Antenatal multiple micronutrient supplements versus iron‐folic acid supplements and birth outcomes: Analysis by gestational age assessment method

Abstract Meta‐analyses consistently have found that antenatal multiple micronutrient supplementation (MMS) compared with iron and folic acid (IFA) alone reduce adverse birth outcomes. In 2020, the World Health Organization (WHO) placed a conditional recommendation for MMS and requested additional trials using ultrasounds to establish gestational age, because the evidence on low birthweight (LBW), preterm birth and small for gestational age (SGA) was considered inconsistent. We conducted meta‐analyses to determine if the effects of MMS on LBW, preterm birth and SGA differed by gestational age assessment method. Using data from the 16 trials in the WHO analyses, we calculated the effect estimates of MMS versus IFA on birth outcomes (generic inverse variance method and random effects model) stratified by method of gestational age assessment: ultrasound, prospective collection of the date of last menstrual period (LMP) and confirmation of pregnancy by urine test and recall of LMP. The effects of MMS versus IFA on birthweight, preterm birth and SGA appeared consistent across subgroups with no evidence of subgroup differences (p > 0.05). When limited to the seven trials that used ultrasound, the beneficial effects of MMS were demonstrated: risk ratios of 0.87 (95% confidence interval [CI] 0.78–0.97) for LBW, 0.90 (95% CI, 0.79–1.03) for preterm birth and 0.9 (95% CI, 0.83–0.99) for SGA. Sensitivity analyses indicated consistency in the results. These results, together with recent analyses demonstrating comparable effects of MMS (vs. IFA) on maternal anaemia outcomes, strengthen the evidence to support a transition from IFA to MMS programmes in low‐ and middle‐income countries.


| INTRODUCTION
Micronutrient deficiencies are highly prevalent, affecting two in three women of reproductive age worldwide (Stevens et al., 2022). During pregnancy, the prevalence is likely to be even higher, as requirements for several micronutrients are increased by up to 50% to accommodate maternal and foetal demands (Gernand, Schulze, et al., 2016).
The deleterious consequences of micronutrient deficiencies during this critical stage of life and development include low birthweight (LBW), preterm birth, being born small for gestational age (SGA), perinatal mortality, maternal mortality, maternal and child cognitive impairment, premature rupture of membranes, insufficient gestational weight gain and congenital anomalies (Bourassa et al., 2019;Gernand, Schulze, et al., 2016).
Antenatal multiple micronutrient supplements (MMS)-containing 13-15 micronutrients including iron and folic acid (IFA)-are designed to reduce the typically wide gap between the requirements for these essential nutrients and average intakes during pregnancy in low-and middle-income countries (LMICs) (Bourassa et al., 2019;World Health Organization [WHO] et al., 1999). MMS are cost-effective for reducing adverse birth outcomes in LMICs Keats et al., 2019Keats et al., , 2022Smith et al., 2017).
In 2016, the antenatal micronutrient intervention recommended by the WHO for all pregnant women was daily elemental IFA supplements (30-60 mg), while MMS was not recommended (WHO, 2016).
Considering the increasing evidence on MMS that emerged subsequently, in 2020 WHO updated the recommendation on using MMS during pregnancy, stating that this intervention was recommended 'in the context of rigorous research' (WHO, 2020). This conditional recommendation acknowledged the benefits of MMS but stipulated a need for additional research.
In describing the additional research needed, WHO noted that 'Evidence on the effects of MMS on the component parts of LBW is inconsistent and controlled clinical trials are needed in which early pregnancy ultrasound is used to establish gestational age with certainty, to understand where the effect on LBW is derived' (Tuncalp et al., 2020).
Given the desire of the WHO to have results for the effects of MMS from trials that assessed gestational age by ultrasound (WHO, 2020), we conducted additional analyses of the large body of existing data that informed the WHO guidelines. In particular, we aimed to (1) describe methods (ultrasound or other methods) used to determine gestational age in all the trials included for analysis in the 2020 WHO guidelines (WHO, 2020) and (2) determine whether the effect of MMS versus IFA on the three birth outcomes of interest (LBW, preterm birth and SGA) differed according to the method used to assess gestational age in each trial.

| METHODS
When examining the effects of antenatal MMS compared with IFA, the 2020 WHO guidelines (WHO, 2020) relied on data from 16 selected trials (out of 20 trials) of a Cochrane systematic review (Keats et al., 2019). The difference between the 20 trials included in the Cochrane review and the 16 trials included in the 2020 WHO guidelines is due to the slightly different inclusion criteria regarding the intervention and control considered for each analysis. The Cochrane review (Keats et al., 2019) included trials that compared MMS (containing at least three micronutrients) to iron (with or without folic acid) or placebo. Meanwhile, the guideline development group (WHO, 2020) decided to include trials that compared MMS with 13-15 micronutrients to IFA. As two trials compared MMS with placebo, one trial evaluated a supplement with eight micronutrients plus IFA and one trial did not provide folic acid to the control group, only 16 trials contributed data to the updated 2020 WHO guidelines (WHO, 2020). Of these 16 trials, 6 evaluated supplements with 13 or 14 micronutrients and 10 evaluated the well-established United Nations International Multiple Micronutrient Antenatal Preparation (UNIMMAP) supplements that included 15 micronutrients (WHO et al., 1999). Thus, the WHO analyses were presented in two ways: comparison 1, which compared MMS with 13-15 micronutrients against IFA, and comparison 2, which compared the UNIMMAP supplements against IFA.
For all the 16 trials included in the analyses of the 2020 WHO guidelines, we extracted data on the method used to determine gestational age when women were enrolled in the trial. When this was unclear, we contacted the authors via e-mail to clarify how gestational age was assessed in their study.
We categorized the methods used for gestational age assessment into: 1. Group 1-gestational age assessed by ultrasound.
2. Group 2-gestational age assessed by prospective collection of the date (1st day) of the last menstrual period (LMP) and confirmation of pregnancy by a urine test.

Key messages
• World Health Organization (WHO) has a conditional recommendation for antenatal multiple micronutrient supplementation (MMS) noting ultrasound to assess gestational age should be used in future trials.
• These meta-analyses of the 16 trials informing the WHO guidelines showed that the benefits of MMS versus iron and folic acid on low birthweight, preterm birth and small-for-gestational-age were not different in subgroups of trials using different gestational age assessment methods.
• In trials that performed ultrasounds for gestational age assessment, the benefits of MMS on birth outcomes were demonstrated.
• These analyses strengthen the evidence for the transition to MMS in antenatal care programmes in low-and middle-income countries.
3. Group 3-gestational age assessed by a recall of the date (1st day) of the LMP.
We then conducted meta-analyses of the effect of MMS versus IFA on LBW (birthweight less than 2500 g), preterm birth (births before 37 weeks of gestation) and SGA (as defined by the authors of the trials [Keats et al., 2019]), stratified by different methods of gestational age assessment.
For this purpose, we extracted the data on the effect estimates (log risk ratio, standard error) in each study arm for each trial on the three outcomes of interest, and then calculated the pooled effect of MMS versus IFA for each outcome: overall, stratified into three subgroups (groups 1, 2 and 3), and stratified into two subgroups (combining groups 1 + 2, i.e., the two best methods, vs. group 3) of gestational age assessment.
Subsequently, we conducted a sensitivity analysis limited to the trials that used UNIMMAP supplements in the intervention arm, following the same trial selection as used for the WHO guidelines (comparison 2, UNIMMAP vs. IFA, pp. 52-53). We repeated the same sensitivity analyses, but added the West trial (West et al., 2014). cluster-randomized trials, using the generic inverse variance methods, as per the methodology used for the 2020 WHO guidelines (WHO, 2020). Subgroup differences were assessed with the χ 2 test for heterogeneity across subgroups, and a p value for heterogeneity of <0.05 was considered statistically significant.

| RESULTS
Supporting Information: Appendix Table 1 presents the methods used for gestational age assessment in each of the 16 trials, including the evidence used to support the classification of each study into one of the three groups.
Ultrasound was the most frequent method used to assess gestational age, reported in 7 of the 16 trials (44%) included for analysis (Adu-Afarwuah et al., 2015;Ashorn et al., 2015;Bhutta et al., 2009;Johnson et al., 2017;Osrin et al., 2005;Persson et al., 2012;Roberfroid et al., 2008). The mean gestational age at enrollment, and the time of the ultrasound scan, varied between 9 and 17 weeks. Four trials (25%) used the date (1st day) of the LMP collected prospectively and confirmation of pregnancy by urine test to determine gestational age (Christian et al., 2003;West et al., 2014;Zagre et al., 2007;Zeng et al., 2008). This method consisted of a pregnancy surveillance system that involved regular visits (e.g., monthly) to the homes of women of reproductive age to ask about LMP dates and conduct urine-based testing to confirm pregnancies, allowing the identification of pregnancies early in gestation (Gernand, Paul, et al., 2016). Five trials (31%) based gestational age on maternal recall of date (1st day) of LMP (Friis et al., 2004;Kaestel et al., 2005;Liu et al., 2013;Shankar et al., 2008;Sunawang et al., 2009).
A summary of the meta-analysis results, including overall analyses, subgroup analyses with three groups and subgroup analyses with two groups, is presented in Table   The results of the sensitivity analyses limited to the trials that provided UNIMMAP supplements in the intervention arm can be found in Supporting Information: Appendix  [West et al., 2014]).
Results of these sensitivity analyses were consistent with the overall findings, with no significant differences across the subgroups of the gestational age assessment method.

| DISCUSSION
The present analyses, based on the data extracted from the 16 trials used for the 2020 WHO guidelines (WHO, 2020), explored the impact of using ultrasound or other methods of gestational age assessment on the effect estimates of MMS versus IFA on the three birth outcomes of interest (LBW, preterm and SGA). We found that: 1. There were three main methods used to assess gestational age, with ultrasound at study enrollment being the most frequently used method.
2. The results of the overall and subgroup analyses for the three outcomes of interest are consistent, pointing towards a risk reduction, across the three methods used to assess gestational age.
3. When the evidence is limited to the trials that used ultrasoundthe gold standard for gestational age dating-the beneficial

| Gestational assessment methods: strengths, limitations and impact on birth outcomes
The most accurate method to establish or confirm gestational age is an ultrasound measurement of the fetus in the first trimester, up to and including 13 weeks and 6 days of gestation; gestational age assessment based on measurement of the crown-rump length has an accuracy of 5-7 days (Committee on Obstetric Practice American

Institute of Ultrasound in Medicine Society for Maternal-Fetal
Medicine, 2017). Second-trimester ultrasound scans conducted T A B L E 1 Summary of results: subgroup analyses of the effect of multiple micronutrient supplementation versus iron and folic acid supplementation on low birthweight, preterm and small for gestational age births, according to method used for gestational age assessment. provision of a calendar where women were instructed to mark the date of the beginning of their menses) and conduct urine-based pregnancy tests (Christian et al., 2003). This method was validated against crown-rump length measured in early pregnancy by ultrasound, showing that gestational age varied by a mean of only 2.8 days (Gernand, Paul, et al., 2016); as such, it can be considered a valid measure for estimating gestational age and preterm birth rates (Gernand, Paul, et al., 2016).
Still, the most widely available method for gestational dating in LMIC is the self-reported or recall (i.e., retrospective collection) of the 1st day of the LMP. This method is based on the assumption that the average menstrual cycle is 28 days in length with ovulation occurring on Day 14, and pregnancy is assumed to last 280 days from the 1st day of the LMP. It has varying levels of accuracy among different populations, and several limitations such as uncertainty of the date due to recall bias, variations in the timing of ovulation and bleeding that can occur for other reasons other than menses (Lynch & Zhang, 2007).
F I G U R E 1 Effect of MMS versus IFA on low birthweight: subgroup analysis by method of gestational age assessment (three groups). CI, confidence interval; GA, gestational assessment; IFA, iron and folic acid supplementation; LMP, (1st day of) last menstrual period; MMS, multiple micronutrient supplementation.
GOMES ET AL.
| 5 of 10 These limitations of the retrospective collection (recall) of the 1st day of the LMP can lead to more errors in the estimation of gestational age, which can affect the estimated prevalence of preterm birth or SGA. Assessment of birthweight is not dependent on gestational age measurements, so there may be less error in estimates of this outcome than is the case for preterm births or SGA.
However, it should be noted that errors in any method used to assess gestational age (or weight) were expected to be the same in both MMS and IFA groups of a masked randomized clinical trial. The potential nondifferential measurement error for outcomes (i.e., if errors are the same in both trial arms) would bias the results towards the null (Rothman et al., 2008 [Smith et al., 2017]) has the potential to benefit an estimated 2.2 million infants in LMIC annually. Moreover, an additional gain in weight or duration of pregnancy would likely be beneficial to many more infants even if it does not change their categorization as LBW, preterm or SGA. This is a significant and important benefit, in addition to the benefit conferred by IFA (the comparison group), as LBW is shown to be associated with increased mortality and morbidity (Gu et al., 2017;UNICEF, 2022).
The benefits of MMS for preterm births and SGA (observed when gestational age is assessed with the best method(s)) are equally important, as these two outcomes are known to increase mortality risk in and beyond the neonatal period (Katz et al., 2013).
Second, we note that several current WHO recommendations on ANC are based largely on reducing LBW including nutrition education on increasing daily energy and protein intake in undernourished populations, and restricting caffeine intake (WHO, 2016). Both of those interventions had little to no effect, and low or very low-quality evidence, on preterm birth and SGA. Daily oral IFA supplementation is also recommended to reduce LBW. Per the WHO guidelines the risk reduction of IFA versus placebo on LBW is not significant (RR, 0.84; 95% CI, 0.69-1.03) and was graded as low-certainty evidence, but was nonetheless interpreted as 'daily iron may reduce the risk of low-birth-weight neonates' (WHO, 2016). None of the recommendations that were based on reduction of LBW were conditional on additional research or requested evidence from studies with ultrasound-based gestational age assessment. In contrast, the risk reduction of MMS versus IFA on LBW is significant (overall analysis: RR, 0.88; 95% CI, 0.86-0.91) and was graded as high-certainty evidence (WHO, 2020).

| CONCLUSION
The present analyses, based on data from the trials used to inform the 2020 WHO guidelines, demonstrated consistent results regarding the effect of MMS versus IFA on birth outcomes F I G U R E 3 Effect of MMS versus IFA on small for gestational age: subgroup analysis by method of gestational age assessment (three groups). CI, confidence interval; GA, gestational assessment; IFA, iron and folic acid supplementation; LMP, (1st day of) last menstrual period; MMS, multiple micronutrient supplementation.
regardless of the method of gestational age assessment. The analyses also showed that 11 of those 16 controlled trials used valid methods to assess gestational age, and that the benefits of MMS (in comparison to IFA) in reducing the risk of adverse birth outcomes were more pronounced in those 11 trials-a finding made more relevant by new estimates for the burden of LBW globally.
These results provide new evidence to support the validity of findings from existing trials, challenging the need for new efficacy trials using ultrasound for gestational age assessment and providing further support for a transition from IFA to MMS programmes in LMICs. Future convenings of the WHO guideline development group should consider these findings, in addition to other recent analyses demonstrating comparable effects of MMS (vs. IFA) on maternal anaemia outcomes (Gomes et al., 2022), when updating recommendations related to using MMS during pregnancy.