine) at the relevant ANC visits. Within 14 days of enrollment date, a woman’s pregnancy was confirmed by the study physician using a portable ultrasound (SonoSite M-Turbo, FUJIFILM SonoSite, Bothell, Washington, USA). GA was estimated by measuring crown-rump length (7 to 13 weeks) or by calculating the mean of 3 to 4 measurements: biparietal diameter, head circumference, abdominal circumference, and femur length (12 to 26 weeks) [25]. In addition to the ultrasound, the physician performed maternal subscapular and tricipital skinfold measurements in triplicate using a Harpenden caliper. At birth, anthropometry of all neonates was assessed in duplicate within the first 72 hours by study midwives (in practice, all were within 12 hours) at the health center. Newborn length was measured to the nearest 1 mm with a Seca 416 Infantometer, whereas birth weight was measured to the nearest 10 g with a Seca 384 scale. Newborn head circumference, thoracic circumference, and MUAC were measured to the nearest 1 mm with a Seca 212 measuring tape. If there was a large discrepancy between measures (e.g., >10 mm for birth length and >200 g for birth weight), a third measurement was taken. The average of the 2 closest measures were used for analyses. The accuracy and precision of anthropometric measurements were established regularly through standardization sessions organized by an expert in anthropometry [26]. MISAME-III data were collected using SurveySolutions (version 21.5) on tablets by the study physician and midwives and were transferred to a central server at Ghent University on a weekly basis. Questionnaire assignments were sent to the field team once a week including preloaded data collected at the previous ANC visit. We programmed generic validation codes to avoid the entry of implausible values and improve the quality of data collection in the field. Additionally, data quality checks and missing or inconsistent data were sent back to the field for revision every 2 weeks. The quality of ultrasound images and estimation of GA was checked for 10% of the examinations on a regular basis by an external gynecologist, using a quality checklist and scoring sheet. The MISAME-III trained village-based project workers collected data on the supplement adherence in both prenatal study groups using smartphones with computer-assisted person interviewing programmed in CSPro (version 7.3.1) on a daily basis. Six field supervisors performed monthly quality checks by verifying a trained village-based project worker’s work, at random, using a Lot Quality Assurance Sampling system [27]. All field staff received extensive training on all standard operating procedures (including Good Clinical Practices) and data collection tools before the start of the trial, with a dry run period of ±3 months for testing and evaluation in the field. The MISAME-III data collection forms are publicly available [28]. The primary study outcome was the prevalence of SGA, defined as the proportion of newborns with a birthweight below the 10th percentile of the International Fetal and Newborn Growth Consortium for the 21st Century [INTERGROWTH-21st] newborn size standards for a given GA at delivery [29]. The secondary outcomes of the prenatal BEP intervention were prevalence of large-for-gestational age (LGA; >90th percentile of INTERGROWTH-21st reference), LBW and PTB, gestation duration (weeks), birth weight (g), birth length (cm), Rohrer’s ponderal index at birth [weight/length3 (g/cm3) × 1,000], head circumference (cm), thoracic circumference (cm), arm circumference (mm), fetal loss (<28 weeks of gestation), and stillbirth (died ≥28 weeks of gestation, before or during birth). Fetal loss is further categorized in (i) <22 weeks of gestation; (ii) between ≥22 weeks and <28 weeks of gestation, according to the “Maternal BEP studies Harmonization Initiative”. Birth length and Rohrer’s ponderal index were measured to distinguish between short and thin newborns. To assess safety and serious adverse events (SAEs), all field staff was trained to recognize pregnancy related health issues to actively refer participants to the health center. All SAEs (i.e., miscarriage, stillbirth, and maternal death) were recorded on a case-by-case basis, and verbal autopsies were conducted for infant and/or maternal deaths that occurred outside a health center. All analyses were documented in the MISAME-III statistical analysis plan prior to analysis, which was validated on October 24, 2019 and published online on November 3, 2020 [28]. We calculated a sample size of 652 pregnant women per prenatal study group (total 1,304 participants) to detect a decrease in SGA of 7 percentage points (pp) between groups, with a power of 80% and a 2-sided significance level (i.e., type I error) of 5%, assuming a SGA prevalence of 32% (estimated from Huybregts and colleagues [15]). In MISAME-I [30] and MISAME-II [15], an approximately 26% loss of information occurred, due to a combination of abortions, miscarriages, stillbirths, multifetal pregnancies, out-migrations, maternal deaths, and incomplete data. Hence, the sample size was increased to 888 pregnant women per prenatal study group to accommodate for these potential losses (total 1,776 participants). Only singleton pregnancies were included in the analysis, as anthropometric measures and fetal loss at birth in multifetal pregnancies are often not primarily nutrition related. The primary analysis followed the intention-to-treat (ITT) principle. Therefore, we conducted multiple imputation by chained equations of missing outcome measures at birth under the “missing at random” assumption. A total of 50 imputations of missing values were done for the lost to follow-up cases to estimate the regression coefficients using the predictors maternal height, BMI, MUAC, Hb, age, GA and primiparity at baseline, and month of inclusion. Descriptive data are presented as percentages or means ± standard deviation (SD). Unadjusted and adjusted group differences were estimated by fitting linear regression models for continuous outcomes to estimate the mean group difference. For binary outcomes, linear probability models with a robust variance estimator were used to estimate risk difference in pp. All models contained health center and randomization block as fixed effect to account for any possible clustering by the study design. The adjusted models contained a priori defined known prognostic factors of study outcomes at birth, including maternal height (cm), BMI (kg/m2), MUAC (mm), Hb (g/dl), age (years) and GA at inclusion (weeks), and primiparity. Due to balanced baseline characteristics across prenatal study groups (i.e., < |2.5| pp difference), no other sociodemographic variables were adjusted for in sensitivity analyses. We conducted the following sensitivity analyses to assess the robustness of the primary findings: (i) complete case analysis (i.e., excluding women lost to follow-up); and (ii) per-protocol analysis restricting the intervention sample to women with BEP adherence of ≥75%. The strict adherence rate was calculated by dividing the total number of BEP supplements effectively taken under direct observation of a trained village-based project worker by the theoretical maximum number of prenatal BEP supplements, i.e., the number of days between study inclusion and delivery. Furthermore, as an exploratory analysis, we tested an interaction term between the intervention group and pre-specified subgroups, including maternal BMI (<18.5 kg/m2), MUAC (<23cm), Hb (<11g/dl), height (<155 cm), age (<20 years), completion of primary education, possible and probable prenatal depression (Edinburgh depression scale ≥10 points and ≥13 points), primiparity, household food insecurity (Household Food Insecurity Access Scale), newborn sex, season of delivery (lean season: June to September), and interpregnancy interval (<18 months). Last, we used the approach by Katz and colleagues [31] and Roberfroid and colleagues [30] to assess whether the treatment effect on birth weight and length was constant over percentiles of children’s birth weight, birth length, and maternal BMI distributions. In this method, differences (and CIs) in birth outcomes between intervention and control groups are estimated as nonlinear smooth functions of the percentiles of birth weight, birth length, or maternal BMI distributions. Statistical significance was set at P < 0.05 for all tests, except for exploratory interactions tests (P < 0.10) as specified in the statistical analysis plan. All analyses were conducted with Stata 17.1 (StataCorp). All SAEs reported by the study physician were evaluated on a continuous basis by the principal study investigators and reported to an independent Data and Safety Monitoring Board (DSMB) when considered related to the supplement. The DSMB (established prior to the start of the efficacy trial) comprised an endocrinologist, 2 pediatricians, a gynecologist, and a medical ethicist of both Belgian and Burkinabè nationalities. Two virtual DSMB meetings were organized, at month 9 and 20 after the start of the trial, to review the study progress and discuss all documented SAEs. The MISAME-III trial was registered on ClinicalTrials.gov (identifier: {"type":"clinical-trial","attrs":{"text":"NCT03533712","term_id":"NCT03533712"}}NCT03533712).