Prospective observational study of clinical outcomes after intravenous magnesium for moderate and severe acute asthma exacerbations in children

Abstract

Background:

There is limited knowledge whether intravenous magnesium (IV-Mg) improves outcomes in children with acute asthma exacerbations.

Objective:

Examine whether IV-Mg improve outcomes in children with moderate and severe exacerbations.

Methods:

We performed a secondary analysis using data from a prospective observational cohort of children aged 5–17 years with moderate and severe exacerbations. Standardized treatment included systemic corticosteroid and inhaled albuterol, with consideration of IV-Mg (75mg/kg) for patients with insufficient response after 20 minutes. Propensity score (PS) models were used to examine associations of IV-Mg treatment with change of the validated Acute Asthma Intensity Research Score (AAAIRS), hospitalization rate, and time to spacing of inhaled albuterol 4 hours or greater (Q4hr) among hospitalized participants.

Results:

Among 301 children, median [IQR] age was 8.1 [6.4, 10.2] years, 170 (57%) Black race, 201 (67%) male sex, and 84 (28%) received IV-Mg. In a PS covariate adjusted multivariable linear regression model, IV-Mg treatment was associated with a 2-hour increase in the AAIRS (β-coefficient 0.98; 95% CI 0.20, 1.77), indicating increased exacerbation severity. Three additional PS-based models yielded similar results. Participants receiving IV-Mg had 5.8-fold (95% CI 2.8, 11.9) and 6.8-fold (95% CI 3.6, 12.9) greater odds of hospitalization in PS-based multivariable regression models. Among hospitalized participants there was no difference in time to Q4hr albuterol in a PS covariate-adjusted Cox proportional hazards model (HR 1.2; 95% CI 0.8, 1.8).

Conclusion:

Among children with moderate and severe exacerbations, IV-Mg is associated with increased exacerbation severity, increased risk of hospitalization, and no acceleration in exacerbation resolution among hospitalized participants.

INTRODUCTION

NHLBI expert guidelines recommend consideration of intravenous magnesium (IV-Mg) in children with moderate or severe asthma exacerbations who have an incomplete response to systemic corticosteroid (CCS) and inhaled albuterol.¹ A 2020 focused update to these guidelines did not change this recommendation.²

The primary mechanism of action of IV-Mg is thought to be competition for the smooth muscle calcium channel, decreased cytosolic Ca²⁺ concentration, and smooth muscle relaxation.³ A Cochrane systematic review of randomized trials concluded that although fewer children treated with IV-Mg required hospital admission, the quality of evidence was poor due to small sample size and varying results between studies.⁴ In addition, the reviewers could not determine whether reduced hospitalizations were associated with exacerbation severity, age, or other treatments.

Moreover, recent studies have shown that IV-Mg is typically administered late in the ED course, and most children are hospitalized after receiving IV-Mg.^{5, 6} Yet despite limited evidence for the safety and efficacy of IV-Mg in children, the use of IV-Mg increased in the United States over the past decade.⁵

In a recently reported study, children with moderate or severe exacerbations after 1 hour of treatment with CCS and nebulized albuterol were randomized to an additional1 hour of nebulized albuterol plus either nebulized Mg or saline placebo.⁷ After this nebulized treatment, ED physicians prescribed additional treatments as deemed necessary, including IV-Mg, and made hospitalization decisions. A post hoc analysis of this post-randomization data indicated that IV-Mg was associated with greatly increased odds of hospitalization.⁸ This association was also present among participants who achieved mild exacerbation severity at the time of disposition decision-making.

Further, a recent study of patients using a large administrative database reported that IV-Mg is not associated with improved in-hospital outcomes such as non-invasive positive pressure ventilation, mechanical ventilation, duration of mechanical ventilation or length of stay.⁹ This study and other retrospectively derived cohorts may be limited by confounding by severity, a type of confounding by indication.^{10, 11} Furthermore, administrative databases are unable to determine more granular clinical outcomes that include validated measures of pre-treatment severity and response-to-treatment.

The limitations of these studies emphasize the need for prospectively derived cohorts that incorporate measures of clinical severity and response-to-treatment to examine the effectiveness of IV-Mg in children with acute asthma exacerbations. The objective of this study was to use data from a large, prospective study of children with moderate and severe acute asthma exacerbations to examine whether IV-Mg administered early in ED management decreases exacerbation severity, hospitalizations, and time to spacing inhaled albuterol to 4 hours or greater in hospitalized patients.

METHODS

Study Population

We performed secondary analyses of data from a prospective, observational cohort of 933 children aged 5 – 17 years with doctor-diagnosed asthma and acute exacerbations in an urban, tertiary children’s hospital emergency department (ED).^{12, 13} These analyses included participants with moderate and severe exacerbations defined using the validated Acute Asthma Intensity Research Score (AAIRS, Figure 1, described below).^{14, 15} An exacerbation was defined as cough, dyspnea, wheezing, and/or chest pain, and need for systemic corticosteroid (CCS) and inhaled albuterol treatments as determined by the clinical team. The primary objective of the parent study was to model and validate a clinical prediction rule with the outcome need for hospitalization, details of which have been previously reported.^{12, 13} The study protocol was reviewed and approved by our institutional review board (protocol #080058). We obtained written, informed consent from a parent and assent from each participant prior to study enrollment.

Figure 1. The Acute Asthma Intensity Research Score (AAIRS).

SCM, Sternocleidomastoid

* Any visible use of accessory muscle group (Yes/No)

Severity levels: mild 1-6; moderate 7-11; severe 12-16

Adapted from: Acad Emerg Med. 2015 Oct;22(10):E25-6.¹⁶

The clinical team had exclusive decision-making for study participants, including decisions for IV-Mg, other treatments, and hospitalization. The attending clinician making treatment decisions (e.g., IV-Mg) and the attending making disposition decisions (e.g., hospitalization) may have differed.

Study data were not made available to the clinical team. Standard treatment directed by ED asthma clinical practice guideline (CPG), based on NHLBI expert asthma guidelines, included CCS and inhaled albuterol for all patients with moderate and severe exacerbations.¹ In accordance with the guidelines, ipratropium was administered during the first hour of nebulized albuterol but was not administered with MDI albuterol. The CPG recommended consideration of IV-Mg, 75 mg/kg (maximum 2.5 gm), for moderate-severity exacerbations if the patient had insufficient response to CCS and albuterol after 20 minutes. For patients with severe exacerbations with insufficient response to CCS and inhaled albuterol, the CPG recommended consideration of IV-Mg, parenteral β-agonist (terbutaline or epinephrine), heliox, ketamine, and/or bilevel positive airway pressure (BiPAP).Insufficient response was not defined further in the CPG but made reference to the NHLBI guidelines, in which the terms “incomplete response” and “no improvement after initial treatment” are used but not further defined.¹

Clinical measures of disease severity, treatment, and response

For each participant we measured and recorded relevant variables before (pre-treatment) and at 2 and 4 hours after administration of CCS, if the participant remained in the ED at those times. These variables included a comprehensive pulmonary examination; the timing of inhaled albuterol, IV-Mg, and other treatments; and the seven components of the AAIRS. The AAIRS assesses multiple domains of exacerbation severity including work of breathing, air entry, expiratory phase prolongation, and oxygenation. It has a range of 0 to 16 points (16 most severe) with decreases indicating clinical improvement. ^{14, 15} Participants were enrolled by the principal investigator (DHA) or an advanced paramedic trained in study measurements, including assessment of each component of the AAIRS.¹⁶

The AAIRS has been validated as a measure of exacerbation severity against the criterion measures %-predicted FEV₁ and %-predicted airway resistance by impulse oscillometry, and as a measure of response-to-treatment against change of these measures.^{14, 16, 17} Additionally, we demonstrated the validity of the AAIRS as the driver of disposition and treatment decisions for children with asthma exacerbations as part of a Clinical Practice Guideline that was associated with reduced admissions, ICU utilization, and hospital length of stay.¹⁸ A 2 point or greater decrease was clinically meaningful for decisions to de-escalate treatment.^{15, 18}

Expert guidelines recommend consideration of IV-Mg treatment only for children with moderate or severe exacerbations who have an incomplete response to standard treatment of systemic CCS and initial inhaled albuterol.¹⁹ With this in mind, our analyses used data only for participants with pre-treatment AAIRS values of 7 or higher, indicating moderate or severe exacerbations.

Outcomes of Interest

The primary outcome was change of the AAIRS after 2 hours of ED-administered treatment. All patients receive CCS according to our ED treatment guideline. Initiation of treatment was defined as the time of CCS dosing. Secondary outcomes included hospitalization rate and, for hospitalized participants, time to spacing inhaled albuterol to 4 hours or greater (Q4hr). Time to Q4hr albuterol is a metric of exacerbation resolution that has been incorporated in expert asthma guidelines and clinical practice guidelines.^{1, 18}

Propensity Score Modelling

Associations of IV-Mg treatment with our outcomes of interest may be distorted by confounding, in particular confounding by indication.^{11, 20} Confounding by indication occurs when some reasons for choosing a treatment (e.g., clinician preferences and experience) also affect treatment outcome. For example, knowledge of prior PICU admission for asthma may influence a clinician’s choice of additional treatments beyond IV-Mg that in turn impact treatment response. Additionally, clinicians who are more likely to use IV-Mg may have different thresholds for admitting a patient to the hospital or ICU. Thus, confounding by indication may occur with IV-Mg treatment because treated patients have risks for the outcomes of interest that are influenced by clinician characteristics or other variables.

Propensity scores (PS) are a tool to reduce the effects of confounding by indication.^{11, 21–24} The PS is the probability that a study participant will receive the treatment of interest based on variables that may influence treatment decisions and outcomes of interest. Propensity score models have been likened to “semi-randomization” in studies of other asthma treatments.²⁵

To model the PS, we fitted a generalized additive mixed-effects logistic regression model with IV-Mg treatment in the first 2 hours of care as the dependent variable. The generalized additive model has the flexibility to model nonlinear relationships by implementing penalized smoothing splines.²⁶ Model predictors are displayed in Table I. The output of the model was a PS with values of 0 to 1 that estimated the probability for each participant of being treated with IV-Mg in the first 2 hours after systemic CCS.

Table I.

Predictor variables in multiple logistic regression to model propensity score

Covariate	Variable type a
Pre-treatment AAIRS	Continuous
Age, years	Continuous
Body mass index	Continuous
Gender	Dichotomous
Race a	Categorical
Ethnicity b	Categorical
Insurance type c	Categorical
Prior endotracheal intubation for asthma	Dichotomous
Prior PICU admission for asthma	Dichotomous
Asthma exacerbation in past year	Dichotomous
Symptom duration, hours	Continuous
Albuterol treatment in 1st 2 hours d	Categorical
Terbutaline, parenteral	Dichotomous
Bilevel Positive Airway Pressure (BiPAP) treatment	Dichotomous
Attending making treatment decisions in 1st 2 hours	Nominal
Attending making disposition decision e	Nominal
Systemic CCS prior to ED arrival, designated as hours of treatment	Continuous
Concurrent influenza	Dichotomous

Abbreviations: AAIRS, Acute Asthma Intensity Research Score; CCS, systemic corticosteroid

^aBlack, Caucasian, other

^bHispanic ethnicity defined according to NIH policy NOT-OD-01-053

^cCommercial, Medicaid, none

^dOne-time nebulized, continuous nebulized, metered-dose inhaler

^eDischarge to home, admit to floor, admit to PICU

Statistical Analysis

Primary Outcome, 2-hour change of the AAIRS

After generating the propensity scores, we modelled the primary outcome, 2-hour change of the AAIRS, using four PS-based models.^{24, 27} Agreement or disagreement of these models would strengthen or diminish, respectively, conclusions that might be drawn from the analyses.

First, we used a PS covariate adjusted, additive linear multivariable regression model. Natural regression cubic splines with five knots were applied for continuous variables to account for non-linear associations. Attending clinician making treatment decisions and attending making disposition decision, were modelled as random effects. Independent variables included IV-Mg treatment as a dependent fixed effect, the PS, pre-treatment AAIRS, clinician making treatment decisions, clinician making disposition decision (e.g., hospital admission), age, gender, race, ethnicity, body mass index (BMI), insurance type, prior PICU admission for asthma, asthma exacerbation in preceding 12 months, treatment with CCS prior to ED arrival, symptom duration, concurrent influenza, and treatments in the first 2 hours of care including IV-Mg, BiPAP, albuterol (intermittent or continuous nebulized or MDI as separate variables), parenteral epinephrine, and parenteral terbutaline. The β-coefficient from the model represented the estimated average treatment effect of IV-Mg measured using change of the AAIRS, with a negative change indicating decreased exacerbation severity.

For the second model we used inverse probability of treatment weighting (IPTW) in which the PS is used to calculate the IPTW for each participant. This makes possible a weighted model with each observation counted differently based on this weighting. For example, some may be counted as 1.5 whereas others may be counted as 0.8 observations. We then used the IPTW in a generalized additive mixed-effects linear regression model with change of the AAIRS as the dependent variable. This model included the covariates of the first model except the PS.

For our third model we used an optimal matching to perform 1:n matching with one participant who received IV-Mg matched to multiple participants who did not. A balance table was created after matching. We then used the same generalized additive mixed-effects multivariable regression model with the covariates of the second model, with IV-Mg treatment as a fixed effect and matched groups as a random effect. This approach would best reflect the average treatment effect of IV-Mg.

Though prior endotracheal intubation (ETI) for asthma was infrequent, this history might disproportionately influence the decision to use IV-Mg and/or bilevel positive airway pressure (BiPAP). To examine potential marginal imbalances of these and other covariates, for the fourth model we first examined covariate balance after forced matching within three groups: those with no prior ETI, those with prior ETI, and those with BiPAP for the current episode. Using data for the entire cohort, each participant who received IV-Mg was then forced matched 1:1 with one who did not, using optimal matching. Matched pairs were then included in a multivariable regression model as described for the third model. This approach would best reflect the average treatment effect for the treated.

Finally, IV-Mg results in smooth muscle relaxation with potential for both bronchodilation and pulmonary vasodilation. This may result in ventilation-perfusion mismatch with a decrease in oxygen saturation (SpO₂) while ventilation and other domains of severity are improving.^{28, 29} Because SpO₂ contributes 0, 1, or 2 points to the overall AAIRS, this change may increase exacerbation severity measured using the AAIRS, though a patient is otherwise improving. With this in mind, we used a two-sample Wilcoxon rank-sum test to examine whether the change in points contributed by the SpO₂ component during the 2-hour treatment interval differed between participants receiving and not receiving IV-Mg.

Secondary outcomes

To examine hospitalization rate, we used a PS-adjusted multivariable logistic regression model and an IPTW logistic regression model, each with hospital admission as the dependent variable. Covariates for the PS-adjusted model included those of the first model for the primary outcome, except BiPAP because all participants receiving this treatment were hospitalized. Covariates for the IPTW model included the same covariates, except the PS.

Time to Q4hr albuterol was examined using a PS covariate adjusted Cox proportional hazards model that included IV-Mg treatment as the independent fixed effect and the PS as a covariate modelled as a cubic spline with five knots. Covariates for this model included those of the first model for the primary outcome.

Continuous data are presented as mean (SD) or median [interquartile range, IQR] as appropriate, dichotomous and categorical data as proportions, output of linear regression models as β-coefficients (95% confidence interval, CI), output of logistic regression models as adjusted odds ratios (aOR) with 95% CI, and output of the Cox proportional hazards model as hazard ratio (HR) with 95% CI.. Analyses were performed using R, version 4.0.3 and R packages “mgcv,” “optmatch,” and “RItools.”^{26, 30, 31} This report adheres to the 22 items of the STROBE guidelines for reporting observational studies.³²

RESULTS

Among the cohort of 933 participants enrolled between April 2008 and February 2013 570 had pre-treatment AAIRS ≥7, and 301 of these met study inclusion criteria and were available in the ED at 2 hours after treatment initiation (Figure 2). These 301 participants had median [IQR] age 8.1 [6.4, 10.2] years, 170 (57%) Black race, 201 (67%) male sex, and 84 (28%) received IV-Mg. Demographic and asthma characteristics of the 301 children according to IV-Mg treatment status are presented in Table II. The majority had an acute asthma exacerbation in the preceding year and had pre-treatment AAIRS indicating moderate severity exacerbations. There were univariate differences by IV-Mg treatment status in age, prior PICU admission for asthma, the AAIRS before and at 2 hours after treatment initiation, change of the AAIRS after 2 hours of treatment, epinephrine treatment, and hospital admission.

Figure 2. Flow diagram of study participants.

AAIRS, acute asthma intensity research score; ED, emergency department; IV-Mg, intravenous magnesium; PICU, pediatric intensive care unit.

Table II.

Demographic and asthma characteristics of 301 children ages 5 to 17 years with moderate and severe acute asthma exacerbations in a pediatric emergency department

Characteristic	IV magnesium treatment status
	Not treated (n = 217)	Treated (n = 84)	P value
Age, median [IQR], years	7.8 [6.2, 9.5]	9.0 [6.7, 11.4]	0.007 a
Male gender	149 (68.7)	52 (61.9)	0.327 b
Race			0.972 c
Black	122 (56.2)	48 (57.1)
Caucasian	83 (38.2)	31 (36.9)
Other	12 (5.5)	5 (6.0)
Insurance			0.727 c
Medicaid	133 (61.3)	55 (65.5)
Commercial	80 (36.9)	27 (32.1)
Asthma exacerbation in past year	133 (61.3)	54 (64.3)	0.728 c
Prior PICU admission for asthma	39 (18.0)	29 (34.5)	0.003 c
Symptom duration, median [IQR], hours	1.0 [0.5, 3.0]	1.8 [0.5, 3.0]	0.398 a
AAIRS before treatment, median [IQR]	8.0 [8.0, 9.0]	9.0 [8.0, 10.0]	<0.001 a
AAIRS change 0-2 hours, median [IQR]	−5.0 [−7.0, −2.0]	−4.0 [−6.0, −1.0]	0.004 a
Concurrent influenza	5 (2.3)	1 (1.2)	0.873 c
Corticosteroid treatment prior to ED arrival	55 (25.3)	28 (33.3)	0.212 c
Treatments applied in 1st 2-hours
Albuterol single nebulization (2.5 mg)	17 (7.8)	0 (0.0)	0.018 c
Albuterol continuous nebulization (10 mg/hr)	194 (89.4)	80 (95.2)	0.172 c
Terbutaline	5 (2.3)	5 (6.0)	0.220 c
Epinephrine	11 (5.1)	13 (15.5)	0.006 c
Time to Q 4-hour albuterol, median [IQR] d	8.2 [2.2, 12.0]	7.5 [4.5, 13.9]	0.330 a
Hospital admission	60 (27.6)	62 (73.8)	<0.001 c

Values are n (%) unless otherwise noted

Abbreviations: AAIRS, Acute Asthma Intensity Research Score; BMI, body mass index; ED, emergency department; IM, intramuscular; IQR, interquartile range; MDI, metered-dose inhaler; PICU, pediatric intensive care unit; SD, standard deviation

^aKruskal-Wallis;

^bOne-way ANOVA;

^cChi-squared with Yates correction for continuity

^dTime to Q 4-hour albuterol after admission order for hospitalized participants

The PS development model achieved an area under curve of 81% and did not require trimming, indicating good fit. The 1:n matched model had good covariate balance for most variables, as indicated by standardized differences ≤0.25 (Table III). Exceptions included age, BMI, and symptom duration, though differences in symptom duration were not clinically meaningful. For the 1:1 force-matched model, similar covariate balance was again observed for most variables when all participants were examined (Table IV). Similar balance was observed for those with no prior ETI, those with prior ETI, and those with BiPAP for the current episode (data not shown).

Table III.

Balance for covariates in 1:n matching by propensity score comparing participants who received IV magnesium with participants who did not

Variable	Value according to IV magnesium treatment status
	Not treated	Treated	Difference	Stratified difference a	Standardized difference a	P value
Age, years	8.3000	9.4844	1.1844	−0.1572	0.4249	0.7114
Male gender	0.6866	0.6190	−0.0676	−0.0131	0.0788	0.8679
Race
Black	0.5622	0.5714	0.0092	0.0122	0.0772	0.8741
White	0.3825	0.3690	−0.0134	−0.0278	0.0766	0.7170
Other	0.0553	0.0595	0.0042	0.0155	0.0343	0.6506
Hispanic ethnicity b	0.0829	0.0595	−0.0234	0.0052	0.0412	0.8994
Asthma exacerbation in past year	0.6129	0.6429	0.0300	0.0338	0.0771	0.6610
BMI	17.6674	19.3956	1.7282	−0.5039	0.6283	0.4226
Prior PICU admission for asthma	0.1797	0.3452	0.1655	−0.0005	0.0663	0.9936
Prior ETI for asthma	0.0092	0.1190	0.1098	0.0035	0.0087	0.6831
Symptom duration, hours	2.4770	2.5000	0.0230	−0.0074	0.7524	0.9921
AAIRS before treatment	8.5161	9.2262	0.7101	−0.1341	0.2303	0.5605
Concurrent influenza	0.0230	0.0119	−0.0111	0.0023	0.0208	0.9114
CCS treatment prior to ED arrival	0.2535	0.3333	0.0799	−0.0521	0.0764	0.4956
Treatments applied in 1st 2-hours
Albuterol MDI	0.0092	0.0119	0.0027	−0.0022	0.0139	0.8744
Albuterol continuous nebulization	0.8940	0.9524	0.0584	0.0084	0.0345	0.8084
Terbutaline	0.0230	0.0595	0.0365	−0.0053	0.0334	0.8736
Epinephrine	0.0507	0.1548	0.1041	−0.0006	0.0406	0.9875
BiPAP	0.0138	0.1071	0.0933	0.0000	0.0000	1.0000

Abbreviation: AAIRS, Acute Asthma Intensity Research Score; BiPAP, bilevel positive airway pressure; BMI, body mass index; CCS, systemic corticosteroid; ETI, endotracheal intubation; IM, intramuscular; MDI, metered-dose inhaler; PICu, pediatric intensive care unit

^aStratified difference, difference between matched pairs; standardized difference, difference between matched pairs divided by standard deviation

^bHispanic ethnicity defined according to NIH policy NOT-OD-01-053

Table IV.

Balance for covariates in 1:1 matching by propensity score comparing participants who received IV magnesium with participants who did not

Variable	Value according to IV magnesium treatment status
	Not treated	Treated	Difference	Standardized difference a	P value
Age, years	9.3141	9.4070	0.0930	0.5262	0.8598
Male gender	0.7042	0.6479	−0.0563	0.0788	0.4748
Race
Black	0.4648	0.5493	0.0845	0.0842	0.3156
White	0.4366	0.3803	−0.0563	0.0828	0.4962
Other	0.0986	0.0704	−0.0282	0.0468	0.5477
Hispanic ethnicity b	0.0704	0.0704	0.0000	0.0431	1.0000
Asthma exacerbation in past year	0.6901	0.6338	−0.0563	0.0797	0.4795
BMI	19.2375	19.0208	−0.2167	0.8654	0.8023
Prior PICU admission for asthma	0.2817	0.2676	−0.0141	0.0752	0.8514
Prior ETI for asthma	0.0282	0.0282	0.0000	0.0279	1.0000
Symptom duration, hours	3.0070	2.3028	−0.7042	0.9401	0.4538
AAIRS before treatment	9.1127	9.1549	0.0423	0.2786	0.8794
Concurrent influenza	0.0000	0.0141	0.0141	0.0141	0.3173
CCS treatment prior to ED arrival	0.3662	0.2958	−0.0704	0.0793	0.3743
Treatments applied in 1st 2-hours
Albuterol MDI	0.0141	0.0141	0.0000	0.0198	1.0000
Albuterol continuous nebulization	0.9437	0.9437	0.0000	0.0388	1.0000
Terbutaline	0.0563	0.0563	0.0000	0.0388	1.0000
Epinephrine	0.1127	0.1268	0.0141	0.0547	0.7967
BiPAP	0.0423	0.0423	0.0000	0.0339	1.0000

Albuterol by one-time nebulized treatment excluded because all participants treated with IV-Mg received this treatment.

Abbreviation: AAIRS, Acute Asthma Intensity Research Score; BiPAP, bilevel positive airway pressure; BMI, body mass index; CCS, systemic corticosteroid; ETI, endotracheal intubation; IM, intramuscular; MDI, metered-dose inhaler; PICU, pediatric intensive care unit

^aStandardized difference, difference between matched pairs divided by standard deviation

^bHispanic ethnicity defined according to NIH policy NOT-OD-01-053

The PS covariate adjusted multivariable linear model examining 2-hour change of the AAIRS yielded a β-coefficient of 0.98 (95% CI 0.20, 1.77; Table V). This positive β-coefficient indicates an increase in acute asthma exacerbation severity associated with IV-Mg treatment during the first 2 hours of treatment. The IPTW and the 1:n matching models yielded similar β-coefficients. There was no difference in change of points contributed by the SpO₂ component to the change of AAIRS (p = 0.63, Wilcoxon rank-sum). This indicates that ventilation-perfusion mismatch did not account for the positive β-coefficient for change of AAIRS in these models. The 1:1 forced matching model yielded a lower coefficient that was not statistically significant.

Table V.

β-coefficients for IV-Mg treatment using propensity score adjusted multivariable linear regression models to predict change of severity measured using the Acute Asthma Intensity Research Score

Propensity-score Adjusted Model	β-coefficient a	95% CI	P value
PS covariate adjusted b	0.98	0.20, 1.77	0.015
Inverse probability of treatment weighting b	0.89	0.27, 1.52	0.006
1:n matching on propensity score c	0.93	0.17, 1.68	0.017
1:1 forced matching on propensity score c	0.73	−0.17, 1.62	0.114

Abbreviations: PS, propensity score

^aβ-coefficient for IV-Mg treatment. Positive β-coefficient indicates increased exacerbation severity during the 2-hour interval.

^bCovariates: PS, pre-treatment AAIRS, clinician making treatment decisions, clinician making disposition decision, age, gender, race, ethnicity, insurance type, prior PICU admission for asthma, treatment with CCS prior to ED arrival, and treatments in the first 2 hours of care including IV-Mg, BiPAP, albuterol by nebulizer, albuterol by MDI, parenteral epinephrine, and parenteral terbutaline

^cCovariates same as PS covariate adjusted model except PS

For the secondary outcome, hospital admission, participants receiving IV-Mg in the first 2 hours of treatment had 5.8-fold (95% CI 2.8, 11.9) greater odds of hospitalization in the PS-adjusted and 6.8-fold (95% CI 3.6, 12.9) greater odds in the IPTW multivariable regression models. The PS covariate-adjusted Cox proportional hazards model did not demonstrate a difference in time to Q4hr albuterol among hospitalized participants (hazard ratio 1.2; 95% CI 0.8, 1.8; p = 0.35).

DISCUSSION

The results of this investigation indicate that IV-Mg treatment is associated with increased exacerbation severity of approximately 1 point on the validated 17-point AAIRS. We have reported that a 2 point or greater decrease is clinically meaningful for decisions to de-escalate treatment but have not examined the threshold for an increase of the AAIRS that warrants escalation of treatment.^{15, 18} Nonetheless, an increase of 1 point indicates an increase of clinical severity and is cause for concern that IV-Mg treatment may be deleterious in children with moderate and severe exacerbations. Moreover, there was no difference in the change of points contributed by the SpO₂ component to the total change of AAIRS between participants receiving and not receiving IV-Mg. This indicates that the increased AAIRS associated with IV-Mg was not a result of VQ mismatch during IV-Mg treatment.

In addition, participants treated with IV-Mg had approximately 6-fold greater odds of hospitalization after adjustment for the PS and relevant clinical covariates. That the increased risk of hospitalization is associated with untoward physiological effects of IV-Mg is supported by the increased clinical severity measured using the AAIRS in our PS-adjusted models.

Our results are consistent with prior reports that IV-Mg for moderate and severe exacerbations is associated with increased risk of hospitalization.^{5, 8} These include a post hoc secondary analysis of data from a large, double-blind randomized clinical trial in 7 Canadian tertiary pediatric EDs.⁸ Participating children with moderate or severe exacerbations after initial treatment with CCS and inhaled albuterol and ipratropium were randomized to receive inhaled albuterol plus nebulized magnesium or saline placebo. After randomization and receipt of this nebulized treatment, physicians administered additional treatments deemed necessary, including IV-Mg, and made hospitalization decisions. Approximately 26% received IV-Mg, similar to our cohort, and more than 88% of those receiving IV-Mg were admitted to hospital. Analysis of this post-randomization data indicated that IV-Mg was associated with greatly increased odds of hospitalization (aOR 22.7 for years 2011-2016 and aOR 4.2 for years 2017-2019). Though these analyses were performed using post-randomization data and did not include PS-adjusted models to mitigate confounding by severity, the results are concerning and consistent with the results of our study.

Finally, IV-Mg treatment was not associated with decreased time to Q4hr albuterol, a widely accepted metric of exacerbation resolution, among hospitalized participants. This suggests that IV-Mg does not accelerate exacerbation resolution in this population. This is consistent with a retrospective cohort study using registry data from U.S. children’s hospitals.⁹ These investigators used PS-matching to compare outcomes in children hospitalized with asthma exacerbations who received IV-Mg with those who did not. Among children receiving IV-Mg, there were no associations with decreased use of non-invasive positive pressure ventilation, mechanical ventilation, or length of stay.

Strengths of our study include use of the AAIRS, a validated measure of exacerbation severity and response to treatment, as the primary outcome. Additionally, we prospectively assessed score components, relevant participant characteristics and clinical variables at defined time points after initiation of treatment. Additionally, the outcomes of clinical severity, time to Q4hr albuterol, and hospitalization are relevant to clinicians caring for children with acute asthma exacerbations and are patient-centered domains of exacerbation severity, response to treatment, and disease burden.

Limitations

Our study has limitations. First, this was a secondary analysis of data from a parent study with the primary objective to model and validate a clinical prediction rule. However, though the data were not recorded for the purpose of the present study, study variables were prospectively measured and recorded. Second, participants were not randomized to receive IV-Mg, though the timing and dose of IV-Mg were standardized in a clinical practice guideline during the study period. The propensity for receipt of IV-Mg would likely have differed, resulting in possible unmeasured residual confounders that were not accounted for in PS modelling. Among unmeasured confounders is clinical momentum to escalate care after placing an IV and administering an adjunct treatment such as IV-Mg. However, clinical momentum does not account for increased exacerbation severity identified in this study, measured using physiologic markers such as the AAIRS and important clinical outcomes such as time to Q4hr albuterol. Third, though the CPG specified consideration of IV-Mg for patients with insufficient response 20 minutes after initiating CCS and inhaled albuterol, we did not record the specific time of IV-Mg treatment. Fourth, enrollment for this research ended in 2013, and there may have been interval changes in exacerbation. However, our institutional asthma CPG was developed in accordance with NHLBI expert panel guidelines in use during the study period and currently.¹ In addition, we would not expect the direction of effect of IV-Mg to change over time, and other treatments for acute exacerbations have not changed over time.

Interpretation

The need to de-implement low-value care has been identified if we are to reduce healthcare costs and improve outcomes.³³ As noted by the Cochrane reviewers, randomized trials of IV-Mg have provided low quality evidence of efficacy to reduce hospitalizations in children with acute asthma exacerbations.⁴ In addition, recent investigations demonstrate no clinical benefit of this treatment.^{8, 9} Though IV-Mg is generally safe and well tolerated, the results of this observational study of real-world use of IV-Mg indicate that it is associated with increased exacerbation severity and hospitalizations and does not accelerate clinical improvement for hospitalized patients. Absent a randomized controlled trial powered to examine important outcomes, there is insufficient evidence to support use of IV-Mg for moderate and severe acute asthma exacerbations in children.

Highlights box.

What is already known about this topic?

IV magnesium use for acute asthma exacerbations is increasing, yet there is limited knowledge whether intravenous magnesium it improves outcomes.

What does this article add to our knowledge?

This research indicates that IV magnesium is associated with increased exacerbation severity, increased hospitalizations, and no change in rate of exacerbation resolution among hospitalized children.

How does this study impact current management guidelines?

Absent data from a randomized controlled trial powered to examine important outcomes, there is insufficient evidence to support use of IV-magnesium for moderate and severe acute asthma exacerbations in children.

Abbreviations:

AAIRS

acute asthma intensity research score

BiPAP

bilevel positive airway pressure

BMI

body mass index

CCS

systemic corticosteroid

CI

confidence interval

CPG

clinical practice guideline

ED

emergency department

ETI

endotracheal intubation

ICU

intensive care unit

IPTW

inverse probability of treatment weighting

IQR

interquartile range

IV-Mg

intravenous magnesium

MDI

metered dose inhaler

PS

propensity score

Q4hr

every 4 hours

SD

standard deviation

SpO₂

oxygen saturation by pulse oximetry