Magnesium supplementation in pregnancy
Abstract
Background
Magnesium is an essential mineral required for regulation of body temperature, nucleic acid and protein synthesis and in maintaining nerve and muscle cell electrical potentials. Many women, especially those from disadvantaged backgrounds, have low intakes of magnesium. Magnesium supplementation during pregnancy may be able to reduce fetal growth restriction and pre‐eclampsia, and increase birthweight.
Objectives
To assess the effects of magnesium supplementation during pregnancy on maternal, neonatal/infant and paediatric outcomes.
Search methods
We searched the Cochrane Pregnancy and Childbirth Group's Trials Register (31 March 2013).
Selection criteria
Randomised and quasi‐randomised trials assessing the effects of dietary magnesium supplementation during pregnancy were included. The primary outcomes were perinatal mortality (including stillbirth and neonatal death prior to hospital discharge), small‐for‐gestational age, maternal mortality and pre‐eclampsia.
Data collection and analysis
Two review authors independently assessed study eligibility, extracted data and assessed the risk of bias of included studies.
Main results
Ten trials involving 9090 women and their babies were included; one trial had a cluster design (with randomisation by study centre). All 10 trials randomly allocated women to either an oral magnesium supplement or a control group; in eight trials a placebo was used, and in two trials no treatment was given to the control group. In the 10 included trials, the compositions of the magnesium supplements, gestational ages at commencement, and doses administered varied, including: magnesium oxide, 1000 mg daily from ≤ four months post‐conception (one trial); magnesium citrate, 365 mg daily from ≤ 18 weeks until hospitalisation after 38 weeks (one trial), and 340 mg daily from nine to 27 weeks' gestation (one trial); magnesium gluconate, 2 to 3 g from 28 weeks' gestation until birth (one trial), and 4 g daily from 23 weeks' gestation (one trial); magnesium aspartate, 15 mmol daily (three trials, commencing from either six to 21 weeks' gestation until birth, ≤ 16 weeks' gestation until birth, or < 12 weeks until birth), or 365 mg daily from 13 to 24 weeks until birth (one trial); and magnesium stearate, 128 mg elemental magnesium from 10 to 35 weeks until birth (one trial).
In the analysis of all trials, oral magnesium supplementation compared to no magnesium was associated with no significant difference in perinatal mortality (stillbirth and neonatal death prior to discharge) (risk ratio (RR) 1.10; 95% confidence interval (CI) 0.72 to 1.67; five trials, 5903 infants), small‐for‐gestational age (RR 0.76; 95% CI 0.54 to 1.07; three trials, 1291 infants), or pre‐eclampsia (RR 0.87; 95% CI 0.58 to 1.32; three trials, 1042 women). None of the included trials reported on maternal mortality.
Considering secondary outcomes, while no increased risk of stillbirth was observed, a possible increased risk of neonatal death prior to hospital discharge was shown for infants born to mothers who had received magnesium (RR 2.21; 95% CI 1.02 to 4.75; four trials, 5373 infants). One trial contributed over 70% of the participants to the analysis for this outcome; the trial authors suggested that the large number of severe congenital anomalies in the supplemented group (unlikely attributable to magnesium) and the deaths of two sets of twins (with birthweights < 750 g) in the supplemented group likely accounted for the increased risk of death observed, and thus this result should be interpreted with caution. Furthermore, when the deaths due to severe congenital abnormalities in this trial were excluded from the meta‐analysis, no increased risk of neonatal death was seen for the magnesium supplemented group. Magnesium supplementation was associated with significantly fewer babies with an Apgar score less than seven at five minutes (RR 0.34; 95% CI 0.15 to 0.80; four trials, 1083 infants), with meconium‐stained liquor (RR 0.79; 95% CI 0.63 to 0.99; one trial, 4082 infants), late fetal heart decelerations (RR 0.68; 95% CI 0.53 to 0.88; one trial, 4082 infants), and mild hypoxic‐ischaemic encephalopathy (RR 0.38; 95% CI 0.15 to 0.98; one trial, 4082 infants). Women receiving magnesium were significantly less likely to require hospitalisation during pregnancy (RR 0.65, 95% CI 0.48 to 0.86; three trials, 1158 women).
Of the 10 trials included in the review, only two were judged to be of high quality overall. When an analysis was restricted to these two trials none of the review's primary outcomes (perinatal mortality, small‐for‐gestational age, pre‐eclampsia) were significantly different between the magnesium supplemented and control groups.
Authors' conclusions
There is not enough high‐quality evidence to show that dietary magnesium supplementation during pregnancy is beneficial.
PICOs
Plain language summary
There is not enough high quality evidence to show that dietary magnesium supplementation during pregnancy is beneficial
Many women, especially those from disadvantaged backgrounds, have intakes of magnesium below recommended levels. Magnesium supplementation during pregnancy may be able to reduce growth restriction of the fetus and pre‐eclampsia (high blood pressure and protein in the urine during pregnancy), and increase birthweight. This review aimed to assess the effects of magnesium supplementation during pregnancy on maternal, neonatal and paediatric outcomes.
We included 10 randomised trials involving 9090 women and their babies in this review. These trials were of a low to moderate quality overall. No difference in the risk of perinatal mortality (stillbirth and death of babies prior to hospital discharge) was found when we compared the group of babies born to mothers who received magnesium during their pregnancy and the group of babies born to mothers who did not receive magnesium. Magnesium supplementation did not reduce the risk of babies being born small for their gestational age, and did not reduce the risk of pre‐eclampsia for the mothers.
We found no convincing evidence that magnesium supplementation during pregnancy is beneficial.
Authors' conclusions
Background
Magnesium is one of the essential minerals needed by humans in relatively large amounts. Magnesium works with many enzymes to regulate body temperature, synthesise nucleic acids and proteins as well as maintaining electrical potentials in nerves and muscle membranes. Magnesium also has an important role in modulating vasomotor tone and cardiac excitability. Magnesium occurs widely in many foods; dairy products, breads and cereals, legumes, vegetables and meats are all good sources. It is therefore not surprising that frank magnesium deficiency has never been reported to occur in healthy individuals who eat varied diets. However, processing of the above foods can lead to marked depletion of magnesium. Common causes of magnesium deficiency include inadequate dietary intake or gastrointestinal absorption, increased losses through the gastrointestinal or renal systems and increased requirement for magnesium, such as in pregnancy.
A study measuring serum magnesium during low‐risk pregnancies reported that both ionised and total serum magnesium were significantly reduced after the 18th week of gestation compared to measurements prior to this time (Arikan 1999). Dietary intake studies during pregnancy consistently demonstrate that many women, especially those from disadvantaged backgrounds, have intakes of magnesium below recommended levels (Inst Med 1990). In a retrospective study of medical records, Conradt 1984 reported that magnesium supplementation during pregnancy was associated with a reduced risk of fetal growth retardation and pre‐eclampsia. A later cross‐sectional study of dietary intake towards the end of the first trimester of pregnancy reported that higher magnesium intake was associated with increased birthweight (Doyle 1989). Stimulated by these encouraging preliminary reports, several randomised clinical trials have been undertaken to evaluate the potential benefits of magnesium supplementation during pregnancy on maternal and infant outcomes.
Why it is important to do this review
This review updates a previously published Cochrane review on magnesium supplementation during pregnancy (Makrides 2001). In this previous version of the review, magnesium supplementation was shown to be associated with a lower frequency of preterm birth, low birthweight and small‐for‐gestational age when compared with placebo. However, of the seven trials included in this previous version, only one was judged to be of high quality. Thus the review authors concluded that there was not enough high‐quality evidence to show that dietary magnesium supplementation during pregnancy is beneficial.
It is important to assess whether magnesium supplementation during pregnancy has benefits for mothers and their infants without causing harm. We believe this review could provide information about the potential for magnesium supplementation to improve neonatal/infant outcomes such as weight and growth and improve maternal outcomes such as pre‐eclampsia, without causing adverse effects for women and their babies.
Objectives
To assess the effects of magnesium supplementation during pregnancy on maternal, neonatal/infant and paediatric outcomes, using the best available evidence.
Methods
Criteria for considering studies for this review
Types of studies
All published, unpublished and ongoing randomised, quasi‐randomised trials or cluster‐randomised trials of dietary magnesium supplementation during pregnancy. For the purpose of this review, a dietary supplement was defined as a product taken by mouth that contains a "dietary ingredient" intended to supplement the diet (US FDA 2009).
Types of participants
Women with normal or high‐risk pregnancies.
Types of interventions
We included studies where magnesium was administered orally at any time during the antenatal period, regardless of dose. We excluded where magnesium was administered intravenously/intramuscularly.
Types of outcome measures
Primary outcomes
Infant
-
Perinatal mortality (including stillbirth and neonatal death prior to hospital discharge)
-
Small‐for‐gestational age
Maternal
-
Maternal mortality
-
Pre‐eclampsia
Secondary outcomes
Infant
-
Stillbirth
-
Neonatal death prior to hospital discharge
-
Miscarriage
-
Gestational age at birth
-
Preterm birth at less than 37 weeks
-
Birthweight and low birthweight
-
Admission to a neonatal intensive care unit
-
Long‐term infant outcomes (disability at paediatric follow‐up)
Infant outcomes not pre‐specified in the original protocol
-
Low Apgar score at one or five minutes
-
Late fetal heart decelerations
-
Meconium‐stained liquor and meconium aspiration
-
Breech presentation
-
Placental abruption
-
Placental weight
-
Hypoxic ischaemic encephalopathy (HIE) and neonatal encephalopathy
-
Congenital abnormalities
Maternal
-
Maternal acceptance of treatment
-
Side effects of therapy (grouped as gastrointestinal and non‐gastrointestinal)
-
Systolic and diastolic blood pressure
-
Pregnancy‐induced hypertension
-
Eclampsia
-
Need for maternal hospitalisation
-
Antepartum and postpartum haemorrhage
-
Length of labour
The methods section of this review is based on a standard template used by the Cochrane Pregnancy and Childbirht Group.
Search methods for identification of studies
Electronic searches
For the search strategy used to identify trials included in the previous version of this review (Other published versions of this review), see Appendix 1.
For this update, we contacted the Trials Search Co‐ordinator to search the Cochrane Pregnancy and Childbirth Group’s Trials Register (31 March 2013).
The Cochrane Pregnancy and Childbirth Group’s Trials Register is maintained by the Trials Search Co‐ordinator and contains trials identified from:
-
monthly searches of the Cochrane Central Register of Controlled Trials (CENTRAL);
-
weekly searches of MEDLINE;
-
weekly searches of Embase;
-
handsearches of 30 journals and the proceedings of major conferences;
-
weekly current awareness alerts for a further 44 journals plus monthly BioMed Central email alerts.
Details of the search strategies for CENTRAL, MEDLINE and Embase, the list of handsearched journals and conference proceedings, and the list of journals reviewed via the current awareness service can be found in the ‘Specialized Register’ section within the editorial information about the Cochrane Pregnancy and Childbirth Group.
Trials identified through the searching activities described above are each assigned to a review topic (or topics). The Trials Search Co‐ordinator searches the register for each review using the topic list rather than keywords.
We did not apply any language restrictions.
Data collection and analysis
For the methods used when assessing the trials identified in a previous version of this review (Other published versions of this review), see Appendix 1.
For the most recent update we used the following methods when assessing the trials identified by the search.
Selection of studies
Two review authors independently assessed for inclusion all the potential studies we identified as a result of the search strategy. We resolved any disagreement through discussion or, if required, we consulted a third person.
Data extraction and management
We designed a form to extract data. For eligible studies, at least two review authors extracted the data using the agreed form. We resolved discrepancies through discussion or, if required, we consulted a third person. We entered data into Review Manager software (RevMan 2012) and checked for accuracy.
When information regarding any of the above was unclear, we attempted to contact authors of the original reports to provide further details. When articles were not written in English, every attempt was made to obtain translations to ensure accurate data extraction and analysis.
Assessment of risk of bias in included studies
Two review authors independently assessed risk of bias for each study using the criteria outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). We resolved any disagreement by discussion or by involving a third author.
(1) Random sequence generation (checking for possible selection bias)
We described for each included study the methods used to generate the allocation sequence in sufficient detail to allow an assessment of whether it should produce comparable groups.
We assessed the methods as:
-
low risk of bias (any truly random process, e.g. random number table; computer random number generator);
-
high risk of bias (any non‐random process, e.g. odd or even date of birth; hospital or clinic record number);
-
unclear risk of bias.
(2) Allocation concealment (checking for possible selection bias)
We described for each included study the method used to conceal the allocation sequence and determined whether intervention allocation could have been foreseen in advance of, or during recruitment, or changed after assignment.
We assessed the methods as:
-
low risk of bias (e.g. telephone or central randomisation; consecutively numbered sealed opaque envelopes);
-
high risk of bias (open random allocation; unsealed or non‐opaque envelopes, alternation; date of birth);
-
unclear risk of bias.
(3.1) Blinding of participants and personnel (checking for possible performance bias)
We described for each included study, the methods, if any, used to blind study participants and personnel from knowledge of which intervention a participant received. We considered studies to be at a low risk of bias if they were blinded, or if we judged that the lack of blinding would be unlikely to affect results. We assessed blinding separately for different outcomes or classes of outcomes.
We assessed the methods as:
-
low, high or unclear risk of bias for participants;
-
low, high or unclear risk of bias for personnel.
(3.2) Blinding of outcome assessment (checking for possible detection bias)
We described for each included study the methods used, if any, to blind outcome assessors from knowledge of which intervention a participant received. We assessed blinding separately for different outcomes or classes of outcomes.
We assessed methods used to blind outcome assessment as:
-
low, high or unclear risk of bias.
(4) Incomplete outcome data (checking for possible attrition bias due to the amount, nature and handling of incomplete outcome data)
We described for each included study and for each outcome or class of outcomes, the completeness of data including attrition and exclusions from the analysis. We stated whether attrition and exclusions were reported, the numbers included in the analysis at each stage (compared with the total randomised participants), reasons for attrition or exclusion where reported, and whether missing data were balanced across groups or were related to outcomes. Where sufficient information was reported or was supplied by the trial authors, we included the missing data in the analyses which we undertook.
We assessed the methods as:
-
low risk of bias (e.g. where there was no missing data or where reasons for missing data were balanced across groups);
-
high risk of bias (e.g. numbers or reasons for missing data imbalanced across groups; 'as treated' analysis done with substantial departure of intervention received from that assigned at randomisation);
-
unclear risk of bias.
(5) Selective reporting bias (checking for reporting bias)
We described for each included study how the possibility of selective outcome reporting bias was examined by us and what we found.
We assessed the methods as:
-
low risk of bias (where it was clear that all of the study’s pre‐specified outcomes and all expected outcomes of interest to the review had been reported);
-
high risk of bias (where not all the study’s pre‐specified outcomes had been reported; one or more reported primary outcomes were not pre‐specified; outcomes of interest were reported incompletely and so could not be used; study failed to include results of a key outcome that would have been expected to have been reported);
-
unclear risk of bias.
(6) Other sources of bias (checking for bias due to problems not covered by (1) to (5) above)
We described for each included study any important concerns we had about other possible sources of bias. We assessed whether each study was free of other problems that could put it at risk of bias:
-
low risk of other bias;
-
high risk of other bias;
-
unclear whether there is risk of other bias.
(7) Overall risk of bias
We made explicit judgements about whether studies were at high risk of bias, according to the criteria given in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). With reference to (1) to (6) above, we assessed the likely magnitude and direction of the bias and whether we considered it is likely to impact on the findings. We explored the impact of the level of bias through undertaking sensitivity analyses ‐ seeSensitivity analysis.
Measures of treatment effect
Dichotomous data
For dichotomous data, we presented results as summary risk ratio with 95% confidence intervals
Continuous data
For continuous data, we used the mean difference when outcomes were measured in the same way between trials. We planned to use the standardised mean difference to combine trials that measured the same outcome, but used different methods.
Unit of analysis issues
Cluster‐randomised trials
We have included one cluster‐randomised trial (Hungary 1988) in the analyses along with individually‐randomised trials. We have adjusted its sample size and event rates using the methods described in the Cochrane Handbook and a conservative estimate of the intracluster correlation co‐efficient (ICC) (0.02) in the primary analysis (Higgins 2011) (Table 1). As estimates of ICCs for mortality outcomes have been shown to be lower than for other perinatal outcomes, we also carried out a sensitivity analysis to investigate the effect of varying the ICC (using a range of ICCs from 0.0002 to 0.02) for the outcomes perinatal mortality, stillbirth, and neonatal death prior to discharge.
| Outcomes | Intervention (original data) | Control (original data) | Intervention (adjusted data)1 | Control (original data)1 | ||||
| Total number | Event number | Total number | Event number | Total number | Event number | Total number | Event number | |
| Perinatal mortality | 6 | 400 | 5 | 396 | 3 | 174 | 2 | 172 |
| Small‐for‐gestational age < 10th percentile | 33 | 400 | 59 | 396 | 14 | 174 | 26 | 172 |
| Stillbirth | 3 | 400 | 3 | 396 | 1 | 174 | 1 | 172 |
| Neonatal death prior to discharge | 3 | 400 | 2 | 396 | 1 | 174 | 1 | 172 |
| Miscarriage (< 20 weeks' gestation) | 28 | 428 | 32 | 428 | 12 | 186 | 14 | 186 |
| Preterm birth < 37 weeks' gestation | 33 | 400 | 54 | 396 | 14 | 174 | 23 | 172 |
| Low birthweight | 18 | 400 | 31 | 396 | 8 | 174 | 13 | 172 |
| Maternal side effects | 140 | 400 | 156 | 396 | 61 | 174 | 68 | 172 |
1. Adjusted data = n / design effect, where:
-
design effect = 1 + (M ‐ 1) x ICC = 2.3
-
M = average cluster size = (total number of intervention + total number of control randomised)/(cluster number of intervention + cluster number of control) = 66
-
ICC = intracluster correlation co‐efficient = 0.02
We acknowledged heterogeneity in the randomisation unit and performed a subgroup analysis to investigate the effects of the randomisation unit.
Dealing with missing data
For included studies, we noted levels of attrition. We planned to explore the impact of including studies with high levels of missing data in the overall assessment of treatment effect by using sensitivity analyses.
For all outcomes, we carried out analyses, as far as possible, on an intention‐to‐treat basis, i.e. we attempted to include all participants randomised to each group in the analyses, and all participants were analysed in the group to which they were allocated, regardless of whether or not they received the allocated intervention. The denominator for each outcome in each trial was the number randomised minus any participants whose outcomes were known to be missing.
Assessment of heterogeneity
We assessed statistical heterogeneity in each meta‐analysis using the Tau², I² and Chi² statistics. We regarded heterogeneity as substantial if an I² was greater than 30% and either the Tau² was greater than zero, or there was a low P value (less than 0.10) in the Chi² test for heterogeneity.
Assessment of reporting biases
In future updates of this review, if there are 10 or more studies in the meta‐analysis, we plan to investigate reporting biases (such as publication bias) using funnel plots. We will assess funnel plot asymmetry visually. If asymmetry is suggested by a visual assessment, we will perform exploratory analyses to investigate it.
Data synthesis
We carried out statistical analysis using the Review Manager software (RevMan 2012). We used fixed‐effect meta‐analysis for combining data where it was reasonable to assume that studies were estimating the same underlying treatment effect: i.e. where trials were examining the same intervention, and the trials’ populations and methods were judged sufficiently similar. If there was clinical heterogeneity sufficient to expect that the underlying treatment effects differed between trials, or where substantial statistical heterogeneity was detected, we used random‐effects meta‐analysis to produce an overall summary, if an average treatment effect across trials was considered clinically meaningful. The random‐effects summary was treated as the average range of possible treatment effects and we have discussed the clinical implications of treatment effects differing between trials. If the average treatment effect was not clinically meaningful, we would not have combined trials.
Where we have used random‐effects analyses, we have presented the results as the average treatment effect with its 95% confidence interval, and the estimates of Tau² and I².
Subgroup analysis and investigation of heterogeneity
If we had identified substantial heterogeneity, we planned to investigate it using subgroup analyses and sensitivity analyses. We planned to consider whether an overall summary was meaningful, and if it was, used random‐effects analysis to produce it.
Maternal characteristics and characteristics of the intervention may affect health outcomes.
We planned to carry out the following subgroup analyses.
-
Composition of magnesium supplement (i.e. magnesium citrate versus magnesium aspartate versus other)
-
Dose of magnesium (i.e. high versus low)
-
Gestational age at commencement of supplementation (i.e. commencement at < 28 weeks versus ≥ 28 weeks)
-
Normal versus high‐risk women
However, we were not able to conduct subgroup analyses based on the gestational age at commencement of supplementation or on the inclusion of normal versus high‐risk women, as only one trial administered magnesium after 28 weeks' gestation and included high‐risk women (China 1997), and this trial did not report on any of the review's primary outcomes. We were not able to conduct a subgroup analysis based on dose in this update.
We restricted subgroup analyses to primary outcomes.
We assessed subgroup differences by interaction tests available within RevMan (RevMan 2012). We have reported the results of subgroup analyses quoting the Chi² statistic and P value, and the interaction test I² value.
Sensitivity analysis
We carried out a sensitivity analysis to explore the effects of trial quality assessed by allocation concealment and sequence generation, by omitting studies rated as 'high risk of bias' and 'unclear' for these components. We restricted this to the primary outcomes.
Results
Description of studies
SeeCharacteristics of included studies; Characteristics of excluded studies and Characteristics of ongoing studies.
Results of the search
The updated search of the Cochrane Pregnancy and Childbirth Group’s Trials Register identified 22 new reports relating to 14 studies.
We have included three new trials (Hungary 1979; Italy 1994; South Africa 2007), and have excluded nine studies (Denmark 1990; Denmark 1991; Detroit 1999; India 2012; ISRCTN03989660; NCT01709968; Norway 2008; Sweden 1987; Sweden 1995).
Two trials are currently ongoing, assessing magnesium supplementation in the second trimester of pregnancy for overweight individuals (NCT01510665) and magnesium citrate for the prevention of increased blood pressure during the final weeks of pregnancy (ISRCTN98365455) (seeCharacteristics of ongoing studies for further details).
We previously included seven trials in this review (Angola 1992; Austria 1997; China 1997; Hungary 1988; Memphis 1989; Mississippi 1992; Zurich 1988). We have therefore included a total of 10 trials (Angola 1992; Austria 1997; China 1997; Hungary 1979; Hungary 1988; Italy 1994; Memphis 1989; Mississippi 1992; South Africa 2007; Zurich 1988), and excluded a total of nine studies from this review.
Included studies
For full details seeCharacteristics of included studies.
Five studies were conducted in Europe (Austria 1997; Hungary 1979; Hungary 1988; Italy 1994; Zurich 1988), two in America (Memphis 1989; Mississippi 1992), two in Africa (Angola 1992; South Africa 2007), and one in Asia (China 1997).
Participants
A total of 9090 women and their babies were included in the 10 trials; four trials had sample sizes of less than 150 women (Angola 1992; China 1997; Italy 1994; Mississippi 1992), and the South Africa 2007 trial randomised 4476 women.
In six trials both primiparous and multiparous women were included (Angola 1992; Austria 1997; Hungary 1988; Mississippi 1992; South Africa 2007; Zurich 1988) while in three trials, only primiparous women were included (China 1997; Italy 1994; Memphis 1989); this was not clearly specified in Hungary 1979.
Women were recruited at various stages in their pregnancies into the included trials. In the Angola 1992 trial, designed for the prevention of pre‐eclampsia, women were recruited at any stage during their first four months of pregnancy, whereas in China 1997, also designed for the prevention of pregnancy‐induced hypertension, normal and high‐risk women were recruited from 22 weeks' gestation. Women with low‐risk pregnancies were recruited into the Austria 1997 trial before their 18th week of pregnancy, for the prevention of preterm labour; in the Mississippi 1992 trial however, women with risk factors for preterm birth were recruited at a mean gestational age at entry between 23 and 24 weeks. In Hungary 1988 and South Africa 2007, women were recruited at their first antenatal clinic/appointment (in South Africa 2007 the gestational age of women varied from 10 to 35 weeks, with a mean of 21 weeks). In Memphis 1989 women were recruited from between 13 and 24 weeks' gestation, whereas in Italy 1994 and Zurich 1988, women were included at no later than 12 and 16 weeks respectively.
Interventions and comparisons
In the 10 included trials, the composition of the magnesium supplements, gestational ages at commencement of supplementation, and doses administered varied. In eight of the trials a placebo supplement (Angola 1992; China 1997; Hungary 1988; Italy 1994; Mississippi 1992; South Africa 2007) or active control (aspartic acid) (Memphis 1989; Zurich 1988) was used; in two trials, no treatment was given to the control group (Austria 1997; Hungary 1979).
Magnesium oxide: in one trial (Angola 1992) women were given two tablets daily of magnesium oxide 500 mg (1000 mg total daily) beginning from no later than four months post‐conception.
Magnesium citrate: in two trials, women received magnesium citrate tablets (Austria 1997; Hungary 1979). Women in Austria 1997 were given 365 mg of magnesium citrate daily from no later than 18 weeks' gestation until hospitalisation after 38 weeks, whereas women in Hungary 1979 received 340 mg of magnesium citrate daily, either before nine weeks' gestation, or from nine to 27 weeks' gestation.
Magnesium gluconate: women in two trials received magnesium gluconate (China 1997; Mississippi 1992). In China 1997 women received 2 g of magnesium gluconate daily from 28 weeks' gestation to 30 weeks; they then received 3 g daily from 30 weeks until delivery. In Mississippi 1992 women received two 500 mg tablets of magnesium gluconate four times daily (4 g daily total, 215 mg elemental magnesium) from 23 weeks' gestation.
Magnesium aspartate: in four trials, women received magnesium aspartate. In Hungary 1988, Italy 1994 and Zurich 1988 women received 15 mmol of magnesium aspartate daily. In Hungary 1988 women received a tablet of 5 mmol three times per day from six to 21 weeks' gestation until birth, whereas in Zurich 1988 women received six tablets daily from no later than 16 weeks' gestation until birth. In Italy 1994 supplementation commenced before 12 weeks' gestation. In Memphis 1989 women received six tablets of magnesium aspartate hydrochloride per day (each containing 60.8 mg elemental magnesium: 365 mg daily) from between 13 to 24 weeks until birth.
Magnesium stearate: in South Africa 2007 women received two 'slow‐release' magnesium tablets daily (64 mg elemental magnesium per tablet) from the time of enrolment (10 to 35 weeks; mean: 21 weeks) until birth.
Outcomes
All of the 10 included trials focused on perinatal outcomes for women and/or their babies (Angola 1992; Austria 1997; China 1997; Hungary 1979; Hungary 1988; Italy 1994; Memphis 1989; Mississippi 1992; South Africa 2007; Zurich 1988). None of the included studies reported on any longer‐term outcomes for the infants.
Excluded studies
Four studies were excluded as intravenous magnesium sulphate was administered as part of the intervention (Denmark 1990; Denmark 1991; Detroit 1999; India 2012). Four further trials were excluded (ISRCTN03989660; NCT01709968; Norway 2008; Sweden 1995), as they assessed the effects of oral magnesium supplementation on pregnancy‐related leg cramps and therefore have been (or are likely to be) considered for inclusion in the relevant Cochrane review (Garrison 2012). One further trial (Sweden 1987) did not assess magnesium, rather assessed calcium and vitamin C for pregnant women with leg cramps.
SeeCharacteristics of excluded studies for further details.
Risk of bias in included studies
We judged the trials to have a moderate risk of bias overall. See Figure 1 and Figure 2.
'Risk of bias' graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.
'Risk of bias' summary: review authors' judgements about each risk of bias item for each included study.
Allocation
Four of the 10 trials used adequate methods to generate a random sequence (Angola 1992; Austria 1997; Memphis 1989; South Africa 2007). Angola 1992 used a table of random numbers; in Austria 1997, Memphis 1989 and South Africa 2007 a computer‐generated randomisation list was used. For five trials, the methods used for random sequence generation were unclear (China 1997; Hungary 1979; Hungary 1988; Italy 1994; Mississippi 1992), and in one trial, allocation was based on the participants' date of birth (Zurich 1988).
Only two trials were judged to have adequate methods to conceal allocation (Memphis 1989; South Africa 2007). In Memphis 1989 allocation was performed through the hospital pharmacy; in South Africa 2007 research assistants not involved in the care of the women issued consecutive numbers to each participant. For the remaining seven trials, the risk of selection bias due to inadequate allocation concealment was judged as unclear (Angola 1992; Austria 1997; China 1997; Hungary 1979; Hungary 1988; Italy 1994; Mississippi 1992).
Blinding
A placebo was used in six of the 10 trials (Angola 1992; China 1997; Hungary 1988; Italy 1994; Mississippi 1992; South Africa 2007); in Memphis 1989 and Zurich 1988 women in the control group received tablets containing aspartic acid. Blinding of participants, personnel and outcome assessors was judged to be adequate in five of these eight trials (China 1997; Italy 1994; Memphis 1989; Mississippi 1992; South Africa 2007). In Angola 1992 whilst a placebo was used, the placebo and magnesium oxide tablets were distributed in different quantities to the women, and it was unclear if the tablets were identical in appearance; thus the risk of performance and detection bias were judged as unclear. In Hungary 1988 the success of blinding of participants and personnel was also judged as unclear, as while a placebo was used, it was unclear whether this was identical in appearance to the magnesium aspartate tablets, and the leader of the study was able to identify of the composition of the tablets by identification marks.
In Zurich 1988, while the study was described as "double‐blind" and a placebo was used, allocation was based on the participants' date of birth, and thus it was unclear as to whether blinding would have been successfully achieved for personnel, participants and outcome assessors.
Austria 1997 was judged to be at a high risk of performance and detection bias with no blinding of participants, personnel or outcome assessors. In Hungary 1979, while no placebo was used, the translation of the manuscript detailed blind outcome assessment.
Incomplete outcome data
Three of the nine trials were judged to be at a low risk of attrition bias, with no or minimal post‐randomisation exclusions or attrition (China 1997; Hungary 1979; Italy 1994). The remaining seven trials were judged to be at an unclear risk of attrition bias (Angola 1992; Austria 1997; Hungary 1988; Memphis 1989; Mississippi 1992; South Africa 2007; Zurich 1988).
In Angola 1992 post‐randomisation exclusions were not reported. In Austria 1997, 7.5% of women (25 from the treatment group (9.7%) and 15 (5.7%) from the control group) were excluded for a variety of reasons including lack of compliance and gastro‐intestinal problems with supplementation. In Hungary 1988, while only 7/507 and 5/490 women were excluded from the placebo and magnesium groups initially (1.2% total), a further 104/500 (20.8%) were excluded from the placebo group and 85/485 (17.5%) from the magnesium group for a number of reasons including loss to follow‐up. In the Memphis 1989 trial, 11/200 (5.5%) women from the placebo group and 15/200 (7.5%) from the treatment group were excluded as they never started medication. In South Africa 2007, 190 women were excluded post‐randomisation, and a further 204 were lost to follow‐up (394/4476) and it was not possible to determine to which groups these women were randomised. Similarly in Mississippi 1992, it was not possible to ascertain from which group(s) the seven (13%) excluded women were originally assigned. Finally in Zurich 1988, the total number of women in each group were not reported in results tables, and it was thus difficult to determine exclusions or attrition; the manuscript detailed that "For various reasons such as refusal to take further tablets, delivery in other hospitals or abortion, some data were not available for analysis".
Selective reporting
Only one trial was judged to be at a low risk of selective reporting, with data reported for all pre‐specified and/or expected outcomes (Memphis 1989). For the remaining nine trials, the risk of reporting bias was judged to be unclear (Angola 1992; Austria 1997; China 1997; Hungary 1979; Hungary 1988; Italy 1994; Mississippi 1992; South Africa 2007; Zurich 1988), with for example, no pre‐specification of outcomes, important or expected outcomes not reported, or outcome data reported in such a way that it could not be included in a meta‐analysis.
Other potential sources of bias
Five of the trials were judged to be at a low risk of other potential of bias, with no obvious sources of bias identified (Austria 1997; China 1997; Hungary 1979; Italy 1994; Mississippi 1992). For the other trials, this was unclear, for example, with high non‐compliance (Hungary 1988; South Africa 2007).
Effects of interventions
Magnesium supplementation versus no magnesium
Primary outcomes
Infant outcomes
Magnesium supplementation compared with no magnesium supplementation was associated with no significant difference in the risk of perinatal mortality (risk ratio (RR) 1.10; 95% confidence interval (CI) 0.72 to 1.67; five trials, 5903 infants) (Analysis 1.1), or small‐for‐gestational age (RR 0.76; 95% CI 0.54 to 1.07; three trials; 1291 infants; I² = 7%) (Analysis 1.2). For the outcome perinatal mortality, data from the Austria 1997 and Memphis 1989 trials have been included in the meta‐analysis; it should be noted, however, that Austria 1997 only reported on stillbirths (and did not report on neonatal deaths), and Memphis 1989 only reported on neonatal deaths (and did not report on stillbirths). Considering the outcome small‐for‐gestational age, the definition provided in Hungary 1988 and Zurich 1988 was birthweight below the 10th percentile for gestational age; in Memphis 1989, no definition was provided.
Maternal outcomes
The effect of magnesium supplementation on maternal death was not reported by any of the included studies. No significant difference in pre‐eclampsia was observed between the magnesium supplemented and control groups (RR 0.87; 95% CI 0.58 to 1.32; three trials, 1042 women) (Analysis 1.3). The definitions for the three trials reporting on pre‐eclampsia varied: in Memphis 1989 pre‐eclampsia was defined as "a systolic blood pressure reading of ≥ 140 mm Hg, diastolic blood pressure of ≥ 90 mm Hg on two occasions at least 6 hours apart with or without proteinuria, or both"; however in Angola 1992, the definition was "the simultaneous occurrence of the clinical triad of hypertension, edema, and proteinuria at any time during the course of pregnancy"; with hypertension defined as "A rise in systolic BP greater than 30 mm Hg and/or a rise in diastolic BP greater than 15 mm Hg"; proteinuria defined as "Protein greater than one determine by test tape" and oedema defined as "Visible fluid accumulation in the ankles and feet; indentation produced by pressure applied by the thumb over the anterior surface of the tibia." In Zurich 1988, the definition was not clear.
No statistical heterogeneity was observed in the meta‐analyses for the primary outcomes (I² = 0%), excluding small‐for‐gestational age, where the I² has been reported above.
Secondary outcomes
Infant outcomes
Magnesium supplementation compared with no magnesium supplementation was associated with no significant difference in the risk of stillbirth (risk ratio (RR) 0.73; 95% confidence interval (CI) 0.43 to 1.25; four trials, 5526 infants) (Analysis 1.4), however, a possible increased risk of neonatal death prior to discharge was seen for the group of infants whose mothers had received magnesium supplementation (RR 2.21; 95% CI 1.02 to 4.75; four trials, 5373 infants) (P = 0.04) (Analysis 1.5). The South Africa 2007 trial contributed the majority of participants (more than 70%) to this outcome, with 17 infant deaths occurring in the magnesium supplemented group and seven in the control group. The authors of the South Africa 2007 trial suggested that the large number of severe congenital anomalies in the supplemented group (accounting for seven of the 17 deaths) and the deaths of two sets of twins (with birthweights of less than 750 g) in the supplemented group likely accounted for the increased risk of death observed with magnesium supplementation, and thus this result should be interpreted with caution. When we excluded the seven deaths due to "congential abnormalities incompatible with life" from the meta‐analysis for this outcome, reassuringly, no increased risk of neonatal death was seen for the magnesium supplemented group (RR 1.45; 95% CI 0.63 to 3.32; four trials, 5373 infants) (Analysis 1.5).
There was no significant difference between the magnesium supplemented group and the control group for the outcome miscarriage (average RR 0.85; 95% CI 0.49 to 1.49; six trials, 3704 women) (Analysis 1.6). As we identified moderate statistical heterogeneity for this outcome (Tau² = 0.19; I² = 44%) a random‐effects model was used. It is possible that the differing types of magnesium supplement (and the differing regimens for administration used) across the six included trials (Austria 1997; Hungary 1979; Hungary 1988; Italy 1994; Memphis 1989; Zurich 1988) contributed to this moderate level of heterogeneity. Similarly, no significant differences between the magnesium supplemented group and the control group were observed for the outcomes gestational age at birth (mean difference (MD) 0.06 weeks; 95% CI ‐0.07 to 0.20; five trials; 5564 women) (Analysis 1.7) and preterm birth (average RR 0.89; 95% CI 0.69 to 1.14; seven trials; 5981 women; Tau² = 0.04; I² = 37%) (Analysis 1.8).
Five trials reported on low birthweight infants of less than 2500 g and revealed no significant difference in the incidence of low birthweight with magnesium supplementation (RR 0.95; 95% CI 0.83 to 1.09; 5577 infants) (Analysis 1.9). One trial (Zurich 1988) reported data relating to very low birthweight infants (less than 1500 g), with no observed effect of maternal magnesium supplementation (RR 0.52; 95% CI 0.13 to 2.07). No difference in mean birthweight between the two groups was observed (MD 22.21 g; 95% CI ‐27.23 to 71.65; five trials, 5564 infants) (Analysis 1.10). As we identified moderate statistical heterogeneity for this outcome (Tau² = 1138.34; I² = 38%) a random‐effects model was used.
Magnesium supplementation was not shown to be associated with a difference in admission to the neonatal intensive care unit (RR 0.74; 95% CI 0.50 to 1.11; three trials, 1435 infants) (Analysis 1.11).
Disability at paediatric follow‐up was not reported by any of the included trials.
Non pre‐specified infant outcomes
We have reported data on a number of outcomes relating to neonatal morbidity that were not pre‐specified in the original review protocol, but that we considered to be clinically relevant and important to include in this update.
Magnesium supplementation was not shown to be associated with any differences considering the outcomes of a one‐minute Apgar score less than five (RR 0.83; 95% CI 0.41 to 1.67; one trial, 377 infants) (Analysis 1.12), meconium aspiration (RR 0.63; 95% CI 0.32 to 1.26; one trial, 4082 infants) (Analysis 1.15), breech presentation (RR 1.25; 95% CI 0.81 to 1.92; one trial, 4082 infants) (Analysis 1.16), placental abruption (RR 0.96; 95% CI 0.48 to 1.94; one trial, 4082 infants) (Analysis 1.17), placental weight (MD ‐0.01 g; 95% CI ‐22.16 to 22.14; two trials, 4459 infants; Tau² = 113.95; I² = 36%) (Analysis 1.18), any hypoxic ischaemic encephalopathy (HIE) (RR 0.70; 95% CI 0.36 to 1.34; one trial, 4082 infants), moderate HIE (RR 1.02; 05% CI 0.26 to 4.09; one trial, 4082 infants), severe HIE (RR 2.56; 95% CI 0.50 to 13.19; one trial, 4082 infants) (Analysis 1.19), or significant congenital abnormalities (RR 2.05; 95% CI 0.77 to 5.45; one trial, 4082 infants) (Analysis 1.20).
Magnesium supplementation was, however, shown to reduce the risk of a five‐minute Apgar score less than seven (RR 0.34; 05% CI 0.15 to 0.80; four trials, 1083 infants) (Analysis 1.12). Significant reductions in late fetal heart rate decelerations (RR 0.68; 95% CI 0.53 to 0.88) (Analysis 1.13), meconium‐stained liquor (RR 0.79; 95% CI 0.63 to 0.99) (Analysis 1.14), and mild HIE (RR 0.38; 95% CI 0.15 to 0.98) (Analysis 1.19) were also observed with magnesium supplementation in one trial of 4082 infants (South Africa 2007).
Maternal outcomes
There were no data available from any of the trials relating to maternal acceptance of treatment. Of the five trials reporting maternal side effects of treatment (Hungary 1988; Italy 1994; Memphis 1989; Mississippi 1992; Zurich 1988), four reported on gastro‐intestinal symptoms and found no significant difference between the magnesium supplemented group and control group (RR 0.88; 95% CI 0.69 to 1.12; 1388 women) (Analysis 1.21). The Mississippi 1992 trial reported that no women in either group had any side effects.
Magnesium supplementation was associated with a significantly higher systolic blood pressure near birth (MD 1.00 mm Hg; 95% CI 0.03 to 1.97; three trials, 1432 women) (Analysis 1.22), however no significant differences in diastolic blood pressure (MD 0.23 mm Hg; 95% CI ‐0.67 to 1.13; three trials, 1432 women) (Analysis 1.23), or pregnancy‐induced hypertension (average RR 0.39; 95% CI 0.11 to 1.41; three trials, 4284 women) (Analysis 1.24) were observed between groups. Substantial statistical heterogeneity was observed for the outcome pregnancy‐induced hypertension (Tau² = 0.98; I² = 77%), and thus a random‐effects model was used. It is possible that the differing types of magnesium supplement (and the differing regimens for administration used) across the three included trials, or variation between trials in the definitions used for this outcome contributed to this level of heterogeneity (with only the Angola 1992 trial providing a clear definition: a rise in systolic blood pressure > 30 mm Hg and/or a rise in diastolic blood pressure > 15 mm Hg during the course of the pregnancy). Furthermore, the Angola 1992 and China 1997 trials, which showed benefit for the outcome pregnancy‐induced hypertension with magnesium supplementation, were of comparatively small sample sizes (150 and 102 women respectively), and judged to be of a lower quality than the South Africa 2007 trial, of 4476 women, which did not show a difference between groups.
Three trials assessed the need for maternal hospitalisation and demonstrated a reduced need with magnesium supplementation compared with no treatment (RR 0.65; 95% CI 0.48 to 0.86; 1158 women) (Analysis 1.26). No difference between magnesium supplementation and no treatment was shown for the outcomes eclampsia (RR 0.14; 95% CI 0.01 to 2.70; one trial, 100 women) (Analysis 1.25), length of labour (MD ‐0.00 hours; 95% CI ‐0.50 to 0.50; two trials, 4650 women) (Analysis 1.28) or antepartum haemorrhage (average RR 0.53; 95% CI 0.09 to 3.15; two trials. 942 women) (Analysis 1.27). It is possible that the Zurich 1988 trial's low quality (being quasi‐randomised, with unclear blinding) contributed to the substantial statistical heterogeneity observed for the outcome antepartum haemorrhage (Tau² = 1.12; I² = 67%).
The incidence of postpartum haemorrhage was not reported by the included studies.
Subgroup analysis based on type of magnesium supplement
Subgroup analysis based on the type of magnesium supplement used, revealed no subgroup differences for the outcome perinatal mortality (Chi² = 0.33; P = 0.85; I² = 0%) (Analysis 1.1). We were not able to perform a subgroup analysis for small‐for‐gestational age, as all the trials that reported on this outcome used magnesium aspartate (Analysis 1.2).
Similarly, subgroup analysis based on the type of magnesium supplement used revealed no subgroup differences for pre‐eclampsia (Chi² = 1.03; P = 0.31; I² = 3.1%) (Analysis 1.3).
Subgroup analysis based on study design
Subgroup analysis based on the study designed used (cluster‐randomised versus individually‐randomised), revealed no subgroup differences for the primary outcomes, perinatal mortality (Chi² = 0.12; P = 0.73; I² = 0%) (Analysis 2.1), and small‐for‐gestational age (Chi² = 1.93; P = 0.16; I² = 48.3%) (Analysis 2.2). We were not able to perform a subgroup analysis for pre‐eclampsia, as the three trials that reported on this outcome were all individually‐randomised.
Sensitivity analysis ‐ varying the ICC for the cluster‐randomised trial
We conducted a sensitivity analysis using less conservative estimates of the ICC for mortality outcomes (0.0002, 0.002) (less conservative when considering the ICC of 0.02 used throughout the main analysis). Although selecting less conservative ICC values (assuming less clustering and thereby increasing the weight of the one included cluster‐randomised trial (Hungary 1988)) narrowed the confidence intervals slightly, overall this did not have a serious impact on the findings (Analysis 3.1; Analysis 3.2; Analysis 3.3).
Sensitivity analysis by quality rating
Only two trials had an allocation concealment and sequence generation rating of 'low risk of bias' (Memphis 1989; South Africa 2007). The sensitivity analysis excluded trials with an allocation concealment and/or sequence generation rating of 'unclear' or 'high risk of bias.' Among the 'high quality studies' there were no significant differences between the magnesium supplemented group and control group for the outcomes perinatal mortality (RR 1.05; 95% CI 0.67 to 1.65; two trials, 4459 infants) (Analysis 4.1) small‐for‐gestational age (RR 0.76; 95% CI 0.33 to 1.77; one trial, 377 infants) (Analysis 4.2) or pre‐eclampsia (RR 0.93; 95% CI 0.60 to 1.44; one trial, 374 women) (Analysis 4.3), as in the main analysis.
Discussion
Summary of main results
The meta‐analysis of trials included in this review indicated no statistically significant effects of magnesium supplementation on the frequency of perinatal mortality or small‐for‐gestational age infants when compared with placebo or no treatment. Similarly, the results of this review indicated that magnesium supplementation during pregnancy had no significant effect on pre‐eclampsia; maternal deaths were not reported by the included trials.
For secondary outcomes, many of the included studies did not provide data, and where they did, mostly we did not detect differences between the magnesium supplemented and control groups. It is important to note that for some outcomes, the definitions used by individual trials were unclear and/or varied, such as for the outcomes pre‐eclampsia, small‐for‐gestational age and pregnancy‐induced hypertension. Some results did appear to show differences between the groups.
Magnesium supplementation was shown to result in fewer maternal hospitalisations during pregnancy; and while higher maternal systolic blood pressure near birth was shown for the magnesium supplemented group of women, this observed mean difference in blood pressure of 1 mm Hg is considered unlikely to be clinically significant. For both outcomes the Zurich 1988 trial significantly influenced the meta‐analysis; and as this trial was of low quality, being quasi‐randomised with women allocated according to their date of birth, and thus these findings should be interpreted with caution. While no significant difference between groups was shown for the outcome stillbirth, on meta‐analysis, a possible increased risk of neonatal death prior to discharge was observed for infants born to mothers who had received magnesium supplementation. It is important to highlight that only four of the 10 trials included in this review reported data to include in the meta‐analysis for this outcome. Furthermore, in the South Africa 2007 trial (which contributed over 70% of the participants to the analysis for this outcome), the trial authors documented a high number of severe congenital anomalies in the supplemented group (unlikely attributable to magnesium, as none of these seven infants had been exposed to magnesium prior to the 19th week of gestation) and the deaths of two sets of twins (with birthweights of less than 750 g) in the supplemented group. When the deaths due to "congenital abnormalities incompatible with life" (including thanatophoric dwarf, anencephaly, hypoplastic left heart, hypoplastic lungs, and multiple abnormalities) were excluded from the meta‐analysis for this outcome, reassuringly, no increased risk of neonatal death was seen for the magnesium supplemented group.
When only high‐quality trials were included in the analysis, there was no effect of magnesium supplementation on the frequency of perinatal mortality, small‐for‐gestational age, or pre‐eclampsia (Memphis 1989; South Africa 2007). One of the highest quality trials, Memphis 1989, of 400 women, demonstrated no effect of maternal magnesium supplementation on blood pressure, pre‐eclampsia or other pregnancy outcomes. The results of this trial may have however been influenced by the fact that all women (both magnesium supplemented and placebo groups) received a multivitamin and mineral preparation containing 100 mg of magnesium. For the outcomes small‐for‐gestational age and pre‐eclampsia, only the Memphis 1989 trial reported data for inclusion in the sensitivity analysis, and this study was underpowered to detect differences between groups for both of these outcomes, with considerable uncertainly about the treatment effects observed; unfortunately the South Africa 2007, of over 4000 women, did not report on small‐for‐gestational age or pre‐eclampsia.
The South Africa 2007 trial found no difference in meconium aspiration, breech presentation, placental abruption, congenital abnormalities, moderate or severe hypoxic ischaemic encephalopathy (HIE) and neonatal encephalopathy between the magnesium supplemented and placebo groups however it found that magnesium supplementation reduced the risk of late fetal heart decelerations, meconium‐stained liquor and mild HIE (South Africa 2007). These outcomes relating to infant morbidity were not pre‐specified in the original protocol for the review, and while the review authors believed their reporting in this update to be important, they acknowledge the potential for bias associated with the reporting of non pre‐specified review outcomes.
Quality of the evidence
There is a lack of high‐quality evidence assessing the use of magnesium supplementation during pregnancy. Only two trials were judged to be at a low risk of selection bias, with adequate methods to conceal allocation and to generate a random sequence (Memphis 1989; South Africa 2007); the remaining trials were largely judged to be at an unclear risk of selection bias. Only four of the 10 trials were judged at a low risk of both performance and detection bias (Italy 1994; Memphis 1989; Mississippi 1992; South Africa 2007), and for the Austria 1997 trial, the risk of performance and detection bias was judged as high, with no control/placebo treatment used. The majority of trials were judged at an unclear risk of attrition bias and reporting bias.
Potential biases in the review process
The evidence for this review is derived from trials identified through a detailed search process. It is possible (but unlikely) that additional trials assessing magnesium supplementation during pregnancy, have been published but not identified. It is also possible that other studies have been conducted but not published. Should such studies be identified, we will include them in future updates of this review.
'Risk of bias' graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.
'Risk of bias' summary: review authors' judgements about each risk of bias item for each included study.
Comparison 1 Magnesium supplementation versus no magnesium, Outcome 1 Perinatal mortality.
Comparison 1 Magnesium supplementation versus no magnesium, Outcome 2 Small‐for‐gestational age (< 10th percentile).
Comparison 1 Magnesium supplementation versus no magnesium, Outcome 3 Pre‐eclampsia.
Comparison 1 Magnesium supplementation versus no magnesium, Outcome 4 Stillbirth.
Comparison 1 Magnesium supplementation versus no magnesium, Outcome 5 Neonatal death prior to hospital discharge.
Comparison 1 Magnesium supplementation versus no magnesium, Outcome 6 Miscarriage (< 20 weeks' gestation).
Comparison 1 Magnesium supplementation versus no magnesium, Outcome 7 Gestational age at birth (weeks).
Comparison 1 Magnesium supplementation versus no magnesium, Outcome 8 Preterm birth < 37 weeks' gestation.
Comparison 1 Magnesium supplementation versus no magnesium, Outcome 9 Low birthweight.
Comparison 1 Magnesium supplementation versus no magnesium, Outcome 10 Birthweight (g).
Comparison 1 Magnesium supplementation versus no magnesium, Outcome 11 Baby admitted to the neonatal unit.
Comparison 1 Magnesium supplementation versus no magnesium, Outcome 12 Apgar score.
Comparison 1 Magnesium supplementation versus no magnesium, Outcome 13 Late fetal heart rate decelerations.
Comparison 1 Magnesium supplementation versus no magnesium, Outcome 14 Meconium‐stained liquor.
Comparison 1 Magnesium supplementation versus no magnesium, Outcome 15 Meconium aspiration.
Comparison 1 Magnesium supplementation versus no magnesium, Outcome 16 Breech presentation.
Comparison 1 Magnesium supplementation versus no magnesium, Outcome 17 Placental abruption.
Comparison 1 Magnesium supplementation versus no magnesium, Outcome 18 Placental weight (g).
Comparison 1 Magnesium supplementation versus no magnesium, Outcome 19 Hypoxic‐ischaemic encephalopathy.
Comparison 1 Magnesium supplementation versus no magnesium, Outcome 20 Significant congenital abnormality.
Comparison 1 Magnesium supplementation versus no magnesium, Outcome 21 Maternal side effects.
Comparison 1 Magnesium supplementation versus no magnesium, Outcome 22 Systolic blood pressure near birth (mm Hg).
Comparison 1 Magnesium supplementation versus no magnesium, Outcome 23 Diastolic blood pressure near birth (mm Hg).
Comparison 1 Magnesium supplementation versus no magnesium, Outcome 24 Pregnancy‐induced hypertension.
Comparison 1 Magnesium supplementation versus no magnesium, Outcome 25 Eclampsia.
Comparison 1 Magnesium supplementation versus no magnesium, Outcome 26 Need for maternal hospitalisation.
Comparison 1 Magnesium supplementation versus no magnesium, Outcome 27 Antepartum haemorrhage.
Comparison 1 Magnesium supplementation versus no magnesium, Outcome 28 Length of labour (hours).
Comparison 2 Subgroup analysis based on study design, Outcome 1 Perinatal mortality.
Comparison 2 Subgroup analysis based on study design, Outcome 2 Small‐for‐gestational age (< 10th percentile).
Comparison 3 Sensitivity analysis based on the ICC, Outcome 1 Perinatal mortality.
Comparison 3 Sensitivity analysis based on the ICC, Outcome 2 Stillbirth.
Comparison 3 Sensitivity analysis based on the ICC, Outcome 3 Neonatal death prior to hospital discharge.
Comparison 4 Sensitivity analysis by quality rating, Outcome 1 Perinatal mortality.
Comparison 4 Sensitivity analysis by quality rating, Outcome 2 Small‐for‐gestational age (< 10th percentile).
Comparison 4 Sensitivity analysis by quality rating, Outcome 3 Pre‐eclampsia.
| Outcomes | Intervention (original data) | Control (original data) | Intervention (adjusted data)1 | Control (original data)1 | ||||
| Total number | Event number | Total number | Event number | Total number | Event number | Total number | Event number | |
| Perinatal mortality | 6 | 400 | 5 | 396 | 3 | 174 | 2 | 172 |
| Small‐for‐gestational age < 10th percentile | 33 | 400 | 59 | 396 | 14 | 174 | 26 | 172 |
| Stillbirth | 3 | 400 | 3 | 396 | 1 | 174 | 1 | 172 |
| Neonatal death prior to discharge | 3 | 400 | 2 | 396 | 1 | 174 | 1 | 172 |
| Miscarriage (< 20 weeks' gestation) | 28 | 428 | 32 | 428 | 12 | 186 | 14 | 186 |
| Preterm birth < 37 weeks' gestation | 33 | 400 | 54 | 396 | 14 | 174 | 23 | 172 |
| Low birthweight | 18 | 400 | 31 | 396 | 8 | 174 | 13 | 172 |
| Maternal side effects | 140 | 400 | 156 | 396 | 61 | 174 | 68 | 172 |
| 1. Adjusted data = n / design effect, where:
| ||||||||
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Perinatal mortality Show forest plot | 5 | 5903 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.10 [0.72, 1.67] |
| 1.1 Magnesium citrate | 1 | 530 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.0 [0.14, 7.05] |
| 1.2 Magnesium aspartate | 3 | 1291 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.58 [0.42, 5.99] |
| 1.3 Magnesium stearate | 1 | 4082 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.05 [0.67, 1.66] |
| 2 Small‐for‐gestational age (< 10th percentile) Show forest plot | 3 | 1291 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.76 [0.54, 1.07] |
| 2.1 Magnesium aspartate | 3 | 1291 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.76 [0.54, 1.07] |
| 3 Pre‐eclampsia Show forest plot | 3 | 1042 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.87 [0.58, 1.32] |
| 3.1 Magnesium aspartate | 2 | 942 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.94 [0.61, 1.44] |
| 3.2 Magnesium oxide | 1 | 100 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.4 [0.08, 1.97] |
| 4 Stillbirth Show forest plot | 4 | 5526 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.73 [0.43, 1.25] |
| 5 Neonatal death prior to hospital discharge Show forest plot | 4 | Risk Ratio (M‐H, Fixed, 95% CI) | Subtotals only | |
| 5.1 All deaths | 4 | 5373 | Risk Ratio (M‐H, Fixed, 95% CI) | 2.21 [1.02, 4.75] |
| 5.2 Excluding deaths due to "congenital abnormalities incompatible with life" | 4 | 5373 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.45 [0.63, 3.32] |
| 6 Miscarriage (< 20 weeks' gestation) Show forest plot | 6 | 3704 | Risk Ratio (M‐H, Random, 95% CI) | 0.85 [0.49, 1.49] |
| 7 Gestational age at birth (weeks) Show forest plot | 5 | 5564 | Mean Difference (IV, Fixed, 95% CI) | 0.06 [‐0.07, 0.20] |
| 8 Preterm birth < 37 weeks' gestation Show forest plot | 7 | 5981 | Risk Ratio (M‐H, Random, 95% CI) | 0.89 [0.69, 1.14] |
| 9 Low birthweight Show forest plot | 6 | Risk Ratio (M‐H, Fixed, 95% CI) | Subtotals only | |
| 9.1 Low birthweight (< 1500 g) | 1 | 568 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.52 [0.13, 2.07] |
| 9.2 Low birthweight (< 2000 g) | 1 | 100 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.4 [0.48, 4.12] |
| 9.3 Low birthweight (< 2500 g) | 5 | 5577 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.95 [0.83, 1.09] |
| 10 Birthweight (g) Show forest plot | 5 | 5564 | Mean Difference (IV, Random, 95% CI) | 22.21 [‐27.23, 71.65] |
| 11 Baby admitted to the neonatal unit Show forest plot | 3 | 1435 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.74 [0.50, 1.11] |
| 12 Apgar score Show forest plot | 4 | Risk Ratio (M‐H, Fixed, 95% CI) | Subtotals only | |
| 12.1 1 minute Apgar < 5 | 1 | 377 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.83 [0.41, 1.67] |
| 12.2 5 minute Apgar < 7 | 4 | 1083 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.34 [0.15, 0.80] |
| 13 Late fetal heart rate decelerations Show forest plot | 1 | 4082 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.68 [0.53, 0.88] |
| 14 Meconium‐stained liquor Show forest plot | 1 | 4082 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.79 [0.63, 0.99] |
| 15 Meconium aspiration Show forest plot | 1 | 4082 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.63 [0.32, 1.26] |
| 16 Breech presentation Show forest plot | 1 | 4082 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.25 [0.81, 1.92] |
| 17 Placental abruption Show forest plot | 1 | 4082 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.96 [0.48, 1.94] |
| 18 Placental weight (g) Show forest plot | 2 | 4459 | Mean Difference (IV, Random, 95% CI) | ‐0.01 [‐22.16, 22.14] |
| 19 Hypoxic‐ischaemic encephalopathy Show forest plot | 1 | Risk Ratio (M‐H, Fixed, 95% CI) | Subtotals only | |
| 19.1 Any HIE | 1 | 4082 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.70 [0.36, 1.34] |
| 19.2 Mild HIE | 1 | 4082 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.38 [0.15, 0.98] |
| 19.3 Moderate HIE | 1 | 4082 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.02 [0.26, 4.09] |
| 19.4 Severe HIE | 1 | 4082 | Risk Ratio (M‐H, Fixed, 95% CI) | 2.56 [0.50, 13.19] |
| 20 Significant congenital abnormality Show forest plot | 1 | 4082 | Risk Ratio (M‐H, Fixed, 95% CI) | 2.05 [0.77, 5.45] |
| 21 Maternal side effects Show forest plot | 5 | Risk Ratio (M‐H, Fixed, 95% CI) | Subtotals only | |
| 21.1 Any gastrointestinal side effects | 4 | 1388 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.88 [0.69, 1.12] |
| 21.2 Any side effects | 1 | 47 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.0 [0.0, 0.0] |
| 22 Systolic blood pressure near birth (mm Hg) Show forest plot | 3 | 1432 | Mean Difference (IV, Fixed, 95% CI) | 1.0 [0.03, 1.97] |
| 23 Diastolic blood pressure near birth (mm Hg) Show forest plot | 3 | 1432 | Mean Difference (IV, Fixed, 95% CI) | 0.23 [‐0.67, 1.13] |
| 24 Pregnancy‐induced hypertension Show forest plot | 3 | 4284 | Risk Ratio (M‐H, Random, 95% CI) | 0.39 [0.11, 1.41] |
| 25 Eclampsia Show forest plot | 1 | 100 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.14 [0.01, 2.70] |
| 26 Need for maternal hospitalisation Show forest plot | 3 | 1158 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.65 [0.48, 0.86] |
| 27 Antepartum haemorrhage Show forest plot | 2 | 942 | Risk Ratio (M‐H, Random, 95% CI) | 0.53 [0.09, 3.15] |
| 28 Length of labour (hours) Show forest plot | 2 | 4650 | Mean Difference (IV, Fixed, 95% CI) | ‐0.00 [‐0.50, 0.50] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Perinatal mortality Show forest plot | 5 | 5903 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.10 [0.72, 1.67] |
| 1.1 Individually‐randomised | 4 | 5557 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.08 [0.70, 1.66] |
| 1.2 Cluster‐randomised | 1 | 346 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.48 [0.25, 8.76] |
| 2 Small‐for‐gestational age (< 10th percentile) Show forest plot | 3 | 1291 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.76 [0.54, 1.07] |
| 2.1 Individually‐randomised | 2 | 945 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.90 [0.60, 1.35] |
| 2.2 Cluster‐randomised | 1 | 346 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.53 [0.29, 0.98] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Perinatal mortality Show forest plot | 5 | Risk Ratio (M‐H, Fixed, 95% CI) | Subtotals only | |
| 1.1 ICC: 0.0002 | 5 | 6343 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.09 [0.73, 1.63] |
| 1.2 ICC: 0.002 | 5 | 6261 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.09 [0.72, 1.64] |
| 1.3 ICC: 0.02 | 5 | 5903 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.10 [0.72, 1.67] |
| 2 Stillbirth Show forest plot | 4 | 17376 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.74 [0.55, 1.01] |
| 2.1 ICC: 0.0002 | 4 | 5966 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.75 [0.45, 1.25] |
| 2.2 ICC: 0.002 | 4 | 5884 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.75 [0.45, 1.25] |
| 2.3 ICC: 0.02 | 4 | 5526 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.73 [0.43, 1.25] |
| 3 Neonatal death prior to hospital discharge Show forest plot | 4 | 16917 | Risk Ratio (M‐H, Fixed, 95% CI) | 2.19 [1.43, 3.36] |
| 3.1 ICC: 0.0002 | 4 | 5813 | Risk Ratio (M‐H, Fixed, 95% CI) | 2.18 [1.05, 4.53] |
| 3.2 ICC: 0.002 | 4 | 5731 | Risk Ratio (M‐H, Fixed, 95% CI) | 2.18 [1.05, 4.53] |
| 3.3 ICC: 0.02 | 4 | 5373 | Risk Ratio (M‐H, Fixed, 95% CI) | 2.21 [1.02, 4.75] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Perinatal mortality Show forest plot | 2 | 4459 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.05 [0.67, 1.65] |
| 2 Small‐for‐gestational age (< 10th percentile) Show forest plot | 1 | 377 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.76 [0.33, 1.77] |
| 3 Pre‐eclampsia Show forest plot | 1 | 374 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.93 [0.60, 1.44] |
