Introduction

Magnesium sulfate (MgSO4) is widely used in obstetric patients for indications such as treatment for preeclampsia, tocolysis, and neuroprotection when preterm delivery is imminent [1, 2]. Recent evidence suggests that antenatal magnesium sulfate is effective in the reduction of moderate to severe cerebral palsy by up to 40% [3, 4]. The American College of Obstetricians and Gynecologists (ACOG) in 2010 and the World Health Organization (WHO) in 2015 endorsed the use of MgSO4 administration to antenatal women to reduce the risk of cerebral palsy (CP) when delivery is anticipated before 32 weeks [5, 6]. A recent meta-analysis found a number needed to treat of 42 to prevent CP [7]. When used as a tocolytic in preterm deliveries, MgSO4 was shown to prolong the delivery despite its safety concerns as compared to other tocolytic agents such as calcium channel blockers and betamimetics [8]. MgSO4 in preeclampsia is known to reduce the risk of seizures by 50% which makes it an effective treatment in women to prevent eclampsia [9].

When administered to antenatal mothers, MgSO4 can rapidly cross the placenta resulting in an increase in fetal and newborn magnesium levels in the serum [10]. Magnesium is known to produce fecal impaction and intestinal atony in infants by depressing the cholinergic activity in mothers exposed to high levels of magnesium [11, 12]. Hypotonia, hypercalciuria, parathyroid gland suppression, abnormal bone mineralization, and neutrophil dysfunction are some of the side effects reported in infants when mothers are exposed to MgSO4 [13,14,15,16]. Intestinal hypomotility with increased meconium overload can lead to increased intraluminal pressure and thinning of intestinal walls which could be exaggerated and increase the risk of NEC or SIP when concomitantly exposed to steroids [17, 18]. MgSO4 acts as a competitive antagonist against calcium and decreases the intracellular calcium and impedes intestinal smooth muscle contractility by interfering with actin-myosin interaction [19]. Studies have shown maternal exposure to MgSO4 resulted in a decrease in intestinal blood flow and an increase in feeding intolerance in infants [11, 20, 21]. The alterations in intestinal blood flow can affect feeding intolerance and subsequent development of necrotizing enterocolitis (NEC) or spontaneous intestinal perforation (SIP) [22,23,24,25,26]. MgSO4 has been associated with increased incidence of NEC in smaller infants and higher doses in recent studies [27, 28]. A few studies also provided evidence to the contrary [29,30,31]. A recent meta-analysis by Sheperd et al. [32] did not show an association between antenatal MgSO4 and neonatal adverse outcomes.

Objective

We conducted a systematic review and meta-analysis to address three questions: Is there an association between antenatal MgSO4 administration and adverse gastrointestinal outcomes (GI) like NEC, SIP, feeding intolerance, time to reach full feeds, and GI-related mortality in preterm infants? In extreme preterm (<28 weeks) and very preterm infants (<32 weeks) does the administration of antenatal MgSO4 have any adverse gastrointestinal outcomes (GI)? And thirdly, is the association based on the different types of indications of MgSO4 like for neuroprotection, preeclampsia, or tocolysis?

Methods

The systematic review was conducted according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) reporting guidelines [33]. Our protocol was registered on the PROSPERO (CRD42022367246), the international prospective register for systematic reviews.

Eligibility criteria

Randomized or quasi-randomized clinical trials and non-randomized studies that reported GI-related morbidities when mothers of preterm infants were given MgSO4 for either neuroprotection, tocolysis, or preeclampsia were included. Studies that did not report prespecified neonatal outcomes were excluded. Preterm infants whose mothers did not receive antenatal magnesium sulfate were considered as comparators. The following outcomes were included in this meta-analysis:

  1. 1.

    Necrotizing enterocolitis (stage ≥ 2 NEC) (defined as per modified Bell’s staging) [34]

  2. 2.

    Surgical necrotizing enterocolitis (those needing surgical intervention)

  3. 3.

    Spontaneous intestinal perforation (SIP) (defined as per international standards) [35]

  4. 4.

    Feeding intolerance—as defined by authors.

  5. 5.

    Time to reach full feeds—Minimum of 150 ml/kg/d

  6. 6.

    Gastrointestinal (GI) related mortality—mortality related to NEC/SIP as reported by authors.

Information sources and search strategy

A systematic literature search was conducted in November 2022 using the following databases: PubMed, CINAHL Plus with Full Text (EBSCOhost), Embase (Elsevier), and the Cochrane Central Register of Controlled Trials, or CENTRAL (Ovid). The references of the identified studies were also screened. No date or language restrictions or search filters were employed. Searches were run by a medical librarian (N.A.) in consultation with authors A.P. and S.D. Details of the search strategy are provided in the online Supplementary File. Results were deduplicated using the SR-Accelerator Deduplicator tool.

Study selection

The deduplicated results were transferred to Rayyan software (www.rayyan.ai) for screening. Title and abstract screening and full-text review of articles were done independently by A.P. and S.D. Any discrepancies were resolved by discussion and consensus (P.C). If necessary, trial authors were contacted by email correspondence to request missing data.

Assessment of risk of bias

All included studies were assessed for methodological quality. The risk of bias was assessed using elements of the Cochrane Collaboration tool 2.0 for randomized studies [36]. For non-randomized studies, the risk of bias for included studies was assessed using a modified Newcastle–Ottawa scale [37]. The following domains were evaluated: selection, comparability, and outcome. A priori, a score of >7/9 was deemed low risk, a score of 4–6/9 was deemed a moderate risk, and a score of ≤3/9 was deemed a high risk of bias. Two authors (A.P. and S.D.) performed the risk of bias independently; conflicts were resolved after discussion and consensus. Similarly, A.P. and S.D. assessed the certainty of evidence (confidence in the estimate of effect) for each outcome based on the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) framework [38]. Any discrepancies were resolved by discussion and consensus.

Data synthesis

All the studies were combined and analyzed using Review Manager V.5.3 (Cochrane Collaboration, Nordic Cochrane Center, Copenhagen, Denmark). The mean difference with 95% CI was calculated for continuous outcomes. For dichotomous outcomes, the OR with 95% CI was calculated from the data provided in the studies. Adjusted odds ratios for potential confounders were extracted from the studies reporting these data. Studies reporting continuous variables as median and range or interquartile range were converted to mean and SD using a published calculator [39]. The random effects model was used to calculate summary statistics owing to anticipated heterogeneity. For some variables like gestational age, a fixed outcome model was used. Statistical heterogeneity was assessed by use of the Cochran Q statistic and by use of the I2 statistic, which is derived from the Q statistic and describes the proportion of total variation that is due to heterogeneity beyond chance. We used the Egger regression test and funnel plots to assess publication bias if more than 10 studies were available for that particular outcome.

Sensitivity analysis was performed for studies based on the different type of indications for MgSO4 (neuroprotection, preeclampsia, tocolysis) and based on gestational age (<28 weeks, <32 weeks)

Results

Study selection and Study characteristics

A total of 6695 articles were identified through database searching. After duplicates were removed, 4332 articles underwent title and abstract screening. Ninety-nine full-text articles were assessed. A final total of 44 articles including 6 RCTs [40,41,42,43,44,45] and 38 non-randomized studies [20, 21, 27, 29,30,31, 46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76] were deemed eligible to be included in the final meta-analysis (Summary Table 1). The PRISMA flow diagram [33] is shown in Fig. 1a. A list of excluded studies and search strategies are provided in the Online Supplementary File.

  1. i.

    Medical necrotizing enterocolitis (Stage ≥ 2 NEC) was reported in 40 included studies [20, 21, 29,30,31, 40,41,42,43,44, 46,47,48,49,50,51, 53,54,55,56,57,58,59,60,61,62,63,64, 68, 72] involving 51,466 infants. The pooled effect estimate favored the MgSO4 group but was not statistically significant between the two groups in non-randomized studies (n = 45,524, OR: 0.95; 95% CI: 0.84–1.08, I2- 5%) and randomized controlled trials (n = 5205 OR:1.24; 95% CI: 0.99–1.56, I2- 0%) (shown in Fig. 2a, b). Sensitivity analysis of infants <28 weeks reported in 5 studies (n = 5133, OR: 1.0; 95% CI: 0.83 to 1.20, I2– 0%) reported no difference between the two groups (shown in Fig. 3a)

  2. ii.

    Surgical NEC was reported in 5 studies [48, 63, 70, 71, 75] of the 44 studies involving 29,506 infants. The incidence of surgical NEC was significantly lower in the MgSO4 group compared with controls (n = 29,506 OR:0.74; 95% CI: 0.62–0.90, ARR: 0.47%) (shown in Fig. 2c). There was no heterogeneity noted between the studies. Sensitivity analysis revealed no significance for infants <32 weeks including 3 studies [70, 71, 75] (n-1163, OR: 0.60; 95% CI: 0.36 to 1.01, I2- 0%) (shown in Fig. 3c).

  3. iii.

    Spontaneous intestinal perforation was reported in 8 out of the 44 studies [27, 29, 31, 48, 63, 70, 71, 75] including 34,186 infants. Antenatal exposure of MgSO4 did not increase the incidence of SIP in preterm infants (n = 34,186, OR: 1.22, 95% CI: 0.94–1.58, I2–30%) (shown in Fig. 2d). Subgroup analysis of the incidence of SIP in infants <32 weeks was analyzed in 5 studies [27, 70, 71, 74, 75] and there was no difference between groups (n-1, 610 OR: 1.54, 95% CI: 0.67 to 3.55, I2 – 34%) (shown in Fig. 3d).

  4. iv.

    Time to reach full feeds was also analyzed and was reported in 8 [21, 52, 59, 65, 68, 71,72,73] of the 44 included studies involving 3,258 infants. Full enteral feeds were defined as a minimum of 150 cc/kg/day for analysis. There was no difference in time taken to achieve full feeds (20.3 d vs 20.7 d in the MgSO4 group vs the control group (MD:1.48: (95% CI: −0.76 to 3.72) (shown in Fig. 2e).

  5. v.

    Feeding intolerance was reported in 3 studies [21, 29, 73] out of the 44 involving 414 infants (n = 414, OR: 1.06 (95% CI: 0.64–1.76, I2- 12%) (shown in Fig. 2f).

  6. vi.

    GI-related mortality (NEC/SIP related) was reported in only one study (n = 4355) [29]. There was no statistically significant difference between the two groups (4.6% vs 3.8%, p - 0.18) in mortality. Stratified analysis based on gestational age (22–25 weeks & 26–27 weeks) did not show any significance either between the two groups.

Table 1 Summary table.
Full size table
Fig. 1
figure 1

Flowchart of search results (adapted from PRISMA 2021) [33] GI-gastrointestinal, NEC-Necrotizing enterocolitis, SIP-Spontaneous intestinal perforation; PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses.

Full size image
Fig. 2: Forest plots for primary outcomes.
figure 2figure 2

a Necrotizing enterocolitis (Stage ≥ 2) of observational studies. b Necrotizing enterocolitis (Stage ≥ 2) of Randomized Controlled Trials. c Surgical Necrotizing enterocolitis. d Spontaneous Intestinal Perforation (SIP). e Time to reach full feeds (150 cc/kg/d). f Feeding intolerance.

Full size image
Fig. 3: Forest plots for various sub-group analysis.
figure 3figure 3

a Necrotizing enterocolitis (Stage ≥ 2) (≤28 weeks). b Necrotizing enterocolitis (Stage ≥ 2) (≤32 weeks). c Surgical Necrotizing enterocolitis (≤32 weeks). d Spontaneous intestinal perforation (SIP) (≤32 weeks). e Necrotizing enterocolitis (Stage ≥ 2) (Neuroprotection & ≤32 weeks). f Necrotizing enterocolitis (Stage ≥ 2) (Neuroprotection & Preeclampsia & ≤32 weeks). g Spontaneous intestinal perforation (Neuroprotection). h Spontaneous intestinal perforation (≤32 weeks). i Time to reach full feeds (150 cc/kg/d) (≤32 weeks). j Time to reach full feeds (150 cc/kg/d) (Neuroprotection).

Full size image

Sensitivity and adjusted analysis

Sensitivity analysis was performed separately for all studies including infants less than 32 weeks of gestation. Seventeen studies involving 10,021 infants were analyzed [40, 43,44,45, 54, 55, 58,59,60,61,62, 64, 66,67,68,69,70,71,72,73,74,75]. The incidence of medical NEC was similar among the MgSO4 group as compared to controls (OR: 0.84; 95% CI: 0.66 to 1.09) (shown in Fig. 3b). Studies that used MgSO4 for neuroprotection alone (13 studies, n = 10,471) were analyzed, and neuroprotection or preeclampsia (5 studies, n = 4870) were analyzed separately. Both analyses did not show any statistical difference between the two groups (shown in Fig. 3e, f). For the outcome of SIP, sensitivity analysis was performed for very preterm infants and studies that used MgSO4 for neuroprotection. A total of 3 studies for neuroprotection [27, 31, 74] (shown in Fig. 3g) and seven studies [27, 29, 31, 70, 71, 74, 75] for <32 weeks were included (shown in Fig. 3h) and there was no difference between the two groups. For other indications like tocolysis and preeclampsia, not enough studies were available for metanalysis. For the outcome of time to reach full feeds subgroup analysis on preterm infants <32 weeks [52, 59, 71,72,73, 77] (shown in Fig. 3i) and studies that used MgSO4 for neuroprotection [52, 59, 77] (shown in Fig. 3j) showed no difference between two groups.

We performed an adjusted odds ratio analysis for medical NEC (5 studies) and surgical NEC (3 studies) and SIP (4 studies) (shown in Fig. 4a–c). None of the analysis showed any statistical significance between the two groups. Analysis for superior mesenteric artery (SMV) blood flow velocity (cm/s) indices like mean velocity (MV), peak systolic velocity (PSV), and end-diastolic velocity (EDV) was reported in only 2 studies [21, 73]. There was no difference between the two groups for all the indices (online supplementary file).

Fig. 4: Adjusted Odds Ratio (aOR) analysis.
figure 4

a Necrotizing enterocolitis (Stage ≥ 2). b Surgical Necrotizing enterocolitis. c Spontaneous intestinal perforation (SIP).

Full size image

Risk of bias (RoB) and Certainty of Evidence (CoE)

All the non-randomized studies were assessed for quality of evidence using the Newcastle-Ottawa scale [37]. All 38 studies were scored as good (>6) (Online supplementary file). The risk of bias assessment for RCTs was performed as per the Cochrane Risk of Bias 2.0 tool [78] in five domains (shown in Fig. 5). Of the six RCTs included, two studies were adjudged as high risk of bias [41, 42], one with some concerns [43], and two studies with low risk of bias (shown Fig. 5) [40, 45].

Fig. 5
figure 5

Risk of Bias 2.0 for randomized controlled trials.

Full size image

The certainty of evidence (CoE) as per GRADE [38] for all outcomes is provided in Table 2. For outcomes of medical and surgical NEC, SIP, and GI-related mortality, the CoE was determined to be “very low”.

Table 2 GRADE: Summary of findings (SoF) for main outcomes.
Full size table

Neither visual inspection of funnel plots nor the Egger test suggested publication or selection bias for the outcome of medical NEC [20, 21, 30, 31, 40,41,42,43,44,45,46,47,48,49,50,51, 53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72, 74,75,76] (41 studies) (shown in Fig. 6). The number of studies was insufficient to evaluate publication bias for the other outcomes.

Fig. 6
figure 6

Funnel plot assessing publication bias for Necrotizing enterocolitis (Stage ≥ 2).

Full size image

Discussion

Our review demonstrates that there is no difference in the incidence of medical necrotizing enterocolitis SIP, feeding intolerance, or time to reach full feeds in infants exposed to antenatal magnesium sulfate as compared to no exposure. We did not find any adverse association between the use of antenatal MgSO4 and GI morbidities in extreme preterm or very preterm infants. On the contrary, infants born to mothers who received antenatal MgSO4 had a lower incidence of surgical NEC by 26%. There was no difference in outcomes of NEC based on the type of indication like neuroprotection, preeclampsia, or tocolysis. Antenatal MgSO4 exposure did not increase gastrointestinal-related mortality in preterm infants.

Magnesium Sulfate has been used in imminent preterm deliveries less than 32 weeks for neuroprotection as a routine practice for the past decade after the large clinical trials [42, 44] and endorsement by ACOG and WHO [5, 6]. Its role in preeclampsia and as a tocolytic had been established long before [8, 9]. Concerns about maternal adverse effects were highlighted for the last 50 years but red flags were raised for its association with NEC and SIP in preterm infants since 2014 [27]. MgSO4 at higher cumulative doses and smaller infants (<25 weeks) increased the odds of NEC and SIP by 2–9 times [27, 28]. Contrasting evidence with absolutely no association was found in large observational studies [29, 63]. This difference could be explained by the adjustment of confounding factors for SIP/NEC in the later studies, for example, the use of steroids, indomethacin, small for gestational, and use of mechanical ventilation. We found no association between MgSO4 and NEC when adjusted odds ratio (aOR) was analysed, with a caveat that the adjusted confounding factors were not uniform across all studies and only 7 studies [29, 53, 54, 63, 66, 70, 75] reported the adjusted data (Shown in Fig. 4).

Various hypotheses were proposed for the association between antenatal MgSO4 and adverse GI-outcomes. These include calcium antagonism and decreasing actin-myosin interaction, intestinal hypomotility, increased resistance, and reduction of blood flow in the mesenteric artery. Our results are consistent with the results of the previous meta-analysis [4, 32]. Zeng et al reported on cerebral palsy and overall mortality with ten studies included and Shepherd et al reported-on NEC, SIP and NEC/SIP-related death as a part of an overall review of neonatal outcomes of antenatal MgSO4. However the previous reviews [4, 32] were not specific to the GI-outcomes, the number of included studies were significantly less compared to the current review which justifies the need for extensive and updated review.

We found that the antenatal MgSO4 group had decreased the incidence of surgical NEC by 26%. This finding is contrary to the conventional reports from previous studies [27, 28]. The reduced incidence of surgical NEC in preterm infants exposed to antenatal MgSO4 had been reported in previous observational studies [31, 63]. This positive effect could be due to the attenuation of peroxide-induced vasoconstriction by inhibiting thromboxane synthesis [79]. Thromboxane is an inciting agent for NEC [34]. MgSO4 also increases the blood flow by vasodilatation [79]. We also attribute the beneficial effects due to the overall benefit of MgSO4 in preterm infants and the fact that infants in the MgSO4 group could be less sick and more hemodynamically stable. This observation needs further exploration by larger studies.

The effect of MgSO4 on intestinal blood flow indices is inconclusive. Theoretically, MgSO4 will induce vasodilatation by calcium antagonism, attenuation of peroxide-induced vasoconstriction, inhibition of platelet activation induced by adenosine diphosphate, and release of vasoactive substances [79, 80]. Studies reporting on blood flow velocities in mesenteric arteries were limited [20, 21, 73] and none reported any significant change in mean velocity, end-diastolic velocity, peak systolic velocity, or the resistivity index between the two groups (online supplementary file).

Conclusions and implications

Our study findings do not support any clear association between antenatal magnesium sulfate administration and adverse gastrointestinal morbidities or gastrointestinal-related mortality in preterm infants. Concerns regarding the adverse effects of antenatal MgSO4 administration on GI morbidity or mortality should not be a hurdle in its routine use when indicated. This review should encourage clinicians to not deviate from feeding practices in the neonatal units when premature infants are exposed to antenatal MgSO4. The biological plausibility of MgSO4 causing NEC or SIP by decreased blood flow in mesenteric arteries needs further studies.

Strengths and limitations

The strength of this meta-analysis was that we used a comprehensive search strategy for our literature search, published our protocol a priori, and assessed the certainty of evidence using GRADE. To our knowledge, this is the largest meta-analysis on adverse neonatal GI outcomes and antenatal MgSO4. Nonetheless, there were several limitations. The majority of the studies were non-randomized, studies were heterogeneous, variable doses of MgSO4 were used in different studies and dose-related associations were not evaluated.