Hostname: page-component-74656f9884-skzmx Total loading time: 0 Render date: 2026-11-06T11:57:45.771Z Has data issue: false hasContentIssue false
  • English
  • Français

Circulatory effects of intravenous magnesium sulphate in children in the cardiac ICU: insights using high-fidelity physiologic monitoring

Published online by Cambridge University Press:  06 November 2026

Jordan Griffiths Open the ORCID record for Jordan Griffiths [Opens in a new window] ,
Aarnav D. Shah ,
Wesam Sourour ,
Raya Safa Open the ORCID record for Raya Safa [Opens in a new window]  and
Rohit S. Loomba
Jordan Griffiths*
Affiliation:
Ann & Robert H Lurie Children’s Hospital of Chicago, USA
Aarnav D. Shah
Affiliation:
Ann & Robert H Lurie Children’s Hospital of Chicago, USA
Wesam Sourour
Affiliation:
Ann & Robert H Lurie Children’s Hospital of Chicago, USA Feinberg School of Medicine, Northwestern University, USA
Raya Safa
Affiliation:
Ann & Robert H Lurie Children’s Hospital of Chicago, USA Feinberg School of Medicine, Northwestern University, USA
Rohit S. Loomba
Affiliation:
Ann & Robert H Lurie Children’s Hospital of Chicago, USA Feinberg School of Medicine, Northwestern University, USA
*
Corresponding author: Jordan Griffiths; Email: jgriffiths@luriechildrens.org
Rights & Permissions [Opens in a new window]

Abstract

Background:

Intravenous magnesium sulphate is routinely administered in paediatric cardiac ICUs for the correction of hypomagnesemia. Its physiologic effects in critically ill children have not been clearly characterised. We aimed to describe circulatory and oximetric responses following intravenous magnesium sulphate administration.

Methods:

A single-centre retrospective, descriptive cohort study of paediatric cardiac ICU patients aged 1–12 years who received intravenous magnesium sulphate. Temporal responses in physiologic indices were characterised using locally estimated scatterplot smoothing and restricted cubic spline regression adjusted for patient characteristics.

Results:

57 magnesium administrations were analysed. Blood pressure demonstrated early and minor relative declines followed by a progressive increase above baseline. Modelled venous resistance reached a nadir of 7.2% below baseline (95% bootstrap confidence interval −24.77% to 25.19%) and modelled mean circulatory filling pressure reached a nadir of 2.2% below baseline (95% bootstrap confidence interval −10.85% to 8.55%) at 124 minutes. Regional oxygen saturation remained above baseline throughout most of the observation period, reaching a maximum increase of 2.6% (95% bootstrap confidence interval −0.38% to 7.55%) at 480 minutes among 30 administrations with regional oxygen saturation data, while modelled oxygen extraction showed a sustained reduction below baseline.

Conclusion:

Findings suggest modest, coordinated changes in venous and arterial vascular physiology, with preserved regional oxygenation. Wide confidence intervals indicate uncertainty and heterogeneity in physiologic response patterns. These findings provide novel insight into the temporal circulatory response to magnesium sulphate in critically ill children and support prospective studies to further characterise these responses and establish causal relationships.

Keywords

magnesium sulphatepaediatric critical carecardiac intensive carehigh fidelity datavenous physiology

Information

Type
Original Article
Information
Cardiology in the Young , First View , pp. 1 - 9
Check for updates
Copyright
© The Author(s), 2026. Published by Cambridge University Press

Introduction

Intravenous magnesium sulphate is frequently administered in the paediatric cardiac ICU for the treatment of hypomagnesemia. Reference Hassan and El Haddad1 Hypomagnesemia is common in paediatric cardiac intensive care settings and has been associated with an increased incidence of atrial dysrhythmias, prolonged intensive care stay, and adverse clinical outcomes. Reference Hassan and El Haddad1–Reference Mildh, Hiippala and Rautiainen5 Consequently, optimisation of serum magnesium has become a routine component of care in the paediatric cardiac ICU.

Magnesium acts as a calcium antagonist at L-type calcium channels within both the myocardium and vascular smooth muscle. In the myocardium, this may produce negative inotropic effects, while action on vascular smooth muscle can promote vasodilation. Reference Fawcett, Haxby and Male6–Reference Wu and Lipsius8 Additional effects on nitric oxide signalling, catecholamine release, and prostacyclin production further contribute to its vasoactive profile. Reference Schutten, Joosten and de Borst9

The physiologic significance of myocardial calcium channel antagonism in paediatric patients is likely to be variable, as immature myocardium depends more heavily on extracellular calcium influx than intracellular calcium stores. Reference Andropoulos, Andropoulos, Stayer, Mossad and Miller-Hance10 In critically ill paediatric cardiac patients, pharmacologic effects are further influenced by underlying cardiovascular pathology, with many patients exhibiting limited cardiovascular reserve and complex circulatory physiology. Consequently, even modest changes in vascular tone or myocardial performance may substantially affect systemic oxygen delivery in patients with CHD, postoperative cardiac physiology, or critical illness. Reference Marino, Tabbutt and MacLaren11,Reference Datt, Wadhhwa and Sharma12 Despite the widespread use of intravenous magnesium sulphate in this population, its integrated circulatory effects remain incompletely characterised.

Previous paediatric studies have primarily evaluated magnesium sulphate for the treatment of asthma, chronic hypertension, or arrhythmia prevention. Reference Hassan and El Haddad1,Reference Han, Fang and Wei13,Reference Graff, Stevenson and Berkenbosch14 Characterisation of its circulatory effects has been typically limited to secondary safety outcomes, and reported findings have been variable. Some studies have demonstrated reductions in heart rate and arterial pressure Reference Graff, Stevenson and Berkenbosch14–Reference Abdulatif, Ahmed and Mukhtar17 , whereas others have reported minimal measurable haemodynamic change. Reference Rower, Liu and Yu7,Reference Dalmora, Lago and Hendler18

The increasing availability of continuous high-fidelity data provides an opportunity to revisit the physiologic effects of commonly administered therapies with substantially greater temporal resolution than has previously been possible. Rather than defining isolated pharmacodynamic effects, these datasets permit characterisation of the integrated haemodynamic response that accompanies therapeutic intervention through time. By incorporating markers of venous physiology, arterial vascular tone, and tissue oxygenation, high-fidelity monitoring offers a framework for understanding the dynamic circulatory consequences of pharmacologic therapies beyond changes in routinely measured vitals.

The aim of the present study was to characterise the temporal circulatory and oximetric responses following intravenous magnesium sulphate administration in critically ill paediatric cardiac patients aged 1 through 12 years using continuous high-fidelity physiologic monitoring. Given the known vasodilatory effects of magnesium, we hypothesised that intravenous magnesium sulphate administration would be associated with reductions in vascular tone, but the overall circulatory response would depend on a balance between vasodilation, potential negative inotropic effects, and the capacity to compensate for such changes. We sought to characterise whether these physiologic responses were associated with preserved or impaired tissue oxygenation and circulatory performance.

Materials and methods

Study design

This study was a retrospective, observational, descriptive cohort study conducted in a single paediatric cardiac ICU. The unit was equipped with the Sickbay platform (Medical Informatics Corporation, Houston, TX), which enabled continuous collection of high-fidelity physiologic data during routine clinical care. The primary objective of the study was to observe and characterise temporal physiologic changes following the administration of intravenous magnesium and assess for evidence of altered circulatory physiology.

Patient selection

This study included patients admitted to the paediatric cardiac ICU between January 1, 2022, and June 1, 2026, who received intravenous magnesium sulphate for the correction of hypomagnesemia. The population was limited to patients aged 1 to 12 years. Children younger than 1 year were excluded for a separate analysis due to known differences in circulatory physiology between infants and older children and thus were not included in the present study.

For each hospital admission during the study period, the first magnesium administration was identified. This administration was excluded if another magnesium administration occurred within the subsequent 8-hour observation window, as the physiologic response could not then be attributed to a single administration. Patients could contribute more than one eligible administration if they had multiple separate hospital admissions during the study period; administrations from separate admissions were considered independently. In total, 57 eligible magnesium administrations from 53 unique patients were included.

Patients needed high-fidelity physiologic monitoring during the observation period to be included. Heart rate and peripheral oxygen saturation were used as surrogate markers of monitoring continuity. Magnesium administrations with 20% or greater missing heart rate or peripheral oxygen saturation data during the observation period were excluded, as substantial gaps suggested that continuous physiologic monitoring was not maintained. Administrations with less than 20% missing data were retained. Intermittent missing observations within retained administrations were left missing and were not imputed or carried forward.

Physiologic variables

Measured physiologic variables included heart rate, systolic blood pressure, diastolic blood pressure, mean arterial blood pressure, central venous pressure, respiratory rate, end-tidal carbon dioxide, pulse oximetry-derived peripheral oxygen saturation, and near-infrared spectroscopy-derived regional oxygen saturation. High-fidelity physiologic data were acquired at 1-second intervals over a9-hour observation period, comprising 60 minutes before and 8 hours after magnesium administration. For analysis, data were aggregated into 30-second bins to reduce computational burden while preserving temporal resolution.

Heart rate was obtained from bedside telemetry. Arterial blood pressure measurements were preferentially sourced from invasive arterial catheters when available; otherwise, noninvasive cuff measurements were used. Pulse pressure was calculated as the difference between systolic and diastolic blood pressure. Central venous pressure measurements were obtained from central venous catheters when available. Respiratory rate and peripheral oxygen saturation were obtained from pulse oximetry monitoring. End-tidal carbon dioxide monitoring was available for observation periods when patients were intubated. Cerebral and renal near infrared spectroscopy measurements were obtained from the INVOS 5100. When more than one near-infrared spectroscopy channel was available, regional oxygen saturation was calculated as the mean of the available near-infrared spectroscopy channels; cerebral and renal measurements were not distinguished in the calculation. Thus, modelled regional oxygen saturation represents a composite measure of the available regional near-infrared spectroscopy measurements for each administration.

Derived physiologic variables included modelled oxygen extraction, modelled systemic vascular resistance, modelled mean circulatory filling pressure, and modelled venous resistance. Oxygen extraction was estimated as the arteriovenous oxygen saturation difference divided by peripheral oxygen saturation. Peripheral oxygen saturation was used as a noninvasive surrogate for arterial oxygen saturation. Composite regional oxygen saturation derived from near-infrared spectroscopy was used as a noninvasive surrogate of venous-weighted regional tissue oxygenation. Prior paediatric studies have demonstrated correlation between near-infrared spectroscopy-derived regional oxygen saturation and venous oxygen saturation in children with CHD. Reference Ghanayem and Hoffman19,Reference Loomba, Rausa and Sheikholeslami20 Modelled systemic vascular resistance was estimated by dividing the pressure gradient between mean arterial pressure and central venous pressure by modelled cardiac output. Modelled cardiac output was estimated using a Fick-based approach, dividing modelled oxygen consumption (estimated using the Seckeler Equation Reference Seckeler, Hirsch and Beekman21 ) by the approximated arteriovenous oxygen content difference.

Modelled mean circulatory filling pressure was estimated using a previously described Guytonian analogue equation (0.96 × central venous pressure + 0.04 × mean arterial pressure + 0.2 × cardiac output). Reference Guyton, Lindsey and Kaufmann22,Reference Parkin and Leaning23 The underlying Guytonian analogue model was originally developed from adult data and has been applied in the present study as a modelled physiologic estimate in a paediatric population. Modelled venous resistance was approximated as the product of systemic vascular resistance and mean circulatory filling pressure divided by mean arterial pressure. Reference Guyton, Lindsey and Kaufmann22

These derived variables were incorporated to facilitate exploratory interpretation of magnesium-associated physiology within a framework that integrates principles of arterial circulation, venous circulation, and tissue oxygenation. Accordingly, these measures were intended to provide modelled approximations of circulatory trends rather than exact physiologic measurements and were derived from shared underlying physiologic inputs; therefore, they should not be interpreted as independent measures of distinct physiologic responses. Derived physiologic variables were calculated only when all required measured variables were available; if a required input was missing, the corresponding derived variable was not calculated for that observation.

Clinical variables

Clinical variables were extracted from the electronic medical record. The time of magnesium administration commencement was recorded and used as the reference time point for all physiologic analyses. Other patient and treatment data extracted included magnesium dose normalised to body weight (mg/kg), patient age, patient weight, body surface area, single ventricle (yes/no), postoperative admission (yes/no), serum magnesium concentration prior to magnesium infusion, haemoglobin level, and vasoactive inotrope score. Reference Gaies, Gurney and Yen24 These variables were used in the derivation of modelled physiologic parameters and as covariates in subsequent analyses.

Statistical analyses

Each magnesium administration was analysed as a unique event. The time of magnesium sulphate administration was defined as time 0. Baseline physiology was defined using the median for each physiologic variable from the 60-minute period immediately preceding each magnesium administration.

Subsequent physiologic data from time 0 until 480 minutes after administration were normalised relative to this baseline period and expressed as percent change from baseline. All analyses were performed using Python version 3.13.15 with pandas 2.2.3, NumPy 2.1.3, SciPy 1.16.3, Matplotlib 3.10.0, statsmodels 0.14.6, and patsy 1.0.2. Analysis code is available from the corresponding author upon reasonable request. P-values less than 0.05 were considered statistically significant. The analyses were considered exploratory, and p values are therefore reported as unadjusted measures of statistical evidence rather than as confirmatory significance tests.

Locally estimated scatterplot smoothing

Locally estimated scatterplot smoothing analyses were used to visualise univariable nonlinear temporal physiologic trajectories following magnesium administration. Mean physiologic trajectories were displayed with corresponding 95% confidence intervals across all administrations.

Spline regressions

Restricted cubic spline regression was performed to evaluate nonlinear temporal trajectories following magnesium administration while adjusting for prespecified covariates. For all spline models, the dependent variable was percent change from baseline for the corresponding physiologic parameter. Time was modelled as the primary independent variable.

Covariates included magnesium dose normalised to body weight, patient age, patient weight, serum magnesium concentration, haemoglobin concentration, vasoactive-inotropic score, postoperative status, and single-ventricle physiology. Continuous covariates were standardised (mean 0, standard deviation 1) before model fitting to improve model convergence and numerical stability.

Vasoactive-ionotropic score was specifically incorporated given its direct effects on vascular tone and hemodynamics. Other concurrent medications, ventilator adjustments, and major physiologic events occurring during the 8-hour observation period were not individually modelled because incorporating multiple time-varying interventions would introduce substantial model complexity and potential collinearity among concurrent therapies.

All physiologic trajectories were aligned to the start time of magnesium administration. Consequently, interventions or clinical events occurring at variable times relative to magnesium administration across individual observations would be expected to contribute to individual-level variability rather than produce a consistent time-aligned trend in the cohort-level spline model.

Restricted cubic spline basis functions with five degrees of freedom were used to permit flexible modelling of nonlinear temporal trajectories. Restricted cubic spline models were fit using ordinary least squares regression with cluster-robust covariance estimation clustered by administration. Spline-predicted trajectories were generated continuously across the full post-administration period and plotted with corresponding 95% confidence intervals. Bootstrap resampling with replacement was performed 200 times to generate maximum and minimum modelled percent change with 95% confidence interval for each physiologic variable.

Physiologic interpretation focused on the modelled trajectory shape, overall spline significance, directionality of change, peak modelled effects, and timing of peak effect. Individual spline basis coefficients were produced but not interpreted independently. Directionality of physiologic change was determined from the modelled spline trajectory.

Results

Cohort characteristics

Fifty-three unique patients were included. Of the 53 unique patients, 50 contributed one eligible administration, 2 contributed two eligible administrations from separate admissions, and 1 contributed three eligible administrations from separate admissions, resulting in 57 eligible magnesium administrations.

The median age was 50 months (interquartile range 23–78), and median weight was 14.0 kg (interquartile range 10.5–19.5). Median pre-administration serum magnesium concentration was 1.7 mg/dL (interquartile range 1.6–1.8). The median magnesium dose was 50 mg/kg (interquartile range 50–50 mg/kg), corresponding to a median absolute dose of 700 mg (interquartile range 500–1000 mg). Four administrations reached the maximum dose of 2000 mg. All magnesium was administered intravenously over 1 hour.

Twenty administrations (35%) occurred during postoperative admissions, and 17 administrations (30%) were in patients with single-ventricle physiology. The median vasoactive-inotropic score was 3 (interquartile range 0–5). An invasive arterial catheter was available during 49 observation periods, and 41 patients were intubated at the start of the observation period.

High-fidelity physiologic monitoring demonstrated excellent data completeness, with heart rate and peripheral oxygen saturation available for > 99% and > 95% of recorded timepoints, respectively. Across the 57 eligible magnesium administrations, heart rate data were available for all administrations, arterial blood pressure for 52, respiratory rate for 56, peripheral oxygen saturation for 55, central venous pressure for 45, end-tidal carbon dioxide for 41, and near-infrared spectroscopy for 30. Modelled oxygen extraction and Fick-estimated cardiac output were available for 30 administrations, while modelled systemic vascular resistance, mean circulatory filling pressure, and venous resistance were available for 25.

Locally estimated scatterplot smoothing

Locally estimated scatterplot smoothing analyses were used to visualise unadjusted temporal physiologic trajectories following magnesium administration (Figure 1). Overall, the trajectories demonstrated coordinated changes in venous physiology, arterial physiology, and oximetric variables following magnesium administration.

Figure 1.

Locally estimated scatterplot smoothing (LOESS) curves demonstrating the unadjusted percent change from baseline in physiologic variables following intravenous magnesium sulphate administration. Time 0 represents magnesium administration. Percent change from baseline was calculated using the median value during the 60-minute pre-administration period and plotted over the subsequent 8 hours. The solid line represents the LOESS estimate and the shaded region the 95% confidence interval. Green and orange markers indicate the minimum and maximum percent change from baseline, respectively, with the corresponding time of occurrence. The number of administrations contributing to each variable was heart rate, n = 57; systolic blood pressure, diastolic blood pressure, and mean arterial pressure, n = 52; pulse pressure, n = 52; central venous pressure, n = 45; respiratory rate, n = 56; peripheral saturation, n = 55; end-tidal carbon dioxide, n = 41; regional oxygen saturation, n = 30; estimated oxygen extraction, n = 30; estimated systemic vascular resistance, n = 25; estimated mean circulatory filling pressure, n = 25; and estimated venous resistance, n = 25.

Multiple graphs depict changes in physiological variables after magnesium administration.

Heart rate had an early decrease below baseline during the early observation period, followed by a gradual increase above baseline during the later half of the observation period. Blood pressure variables were initially above baseline at time 0, lowered towards baseline during the first hour following magnesium administration and subsequently increased throughout the remainder of the observation period.

Indices of venous physiology demonstrated a pattern of early reduction followed by delayed recovery. Mean circulatory filling pressure and venous resistance both decreased during the first 3 hours following magnesium administration before subsequently increasing above baseline later in the observation period. Central venous pressure remained near baseline during the first hour before increasing above baseline for the remainder of the observation period. Systemic vascular resistance remained below baseline for several hours following magnesium administration. Pulse pressure demonstrated a sustained upward trajectory throughout the observation period.

Oximetric variables demonstrated coordinated changes following magnesium administration. Regional oxygen saturation increased early and remained above baseline throughout most of the study period, while oxygen extraction demonstrated a reciprocal decline. Peripheral oxygen saturation remained largely unchanged throughout the observation period. End-tidal carbon dioxide and respiratory rate both demonstrated progressive increases over time. Peak and nadir percent changes are annotated on the Locally estimated scatterplot smoothing analyses plots in Figure 1

Spline regression analysis

Restricted cubic spline analyses demonstrated evidence of nonlinear temporal association (all unadjusted global p < 0.01). Spline plots for all discussed variables are presented in Figure 2

Figure 2.

Restricted cubic spline analyses demonstrating the adjusted percent change from baseline in studied variables. Time 0 represents magnesium administration. The solid blue line represents the adjusted spline estimate, with shaded areas indicating the 95% confidence interval. Green and orange markers indicate the minimum and maximum percent change from baseline, respectively. The number of administrations contributing to each variable was: heart rate, n = 57; systolic blood pressure, diastolic blood pressure, and mean arterial pressure, n = 52; pulse pressure, n = 52; central venous pressure, n = 45; respiratory rate, n = 56; peripheral saturation, n = 55; end-tidal carbon dioxide, n = 41; regional oxygen saturation, n = 30; estimated oxygen extraction, n = 30; estimated systemic vascular resistance, n = 25; estimated mean circulatory filling pressure, n = 25; and estimated venous resistance, n = 25.

Multiple line graphs depict changes in various physiological variables over time after magnesium administration.

Heart rate demonstrated minimal early change, with a nadir of 1.9% below baseline (95% bootstrap confidence interval −4.29% to −0.43%) at 88 minutes, before gradually returning to baseline and subsequently increasing to 2.0% above baseline (95% bootstrap confidence interval −0.73% to 6.59%) by the end of the observation period. Systolic, diastolic, and mean arterial blood pressure all demonstrated early relative declines over the first 90 minutes, then progressively increased, peaking at 6.43% (95% bootstrap confidence interval 2.72% to 11.17%), 5.49% (95% bootstrap confidence interval 2.30% to 9.94%), and 5.77% (95% bootstrap confidence interval 2.72% to 9.66%) above baseline at 447, 426, and 423 minutes, respectively.

Indices of venous physiology demonstrated early reductions then later increased above baseline. Central venous pressure initially declined towards baseline during the first hour before progressively increasing and peaking at 14.6% (95% bootstrap confidence interval 5.57% to 25.13%) above baseline at 410 minutes. Modelled mean circulatory filling pressure reached a nadir of 2.2% (95% bootstrap confidence interval −10.85% to 8.55%) below baseline at 124 minutes, at which time trajectory changed and a peak of 10.4% (95% bootstrap CI 0.55% to 23.23%) above baseline was reached at 431 minutes. Modelled venous resistance reached a nadir of 7.2% (95% bootstrap confidence interval −24.77% to 25.19%) below baseline at 255 minutes before increasing to 8.9% (95% bootstrap confidence interval −8.83% to 42.27%) above baseline at 445 minutes.

Modelled systemic vascular resistance showed a relative reduction below baseline for approximately 5 hours following magnesium administration, reaching a modelled nadir of 6.4% (95% bootstrap confidence interval −26.46% to 29.08%) before increasing modestly to 4.4% (95% bootstrap confidence interval −11.58% to 42.59%) above baseline during the final 2 hours of observation. Pulse pressure remained elevated throughout the observation period, increasing progressively from 7.2% above baseline at time 0 to 18.1% above baseline by the end of the study window.

Oximetric variables demonstrated coordinated changes following magnesium administration. Regional oxygen saturation showed an early increase and remained above baseline throughout most of the observation period, reaching a maximal modelled increase of 2.6% (95% bootstrap confidence interval −0.38% to 7.55%) at 480 minutes. Modelled oxygen extraction showed a reciprocal early decline, reaching a nadir of 6.6% (95% bootstrap confidence interval −20.70% to 2.88%) below baseline at 74 minutes and remaining below baseline for most of the observation period.

Respiratory rate demonstrated an early and sustained increase above baseline, reaching a maximal increase of 10.2% (95% bootstrap confidence interval 3.63% to 21.22%) at 480 minutes. Peripheral oxygen saturation remained stable throughout the study period, fluctuating within 1% of baseline. End-tidal carbon dioxide increased progressively from 2.4% above baseline at time 0 to 10.3% (95% bootstrap confidence interval 4.38% to 19.67%) above baseline at 411 minutes.

Discussion

This study used high-fidelity physiologic monitoring to describe circulatory responses and effects on tissue oxygenation following intravenous magnesium sulphate administration in critically ill children admitted to a cardiac ICU.

The following discussion of the observed physiologic trajectories represents the modelled mean response over time following magnesium administration. Several trajectories had wide confidence intervals that crossed baseline, indicating considerable physiologic variability around the estimated mean response. Further, multiple trajectories do not begin at baseline, as the baseline value was derived from a pre-administration median rather than the single value at time 0, and the spline represents a best-fit model of the observed data rather than being constrained to pass through the baseline at time 0. Many of the late stage physiologic changes observed in the spline trajectories warrant cautious interpretation, as they may represent a later compensatory response following magnesium administration, reflect changes in serum magnesium concentration or indicate other time-varying clinical factors. Despite these considerations, the spline trajectories provide a useful representation of the overall population-level physiologic response over time, allowing characterisation of the expected temporal pattern following magnesium administration.

When observing the modelled spline trajectories, magnesium administration was associated with early reductions in heart rate, mean circulatory filling pressure, central venous pressure and venous resistance, sustained reductions in systemic vascular resistance, and increased regional oxygen saturation with reciprocal reductions in modelled oxygen extraction. Collectively, these findings may suggest temporal changes across both the venous and arterial circulations, accompanied by changes in regional oxygenation.

Magnesium administration was associated with an early decline in mean circulatory filling pressure and sustained reduction in modelled venous resistance. Within a Guytonian framework, mean circulatory filling pressure represents the pressure driving venous return, while venous resistance represents opposition to venous flow. Reference Guyton, Lindsey and Kaufmann22 The reduction in mean circulatory filling pressure may reflect reduced stressed venous volume, consistent with venous dilation, thereby decreasing the pressure gradient for venous return. Concurrent reduction in modelled venous resistance may mitigate the reduction in driving pressure by reducing resistance to venous return. However, because these modelled indices are derived from shared physiologic inputs, their concurrent changes should not be interpreted as independent evidence of venous vasodilation.

In addition to its effects on venous physiology, magnesium administration was associated with temporal changes in the arterial circulation. Modelled systemic vascular resistance remained below baseline for several hours following magnesium administration; however, the trajectory was already 6.4% below baseline at time 0, making it difficult to determine whether the sustained reduction was attributable to magnesium or reflected pre-existing physiologic changes. Similarly, pulse pressure progressively increased throughout the observation period, from 7.2% above baseline at time 0 to a peak of approximately 18%, which may be consistent with a progressive reduction in arterial vascular tone. Modelled arterial blood pressure showed a relative reduction during the first 100 minutes, although the trajectory also began above baseline. Taken together, these findings suggest temporal alterations in the arterial vascular compartment following magnesium administration, although the magnitude and clinical significance of these changes remain difficult to determine.

Oximetric variables provide important context for interpretation of the observed vascular responses. Regional oxygen saturation increased early following magnesium sulphate administration and remained above baseline throughout most of the observation period, while oxygen extraction demonstrated a reciprocal reduction. These findings are consistent with preserved or potentially improved regional oxygenation; Reference Kissack, Garr and Wardle25 however, the magnitude of the increase in regional oxygen saturation was small.

Together, the modelled trajectories demonstrate early changes in mean circulatory filling pressure and venous resistance that may reflect a temporal pattern of altered venous tone. Early relative reductions in arterial blood pressure and progressive increases in pulse pressure may be consistent with changes in arterial vascular tone, although interpretation is limited by trajectories that began above baseline. Importantly, regional oxygenation was maintained despite these observed vascular changes. Thus, while the arterial and venous trajectories suggest alterations in vascular physiology following magnesium administration, there was no corresponding evidence of impaired regional tissue oxygenation, suggesting that these changes did not result in an adverse effect on the oxygenation of monitored tissues.

Experimental data provide additional context for the observed vascular and myocardial responses. In an anaesthetised canine model, intravenous bolus magnesium sulphate at doses of 60, 90, and 120 mg/kg produced dose-dependent reductions in arterial pressure, heart rate, and myocardial contractility, while systemic vascular resistance decreased and cardiac output was maintained. These findings provide evidence that magnesium can produce both vascular and myocardial effects without impairing cardiac output, although direct comparison with the present study is limited by differences in species, anaesthesia, dose, and duration of administration. Reference Nakaigawa, Akazawa and Shimizu26

The observed responses are grounded in physiologic plausibility given the known pharmacologic action of magnesium as a calcium channel antagonist within vascular smooth muscle. Reference Fawcett, Haxby and Male6–Reference Wu and Lipsius8 In addition, magnesium suppresses catecholamine release and augments nitric oxide- and prostacyclin-mediated vasodilatory pathways, providing a strong pharmacological basis for the observed coordinated arterial and venous tone. Reference Schutten, Joosten and de Borst9

Magnesium also has the potential to exert direct negative inotropic effects through antagonism of myocardial L-type calcium channels. Reference Nakaigawa, Akazawa and Shimizu26 Although we observed transient reductions in heart rate following magnesium administration, tissue oxygenation did not appear to be adversely affected, thus, we found no evidence of impaired circulatory performance. This response may be different in a younger cohort, as the physiologic significance of calcium channel antagonism varies with developmental maturity. Immature myocardium relies more heavily on extracellular calcium influx because sarcoplasmic reticulum calcium storage and release mechanisms are incompletely developed. Reference Andropoulos, Andropoulos, Stayer, Mossad and Miller-Hance10 Consequently, younger infants may be more susceptible to the negative inotropic effects of magnesium, whereas older children are likely to better tolerate these effects.

Beyond developmental age, the underlying cardiovascular physiology of the individual patient is anticipated to affect circulatory response. Postoperative physiology, congenital cardiac anatomy, altered ventricular compliance, abnormal pulmonary vascular physiology, systemic inflammation, vasoactive support, and coexisting medical therapies all have the potential to modify vascular responses and alter compensatory mechanisms following magnesium administration. Reference Wu and Lipsius8,Reference Marino, Tabbutt and MacLaren11,Reference Datt, Wadhhwa and Sharma12,Reference Lee, Wu and Chan27 This is likely being reflected in the wide confidence intervals demonstrated in the model.

The present study extends current understanding of the haemodynamic effects of magnesium sulphate in critically ill children. Previous cardiology-focused studies have examined its anti-arrhythmic efficacy and the occurrence of hypotension as a side effect following administration, with relatively little attention given to the integrated circulatory response to the drug. Reference He, Sznycer-Taub and Cheng2,Reference Manrique, Arroyo and Lin28,Reference Verma, Chauhan and Gharde29

Our findings are consistent with changes in vascular tone following magnesium administration, while tissue oxygenation remained preserved. Interpretation of these responses required simultaneous assessment of venous, arterial, respiratory, and oximetric variables, as no single physiologic parameter adequately captured the coordinated physiologic response observed. Likewise, incorporation of estimated venous physiologic indices, including modelled mean circulatory filling pressure and venous resistance, provided important mechanistic insight that would not have been apparent from conventional haemodynamic monitoring alone. The use of high-fidelity physiologic monitoring was fundamental to these observations. Continuous high-resolution data enabled characterisation of coordinated temporal relationships across venous, arterial, respiratory, and oximetric domains that would likely not have been apparent using conventionally charted intermittent observations.

The present study has several limitations. The retrospective observational design precludes causal inference, and residual confounding related to illness severity, ventilatory changes, concurrent therapies, and clinician decision-making may have influenced the observed physiologic trajectories. Adjustment was performed for vasoactive support using the vasoactive-inotropic score; however, it was not feasible to account for all concurrent medications and interventions, as critically ill paediatric cardiac patients rarely receive magnesium in isolation. Measurement variability related to monitor and sensor accuracy cannot be excluded, particularly for small changes in physiologic variables. Further, variation in physiologic trends may be driven by fluctuations in serum magnesium level rather than being secondary to drug effect alone. In addition, the study was conducted within a single paediatric cardiac ICU in children aged 1–12 years, and the findings may not be generalisable to infants, other paediatric populations, or non-cardiac patient groups.

Several variables, including systemic vascular resistance, mean circulatory filling pressure, and venous resistance, were secondary derivations using physiologic equations rather than direct invasive measurements and share underlying physiologic inputs. They should therefore be interpreted as modelled indices rather than independent measurements. Accordingly, interpretation focused on the temporal behaviour of the modelled vascular indices rather than their absolute magnitude. Prospective studies incorporating a contemporaneous control period and serial serum magnesium measurements would help determine the extent to which these temporal patterns are attributable to magnesium administration.

Conclusion

While intravenous magnesium sulphate is routinely administered in the paediatric cardiac ICU, its integrated circulatory effects remain incompletely characterised. In this retrospective study of 57 magnesium administrations in critically ill paediatric cardiac patients, high-fidelity physiologic data demonstrated temporal alterations in arterial blood pressure, modelled mean circulatory filling pressure, and modelled venous resistance suggesting concurrent changes in vascular tone. These vascular changes were not accompanied by evidence of worsening regional oxygenation, with regional oxygen saturation remaining above baseline and modelled oxygen extraction generally reduced during the observation period. Although heterogeneity and uncertainty in the modelled responses limit conclusions regarding the magnitude and clinical significance of these changes, the observed trajectories did not demonstrate an adverse pattern of circulatory or regional oxygenation response following magnesium administration.

High-fidelity physiologic monitoring enabled characterisation of temporal relationships across venous, arterial, respiratory, and oximetric variables that may not be apparent from intermittently recorded vital signs. These findings provide a descriptive characterisation of the physiologic response to magnesium administration and support further prospective investigation incorporating a contemporaneous control period and serial assessment of serum magnesium concentration.

Acknowledgements

None.

Financial support

This research received no specific grant from any funding agency, commercial, or not-for-profit sectors.

Competing interests

The authors declare none.

Ethical standard

The study protocol was reviewed and approved by the Stanley Manne Children’s Research Institute institutional review board (IRB ID: STUDY00000419) and conducted in accordance with the Declaration of Helsinki.

Artificial Intelligence use statement

The authors used Gemini (Google) between June and July 2026 to assist with Python code development, debugging, and optimisation for data processing, and statistical analyses. Gemini was used in its standard publicly available form without additional training or modification. All AI-generated codes were independently reviewed, modified where necessary, and validated by the authors, who take full responsibility for the analyses and conclusions. The authors declare no competing interests related to the use of this tool.

References

Hassan, PF, El Haddad, AM. Dexmedetomidine and magnesium sulfate in preventing junctional ectopic tachycardia after pediatric cardiac surgery. Paediatr Anaesth 2024; 34(5):459–466. DOI: 10.1111/pan.14848.10.1111/pan.14848CrossRefGoogle Scholar
He, D, Sznycer-Taub, N, Cheng, Y, et al. Magnesium lowers the incidence of postoperative junctional ectopic tachycardia in congenital heart surgical patients: is there a relationship to surgical procedure complexity? Pediatr Cardiol 2015; 36(6):1179–1185. DOI: 10.1007/s00246-015-1141-5.10.1007/s00246-015-1141-5CrossRefGoogle Scholar
Saleem, AF, Haque, A. On admission hypomagnesemia in critically ill children: risk factors and outcome. Indian J Pediatr 2009; 76(12):1227–1230. DOI: 10.1007/s12098-009-0258-z.10.1007/s12098-009-0258-zCrossRefGoogle Scholar
Munoz, R, Laussen, PC, Palacio, G, et al. Whole blood ionized magnesium: age-related differences in normal values and clinical implications of ionized hypomagnesemia in patients undergoing surgery for congenital cardiac disease. J Thorac Cardiovasc Surg 2000; 119(5):891–899. DOI: 10.1016/S0022-5223(00)70083-3.10.1016/S0022-5223(00)70083-3CrossRefGoogle Scholar
Mildh, L, Hiippala, A, Rautiainen, P, et al. Junctional ectopic tachycardia after surgery for congenital heart disease: incidence, risk factors and outcome. Eur J Cardiothorac Surg 2011; 39(1):75–80. DOI: 10.1016/j.ejcts.2010.04.002.10.1016/j.ejcts.2010.04.002CrossRefGoogle Scholar
Fawcett, WJ, Haxby, EJ, Male, DA. Magnesium: physiology and pharmacology. Br J Anaesth 1999; 83(2):302–320. DOI: 10.1093/bja/83.2.302.10.1093/bja/83.2.302CrossRefGoogle Scholar
Rower, JE, Liu, X, Yu, T, et al. Clinical pharmacokinetics of magnesium sulfate in the treatment of children with severe acute asthma. Eur J Clin Pharmacol 2017; 73(3):325–331. DOI: 10.1007/s00228-016-2165-3.10.1007/s00228-016-2165-3CrossRefGoogle Scholar
Wu, JY, Lipsius, SL. Effects of extracellular Mg2+ on T- and L-type Ca2+ currents in single atrial myocytes. Am J Physiol 1990; 259(6 Pt 2):H1842–1850. DOI: 10.1152/ajpheart.1990.259.6.H1842.Google Scholar
Schutten, JC, Joosten, MM, de Borst, MH, et al. Magnesium and blood pressure: a physiology-based approach. Adv Chronic Kidney Dis 2018; 25(3):244–250. DOI: 10.1053/j.ackd.2017.12.003.10.1053/j.ackd.2017.12.003CrossRefGoogle Scholar
Andropoulos, DB. Physiology and cellular biology of the developing circulation, Anesthesia for Congenital Heart Disease. In: Andropoulos, DB, Stayer, S, Mossad, EB, Miller-Hance, WC. 3rd ed. John Wiley & Sons, 2015. p.84–105.10.1002/9781118768341.ch5CrossRefGoogle Scholar
Marino, BS, Tabbutt, S, MacLaren, G, et al. Cardiopulmonary resuscitation in infants and children with cardiac disease. Circulation 2018; 137(22):e691–e782. DOI: 10.1161/CIR.0000000000000524.10.1161/CIR.0000000000000524CrossRefGoogle Scholar
Datt, V, Wadhhwa, R, Sharma, V, et al. Vasoplegic syndrome after cardiovascular surgery: a review of pathophysiology and outcome-oriented therapeutic management. J Card Surg 2021; 36(10):3749–3760. DOI: 10.1111/jocs.15805.10.1111/jocs.15805CrossRefGoogle Scholar
Han, H, Fang, X, Wei, X, et al. Dose–response relationship between dietary magnesium intake, serum magnesium concentration and risk of hypertension: a systematic review and meta-analysis of prospective cohort studies. Nutr J 2017; 16(1):26. DOI: 10.1186/s12937-017-0247-4.10.1186/s12937-017-0247-4CrossRefGoogle Scholar
Graff, DM, Stevenson, MD, Berkenbosch, JW. Safety of prolonged magnesium sulfate infusions during treatment for severe pediatric status asthmaticus. Pediatr Pulmonol 2019; 54(12):1941–1947. DOI: 10.1002/ppul.24499.10.1002/ppul.24499CrossRefGoogle Scholar
Taher, KW, Johnson, PN, Miller, JL, et al. Efficacy and safety of prolonged magnesium sulfate infusions in children with refractory status asthmaticus. Front Pediatr 2022; 10: 860921. DOI:10.3389/fped.2022.860921.10.3389/fped.2022.860921CrossRefGoogle Scholar
Egelund, TA, Wassil, SK, Edwards, EM, et al. High-dose magnesium sulfate infusion protocol for status asthmaticus: a safety and pharmacokinetics cohort study. Intensive Care Med 2013; 39(1):117–122. DOI: 10.1007/s00134-012-2734-6.10.1007/s00134-012-2734-6CrossRefGoogle Scholar
Abdulatif, M, Ahmed, A, Mukhtar, A, et al. The effect of magnesium sulphate infusion on the incidence and severity of emergence agitation in children undergoing adenotonsillectomy using sevoflurane anaesthesia. Anaesthesia 2013; 68(10):1045–1052. DOI: 10.1111/anae.12380.10.1111/anae.12380CrossRefGoogle Scholar
Dalmora, DDÁ., Lago, PM, Hendler, JV, et al. Continuous infusion of magnesium sulfate to treat acute severe asthma in a Brazilian pediatric emergency department. Pediatr Emerg Med J 2024; 11(3):129–135. DOI: 10.22470/pemj.2024.00948.10.22470/pemj.2024.00948CrossRefGoogle Scholar
Ghanayem, NS, Hoffman, GM. Near infrared spectroscopy as a hemodynamic monitor in critical illness. Pediatr Crit Care Med 2016; 17(8):S201. DOI: 10.1097/PCC.0000000000000780.10.1097/PCC.0000000000000780CrossRefGoogle Scholar
Loomba, RS, Rausa, J, Sheikholeslami, D, et al. Correlation of near-infrared spectroscopy oximetry and corresponding venous oxygen saturations in children with congenital heart disease. Pediatr Cardiol 2022; 43(1):197–206. DOI: 10.1007/s00246-021-02718-7.10.1007/s00246-021-02718-7CrossRefGoogle Scholar
Seckeler, MD, Hirsch, R, Beekman, RH, et al. A new predictive equation for oxygen consumption in children and adults with congenital and acquired heart disease. Heart 2015; 101(7):517–524. DOI: 10.1136/heartjnl-2014-306378.10.1136/heartjnl-2014-306378CrossRefGoogle Scholar
Guyton, AC, Lindsey, AW, Kaufmann, BN. Effect of mean circulatory filling pressure and other peripheral circulatory factors on cardiac output. Am J Physiol 1955; 180(3):463–468. DOI: 10.1152/ajplegacy.1955.180.3.463.10.1152/ajplegacy.1955.180.3.463CrossRefGoogle Scholar
Parkin, WG, Leaning, MS. Therapeutic control of the circulation. J Clin Monit Comput 2008; 22(6):391–400. DOI: 10.1007/s10877-008-9147-7.10.1007/s10877-008-9147-7CrossRefGoogle Scholar
Gaies, MG, Gurney, JG, Yen, AH, et al. Vasoactive–inotropic score as a predictor of morbidity and mortality in infants after cardiopulmonary bypass. Pediatr Crit Care Med 2010; 11(2):234. DOI: 10.1097/PCC.0b013e3181b806fc.10.1097/PCC.0b013e3181b806fcCrossRefGoogle Scholar
Kissack, CM, Garr, R, Wardle, SP, et al. Cerebral fractional oxygen extraction is inversely correlated with oxygen delivery in the sick, newborn. Preterm Infant. J Cereb Blood Flow Metab 2005; 25(5):545–553. DOI: 10.1038/sj.jcbfm.9600046.10.1038/sj.jcbfm.9600046CrossRefGoogle Scholar
Nakaigawa, Y, Akazawa, S, Shimizu, R, et al. Effects of magnesium sulphate on the cardiovascular system, coronary circulation and myocardial metabolism in anaesthetized dogs. Br J Anaesth 1997; 79(3):363–368. DOI: 10.1093/bja/79.3.363.10.1093/bja/79.3.363CrossRefGoogle Scholar
Lee, EP, Wu, HP, Chan, OW, et al. Hemodynamic monitoring and management of pediatric septic shock. Biomed J 2022; 45(1):63–73. DOI: 10.1016/j.bj.2021.10.004.10.1016/j.bj.2021.10.004CrossRefGoogle Scholar
Manrique, AM, Arroyo, M, Lin, Y, et al. Magnesium supplementation during cardiopulmonary bypass to prevent junctional ectopic tachycardia after pediatric cardiac surgery: a randomized controlled study. J Thorac Cardiovasc Surg 2010; 139(1):162–169.e2. DOI: 10.1016/j.jtcvs.2009.07.064.10.1016/j.jtcvs.2009.07.064CrossRefGoogle Scholar
Verma, YS, Chauhan, S, Gharde, P, et al. Role of magnesium in the prevention of postoperative arrhythmias in neonates and infants undergoing arterial switch operation. Interact CardioVasc Thorac Surg 2010; 11(5):573–576. DOI: 10.1510/icvts.2010.239830.10.1510/icvts.2010.239830CrossRefGoogle Scholar
Figure 0

Figure 1. Locally estimated scatterplot smoothing (LOESS) curves demonstrating the unadjusted percent change from baseline in physiologic variables following intravenous magnesium sulphate administration. Time 0 represents magnesium administration. Percent change from baseline was calculated using the median value during the 60-minute pre-administration period and plotted over the subsequent 8 hours. The solid line represents the LOESS estimate and the shaded region the 95% confidence interval. Green and orange markers indicate the minimum and maximum percent change from baseline, respectively, with the corresponding time of occurrence. The number of administrations contributing to each variable was heart rate, n = 57; systolic blood pressure, diastolic blood pressure, and mean arterial pressure, n = 52; pulse pressure, n = 52; central venous pressure, n = 45; respiratory rate, n = 56; peripheral saturation, n = 55; end-tidal carbon dioxide, n = 41; regional oxygen saturation, n = 30; estimated oxygen extraction, n = 30; estimated systemic vascular resistance, n = 25; estimated mean circulatory filling pressure, n = 25; and estimated venous resistance, n = 25.

Figure 1

Figure 2. Restricted cubic spline analyses demonstrating the adjusted percent change from baseline in studied variables. Time 0 represents magnesium administration. The solid blue line represents the adjusted spline estimate, with shaded areas indicating the 95% confidence interval. Green and orange markers indicate the minimum and maximum percent change from baseline, respectively. The number of administrations contributing to each variable was: heart rate, n = 57; systolic blood pressure, diastolic blood pressure, and mean arterial pressure, n = 52; pulse pressure, n = 52; central venous pressure, n = 45; respiratory rate, n = 56; peripheral saturation, n = 55; end-tidal carbon dioxide, n = 41; regional oxygen saturation, n = 30; estimated oxygen extraction, n = 30; estimated systemic vascular resistance, n = 25; estimated mean circulatory filling pressure, n = 25; and estimated venous resistance, n = 25.