https://orcid.org/0000-0002-1338-0829
Figures
Abstract
Pulmonary complications in non-pulmonary sepsis (PC-NPS) are the leading cause of morbidity and mortality in the intensive care unit. Early prevention and monitoring are paramount since the prevention strategies remain limited yet. Magnesium, an essential electrolyte involved in inflammation and vascular regulation, may influence the development of such complications. This retrospective cohort study used data from the MIMIC-IV database to explore the relationship between baseline serum magnesium levels and PC-NPS among 4,836 patients with non-pulmonary sepsis. Survival analysis demonstrated that patients who developed PC-NPS had significantly higher 90-day mortality compared with those without lung injury. When stratified by baseline serum magnesium quartiles, patients in the highest quartile (>2.1 mg/dL) showed the poorest survival. Multivariable logistic regression confirmed that elevated magnesium was independently associated with increased risk of PC-NPS, and restricted cubic spline modeling revealed a U-shaped, nonlinear association between baseline magnesium concentration and PC-NPS risk. Inflection points at 1.26 and 1.91 mg/dL identified a range of relatively lower risk. These findings suggest that baseline serum magnesium levels exhibit a U-shaped relationship with the risk of PC-NPS. Evaluating these levels may aid in clinical prognostication and the exploration of underlying mechanisms.
Citation: Peng T, Li Y, Ren Y, Yang M, Long Z, Zuo D, et al. (2026) The safety window of blood magnesium in pulmonary complications of non-pulmonary sepsis: A U-shaped risk and prognostic analysis based on MIMIC-IV. PLoS One 21(6): e0351216. https://doi.org/10.1371/journal.pone.0351216
Editor: Tomasz W. Kaminski, Versiti Blood Research Institute, UNITED STATES OF AMERICA
Received: October 13, 2025; Accepted: May 25, 2026; Published: June 15, 2026
Copyright: © 2026 Peng et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: We are unable to provide the raw dataset publicly because this study utilizes the MIMIC-IV database, which contains sensitive patient information and is subject to a strict Data Use Agreement (DUA) that prohibits public sharing. Data are available from PhysioNet (https://physionet.org/content/mimiciv/) for researchers who meet the criteria for access to confidential data. By applying our provided codes to the MIMIC-IV database, the study findings can be replicated in their entirety.
Funding: This research was financially supported by the National Natural Science Foundation of CHINA (Grant No. 2024KYS01803 to Zonghong Long). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Sepsis, a syndrome of dysregulated host response induced by systemic infection, has become one of the leading causes of death in intensive care units worldwide [1]. Among its complications, pulmonary complications in non-pulmonary sepsis(PC-NPS) are one of the most common and lethal, with an incidence ranging from 25% to 45% and a mortality rate as high as 30% to 40% [2, 3]. Despite continuous advances in critical care medicine and supportive therapies, clinical treatments for sepsis-associated PC-NPS remain limited and often ineffective, severely impacting patient outcomes. The primary pathological mechanisms underlying these pulmonary complications involve a vicious cycle of inflammatory storms, disruption of the capillary-alveolar endothelial barrier, coagulation abnormalities, and oxidative stress. Currently, there is a lack of specific biomarkers for early warning of alveolar epithelial cell death in sepsis risk, leading to delayed intervention and further exacerbating the disease burden [4, 5, 6, 7].
In recent years, the role of electrolyte disturbances in the progression of sepsis has garnered increasing attention. As the second most abundant intracellular cation in the human body, magnesium ions are indispensable for maintaining cellular homeostasis and regulating immune responses. Dysregulation of magnesium levels has been proven to be closely associated with the pathogenesis and prognosis of sepsis, indicating its potential critical role in this pathological process. Hypomagnesemia exacerbates the inflammatory cascade by activating the HMGB1/TLR4/NF-κB pathway [8], while also inhibiting P2X7 receptor-mediated calcium signaling abnormalities, aggravating pyroptosis and mitochondrial dysfunction [9]. Clinical evidence further confirms that blood magnesium levels are closely related to microcirculatory dysfunction, lactate clearance rate, and coagulopathy risk in septic patients, exhibiting a J-shaped dose-effect relationship—both hypomagnesemia (<1.6 mg/dL) and hypermagnesemia (>2.4 mg/dL) are associated with increased mortality [10, 11, 12, 13, 14]. Notably, ferroptosis, a key mechanism of alveolar epithelial cell death in septic lung injury, is regulated by magnesium ions through modulation of NLRP3 inflammasome activity and the SLC7A11/GPX4 pathway [15, 16, 17, 18]. These findings suggest that dynamic changes in blood magnesium may influence the occurrence and progression of PC-NPS through immune-metabolic reprogramming.
Existing studies have indicated that electrolyte disturbances may affect the prognosis of sepsis, but research on biomarkers for PC-NPS has paid insufficient attention to blood magnesium homeostasis. Its predictive efficacy, optimal cutoff values, and mechanistic explanations remain unclear. It is worth noting that septic patients often experience fluctuations in blood magnesium due to interventions such as renal replacement therapy and parenteral nutrition. Therefore, studies on baseline monitoring strategies and their spatiotemporal relationship with lung injury are urgently needed. Based on this, this study aims to investigate the dose-effect relationship between baseline blood magnesium levels and PC-NPS, to evaluate its predictive value, thereby providing evidence for personalized risk stratification and targeted magnesium supplementation strategies in sepsis.
Materials and methods
Study participants
This study utilized the MIMIC-IV electronic database (version 3.0), jointly established by the Massachusetts Institute of Technology (MIT) and Beth Israel Deaconess Medical Center (BIDMC), to include hospitalized patients diagnosed with sepsis. Sepsis was identified using International Classification of Diseases (ICD-9 and ICD-10) codes (code range: 0380, A40, A400, A401, A408, A409, 0382, A403, 03810, 03811, 03812, 03819, A410, A4101, A4102, A411, A412, 0223, A227, A267, A327, 03842, A4151, 03843, A4152, 03844, A4153, 0031, A021, 03841, A413, A5486, A4181, A427, 0383, A414, B377, 0545).
Exclusion criteria included: ① Hospital stay < 24 hours; ② No baseline serum magnesium concentration measured within 24 hours of admission; ③ Missing follow-up data; ④ Patients with concurrent pneumonia (to avoid confounding effects of pneumonia-induced sepsis on pulmonary outcomes). Specifically, a total of 359,791 patients were excluded due to missing baseline serum magnesium measurements.The patient screening process is illustrated in Fig 1. All laboratory data, including components for the Sequential Organ Failure Assessment (SOFA) score, were obtained from the first test after admission. SOFA scores were calculated based on available clinical and laboratory data; patients with missing data for one or more SOFA components were recorded as SOFA-N, while those with complete data were recorded as SOFA-Y. The distribution of patients with available and missing SOFA scores across PC-NPS groups and magnesium quartiles is provided in Supplementary S1 Table.
This flowchart outlines the steps involved in selecting participants for the analysis cohort, starting from the initial dataset of 364,627 patients in the MIMIC-IV database. Patients were excluded based on various criteria, including the absence of sepsis at admission, death within 24 hours of admission, missing serum magnesium index, and missing follow-up data. The final analysis cohort consists of 4,836 participants.
Data extraction
The following variables were extracted from the MIMIC-IV database:
Demographic characteristics: age, sex;
Blood tests and vital signs: baseline magnesium (Mg²⁺), red blood cells (RBC), red cell distribution width (RDW), hematocrit (Hct), hemoglobin (Hb), platelets (PLT), activated partial thromboplastin time (APTT), prothrombin time (PT), anion gap (AG), bicarbonate (HCO₃⁻), alkaline phosphatase (ALP), alanine aminotransferase (ALT), aspartate aminotransferase (AST), total bilirubin (TBIL), blood urea nitrogen (BUN), creatinine (Cr), calcium (Ca²⁺), chloride (Cl⁻), potassium (K⁺), sodium (Na⁺), phosphate (PO₄³⁻), glucose;
Diagnostic indicators for PC-NPS: hypoxemia, pulmonary edema, acute respiratory failure, pulmonary embolism, pneumothorax. We explicitly defined these composite criteria to standardize the diagnosis within our cohort, thereby ensuring a robust, clinically relevant assessment of the primary endpoint.
Comorbidities: chronic obstructive pulmonary disease (COPD), heart failure (HF), angina pectoris, myocardial infarction (MI), chronic myocardial ischemia, diabetes, chronic kidney disease (CKD), obesity, pneumonia history, HIV infection;Current ICU admission status;History of prior ICU admissions;Follow-up survival status and survival time.
All laboratory data were obtained from the first test after admission. Variables with missing values exceeding 20% were excluded. Variables with missing values below 20% were handled using multiple imputation (R language MICE package).
Ethical statement
The study adhered to the Declaration of Helsinki. The use of the MIMIC-IV database was approved by the ethics committees of MIT and BIDMC. As the data are publicly available, the requirement for informed consent was waived.
Endpoints and grouping
The primary endpoint was the occurrence of PC-NPS during the current hospitalization. Secondary endpoints included 90-day all-cause mortality and ICU admission during the current hospitalization.
Patients were first divided into PC-NPS and non-PC-NPS groups based on the primary endpoint to evaluate differences in 90-day all-cause mortality and ICU admission rates. They were then stratified into four groups based on baseline magnesium quartiles to assess differences in PC-NPS incidence, mortality, and ICU admission rates across different magnesium levels.
Statistical analysis
Before analysis, continuous variables were checked for normality. Normally distributed data are shown as mean ± SD, and t tests were applied for comparisons. For variables not meeting normality, data are summarized as median (IQR) and compared with Mann–Whitney U tests. Categorical variables are presented as number and percentage, with group comparisons based on chi-square tests.
To identify independent risk factors for the incidence of PC-NPS, we performed univariate and multivariate logistic regression analyses. Survival was analyzed using Kaplan–Meier estimates, and log-rank tests were used for group comparisons. For survival analysis regarding 90-day mortality, Kaplan-Meier curves were generated and compared using the log-rank test. Hazard ratios were not calculated as our regression models focused exclusively on the primary endpoint. We also assessed multicollinearity through variance inflation factors; variables with VIF above 5 were excluded. Analyses were performed with R software (version 4.3.1), and statistical significance was defined as a two-tailed p < 0.05.
Results
Baseline characteristics
A total of 4,836 septic patients were included in this study. The median age of the patients was 67 years (IQR: 56–78), and 54.28% were male. The incidence of PC-NPS during the current hospitalization was 26.28%, the 90-day mortality rate was 20.72%, and the ICU admission rate was 81.12%. The median magnesium level was 1.9 mg/dL (IQR: 1.6–2.1) (Table 1).
Differences in baseline characteristics between the PC-NPS and Non-PC-NPS groups
Compared with the non-PC-NPS group, patients in the PC-NPS group were older (median 68 [IQR: 57–80] vs. 67 [IQR: 55–78] years, P = 0.003), had higher 90-day mortality (32.26% vs. 16.61%, P < 0.001), and although they had a lower current ICU admission rate (74.82% vs. 83.37%, P < 0.001), they had a higher proportion of prior ICU admission history (85.84% vs. 57.53%, P < 0.001). Additionally, the PC-NPS group had significantly higher proportions of comorbidities, including chronic obstructive pulmonary disease (COPD), heart failure (HF), myocardial infarction (MI), chronic kidney disease (CKD), obesity, and a history of pneumonia.
In terms of blood indicators, the PC-NPS group exhibited significantly higher levels of magnesium, red cell distribution width (RDW), white blood cell count (WBC), coagulation parameters (APTT, PT), anion gap (AG), liver function markers (alanine aminotransferase [ALT], aspartate aminotransferase [AST], total bilirubin [TBIL]), renal function markers (blood urea nitrogen [BUN], creatinine [Cr]), serum potassium (K⁺), phosphate (PO₄³⁻), and blood glucose. In contrast, platelet count (PLT), bicarbonate (HCO₃⁻), and calcium (Ca²⁺) levels were significantly lower (Table 2, Fig 2A,B).
A: Differences in serum magnesium levels under PC-NPS conditions; B: Patient survival curves under PC-NPS conditions.
Baseline characteristics stratified by magnesium quartiles
Patients were divided into four groups based on magnesium quartiles: Q1 (<1.6 mg/dL), Q2 (1.6–1.9 mg/dL), Q3 (1.9–2.1 mg/dL), and Q4 (>2.1 mg/dL). From Q1 to Q4, patient age gradually increased(median: 66 [IQR: 54–76], 67 [IQR: 56–79], 68 [IQR: 55–80], and 69 [IQR: 58–79] years, P < 0.001). Among comorbidities, the prevalence of HF, MI, and CKD showed an increasing trend (all P < 0.001). COPD was slightly lower only in Q3 but overall increased with rising magnesium concentrations (P = 0.031).
Regarding laboratory parameters, RDW increased from Q1 to Q4 (P < 0.001). Platelet (PLT) levels generally trended upward, peaking in Q3 (P < 0.001). Bicarbonate (HCO₃⁻) levels increased from Q1 to Q3 (P < 0.001). Among liver and kidney function markers, ALT and AST were significantly elevated in Q4 (both P < 0.001). BUN and Cr showed a marked increasing trend, particularly in Q4 (both P < 0.001). For electrolyte indicators, Ca² ⁺ , K ⁺ , and PO₄³ ⁻ increased across quartiles (all P < 0.001), while Na ⁺ was significantly higher in Q4 (P = 0.009). Conversely, Cl⁻ decreased across quartiles (P < 0.001).
Additionally, the current ICU admission rate significantly decreased from Q1 to Q4 (86.32% → 81.55% → 77.46% → 76.92%, P < 0.001), while mortality (15.35% → 20.13% → 20.79% → 28.95%, P < 0.001) and PC-NPS incidence (25.54% → 24.35% → 24.29% → 32.8%) significantly increased (all P < 0.001) (Table 1, Fig 3).
Q1–Q4 represent magnesium concentrations <1.6, 1.6–1.9, 1.9–2.1, and >2.1 mg/dL, respectively. Data shown as proportion (%) of patients within each quartile developing PC-NPS.
Survival analysis
Regarding clinical outcomes, Kaplan-Meier survival curves demonstrated a significant separation between the PC-NPS and non-PC-NPS groups, with the PC-NPS group exhibiting a higher cumulative 90-day mortality (log-rank P < 0.001; Fig 2B). Furthermore, when stratified by serum magnesium quartiles, the highest quartile (Q4, > 2.1 mg/dL) showed the poorest survival, with the highest cumulative 90-day mortality (log-rank P < 0.001; Fig 4).
Kaplan-Meier survival curves of 90-day mortality stratified by baseline serum magnesium quartiles (Q1–Q4). Shaded areas indicate 95% confidence intervals. Numbers at risk are shown below the x-axis.
To identify risk factors for the development of PC-NPS, univariate logistic regression analysis was performed (Table 3). The results demonstrated that elevated serum magnesium levels were significantly associated with an increased risk of PC-NPS (Odds Ratio [OR] = 1.322, 95% CI:1.117–1.566, P = 0.001). Additionally, age, history of ICU admission, and comorbidities such as COPD, asthma, and heart failure were also identified as significant risk factors for the occurrence of PC-NPS (P < 0.05).
Primary outcome
Multivariate logistic regression analysis, using the backward stepwise method, revealed that an increase in magnesium concentration was significantly associated with a higher risk of PC-NPS significantly rising (OR=1.252, 95% CI: 1.04–1.506, P = 0.017) (Table 4). Restricted cubic spline analysis uncovered a “U-shaped” nonlinear relationship between serum magnesium concentration and risk (OR), which has two critical inflection points (1.26 mg/dL and 1.91 mg/dL).
In the low magnesium risk zone (<1.26 mg/dL), with the decreasing of magnesium concentration, the OR value increase rapidly. Noteworthy, the confidence interval crossed 1 remind that the trend of increased risk was not significant when magnesium concentration was < 1.5 mg/dL. In the safety window interval (1.26–1.91 mg/dL), the OR value fluctuated around 1.0, finding the lowest risk within this range. In the high magnesium risk zone (>1.91 mg/dL), the OR value gradually increased with rising magnesium concentration. With the confidence interval did not cross 1, demonstrating a significant increase in risk when magnesium concentration exceeded 1.91 mg/dL (Fig 5, Table 5).
Restricted cubic spline analysis showing U-shaped relationship between baseline serum magnesium concentration (mg/dL) and odds ratio (OR) for PC-NPS. Shaded area represents 95% confidence interval; dashed line indicates OR=1; vertical lines denote inflection points at 1.26 and 1.91 mg/dL.
Discussion
This study is the first to reveal a nonlinear U-shaped dose-response relationship between serum magnesium levels and PC-NPS in a large sepsis cohort. We identified a potential safety range for serum magnesium in clinical management, 1.26–1.91 mg/dL. Beyond this range, particularly in the highest quartile (Q4, > 2.1 mg/dL), the incidence of PC-NPS increased from 25.54% (Q1) to 32.8%, and 90-day mortality rose from 15.35% to 28.95%. Our findings underscore the importance of magnesium level control in sepsis management, which has emphasized that the serum magnesium concentration remain inferior to 1.91 mg/dL and also proposed to maintain it within the safe range of 1.26–1.91 mg/dL. Notably, the significantly lower ICU admission rate in the hypermagnesemia group (76.92% vs. 86.32%) may indicate rapid deterioration or even death before ICU transfer in some patients. Overall, these results warrant enhanced early warning and dynamic monitoring for patients with magnesium dysregulation, particularly hypermagnesemia.
The pathogenesis of hypermagnesemia may involve multiple pathophysiological processes. Previous studies reported anti-inflammatory effects of magnesium via inhibition of the HMGB1/TLR4/NF-κB pathway; however, other studies have observed elevated levels of pro-inflammatory cytokines (e.g., IL-6, TNF-α) in hypermagnesemia8. Regard as coagulation abnormalities, we observed a paradoxical phenomenon: prolonged APTT and PT alongside thrombocytopenia in the hypermagnesemia group. Which might reflect a direct inhibition of coagulation factor activity by magnesium ions (e.g., competitive antagonism of calcium-dependent thrombin generation) combined with endothelial injury effects. Similarly, the high concentration of magnesium may expedite the mitochondrial dysfunction by modulating P2X7 receptor signaling. Hypermagnesemia promots ferroptosis-related lipid peroxidation [9], which could be a key mechanism in alveolar epithelial damage during PC-NPS. Therefore, we had found elevated phosphate levels in the hypermagnesemia group, which may promote mitochondrial membrane permeability transition pore opening, aggravating ferroptosis-related injury [9], providing new directions for future mechanistic research.
Our results offer new perspectives on magnesium management in septic patients. Consistent with previous studies showing hypomagnesemia increases sepsis mortality risk, the lowest magnesium quartile (Q1) in our study had a mortality rate of 15.35%. However, the hypermagnesemia group (Q4) showed a further increase to 28.95%, suggesting a “bidirectional” association between magnesium concentration and sepsis prognosis. Notably, while hypomagnesemia typically worsens prognosis by exacerbating microcirculatory dysfunction and lactic acidosis, we found significantly elevated anion gap and lactate levels in the hypermagnesemia group, indicating that metabolic acidosis may serve as a critical mediator of mortality risk regardless of magnesium status [10, 11, 12, 13, 14]. Therefore, dynamic monitoring of serum magnesium levels shows potential practical value for predicting sepsis-associated PC-NPS. For patients whose magnesium levels fall outside the safety window, strict monitoring and prevention of metabolic acidosis are essential.
There are still several limitations need to careful consideration. The retrospective design can’t fully eliminate the residual confounding factors, which influenced the causal relationship between serum magnesium levels and outcomes, such as the dose and timing of magnesium supplementation. Similarly, the MIMIC-IV database lacks the localized microenvironment indicators of patients (e.g., bronchoalveolar lavage fluid), making it difficult to distinguish the distribution of magnesium in serum or lung tissue. Moreover, the study did not dynamically assess the impact of magnesium fluctuations on prognosis. Also, the magnesium homeostasis during sepsis may hold greater clinical significance than single measurements. Additionally, as a common cause of sepsis, pneumonia patients had been excluded to reduce the confouding bias, the underestimate of true correlation between magnesium levels and PC-NPS is still remaind. Furthermore, our study relied on ICD-9 and ICD-10 administrative codes to identify patients with sepsis from the MIMIC-IV database. Previous validation studies have demonstrated that while administrative claims data possess adequate specificity, their sensitivity for identifying true sepsis cases is relatively low and variable compared to clinical criteria [19, 20]. Consequently, a proportion of patients who clinically met the Sepsis-3 criteria might not have been captured in our cohort. This reliance on billing codes may introduce selection bias and could affect the generalizability of our findings to clinical settings that utilize real-time physiological and laboratory criteria for sepsis identification.
Future research should: 1) Conduct prospective cohort studies to validate the magnesium safety window, combined with multi-omics approaches to elucidate hypermagnesemia-lactate metabolism interactions; 2) Use animal models to investigate magnesium’s regulatory mechanisms on alveolar epithelial ferroptosis and pyroptosis, particularly focusing on P2X7 receptor signaling pathways [9, 10]; 3) Develop risk models based on dynamic magnesium changes for prognostic stratification; and 4) Evaluate how different magnesium concentrations affect responses to anticoagulant therapy to inform personalized coagulation management.
In summary, our study shows a U-shaped association between serum magnesium levels and sepsis-associated PC-NPS risk/mortality: the 1.26–1.91 mg/dL safety window for preventing sepsis-associated pulmonary complications. These findings suggest strengthening dynamic monitoring of magnesium levels in septic patients.It warns us to avoid blindly supplementation that leads to excessive concentrations. While closely monitoring early organ dysfunction and fluid-electrolyte/acid-base balance changes in patients with dysmagnesemia. Future research should explore how magnesium concentration thresholds influence immunometabolic reprogramming. Such work could offer fresh insights into precision medicine for sepsis and help shape more individualized treatment strategies.
Conclusion
In this current study, we identified a U-shaped, nonlinear association between serum magnesium levels and the risk of PC-NPS of the large retrospective cohort of septic patients which from the MIMIC-IV database. A safety window of 1.26–1.91 mg/dL was associated with the lowest risk of PC-NPS and mortality by the adverse outcomes of both hypomagnesemia and hypermagnesemia. These findings highlight the importance of dynamic monitoring of magnesium levels during sepsis, while revealing that avoiding excessive supplementation and maintaining serum concentrations within the optimal range may reduce the risk of sepsis-associated pulmonary complications. Future prospective studies and mechanistic investigations are warranted to validate these thresholds, to further investigate the immunometabolic mechanism through which magnesium regulates PC-NPS.
Supporting information
S1 Table. Availability of SOFA scores in the cohort.
https://doi.org/10.1371/journal.pone.0351216.s002
(DOCX)
Acknowledgments
We acknowledge the support of all investigators of the MlMIC-IV group and extend their sincere thanks to all the institutions and the medical staff providing helps and participating in making this study possible.
References
- 1. Singer M, Deutschman CS, Seymour CW, Shankar-Hari M, Annane D, Bauer M, et al. The Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3). JAMA. 2016;315(8):801–10. pmid:26903338
- 2. Rudd KE, Johnson SC, Agesa KM, Shackelford KA, Tsoi D, Kievlan DR, et al. Global, regional, and national sepsis incidence and mortality, 1990-2017: analysis for the Global Burden of Disease Study. Lancet. 2020;395(10219):200–11.
- 3. Gotts JE, Matthay MA. Sepsis: pathophysiology and clinical management. BMJ. 2016;353:i1585.
- 4. Prescott HC, Angus DC. Enhancing Recovery From Sepsis: A Review. JAMA. 2018;319(1):62–75. pmid:29297082
- 5. Ware LB, Matthay MA. The acute respiratory distress syndrome. N Engl J Med. 2000;342(18):1334–49. pmid:10793167
- 6. Meyer NJ, Gattinoni L, Calfee CS. Acute respiratory distress syndrome. Lancet. 2021;398(10300):622–37.
- 7. Bos LDJ, Ware LB. Acute respiratory distress syndrome: causes, pathophysiology, and phenotypes. Lancet. 2022;400(10358):1145–56.
- 8. Huang Y, Wang X-X, Sun D-D, Zhang Z-X, Yang W-W, Shao T, et al. Sub-anesthesia Dose of Isoflurane in 60% Oxygen Reduces Inflammatory Responses in Experimental Sepsis Models. Chin Med J (Engl). 2017;130(7):840–53. pmid:28345549
- 9. Saguil A, Fargo MV. Acute Respiratory Distress Syndrome: Diagnosis and Management. Am Fam Physician. 2020;101(12):730–8. pmid:32538594
- 10. Mayow AH, Ahmad F, Afzal MS, Khokhar MU, Rafique D, Vallamchetla SK. A systematic review and meta-analysis of independent predictors for acute respiratory distress syndrome in patients presenting with sepsis. Cureus. 2023;15(4):e37055.
- 11. Rello J, Valenzuela-Sánchez F, Ruiz-Rodriguez M, Moyano S. Sepsis: A Review of Advances in Management. Adv Ther. 2017;34(11):2393–411. pmid:29022217
- 12. Komarova YA, Kruse K, Mehta D, Malik AB. Protein interactions at endothelial junctions and signaling mechanisms regulating endothelial permeability. Circulation Research. 2017;120(1):179–206.
- 13. Ma S, Zhao ML, Wang K, Yue YF, Sun RQ, Zhang RM. Association of Ang-2, vWF, and EVLWI with risk of mortality in sepsis patients with concomitant ARDS: A retrospective study. J Formos Med Assoc. 2020;119(5):950–6.
- 14. Fan Y, Ye Z, Tang Y. Serum HMGB1 and soluble urokinase plasminogen activator receptor levels aid diagnosis and prognosis prediction of sepsis with acute respiratory distress syndrome. Biomark Med. 2023;17(4):231–9. pmid:37158106
- 15. Zhang Y, Xin J, Zhao D, Chen G, Ji P, Liu P, et al. Magnesium hexacyanoferrate mitigates sepsis-associated encephalopathy through inhibiting microglial activation and neuronal cuproptosis. Biomaterials. 2025;321:123279. pmid:40164040
- 16. Velissaris D, Karamouzos V, Pierrakos C, Aretha D, Karanikolas M. Hypomagnesemia in Critically Ill Sepsis Patients. J Clin Med Res. 2015;7(12):911–8. pmid:26566403
- 17. Jiang P, Lv Q, Lai T, Xu F. Does Hypomagnesemia Impact on the Outcome of Patients Admitted to the Intensive Care Unit? A Systematic Review and Meta-Analysis. Shock. 2017;47(3):288–95. pmid:27753792
- 18. Lee C-Y, Jan W-C, Tsai P-S, Huang C-J. Magnesium sulfate mitigates acute lung injury in endotoxemia rats. J Trauma. 2011;70(5):1177–85; discussion 1185. pmid:21610432
- 19. Rhee C, Dantes R, Gruber S, Klompas M. JAMA. 2017;318(13):1241–9.
- 20. Jolley RJ, Quan H, Jetté N, Sawka KJ, Diep L, Goliath J, et al. Validation and optimisation of an ICD-10-coded case definition for sepsis using administrative health data. BMJ Open. 2015;5(12):e009487. pmid:26700284