Effect of Multiple Micronutrient Supplementation on Pregnancy and Infant Outcomes: A Systematic Review

Authors

  • Usha Ramakrishnan,

    Corresponding author
    1. Hubert Department of Global Health, Rollins School of Public Health, Emory University
    2. Nutrition & Health Sciences Program, Graduate Division of Biological & Biomedical Sciences, Emory University, Atlanta, GA, USA
    Search for more papers by this author
  • Frederick Kobina Grant,

    1. Hubert Department of Global Health, Rollins School of Public Health, Emory University
    2. Nutrition & Health Sciences Program, Graduate Division of Biological & Biomedical Sciences, Emory University, Atlanta, GA, USA
    Search for more papers by this author
  • Tamar Goldenberg,

    1. Hubert Department of Global Health, Rollins School of Public Health, Emory University
    Search for more papers by this author
  • Vinh Bui,

    1. Nutrition & Health Sciences Program, Graduate Division of Biological & Biomedical Sciences, Emory University, Atlanta, GA, USA
    Search for more papers by this author
  • Aamer Imdad,

    1. Division of Women & Child Health, The Aga Khan University, Karachi, Pakistan
    Search for more papers by this author
  • Zulfiqar Ahmed Bhutta

    1. Division of Women & Child Health, The Aga Khan University, Karachi, Pakistan
    Search for more papers by this author

Usha Ramakrishnan, Hubert Department of Global Health, Rollins School of Public Health, Emory University, 1518 Clifton Road NE, Room 7009, Atlanta, GA 30322, USA. Email: uramakr@sph.emory.edu

Abstract

Supplementation with multiple micronutrients (MM) during pregnancy may result in improved pregnancy and infant outcomes. We conducted meta-analyses of randomised controlled trials that evaluated the effects of prenatal supplementation with MM (defined as containing at least five micronutrients and typically included iron or iron and folic acid). The outcomes of interest were low birthweight (<2500 g), birthweight, small-for-gestational age (SGA), gestational age, preterm birth (<37 weeks' gestation), stillbirth and neonatal death, maternal morbidity and mortality. We identified eligible studies through PubMed and EMBASE database searches. Meta-analyses were performed by pooling results for outcomes that were reported from more than one trial and sub-analyses were conducted to evaluate the effect of timing of intervention and amount of iron. We included published results from 16 trials in this review. Compared with control supplementation that was usually iron plus folic acid in most studies, MM supplementation resulted in a significant reduction in the incidence of low birthweight [pooled risk ratio (RR) 0.86; 95% confidence interval (CI) 0.81, 0.91] and SGA (pooled RR 0.83 [95% CI 0.73, 0.95]) and an increase in mean birthweight (weighted mean difference (WMD) 52.6 g [95% CI 43.2 g, 62.0 g]). There was no significant difference in the overall risk of preterm birth, stillbirth, and maternal or neonatal mortality, but we found an increased risk of neonatal death for the MM group compared with iron–folate in the subgroup of five trials that began the intervention after the first trimester (RR 1.38 [95% CI 1.05, 1.81]). None of the studies evaluated maternal morbidity. Compared with iron plus folic acid supplementation alone, prenatal maternal supplementation with MM resulted in a reduction in the incidence of low birthweight and SGA but increased risk of neonatal death in the subgroup of studies that began the intervention after the first trimester.

Maternal nutrition, before and during pregnancy, is very important for optimising pregnancy outcomes. Poor nutrition during pregnancy has been associated with adverse maternal and child outcomes such as increased risks of infertility, abortion, fetal intrauterine growth restriction and perinatal mortality.1 Besides inadequate intakes of energy and protein due to limited food availability and/or practices that restrict food intake during pregnancy, the diets of pregnant women in many developing countries are poor in several micronutrients such as iron, zinc and vitamin A, due to low intakes of animal foods and/or poor bioavailability of micronutrients due to the presence of inhibitors like phytates.2 To ensure optimal pregnancy outcomes, public health measures including adequate access to reproductive health services such as family planning, prenatal health care and nutritious food, prevention of gestational malarial infection, and provision of iron and folic acid supplements to all pregnant women have been recommended.3 However, adverse pregnancy outcomes, such as low birthweight, preterm delivery, and maternal and neonatal mortality, continue to be significant public health problems in many resource-poor environments.4,5 Because provision of food supplements to pregnant women is both logistically and economically challenging in resource-poor settings, several studies were undertaken to examine the potential benefit of providing supplements6 containing several micronutrients in addition to iron and folic acid.

The objective of this paper is to conduct a systematic review and meta-analysis of intervention trials that evaluated the benefits of multiple micronutrient (MM) supplements during pregnancy. Although several reviews and meta-analyses7–10 have been published on this topic, the findings are mixed. Fall et al. reported that MM supplementation was associated with an increase of ∼60 g in mean birthweight compared with controls who received iron and/or folic acid, as well as reductions in the prevalence of low birthweight and small-for-gestational age (SGA).10 This meta-analysis does not, however, include some recent trials. The most recent publication by Haider et al.6 that includes all trials is an update of an earlier Cochrane review and meta-analysis.7 In this review, we examine a wide range of maternal, neonatal and child health outcomes [low birthweight (<2500 g), birthweight, SGA, gestational age, preterm birth (<37 weeks' gestation), stillbirth and neonatal death, maternal mortality]; and dose–response relationships that relate to (i) the amount of iron received by the intervention and control groups, and (ii) the time of initiation.

Methods

Search strategy

We identified published studies using Medline through searches using the PubMed and EMBASE databases in April 2011. The search included studies with no language restriction from 1950 to July 2011 and key words such as ‘randomized control trials’ or ‘control trials’, ‘pregnancy’ or ‘pregnant’, ‘vitamin supplementation’ or ‘vitamin’‘folic acid’ or ‘folate’, and ‘iron supplementation’. The search was limited to ‘humans’ and ‘females’. Additional studies were identified through hand search of references from previous review articles and from personal communications.

Inclusion criteria

Inclusion criteria were as follows: (1) randomised, controlled intervention trial; (2) pregnant women (including non-symptomatic HIV-positive pregnant women); (3) studies that compared MM supplementation (≥5 micronutrients) with control (≤3 micronutrients including iron, folic acid and/or only one additional vitamin/mineral); and (4) reported findings for key outcomes of interest, namely low birthweight (<2500 g), SGA, birthweight, preterm birth (<37 weeks gestation), gestational age, stillbirth, neonatal death and/or maternal mortality.

Exclusion criteria

Exclusion criteria were as follows: (1) animal studies; (2) review articles, cross-sectional, case–control, cohort studies, commentary letter or editorial; (3) non-intervention studies; (4) non-healthy pregnant women (hospitalised patients, symptomatic HIV-positive women); (5) studies without MM interventions (<5 micronutrients, fortified food); (6) studies of preconceptual or periconceptual interventions; (7) non-accessible full texts.

Data abstraction and statistical analyses

Details of all included studies were abstracted by two independent co-authors (TG, VB) using the standard tables that were developed for this review and are based on what has been used previously.11 The details were verified by other co-authors (FG, UR) and effect sizes were calculated for each study outcome by dividing the difference between the mean values for the treatment (i.e. MM supplements) and control groups by the pooled SD. If the study reported only the SE or 95% CI, we calculated the SD from the SE or 95% CI. Pooled analyses were conducted where data were available from more than one study for an outcome and the results are presented as risk ratios (RR) for dichotomous variables or weighted mean difference (WMD) for continuous variables with 95% confidence intervals (95% CIs). In the case of trials with more than one comparison that shared the same treatment or control group, we selected the most appropriate comparison; we also conducted sensitivity analysis in which we used pooled estimates that combined the results for more than one arm that shared the same control or treatment group.12 The overall mean effect size and 95% CI across studies was estimated by a fixed or random effects model that used the weighted mean effect size for each study where the weight was the inverse of the intra-study variance. For all outcomes of interest, we conducted an overall comparison of all studies that provided MM (≥5 micronutrients) with control (≤3 micronutrients). To examine dose–response relationships, we carried out subanalysis for the following comparisons: MM (including 30 mg iron) vs. control (60 mg iron with or without folic acid); MM (including ≥60 mg iron) vs. control (≥60 mg iron with or without folic acid). We also compared subanalysis by timing of initiation of the intervention namely began intervention in the (i) first trimester (≤12 weeks gestation) and (ii) second or third trimester (>12 weeks gestation).

In all cases, we assessed heterogeneity of effect sizes using the chi-squared test of homogeneity; P < 0.05 was considered as evidence of heterogeneity. The I2 values were also examined and values >50% were deemed as exhibiting substantial statistical heterogeneity. Estimates based on the random effects model were used in cases with significant heterogeneity. We evaluated the presence of publication bias using funnel plots.13 The Begg's rank correlation test was used for statistical testing of funnel plot symmetry.14 Finally, we assessed the overall quality of evidence for the outcomes by the GRADE criteria.11,15

All statistical analyses were conducted using Review Manager Version 5.0 (Cochrane Collaboration, Copenhagen, Denmark). All statistical tests were 2-sided, with P < 0.05 reported as significant.

Results

The flow of the number of studies that were included in the meta-analyses is shown in Figure 1. We identified a total of 827 titles for searches on Medline database and other sources. An initial 52 publications were considered for inclusion out of which six were excluded after careful examination: two studies had neural tube defects as outcome, two with anaemic pregnant women as study participants, two studies either did not have a control group or were not blinded. The results from 46 eligible publications based on 16 trials were included in the analysis.

Figure 1.

Studies excluded and included in the meta-analyses of the effects of multiple micronutrient (MM) on pregnancy outcomes. RDA, recommended daily allowances; RCT, randomised controlled trial.

Study characteristics

A summary of the key characteristics of the 16 randomised controlled trials that are included in this review is provided in Table 1 (The detailed abstraction tables are available on request). The MM group received supplements containing ≥5 micronutrients (Table 2) including iron and the control group received ≤3 micronutrients that typically included iron alone or iron plus folic acid (IFA). Initiation of supplementation ranged between 4 and 30 weeks of gestation and continued daily until delivery. Nine of the included studies were from Asia,16–24 six from sub-Saharan Africa,25–30 one from Europe,31 and one from Central America.32 Only one trial was conducted among HIV-positive pregnant women in Tanzania in which women were assigned to one of four groups, namely MM with/or without vitamin A, vitamin A only and placebo. All women received IFA.25,26 One trial from Nepal had five groups; folic acid alone, folic acid + iron (60 mg), folic acid + zinc + iron (60 mg), MM (including 60 mg iron) and placebo, and all groups received vitamin A.17,18 For the overall pooled analysis, the comparison of interest was ‘MM vs. folic acid + iron’. The trial in Guinea-Bissau18 had two types of MM supplements containing ∼1 and 2 times the recommended daily allowances (RDA) of several micronutrients whereas the one in Bangladesh27 had two types of IFA supplements containing 30 and 60 mg iron; we selected the groups that received the MM supplement containing 1 RDA and the 30 mg of iron, respectively. The studies by Gupta et al. (60 mg iron) and Friis et al. (dose of iron unspecified) compared MM with placebo.20,27 Since all the study participants received iron and folic acid, we considered the placebo as the control group. The overall pooled estimates (Table 3) and results for stratified analysis by amount of iron (Table 4) and timing of initiation for the key outcomes of interest (Table 5) are described by outcome in the following sections.

Table 1.  Characteristics of included trials in the meta-analyses of multiple micronutrient (MM) supplementation during pregnancy
CountryAuthorsYearSample sizeIntervention receivedControl receivedOutcome(s) reportedComments
  1. MMN, multiple micronutrient; UNIMMAP, United Nations International Multiple Micronutrient Preparation; FA, folic acid; LBW, low birthweight; SGA, small-for-gestational age; BW, birthweight; PTD, preterm birth; GA, gestational age; SB, stillbirth; ND, neonatal death; MD, maternal death.

TanzaniaFawzi et al.2519981075MMN ± vitamin APlaceboLBW, SGA, PTD, SBAll participants (HIV-positive women) received iron (120 mg ferrous iron) and FA. Compared MM with vitamin A vs. placebo.
NepalChristian et al.17,1820034926MMN (60 mg Fe)Vitamin A, or FA + Fe (60 mg), or FA + Zn + Fe (60 mg)LBW, BW, SGA, PTD, SB, NDCompared MM vs. FA + iron. Iron was ferrous fumarate (60 mg). All participating women received vitamin A.
MexicoRamakrishnan et al.322003873MMN (60 mg Fe)Fe only (60 mg)LBW, SGACompared MM of 13 nutrients (including 60 mg ferrous sulfate as iron) with iron only (60 mg ferrous sulfate).
ZimbabweFriis et al.2720041669MMNPlaceboLBW, SGAAll participating women received FA and iron as part of routine prenatal care. Compared 13 MM with placebo.
FranceHininger et al.312004100MMN (no Fe)PlaceboLBWCompared 12 MM (without iron) vs. placebo. The supplement was iron free, due to the oxidative potential effects of iron.
Guinea-BissauKaestel et al.2820052100MMN (30 mg Fe)Fe + FA (60 mg Fe)LBW, BW, SB, ND, MDCompared two types of MM with control. Controls received FA and iron (60 mg ferrous fumarate). Dose of iron (ferrous fumarate) in the MM was 30 mg.
NepalOsrin et al.2120051200MMN (30 mg Fe)Fe + FA (60 mg Fe)LBW, BW, PTD, SB, NDControls received FA with iron (60 mg ferrous fumarate). Compared 15 MM (including 30 mg iron as ferrous fumarate) vs. control
IndiaGupta et al.202007200MMN (10 mg Fe)Placebo (calcium)LBW, SGA, GAAll participating women received FA with iron (60 mg). Compared 29 MM (with 10 mg iron) vs. placebo (calcium).
NigerZagre et al.2920072550MMNFe + FALBW, BWCompared 15 MM (30 mg iron as ferrous fumarate) vs. FA + iron (60 mg iron as ferrous fumarate).
TanzaniaFawzi et al.2620071078MMN ± vitamin AFe + FALBW, SGA, NDAll participants (HIV-negative women) received FA + iron (120 mg). Compared MM with/without vitamin A vs. vitamin A alone or placebo.
Burkina-FasoRoberfroid et al.3020081426UNIMMAP (30 mg Fe)Fe + FA (60 mg Fe)LBW, SGA, BW, PTD, GA, SB, NDCompared MM (UNIMMAP, containing 30 mg iron as ferrous fumarate) vs. FA + iron (60 mg ferrous fumarate).
IndonesiaShankar et al.22200831 290MMN (30 mg Fe)Fe + FA (30 mg Fe)LBW, BW, SB, ND, MDAll participants received FA + iron (30 mg ferrous fumarate). Compared MM with placebo.
ChinaZeng et al.2420085828MMN (30 mg Fe)FA, or FA + Fe (60 mg)LBW, SGA, BW, PTD, GA, SB, NDCompared MM (containing 15 nutrients, 30 mg iron as ferrous fumarate) vs. FA + iron (60 mg ferrous fumarate) and placebo (400 µg FA).
PakistanBhutta et al.1620092378MMNFe + FALBW, BW, GA, SB, NDCompared MM (UNIMMAP formulation, 30 mg iron as ferrous fumarate) vs. FA + iron (60 mg ferrous fumarate).
IndonesiaSunawang et al.232009843MMNFe + FALBW, BW, GA, SBCompared 15 MM (30 mg iron as ferrous fumarate) with FA + iron (60 mg ferrous fumarate).
BangladeshEneroth et al.1920104436MMN (30 mg Fe)Fe + FA (30 mg Fe)LBW, BWParticipants of food supplementation program. Compared MM (30 mg iron as ferrous fumarate) vs. FA + iron (30 mg ferrous fumarate), or FA + iron (60 mg ferrous fumarate).
Table 2.  Composition of the prenatal multiple micronutrient (MM) supplements used in the trials included in the meta-analysis
StudyIron (mg)Folic acid (µg)Zinc (mg)Vitamin A (IU)Thiamine (mg)Vitamin B6 (mg)Niacin (mg)Number of micronutrients
  • a

    One group received MM without vitamin A.

Fawzi et al, 1998258005000a202510010
Christian et al, 200317,18604003010001.62.22016
Ramakrishnan et al, 20033262.421512.921500.931.9415.513
Friis et al, 2004271530001.52.21713
Hininger et al, 200431200151.41512
Kaestel et al, 200528 (MM-1)30400158001.41.91815
Kaestel et al, 200528 (MM-2)308003016002.83.83615
Osrin et al, 200521304001526401.41.91815
Gupta et al, 200720101501525001.01.02029
Zagre et al, 200729304001526401.41.91815
Fawzi et al, 20072680020251008
Roberfroid et al, 200830304001526401.41.91815
Shankar et al, 200822304001526401.41.91814
Zeng et al, 200824304001526401.41.91815
Bhutta et al, 200916304001526401.41.91815
Sunawang et al, 200923304001526401.41.91815
Eneroth et al, 201019304001526401.41.91815
Table 3.  GRADE table for multiple micronutrient (MM) supplements during pregnancy Thumbnail image of
Table 4.  Summary results from meta-analysis of multiple micronutrients (≥5 micronutrients) vs. controls (≤3 micronutrients) of studies stratified by the amount of iron in the multiple micronutrient (MM) supplementsa
Outcome1. MM containing 30 mg Fe2. MM containing 60 mg Fe
Effect size [95% CI]Effect size [95% CI]
No. of studies/sample size; I2 (%)No. of studies/sample size; I2 (%)
  • a

    Effect sizes were estimated by risk ratios for all outcomes except birthweight and gestational age which were estimated by weighted mean difference where multiple micronutrients are compared with controls who received supplements containing ≥60 mg iron.

Low birthweight0.87 [0.81, 0.93]0.79 [0.63, 0.97]
9/23 179; 06/12 112; 72
Birthweight (g)52.65 [41.59, 63.71]43.44 [7.8, 79.08]
9/23 165; 405/11 091; 58
Small-for-gestational age0.90 [0.80, 1.00]0.78 [0.63, 0.96]
2/3906; 06/11 891; 80
Gestational age (weeks)0.07 [−0.01, 0.15]0.09 [−0.02, 0.21]
5/8518; 04/9890; 2
Preterm birth1.01 [0.84, 1.21]0.97 [0.90, 1.05]
3/5381; 05/12 102; 0
Stillbirths1.13 [0.93, 1.38]0.90 [0.58, 1.41]
6/8793; 343/10 018; 77
Neonatal death1.24 [0.96, 1.60]1.08 [0.61, 1.90]
6/8679; 162/9393; 79
Table 5.  Summary results from meta-analysis of multiple micronutrients (≥5 micronutrients) vs. controls (≤3 micronutrients) of studies that initiated supplementation early (≤12 weeks' gestation) and late (>12 weeks' gestation)a
Outcome≤12 weeks' gestation>12 weeks' gestation
Effect size [95% CI]Effect size [95% CI]
No. of studies/sample size; I2 (%)No. of studies/sample size; I2 (%)
  • a

    Effect sizes were estimated by risk ratios for all outcomes except birthweight and gestational age which were estimated by weighted mean difference.

Low birthweight0.86 [0.80, 0.93]0.86 [0.78, 0.94]
8/25 301; 488/10 055; 35
Birthweight (g)43.26 [19.34, 67.18]45.49 [27.51, 63.46]
7/24 280; 658/10 041; 14
Small-for-gestational age0.82 [0.64, 1.05]0.86 [0.78, 0.96]
4/10 639; 854/5158; 27
Gestational age (weeks)0.08 [−0.04, 0.20]0.09 [0.01, 0.16]
4/9462; 06/9011; 21
Preterm birth1.00 [0.96, 1.03]0.95 [0.82, 1.10]
5/39 422; 04/6487; 0
Stillbirths0.92 [0.73, 1.15]1.14 [0.93, 1.40]
5/40 800; 575/7950; 46
Neonatal death0.92 [0.73, 1.17]1.38 [1.05, 1.81]
4/38 267; 535/7835; 0

Low birthweight

Sixteen trials reported data on the effect of MM during pregnancy on the risk of low birthweight in the offspring when compared with a control group which in most studies received supplements containing varying doses of iron with or without folic acid.16,17,19,32 The pooled analysis indicated a 14% significant reduction in incidence of low birthweight in the intervention group compared with the control (RR 0.86 [95% CI 0.81, 0.91]) (Figure 2); The reduction in the risk of delivering low birthweight infants was greater (RR 0.79 [95% CI 0.63, 0.97]) for the subgroup of six trials17,18,20,25–27,32 that compared MM supplements containing ≥60 mg iron to the control group receiving ≥60 mg iron with or without folic acid (Table 4), but was not significantly different from the subgroup of nine trials16,19,21–24,28–30 that compared MM supplements containing 30 mg with the control group who received ≥60 mg iron with or without folic acid (RR 0.87 [95% CI 0.81, 0.93]). There were no differences by timing of initiation of supplementation (Table 5).

Figure 2.

Effect sizes for low birthweight in the groups receiving multiple micronutrients (including iron) vs. control (including iron) with or without folic acid.

Birthweight

The impact on mean birthweight was reported for 15 trials that compared MM supplementation (containing varying dose of iron) with iron alone or IFA supplementation (varying dose of iron).16,17,19,26–32 The effects on birthweight ranged from 4.0 g in Mexico32 to 251.0 g in France.31 Eight trials reported significantly positive effects,16,20,21,26–29,31 seven trials reported null findings,17,19,22–24,30,32 and none reported negative effects of MM supplements. In a pooled analysis, birthweight of infants of women supplemented with MM was 53 g higher compared with those of women given iron or IFA supplements (Figure 3). Stratified analysis that compared MM containing 30 mg iron vs. control containing 60 mg iron with folic acid17,19,21,23,24,28–30 showed a similar effect in infant birthweight in the MM group (Table 4). The effects of MM supplements in the subgroup of 7 trials that began the intervention after 12 weeks of gestation were not significantly different compared with the estimate from the 8 trials that began the intervention in the first trimester (Table 5).

Figure 3.

Effect sizes for birthweight in the groups receiving multiple micronutrients (including iron) vs. control (including iron) with or without folic acid.

Small-for-gestational age

Data on SGA were reported for eight trials17,20,24–27,30,32 and the pooled analysis showed a significant reduction of 17% (RR 0.83 [95% CI 0.73, 0.95]) (Figure 4) in the MM group compared with the controls who received varying amounts of iron with or without folic acid. The effect size was larger for the subset of six trials that used at least 60 mg iron in their supplements17,20,25–27,32 (RR 0.78 [95% CI 0.63, 0.96], Table 4), but not significantly different from the overall estimate. There were no differences by timing of initiation as well (early initiation: RR 0.82 [95% CI 0.64, 1.05]; vs. late initiation: RR 0.86 [95% CI 0.78, 0.96], Table 5).

Figure 4.

Effect sizes for small-for-gestational age in the groups receiving multiple micronutrients (including iron) vs. control (including iron) with or without folic acid.

Gestational age

The effect on gestational age was reported in 10 trials.16,20,21,23,24,26,27,30–32 Although none of the individual studies reported statistically significantly positive effects, the pooled analysis indicated a significant effect of MM supplementation on gestational age in infants of women given MM supplementation compared with those given iron or IFA supplementation (WMD 0.08 weeks [95% CI 0.01 weeks, 0.14 weeks]) (Figure 5). The estimates did not differ by amount of iron in the supplements (Table 4) or timing of initiation (Table 5).

Figure 5.

Effect sizes for gestational-age in the groups receiving multiple micronutrients (including iron) vs. control (including iron) with or without folic acid.

Preterm birth

A pooled analysis of nine trials17,21,22,24–27,30,32 on preterm birth with different levels of iron in the MM supplements showed no beneficial effect in the intervention group compared with the control group (RR 0.99 [95% CI 0.96, 1.03]) (Figure 6). This result was not altered in the stratified analysis of MM vs. control by amount of iron (RR 0.97 [95% CI 0.90, 1.05] for MM containing ≥60 mg iron; vs. RR 1.01 [95% CI 0.84, 1.21] for MM containing 30 mg iron;) and timing of initiation.

Figure 6.

Effect sizes for preterm birth in the groups receiving multiple micronutrients (including iron) vs. control (including iron) with or without folic acid.

Stillbirth

Data on stillbirth were included from 10 trials16,18,21–26,28,30 that compared MM (including iron) with iron and folic acid controls. The pooled analysis showed no significant differences between the intervention and control group (RR 1.00 [95% CI 0.84, 1.21]) (Figure 7). Stratified analysis did not reveal any differences by varying levels of iron in the MM and control groups or timing of initiation (early initiation: RR 0.92 [95% CI 0.73, 1.15]; vs. late initiation: RR 1.14 [95% CI 0.93, 1.40]).

Figure 7.

Effect sizes for stillbirths in the groups receiving multiple micronutrients (including iron) vs. control (including iron) with or without folic acid.

Neonatal death

Nine trials reported data on neonatal death16,18,21–24,26,28,30 and pooled analysis did not reveal significant differences when MM supplements were compared with iron or IFA (RR 0.97 [95% CI 0.87, 1.09]) (Figure 8). Stratified analysis revealed no differences by the amount of iron in the supplements but we found evidence of increased risk of neonatal death in the five trials that began the intervention after 12 weeks of comparison (RR 1.38 [95% CI 1.05, 1.81]), which was significantly different from the effect seen in studies that began the intervention in the first trimester (RR 0.92 [95% CI 0.73, 1.17]).

Figure 8.

Effect sizes for neonatal death in the groups receiving multiple micronutrients (including iron) vs. control (including iron) with or without folic acid.

Maternal Mortality and Morbidity

We found only 2 trials22,28 that reported the effects on maternal mortality. The pooled estimate revealed no significant differences (RR 0.96 [95% CI 0.64, 1.45], most of which was explained by the large trial in Indonesia28.

Publication bias

Apart from the studies of effect of MM on birthweight, the funnel plot in each meta-analysis was symmetrical, an indication of the absence of publication bias (data not shown). The symmetrical observation of the funnel plots were further confirmed by Egger's weighted regression method and Begg's rank correlation method with P-values of >0.05.

Comments

In this review, supplementation of women during pregnancy with five or more micronutrients was more effective in reducing the risk of delivering low birthweight and/or SGA infants when compared with supplementation with three or fewer micronutrients including iron and folic acid in most cases. Although the reductions were greater in the trials in which MM supplements contained at least 60 mg iron, the differences were not statistically different when compared with the subgroup of trials using MM supplements containing 30 mg iron (RR 0.79 [95% CI 0.63, 0.97] vs. RR 0.87 [95% CI 0.81, 0.93] for low birthweight; RR 0.78 [95% CI 0.63, 0.96] vs. RR 0.90 [95% CI 0.80, 1.00] for SGA). The birthweight of infants whose mothers were given MM was on average 53 g greater than the controls given IFA supplementation. MM supplementation increased gestational age, but there were no significant differences in the risk of preterm birth, stillbirth and neonatal death. We did not find significant differences in the effect of MM interventions by timing of intervention except for an increased risk of neonatal death in the subgroup of trials that began intervention after the first trimester, which is a major concern.

The included trials were of high quality and the overall quality of evidence for our outcomes were from moderate to high (Table 3). The quality of evidence for the subanalysis by time of initiation and amount of iron (provided as Tables S1–3), however, was more variable and ranged from low to moderate for most outcomes except birthweight. Most of the trials started supplementation in the first or second trimester of pregnancy with a few studies starting in the third trimester and the sample sizes ranged from 100 in the study in France31 to 31 290 in Indonesia.22 The amount of iron in the MM supplement was mostly up to 30 mg although some studies provided at least 60 mg iron to both intervention and control groups. Some of the trials included in the current review provided nutrients such as vitamin A or D or calcium to both the control group (in addition to the iron/or iron + folic acid) and the intervention group.17,18,20,25,26 This resulted in the control group receiving three micronutrients (i.e. iron/or iron + folic acid and one additional vitamin/mineral), which has sometimes been used as a definition for MM interventions. The definition of the control group is a concern, and although earlier reviews have defined the control group as receiving two or fewer micronutrients, careful examination of these reviews and the relevant trials reveal that this is not necessarily the case. Another problem is that some studies had several arms and may have provided a placebo that contained ≤1 micronutrient, and iron or IFA supplements may have been provided separately to all women. However, we found similar results when we conducted a sensitivity analysis that used pooled estimates for trials that had more than one comparison but shared the same control or intervention arm, namely: (1) MM vs. IFA and placebo in the trial in Nepal by Christian et al.,17,18 (2) MM vs. IFA with 30 mg Fe and IFA with 60 mg Fe in Bangladesh,19 (3) MM(1 RDA) and MM (2 RDA) vs. IFA in Guinea-Bissau,28 and (4) MM with vitamin A and MM without vitamin A vs. (vitamin A and Placebo) in the Tanzanian trial among HIV + women.25

Overall, our findings confirm some of the results from earlier meta-analyses conducted by Shah et al.9 and by Haider et al.6,7 regarding the benefits of MM supplementation. In addition to confirming earlier findings, we have additional findings, which compared the effect of MM supplementation by amount of iron in the supplements provided to the treatment and control groups and differences by the timing of initiation of supplementation. Our findings indicated that MM supplementation during pregnancy was more efficacious than IFA supplements alone in reducing the incidence of low birthweight and SGA. This is in contrast with the earlier review by Haider and Bhutta,7 which found a beneficial effect only when compared with placebo or supplementation of two or fewer micronutrients in nine trials, but in agreement with the recent review by the same authors.6 Shah and colleagues9 reported similar observations of beneficial findings on risk of low birthweight and increases in mean birthweight as in our review when they compared MM supplementation with IFA alone; however, in contrast to our observation, they did not find any significant effect on the risk of delivering SGA infants. Although we used similar methods, we included findings from seven trials (n = 16 839) that were absent in the review by Shah et al.9 The direction and magnitude of effect size for low birthweight, SGA and birthweight in our review are comparable to those of Fall et al.,10 who reviewed 12 studies that compared the UNIMMAP MM supplementation with IFA, and similar for SGA in a recent review by Haider et al.6

Our finding of increased risk of neonatal death with MM supplementation in the subgroup of trials that began the intervention after the first trimester is new and presents a major concern especially in many developing country settings, where women do not receive prenatal care in the first trimester. Although these findings may have been by chance, especially since several comparisons were made and the overall pooled estimates of the effects of MM supplementation on stillbirth and neonatal death were non-significant, they cannot be disregarded and have important policy implications.34–36 The reasons remain unclear and may be related to timing and/or dose. The five trials that began the intervention later compared MM supplements containing 30 mg iron to IFA supplements that contained 60 mg iron and were conducted in developing countries. The quality of evidence was also graded ‘moderate’ for the estimates of the effect of MM after 12 weeks' gestation on neonatal death but ‘low’ for the estimates based on studies that began the intervention in the first trimester. It should also be noted that in the large trial in Indonesia22, women began the intervention at varying times and the authors found no evidence of increased risk among those who began later. The findings from ongoing large trials in Bangladesh and China are awaited and may help clarify this issue. Our subanalysis by the amount of iron in the MM supplement also showed increased risk of stillbirth and neonatal mortality in the subgroup of six trials in which the MM supplement contained 30 mg iron but the differences were not significantly different from the overall estimates or subgroup of four trials that had MM supplements with 60 mg iron. The quality of evidence, however, was ‘moderate’ and ‘low’ for the estimates of effects for these outcomes for the subanalysis of studies providing MM supplements containing 30 and 60 mg iron, respectively.

The results of our review confirm the benefits of MM supplementation in reducing the risk of low birthweight under controlled conditions but raise new concerns about the increased risk of neonatal death. We did not include anaemia as an outcome in the current review and although anaemia could be a mediator for the reported effects of the MM interventions on birth outcomes, recent reviews indicate no additional benefit of MM supplementation on third trimester maternal anaemia compared with iron-folic acid.6–8,37 Further research and reviews should include data on baseline nutrient status that would help identify which target groups would benefit the most. Information on the acceptability of the MM supplements and compliance in programmatic settings would also be useful in light of the problems that have been reported with routine IFA supplementation in many settings.16,38–40

In conclusion, MM supplementation during pregnancy reduces the incidence of low birthweight and SGA. Although there was a suggestion of increased gestational age, we found no significant differences in preterm birth. The increased risk for neonatal death in the subgroup of studies that began intervention after the first trimester, however, calls for caution before making policy recommendations to change to MM supplementation from current standard of care, namely IFA supplementation, in settings where prenatal care is not started early and is suboptimal.

Conflicts of interest

The authors declare no conflicts of interests.

Ancillary