https://orcid.org/0009-0004-3265-8533
Figures
Abstract
Background
Few studies have investigated the relationship between the anion gap, including the corrected anion gap, and patient mortality in intensive care units (ICUs) without restricting the analysis to specific diseases or medical specialties. Our primary objective was to investigate the association between the anion gap and ICU mortality using multiple open-access databases.
Methods
We identified 4229 subjects from the Medical Information Mart for Intensive Care IV (MIMIC-IV) database, whose entries were from between 2008 and 2019. For each patient, the anion gap and corrected anion gap were calculated, and the study sample was divided into tertile groups (T) according to these levels. The association between the anion gap and in-hospital mortality was assessed using hazard ratios (HRs) and 95% confidence intervals (CIs) derived from a multivariable-adjusted Cox proportional hazards model. Besides MIMIC-IV, we also incorporated study samples from two other databases (MIMIC-III and electronic ICU) to calculate summary HRs using a random-effects meta-analysis.
Results
Within MIMIC-IV, 1015 patients (24%) died during an average follow-up period of 15.5 days. The fully adjusted HRs and 95% CIs for T2 and T3, relative to T1, were 1.31 (95% CI 1.08–1.58) and 1.54 (95% CI 1.24–1.90), respectively. When grouped by corrected anion gap, the results remained statistically significant. In the meta-analysis, the summary HRs and 95% CIs for T2 and T3 were 1.24 (95% CI 1.08–1.43) and 1.55 (95% CI 1.33–1.82), respectively.
Citation: Yang DE, Jo S, Lee DH, An WS, Jeong MJ, Son M (2024) Dynamics of serum anion gaps with in-hospital mortality: Analysis of the multi-open databases. PLoS ONE 19(4): e0302206. https://doi.org/10.1371/journal.pone.0302206
Editor: Samuele Ceruti, Sant Anna Hospital: Clinica Sant’Anna, SWITZERLAND
Received: January 13, 2024; Accepted: March 31, 2024; Published: April 16, 2024
Copyright: © 2024 Yang 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: The data that support the findings of this study are openly available in the Medical Information Mart for Intensive Care IV (MIMIC-IV) database at https://mimic.mit.edu/.
Funding: This research was supported by Research Grant through the Seokchun Caritas Foundation (SCY2304P). 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
Patients admitted to the intensive care unit (ICU) are at a high risk of death, making it crucial to assess the degree of severity to enhance ICU quality [1]. Determining the mortality risk in critically ill patients, who have diverse diagnoses and comorbidities, is challenging. Various prognostic biomarkers exist to estimate mortality among ICU patients, and these markers assist clinicians in stratifying patients based on the risk of specific outcomes [1].
Acid-base imbalances can indicate disease severity and are associated with poor outcomes, including profound effects on the cardiovascular system. As such, an elevated anion gap can be an initial prognostic indicator in critically ill patients [2–4]. While acid-base disorders intuitively reflect the concentration of unmeasured anions, the calculated anion gap can be unreliable due to the influence of albumin levels, as albumin constitutes a substantial portion of the unmeasured anions [5]. Given that hypoalbuminemia is commonly observed in ICU patients, some researchers have recommended using the corrected anion gap for albumin in all critically ill patients [6, 7].
Some studies have explored the relationship between the anion gap and mortality in ICU units caring for specific categories of patients with specific types of diseases [3, 7–16]. However, few studies have investigated the relationship between the anion gap, including the corrected anion gap, and mortality in ICU patients not restricted to particular diseases or hospital departments. This study aimed to investigate the predictive value of the anion gap regarding ICU mortality across different anion gap values, using multiple open-access databases, including the Medical Information Mart for Intensive Care IV (MIMIC-IV).
Materials and methods
Data source
We extracted data from MIMIC-IV, an open-access, single-center database of critical care units. The MIMIC-IV database comprises ICU patients admitted to Beth Israel Deaconess Medical Center between 2008 and 2019 [17]. We obtained authorization to access the MIMIC-IV database (date of agreement: 2022.8.3). This study was approved by the Institutional Review Board of Dong-A University Hospital (DAUHIRB-EXP-22-032). All patient information in the MIMIC-IV database was anonymized, and the requirement for informed consent was waived.
Data extraction
We excluded patients with missing data and those under 20 years of age. For patients with multiple ICU admissions, only data from the first ICU admission were considered. Patients with an ICU stay of fewer than 24 hours were also excluded. Data regarding the following variables were collected within the first 24 hours of ICU admission: demographics (age and sex), clinical factors (body mass index [BMI], mean arterial pressure [MAP], pulse rate [PR], and respiratory rate [RR]), laboratory findings (white blood cell [WBC] count, platelet count, hemoglobin, prothrombin time [PT], activated partial thromboplastin time [aPTT], blood urea nitrogen [BUN], glomerular filtration rate [GFR], and anion gap, as well as serum levels of glucose, albumin, total bilirubin, and lactate), and arterial blood gas measurements (pH, partial pressure of oxygen [pO2], and partial pressure of carbon dioxide [pCO2]). Means were calculated for any laboratory values measured multiple times within the initial 24 hours. The Charlson Comorbidity Index (CCI) scores were determined based on underlying diseases using ICD-9 codes, and APACHE (Acute Physiology and Chronic Health Evaluation) III scores were also calculated. Data were collected using SQL (Structured Query Language) via Google BigQuery on the Google Cloud Platform (Alphabet Inc., Mountain View, CA, USA). The study’s final participant count was 4229.
Definition of the anion gap and outcome
The anion gap was determined using the following formula: [Na+] + [K−]–[Cl−]–[HCO3−] [18]. Study patients were divided into tertile groups (T1–T3) based on their anion gaps. Additionally, the corrected anion gap was computed as follows: corrected anion gap = anion gap + 2.5 x [4.4-albumin (g/dL)] [19]. The study’s primary endpoint was in-hospital mortality. Participants were tracked from the day of ICU admission to either the day of in-hospital death or discharge, whichever occurred first.
Statistical analysis
Baseline characteristics of the study sample were evaluated as means with standard deviations for continuous variables and as numbers with percentages for categorical variables. Comparisons among study groups were performed using analysis of variance for continuous variables and the chi-squared test or Fisher’s exact test for categorical variables. The in-hospital mortality outcome was calculated by dividing the number of deaths by the total follow-up duration (person-days). Survival probability based on the anion gap group was estimated using the Kaplan-Meier method and evaluated using the log-rank test. To evaluate the association between the anion gap and in-hospital mortality, hazard ratios (HRs) and 95% confidence interval (CIs) for in-hospital mortality were calculated using Cox proportional hazards modeling. Both crude and multivariable-adjusted models were analyzed: Model 1 adjusted for age, sex, BMI, CCI, and APACHE III score; Model 2 adjusted for age, sex, BMI, CCI, APACHE III score, MAP, PR, and RR; and Model 3 (fully adjusted) adjusted for age, sex, BMI, CCI, APACHE III score, MAP, PR, RR, WBC count, platelet count, hemoglobin, PT, aPTT, BUN, GFR, glucose, albumin, total bilirubin, lactate, arterial blood pH, pO2, and pCO2 level. Subgroup analyses were performed to further evaluate the reliability of the results in terms of in-hospital mortality. We analyzed subgroups according to sex (male and female) and age (≥60 years and <60 years). Moreover, our findings were validated using two additional open databases of critical care units: (1) the MIMIC-III database from the Beth Israel Deaconess Medical Center, spanning 2001 through 2012 [20]; and (2) the eICU (Electronic ICU) Collaborative Research Database, collected from 206 US hospitals between 2014 and 2015 [21]. We extracted the subjects from these two databases using the same criteria and variables described above. The HRs and 95% CIs for in-hospital mortality in each database were calculated using multivariable-adjusted Cox proportional hazards modeling. The three database-level HRs were then combined to calculate summary HRs through a random-effects meta-analysis. All statistical analyses were performed using SPSS Statistics for Windows, version 22 (IBM Corp., Armonk, NY, USA) and R, version 4.3.0 (R Foundation for Statistical Computing, Vienna, Austria). A p-value < 0.05 was considered statistically significant.
Results
Baseline characteristics of the study sample
The baseline characteristics of the study patients, divided into three groups based on anion gap values, are shown in Table 1. The mean anion gap values for the three groups were 11.8, 15.4, and 21.4, respectively. A higher anion gap was associated with higher BMI, PR, and RR. In addition, a higher anion gap was associated with higher WBC count, PT, aPTT, and BUN, as well as serum levels of glucose, total bilirubin, potassium, and lactate. Conversely, a higher anion gap was associated with lower GFR, sodium, chloride, bicarbonate, pO2, and pCO2 values. A higher proportion of elevated CCI and APACHE III scores were observed in the third tertile group.
Association between anion gap and in-hospital mortality
Of the study sample, 1015 patients died during a mean follow-up of 15.46 days. Fig 1 illustrates the Kaplan-Meier curves; log-rank tests indicated that in-hospital mortality was higher in association with higher anion gap values. As per Table 2, a significant association between the anion gap and in-hospital mortality was observed in the Cox proportional hazards analysis. The fully adjusted HRs and 95% CIs for T2 and T3, relative to T1, were 1.31 (95% CI 1.08–1.58) and 1.54 (95% CI 1.24–1.90), respectively. When the corrected anion gap values were analyzed, the fully adjusted HRs and 95% CIs remained significant: 1.23 (95% CI 1.01–1.49) and 1.47 (95% CI 1.18–1.84), respectively, for T2 and T3 relative to T1.
T, tertile.
Subgroup analysis according to sex and age
The analysis according to sex and age, in relation to anion gap value, is depicted in Fig 2. When categorized by sex, the HRs and 95% CIs for in-hospital mortality were significant, particularly for males: 1.37 (95% CI 1.06–1.77) and 1.69 (95% CI 1.28–2.24), respectively, for T2 and T3 relative to T1. In the age-based subgroups, all of the HRs and 95% CIs for in-hospital mortality were statistically significant.
HR, hazard ratio; CIs, confidence intervals.
Meta-analysis using other open databases
The MIMIC-III and eICU databases contained 3069 and 697 patients, respectively. Within the MIMIC-III database, the mean anion gap values across the three groups were 11.4, 14.8, and 20.3, respectively. In the eICU database, the values were 7.3, 11.7, and 18.8, respectively. Summary HRs and CIs were calculated after pooling the data from all three databases. The meta-analysis revealed a significant association between anion gap values and in-hospital mortality in Fig 3. For the anion gap groups, the summary HRs and 95% CIs for T2 and T3, relative to T1, were 1.24 (95% CI 1.08–1.43) and 1.55 (95% CI 1.33–1.82), respectively.
HR, hazard ratio; CIs, confidence intervals.
Discussion
This study demonstrated that higher anion gap values were associated with higher in-hospital mortality in the MIMIC-IV database. Even after adjusting for multiple covariates, a higher anion gap remained a significant predictor of in-hospital mortality. When the corrected anion gap was used to adjust for albumin levels, the results remained consistent. A meta-analysis incorporating the MIMIC-III and eICU databases also showed the anion gap to be associated with in-hospital mortality.
Many previous studies have used the anion gap to predict mortality rates for specific patient groups or designated hospital departments [3, 7–16]. These studies have demonstrated associations in populations such as patients with cardiogenic shock, cardiac disease, post–cardiac arrest complications, disseminated intravascular coagulation, postoperative status, and advanced kidney disease. Ji and Peng demonstrated a positive correlation between serum anion gap values and all-cause mortality among unselected adult patients using only the MIMIC-III database [22]. While many studies have concentrated on specific patient groups and relied on single databases or a limited set of covariates, our research has delved into the association between the anion gap (including the corrected anion gap) and in-hospital mortality among general ICU patients. This was achieved by using multiple data sources (encompassing as many covariates as possible): the MIMIC-IV, MIMIC-III, and eICU databases.
The anion gap quantifies the difference between cations(Na+ and K+) and anion (Cl- and HCO3-) concentrations in the blood [18]. In a theoretically electro-neutral human body, unquantifiable plasma anions maintain an anion gap ranging from 4 to 12 mmol/L under normal conditions. Deviations from this range suggest possible imbalances between measured and unmeasured ions. Historically, the anion gap has been used as an indicator for evaluating acid–base imbalances and metabolic abnormalities [23]. An elevated serum anion gap typically points to increased production of organic acid anions or diminished anion excretion. While the exact physiological mechanisms underlying acid–base disturbances are not fully understood, acid–base equilibrium is known to reflect disease prognosis and severity, including organ dysfunction—a vital concern for ICU patients [24, 25].
In our analysis, lactate values were also considered as a covariate; however, even after this adjustment, the anion gap’s significance persisted. This might be because of the presence of other unmeasured anions, such as ketones and pyruvate. ICU patients often exhibit elevated metabolic rates, marked by sympathetic activation, accelerated glycolysis, and increased serum concentrations of unmeasured anions, including lactate and ketones [26, 27]. Furthermore, impaired kidney function is common in ICU patients, and this can lead to an accumulation of unmeasured anions due to the build-up of acidic compounds [14]. Collectively, these findings strongly suggest that a higher anion gap may be associated with prognosis.
This study had several limitations. Firstly, it was retrospective in nature, which can introduce selection bias and misclassification bias. Secondly, the identification of underlying diseases relied on ICD codes, which may introduce misclassification bias. Lastly, while we incorporated numerous potential covariates, there may have been additional variables not considered. Consequently, future prospective studies are warranted to validate the relationship between the anion gap and mortality.
Conclusions
This study demonstrated an association between an increased anion gap levels and elevated in-hospital mortality. Furthermore, the anion gap might hold potential as a prognostic indicator for mortality risk in a broad patient population, which aligns with the overarching goal of identifying measurable biomarkers to forecast patient outcomes and guide treatment approaches.
Supporting information
S1 Checklist. Human participants research checklist.
https://doi.org/10.1371/journal.pone.0302206.s001
(DOCX)
References
- 1.
Pirracchio R. Mortality Prediction in the ICU Based on MIMIC-II Results from the Super ICU Learner Algorithm (SICULA) Project. Secondary Analysis of Electronic Health Records. Cham (CH)2016. p. 295–313.
- 2. Adrogué HJ, Madias NE. Management of life-threatening acid–base disorders. New England Journal of Medicine. 1998;338(1):26–34. pmid:9414329.
- 3. Sahu A, Cooper HA, Panza JA. The initial anion gap is a predictor of mortality in acute myocardial infarction. Coron Artery Dis. 2006;17(5):409–12. Epub 2006/07/18. pmid:16845247.
- 4. Schricker S, Schanz M, Alscher MD, Kimmel M. [Metabolic acidosis: Diagnosis and treatment]. Med Klin Intensivmed Notfmed. 2020;115(4):275–80. Epub 2019/02/07. pmid:30725274.
- 5. Pratumvinit B, Lam L, Kongruttanachok N, Hemthong C, Kost GJ, Kamkang P, et al. Anion gap reference intervals show instrument dependence and weak correlation with albumin levels. Clin Chim Acta. 2020;500:172–9. Epub 2019/11/02. pmid:31669932.
- 6. Fencl V, Jabor A, Kazda A, Figge J. Diagnosis of metabolic acid-base disturbances in critically ill patients. Am J Respir Crit Care Med. 2000;162(6):2246–51. Epub 2000/12/09. pmid:11112147.
- 7. Zhong L, Xie B, Ji XW, Yang XH. The association between albumin corrected anion gap and ICU mortality in acute kidney injury patients requiring continuous renal replacement therapy. Intern Emerg Med. 2022;17(8):2315–22. Epub 2022/09/17. pmid:36112320.
- 8. Lee SW, Kim S, Na KY, Cha RH, Kang SW, Park CW, et al. Serum Anion Gap Predicts All-Cause Mortality in Patients with Advanced Chronic Kidney Disease: A Retrospective Analysis of a Randomized Controlled Study. PLoS One. 2016;11(6):e0156381. Epub 2016/06/02. pmid:27249416.
- 9. Gao Y, Hong Z, Shen R, Zhang S, You G, Chen J, et al. Association between anion gap and mortality of aortic aneurysm in intensive care unit after open surgery. BMC Cardiovasc Disord. 2021;21(1):458. Epub 2021/09/25. pmid:34556051.
- 10. Zhang T, Wang J, Li X. Association Between Anion Gap and Mortality in Critically Ill Patients with Cardiogenic Shock. Int J Gen Med. 2021;14:4765–73. Epub 2021/09/02. pmid:34466021.
- 11. Hu B, Cao J, Hu Y, Qin Z, Wang J. The Association Between Serum Anion Gap and All-Cause Mortality in Disseminated Intravascular Coagulation Patients: A Retrospective Analysis. Int J Gen Med. 2021;14:4535–44. Epub 2021/08/26. pmid:34429638.
- 12. Zhang H, Tian W, Sun Y. The value of anion gap for predicting the short-term all-cause mortality of critically ill patients with cardiac diseases, based on MIMIC-III database. Heart Lung. 2022;55:59–67. Epub 2022/04/28. pmid:35477032.
- 13. Chen J, Dai C, Yang Y, Wang Y, Zeng R, Li B, et al. The association between anion gap and in-hospital mortality of post-cardiac arrest patients: a retrospective study. Sci Rep. 2022;12(1):7405. Epub 2022/05/07. pmid:35524151.
- 14. Asahina Y, Sakaguchi Y, Kajimoto S, Hattori K, Doi Y, Oka T, et al. Time-updated anion gap and cardiovascular events in advanced chronic kidney disease: a cohort study. Clin Kidney J. 2022;15(5):929–36. Epub 2022/05/03. pmid:35498899.
- 15. Sun X, Lu J, Weng W, Yan Q. Association between anion gap and all-cause mortality of critically ill surgical patients: a retrospective cohort study. BMC Surg. 2023;23(1):226. Epub 2023/08/10. pmid:37559030.
- 16. Lu J, Zhong L, Yuan M, Min J, Xu Y. Association between serum anion gap and all-cause mortality in patients with acute myocardial infarction: A retrospective study based on MIMIC-IV database. Heliyon. 2023;9(7):e17397. Epub 2023/08/04. pmid:37539277.
- 17. Johnson AEW, Bulgarelli L, Shen L, Gayles A, Shammout A, Horng S, et al. MIMIC-IV, a freely accessible electronic health record dataset. Sci Data. 2023;10(1):1. Epub 2023/01/04. pmid:36596836.
- 18. Berend K, de Vries AP, Gans RO. Physiological approach to assessment of acid-base disturbances. N Engl J Med. 2014;371(15):1434–45. Epub 2014/10/09. pmid:25295502.
- 19. Figge J, Jabor A, Kazda A, Fencl V. Anion gap and hypoalbuminemia. Crit Care Med. 1998;26(11):1807–10. Epub 1998/11/21. pmid:9824071.
- 20. Johnson AE, Pollard TJ, Shen L, Lehman LW, Feng M, Ghassemi M, et al. MIMIC-III, a freely accessible critical care database. Sci Data. 2016;3:160035. Epub 2016/05/25. pmid:27219127.
- 21. Pollard TJ, Johnson AEW, Raffa JD, Celi LA, Mark RG, Badawi O. The eICU Collaborative Research Database, a freely available multi-center database for critical care research. Sci Data. 2018;5:180178. Epub 2018/09/12. pmid:30204154.
- 22. Ji X, Peng S. The association between serum anion gap and all-cause mortality of unselected adult patients: A retrospective cohort study of >20,000 patients. J Clin Lab Anal. 2023;37(1):e24818. Epub 2022/12/24. pmid:36550640.
- 23. Kraut JA, Madias NE. Serum anion gap: its uses and limitations in clinical medicine. Clin J Am Soc Nephrol. 2007;2(1):162–74. Epub 2007/08/21. pmid:17699401.
- 24. Adrogue HJ, Madias NE. Management of life-threatening acid-base disorders. First of two parts. N Engl J Med. 1998;338(1):26–34. Epub 1998/01/01 21:21. pmid:9414329
- 25. Adrogue HJ, Madias NE. Management of life-threatening acid-base disorders. Second of two parts. N Engl J Med. 1998;338(2):107–11. Epub 1998/01/08. pmid:9420343
- 26. Gabow PA, Kaehny WD, Fennessey PV, Goodman SI, Gross PA, Schrier RW. Diagnostic importance of an increased serum anion gap. N Engl J Med. 1980;303(15):854–8. Epub 1980/10/09. pmid:6774247.
- 27. Lazzeri C, Valente S, Chiostri M, Gensini GF. Clinical significance of lactate in acute cardiac patients. World J Cardiol. 2015;7(8):483–9. Epub 2015/09/01. pmid:26322188.