Postoperative anion gap associates with short- and long-term mortality after cardiac surgery: A large-scale cohort study

Jiajing Li, Yu Tian, Lingzhi Wang, Jiayue Chen, Xiaoshu Chen, Huansen Huang, Yihao LiFrontiers in Cardiovascular Medicine 2022 · DOI

Abstract

This study found a positive association between postoperative anion gap (AG) levels and short- and long-term mortality among patients after cardiac surgery. This relationship suggests that AG, a routine and inexpensive blood test, could serve as a valuable prognostic marker in clinical settings.

Introduction

Cardiac surgery remains a high-stakes medical intervention, with global mortality rates typically ranging between 2% and 6%. While surgical techniques and perioperative care have advanced significantly, identifying which patients are at the highest risk for complications or death remains a persistent challenge. Many existing prognostic tools are either too complex for rapid clinical application or focus primarily on short-term survival, failing to account for long-term patient outcomes.

The authors of this study investigated the clinical utility of the anion gap (AG), a standard measure of the difference between measured cations (such as sodium) and anions (such as chloride and bicarbonate) in the blood. While AG is traditionally used to assess acid-base balance, its specific role as a predictor for cardiac surgery outcomes—particularly long-term mortality—had not been extensively explored in a large-scale cohort until now.

Methodology

The researchers conducted a retrospective cohort study using data from the Medical Information Mart for Intensive Care III (MIMIC-III v1.4). This massive, de-identified database contains high-granularity clinical data from over 40,000 patients admitted to the Beth Israel Deaconess Medical Center, making it a gold standard for retrospective ICU research.

The study focused on a cohort of 6,410 adult patients who underwent on-pump cardiac surgery. To ensure a rigorous analysis, the researchers employed several advanced statistical techniques:

  • Generalized Additive Models (GAM): These were used to identify non-linear relationships between AG levels and mortality.
  • Regression Analysis: Logistic and Cox regression models calculated the Odds Ratio (OR) and Hazard Ratio (HR) for mortality while adjusting for a wide array of confounders, including age, sex, comorbidities, and surgical type.
  • E-value Analysis: A sensitivity analysis was performed to determine the potential impact of unmeasured confounding factors on the study's findings.
  • Tertile Classification: Patients were categorized into three groups based on their initial AG: Tertile 1 (≤ 11 mmol/L), Tertile 2 (11–13 mmol/L), and Tertile 3 (> 13 mmol/L).
Figure 1: Flowchart of study cohort selection.

Results

The study revealed a clear "dose-response" relationship: as postoperative AG levels increased, survival rates decreased significantly across all measured timeframes.

Short-Term and Long-Term Mortality

The association between AG and mortality was robust. Compared to the lowest group (Tertile 1), patients in the highest group (Tertile 3) were 4.51 times more likely to die during their hospital stay (OR 4.51, 95% CI 2.57–7.91). This predictive power extended well beyond the immediate postoperative period; for 4-year mortality, patients in Tertile 3 faced a 48% higher risk of death compared to those in Tertile 1 (HR 1.48, 95% CI 1.20–1.82).

Figure 2: Association between initial anion gap and the probability of in-hospital death.

Predictive Performance and Efficiency

While initial AG was a strong predictor, the maximum AG recorded during the ICU stay showed even higher diagnostic performance for in-hospital mortality, achieving an Area Under the Curve (AUC) of 0.918—indicating excellent predictive accuracy. Furthermore, higher AG levels were linked to longer stays in both the ICU and the hospital, highlighting the increased resource burden associated with these high-risk patients.

Discussion

The findings suggest that AG is more than just a snapshot of blood chemistry; it serves as a proxy for metabolic distress. High AG often indicates tissue hypoperfusion and the accumulation of unmeasured organic acids, which are common physiological responses to the "insult" of cardiopulmonary bypass.

Perhaps the most significant insight is the long-term predictive power of a measurement taken shortly after surgery. The correlation with survival four years later suggests that high postoperative AG may reflect a patient's underlying physiological frailty or the severity of the systemic inflammatory response triggered by the surgical procedure.

Limitations and Future Directions

Despite the large sample size, the study has several limitations. As a single-center retrospective study, it demonstrates association rather than direct causation. Important biomarkers like lactate and ketone bodies were excluded due to missing data in the MIMIC-III database, which might have provided more granular insight into the causes of elevated AG. Additionally, the database lacks specific intraoperative details, such as exact bypass time, which could influence postoperative results.

In conclusion, postoperative AG is a powerful, inexpensive, and readily available marker for predicting mortality after cardiac surgery. Future multi-center prospective studies are needed to validate these findings and explore whether active clinical management of AG levels could directly improve patient survival.

Summary updated

Abstract

This study found a positive association between postoperative AG levels and short- and long-term mortality among patients after cardiac surgery. This relationship warrants further research.
BibTeX
@inproceedings{li2022postoperative,
  author = {Jiajing Li and Yu Tian and Lingzhi Wang and Jiayue Chen and Xiaoshu Chen and Huansen Huang and Yihao Li},
  title = {{Postoperative anion gap associates with short- and long-term mortality after cardiac surgery: A large-scale cohort study}},
  booktitle = {Frontiers in Cardiovascular Medicine},
  year = {2022},
  doi = {10.3389/fcvm.2022.1024484},
}
Figure