Summary of main results

Children with predominantly moderate to severe asthma presenting to the ED who were treated with IV MgSO₄ in addition to standard therapy showed a 68% reduction in the odds of hospital admission (OR 0.32, 95% CI 0.14 to 0.74). This result represents a dramatic reduction in admission rates, however confidence in the result was low due to the small number of children included in the analysis (n = 115) and variability between study results.

The variability in study results was reflected in the estimate of heterogeneity for the primary outcome. Examining the primary analysis, the data from Scarfone 2000 appears to lie outside those reported in the other two studies. It is difficult to explain the heterogeneity, as the studies were very similar in design and study population included. All of the studies were relatively small in size, increasing the imprecision in the reporting of effects. Repeating the analysis using a random‐effects model gave a result that was not statistically significant (OR 0.18, 95% CI 0.02 to 1.59). This again highlights that although the results favour the use of MgSO₄, the strength of that conclusion is very limited.

Data for the secondary outcomes was also limited. Two studies provided data for the outcome 'return to the emergency department within 48 hours' (Ciarallo 1996; Scarfone 2000), which favoured IV MgSO₄, but the confidence intervals did not exclude no difference (OR 0.40, 95% CI 0.02 to 10.30). Several studies reported improvements in spirometric data with treatment, but did not include the raw data for meta‐analysis. The message from the narrative description of results for symptoms score and spirometry in all but one case favoured treatment with IV MgSO₄. However, interpreting these results requires even greater caution.

Traditionally asthma has been referred to in the literature in terms of severity, and as such national guidelines are often framed in this manner. We intended to perform a subgroup analysis by severity subgroup. We had proposed to get a blinded review author to rate the severity of each study based on the inclusion criteria, but the limited amount of available data prevented this subgroup analysis. However, examining the inclusion criteria for the three studies that presented data for the primary outcome (Ciarallo 1996; Ciarallo 2000; Scarfone 2000), we can see that all children had a peak expiratory flow (PEF) less than 70%, which would have traditionally been referred to as moderate to severe exacerbations.

While the evidence is based on a moderate‐to‐severe cohort, it is acknowledged that severity in paediatric practice is often a clinical judgement as spirometry in young children is frequently limited in its accuracy.

Again, data on side effects was inconsistently reported, which limited the meta‐analysis, but on the whole the number of side effects was small. In fact, the only reported side effect was a relaxed sensation in two children (Ciarallo 1996). There was no reported evidence of the haemodynamic instability that is often historically reported with the use of MgSO₄. The lack of any reported harm is encouraging, but the ability to extrapolate this result is again limited by the small number of studies included in the review.

If we accept that there is weak evidence of the effectiveness of MgSO₄, the question of timing of administration needs to be addressed. The included study protocols administered IV MgSO₄ after failure to improve with first‐line nebulised bronchodilator therapy. We would agree that MgSO₄ remains a second‐line therapy after patients have failed to respond to more evidence‐based nebulised bronchodilator therapies. In clinical practice IV MgSO₄ is often used in conjunction with other IV bronchodilators (salbutamol and aminophylline). As the studies included in this review only examined MgSO₄ as a single IV agent, evaluating any possible synergistic interaction was beyond the scope of this review.

Overall completeness and applicability of evidence

Three studies contributed data to the primary outcome assessing the potential effect of IV MgSO₄ on preventing hospital admission (Ciarallo 1996; Ciarallo 2000; Scarfone 2000). The studies used between 25 and 75 mg/kg of MgSO₄, but all were given in a 20‐minute infusion. We cannot comment on optimal dosing given the limited data in the review; at this time it would be prudent to comply with national guidance which is in line with the range in this review. All three studies were conducted in the USA, which has implications on the applicability of the evidence to other healthcare systems, especially with regard to the comedications administered previous to MgSO₄. The US studies gave children nebulised bronchodilators (albuterol, ipratropium bromide, or both) and IV methylprednisolone, which may not be the practice in other countries.

A common criticism of hospital admission as an outcome is that the criteria to make this decision is often not standardised, and it is widely acknowledged that admission rates vary by country and even region. However, where randomisation is appropriate it should provide a reliable between‐group difference in moderate‐to‐severe patients. It may be argued that severe asthma by its nature requires admission to hospital, in which case admission to paediatric intensive care unit (PICU) may be a possible outcome. However, it must also be acknowledged that in some institutions some therapies mandate admission to a PICU (certain IV therapies), in which case the need for intubation and mechanical ventilatory support may represent an alternative. Rightly or wrongly, burden of disease is widely reported as a rising admission rate, therefore it stands to reason we should attempt to use this as an outcome of treatment efficacy.

The children in the studies had a mean age of around 11 in Ciarallo 1996 and Ciarallo 2000 and were younger on average in Scarfone 2000, at around 7 in the active group and 5 in the placebo group. This mean age would support the recommendation in British Thoracic Society guidance, however younger children were also included and treated safely in the studies.

The age imbalance within Scarfone 2000 may have affected the results and suggests that the sample sizes of the included studies, of which Scarfone 2000 is the largest at 54, are unlikely to have been sufficiently large enough to assume that important baseline variables were evenly distributed by randomisation. These differences and limitations in the design and conduct of the studies contributing data means that the implications for practice must be carefully considered.

No more than two studies contributed to any of the secondary outcomes, and we were unable to conduct a meta‐analysis for five of the eight secondary outcomes. As such, even the three secondary outcomes for which we could include data (duration in ED, return to ED within 48 hours, and duration of hospital stay) included data for no more than 88 children, which severely limits the applicability of any conclusions that can be drawn. Outcomes that could not be supported by analyses included admission to intensive care unit, spirometry, and adverse events, which we fully expected to be better reported in studies conducted in an emergency setting. For these outcomes, we tried to describe narratively information in the studies that was not fully reported in order to give the fullest picture possible, but they represent an important gap in the evidence base.

Previous reviews have examined subgroups based on exacerbation severity and concluded that IV MgSO₄ has a role primarily in the treatment of severe exacerbations. We have already alluded to the difficulties of assessing asthma severity in children, but our results are most applicable to children with moderate or severe asthma attacks. This group of patients is more likely to present to the ED, therefore we would agree that currently the role of IV MgSO₄ would be in moderate to severe patients who have failed to respond to inhaled therapies, and not to children presenting in primary care. Clinically MgSO₄ is also used prior to or concurrently with nebuliser therapy in severe life‐threatening episodes where severe obstruction to air flow limits the effectiveness of inhaled therapies. Studies in this review did not test IV MgSO₄ in these situations, but we excluded a study recruiting children in status asthmaticus that did not include a placebo group (Irazuzta 2016). Prioritisation of therapies in extremis needs to be based on sound clinical judgement and experience. It is important to recognise that asthma attacks run a continuum, often lapsing and relapsing over several days, and what role magnesium has in altering the course of an exacerbation is unknown.

Quality of the evidence

While there were some uncertainties in the study procedures, particularly with selection and attrition bias, we considered the included studies to be of good methodological quality overall, and so none of the outcomes were downgraded for risk of bias. Additionally, the five studies closely matched the inclusion criteria set out in the protocol, therefore we did not consider the evidence to be compromised by indirectness.

However, our confidence in the findings was reduced by serious imprecision in the estimates, largely due to very small numbers of studies and participants. In the case of the primary outcome, serious differences in what the three studies found resulted in the effect becoming imprecise when we performed a random‐effects sensitivity analysis, and this reduced our confidence in the main finding.

Incomplete outcome reporting in two studies affected some of the secondary outcomes that we were not able to meta‐analyse (spirometry, intensive care admissions, vital signs). As mentioned above, the small number of included studies overall and the very small amount of data suitable for meta‐analysis may suggest publication bias, and limits the conclusions that can be drawn.

Potential biases in the review process

This review examined included studies for bias against predetermined criteria as specified by current Cochrane methodology. We identified internal reporting biases in two of the included studies (Ciarallo 1996; Ciarallo 2000), which may have impacted on the completeness of our results, but this was considered in the relevant GRADE ratings. Ciarallo 1996 reported in their methodology a wide range of parameters that were collected on participants but only reported some of these outcomes, which all had significant results. The authors stated that the trial recruitment period was cut short because of a change in ED practice. Their original power calculation stipulated 40 participants to detect a 25% difference in PEFR; only 31 participants were enrolled, and as such the estimate of effect can be exaggerated. However, we did not include PEFR in our meta‐analysis, and so our results were not affected. The authors also note there was a difference in the baseline characteristic (FEV1) between the groups, which could again have led to an overestimate of the effect in the treatment group.

We implemented the planned methods as far as possible, but in some cases the small number of studies meant this was not possible or valid. We have listed these instances in Differences between protocol and review.

Agreements and disagreements with other studies or reviews

Having urged caution in the interpretation of the primary analysis, it can be seen that the result favouring admission reduction in the IV MgSO₄ group is in keeping with the recently published adult review. A recent review of IV MgSO₄ for acute asthma in the ED in adult patients reported a reduction in the odds of hospital admission of 25% (OR 0.75, 95% CI 0.60 to 0.92) based on 11 studies including 1769 participants (Kew 2014). This systematic review of the adult evidence contains new data from the largest adult study to date (Goodacre 2013, n = 752). They also found an improvement in spirometric parameters in participants treated with IV MgSO₄. An older Cochrane review that assessed adults and children together only found a treatment effect in a severe patient subgroup (Rowe 2000).

Caution is required in extrapolating results from adult asthma studies, as childhood asthma is reported to respond differently to some established therapies.

Cheuk 2005 published a meta‐analysis of paediatric patients, which reported hospital admission as its primary outcome. It included four studies in this analysis, three of the studies presented here and also admission data for Devi 1997 (admission data not included in published article and was not made available despite attempts to contact the study author). The meta‐analysis concluded that magnesium was likely to reduce hospital admission and improve bronchospasm.

Most recently in the paediatric literature Ohn 2014 published a review stating magnesium should be given to all children presenting to hospital with acute severe asthma. This conclusion was based on two meta‐analyses that included adult and paediatric patients, Rowe 2001 and Shan 2013, and one recent randomised controlled trial, Torres 2012, which was excluded from this review as it compared IV MgSO₄ versus standard care and not placebo. Torres 2012 reported a statistically significant reduction in the need for mechanical ventilation in the IV MgSO₄ group.

MgSO₄ can also be administered as an aerosol by nebuliser devices. There has been increasing interest in its use in adults and children with acute asthma, and a Cochrane review in 2012 found no improvement in lung function and no decrease in hospital admission with its use (Powell 2012). A large paediatric randomised controlled trial examining its use in children in the ED (N = 508) was recently published (Powell 2013), and while the new study is yet to be incorporated in to their Cochrane review (Powell 2012), the study did not find a clinical difference in asthma severity score and did not report hospital admission data. Admission to PICU/high dependency unit was reported and was required by 35 out of 508 children (9% in the treatment group versus 6% in the placebo group).

IV MgSO₄ is commonly used in paediatric practice. In a survey of 183 ED consultants in the UK and Ireland, 94.5% report using it in their management of acute wheeze, and nearly one‐third (28.4%) use it as their first‐line intravenous agent (Lyttle 2015). The overwhelming narrative from the paediatric and adult literature supports the use of IV MgSO₄. This conclusion, while in keeping with this review, has at times been based on a very limited evidence base. Despite the largest adult trial to date showing no statistical difference, its inclusion in the most recent Cochrane analysis has provided a more robust base for the use of IV MgSO₄ in adults. In paediatric practice the literature has often drawn from adult data and paediatric data not obtained from randomised controlled trials. This review does draw a similar conclusion to the previous paediatric meta‐analysis, but highlights the very severe limitations of the evidence.