A multicountry randomized controlled trial of comprehensive maternal nutrition supplementation initiated before conception: the Women First trial

Abstract

Background: Reported benefits of maternal nutrition supplements commenced during pregnancy in low-resource populations have typically been quite limited.

Objectives: This study tested the effects on newborn size, especially length, of commencing nutrition supplements for women in low-resource populations ≥3 mo before conception (Arm 1), compared with the same supplement commenced late in the first trimester of pregnancy (Arm 2) or not at all (control Arm 3).

Methods: Women First was a 3-arm individualized randomized controlled trial (RCT). The intervention was a lipid-based micronutrient supplement; a protein-energy supplement was also provided if maternal body mass index (kg/m2) was <20 or gestational weight gain was less than recommendations. Study sites were in rural locations of the Democratic Republic of the Congo (DRC), Guatemala, India, and Pakistan. The primary outcome was length-for-age z score (LAZ), with all anthropometry obtained <48 h post delivery. Because gestational ages were unavailable in DRC, outcomes were determined for all 4 sites from WHO newborn standards (non-gestational-age-adjusted, NGAA) as well as INTERGROWTH-21st fetal standards (3 sites, gestational age-adjusted, GAA).

Results: A total of 7387 nonpregnant women were randomly assigned, yielding 2451 births with NGAA primary outcomes and 1465 with GAA outcomes. Mean LAZ and other outcomes did not differ between Arm 1 and Arm 2 using either NGAA or GAA. Mean LAZ (NGAA) for Arm 1 was greater than for Arm 3 (effect size: +0.19; 95% CI: 0.08, 0.30, P = 0.0008). For GAA outcomes, rates of stunting and small-for-gestational-age were lower in Arm 1 than in Arm 3 (RR: 0.69; 95% CI: 0.49, 0.98, P = 0.0361 and RR: 0.78; 95% CI: 0.70, 0.88, P < 0.001, respectively). Rates of preterm birth did not differ among arms.

Conclusions: In low-resource populations, benefits on fetal growth-related birth outcomes were derived from nutrition supplements commenced before conception or late in the first trimester. This trial was registered at clinicaltrials.gov as NCT01883193.

FIGURE 1

CONSORT diagram. Overall screening, random assignment, and obtainment of primary outcome by treatment arm. ¹Percentage of those randomly assigned. Excludes women who became pregnant <3 mo into the study. The women who had eligible pregnancies may have had delivery data obtained or they may have exited the study before delivery. ²Primary outcome was obtained for live newborns with 3 length measurements taken within 48 h of delivery. Among women, primary outcome obtained from ≥1 infants of the woman. ³LAZ for birth length, based on actual birth length measured by 48 h of age, calculated using the expanded tables of the Child Growth Standards published by the WHO (30) that provide scores by day of measurement. The same standards were used to calculate the weight-for-age, head circumference-for-age, and BMI-for age z scores (WAZ, HCAZ, and BMIAZ). ⁴GA at birth is defined as the age at the time of the ultrasound based on the ultrasound plus time until birth if the ultrasound was done between 6 wk + 0 d and 13 wk + 6 d and the GA at birth was between 24 wk + 0 d and 42 wk + 6 d. If the ultrasound was not conducted during the GA previously mentioned, then the GA at birth is missing. ⁵LAZ, WAZ, HCAZ, and weight to length-ratio-for age (WLRAZ) z scores and percentiles based on measurements within 48 h of delivery are calculated using the INTERGROWTH-21st International Standards for Newborn Size (32) and International Standards for Newborn Size for Very Preterm Infants (33), which provide z scores by sex and GA at birth for infants between 33 wk + 0 d and 42 wk + 6 d GA at birth and between 24 wk + 0 d and 32 wk + 6 d GA at birth, respectively. CONSORT, Consolidated Standards of Reporting Trials; GA, gestational age; LAZ, length-for-age z score; MTP, medical termination of pregnancy.

FIGURE 2

Mean ± SEM length-, weight-, and BMI-for-age z scores for each site by treatment arm for non-gestational-age-adjusted data. P values comparing mean z scores for pairwise comparisons of the treatment effect were obtained from linear models for the outcome of interest, adjusted for cluster. Pairwise comparisons of treatment arms within an individual site are evaluated at a significance level of α = 0.00625. *Mean length-for-age z score for Arm 1 differs from Arm 3 in Pak (P = 0.0057) and in DRC (P = 0.0042). **Arm 1 differs from Arm 2 in Guat (P = 0.0044). Sample size for each arm within a site ranged as follows: n = 183–199 (DRC), n = 201–236 (Pak), n = 199–200 (Ind), and n = 189–216 (Guat). DRC, Democratic Republic of the Congo; Guat, Guatemala; Ind, India; Pak, Pakistan.

FIGURE 3

Effect sizes and 95% CIs for pairwise comparisons of the difference in mean length-for-age z scores at birth by treatment arm, by site and for combined sites. Effect sizes with corresponding 95% CIs obtained from linear model, adjusted for clusters. The combined site analysis is adjusted for country and cluster-nested within country. GAA data (31) presented for all individual sites except for the DRC which is not adjusted for gestational age (30). “All” represents combined data from the 3 sites with GAA data available (Pak, Ind, and Guat; n = 459–517 per arm). Sample size for each arm within a site ranged as follows: n = 141–160 (Pak), n = 158–184 (Ind), n = 156–177 (Guat), and n = 183–199 (DRC). DRC, Democratic Republic of the Congo; GAA, gestational-age adjusted; Guat, Guatemala; Ind, India; Pak, Pakistan.

FIGURE 4

RRs and 95% CIs for pairwise comparisons of SGA by treatment arm, by site, and for combined sites (31). RRs with corresponding 95% CIs comparing proportions of SGA for the pairwise comparisons are obtained from generalized linear models with generalized estimating equations to estimate parameters while controlling for cluster correlations. For combined data, models are adjusted for country. Women First newborn measurements based on those with gestational age data. "All" comprised combined data from Pak, Ind, and Guat. Sample sizes for each arm within a site ranged as follows: n = 141–160 (Pak), n = 158–184 (Ind), n = 156–177 (Guat), and n = 459–517 (All). Guat, Guatemala; Ind, India; Pak, Pakistan; SGA, small for gestational age.