Sepsis is characterized by an excessive systemic inflammatory response to infection, leading to physiological, pathological, and biochemical dysregulation that can cause multi-system and multi-organ dysfunction or failure (1). Despite significant advancements in diagnosis and treatment, sepsis remains highly prevalent, especially in high-income countries, with an incidence of approximately 270 cases per 100,000 people per year and an in-hospital mortality rate of 26% (2, 3). Globally, the incidence is estimated at 20 million cases annually, resulting in around 6 million deaths, making sepsis one of the leading causes of intensive care unit (ICU) admissions and in-hospital mortality (4, 5). The annual cost of sepsis care in the United States is estimated at $16.7 billion (6).

Sepsis is a major reason for hospital admission among patients with epilepsy (7), a common chronic neurological disorder associated with persistent risk of seizures and affecting approximately 70 million people worldwide. Epilepsy often coexists with cognitive, neuropsychiatric, and psychosocial impairments, elevating the risk of mental health conditions such as depression and anxiety, as well as increasing mortality from various diseases (8–10). Studies have reported that the relative risk of mortality in epilepsy patients is 2.87 [90% confidence interval (CI) 2.16–3.81; I² 99.7%] and 3.66 (90% CI 2.87–4.66; I² 97.5%), suggesting that high mortality may be associated with chronic degenerative diseases, cerebrovascular diseases, tumors, and potentially with pathophysiological mechanisms leading to sudden unexpected death in epilepsy (11). Epilepsy not only results in health complications but also imposes significant economic burdens on society and families. Epidemiological surveys estimate that the annual expenditure on epilepsy in the United States is approximately $12.5 billion (12). Multifactorial regression analysis of national hospital admissions in the United States has shown that patients with epilepsy are more likely to be transferred between hospitals [odds ratio (OR) = 1.77, 95% CI: 1.72–1.81, p < 0.0001], with overall costs being 4% higher (p < 0.0001). In light of the increased adverse outcomes of both conditions, identifying biological predictors of short-term and long-term mortality in patients with epilepsy and concurrent sepsis in the ICU is crucial for risk stratification and optimization of medical resources.

Patients with sepsis often experience hemodynamic instability, leading to inadequate tissue and organ perfusion, which in turn affects serum bicarbonate levels. Serum bicarbonate has recently been recognized as a clinical biomarker for assessing the body’s acid–base balance and predicting disease prognosis (13). Multiple studies have confirmed its role as a predictor of poor outcomes in various diseases, including all-cause mortality in type 2 diabetes (14), acute and chronic kidney diseases (15, 16), non-alcoholic liver disease (17), cardiovascular events (18), tumors (19), and ischemic cerebrovascular disease (13). However, limited research has explored the relationship between serum bicarbonate and prognosis in patients with epilepsy and concurrent sepsis. To address this gap, we conducted a retrospective cohort analysis using data from patients with epilepsy and concurrent sepsis in the ICU from the Medical Information Mart for Intensive Care (MIMIC)-IV database. This study aims to determine whether serum bicarbonate levels serve as a reliable predictor of short-term and long-term mortality in ICU patients with epilepsy and sepsis.

This retrospective cohort study demonstrated that serum bicarbonate levels measured at time of ICU admission are an independent prognostic factor for 28-day and 90-day mortality rates in patients with epilepsy and concurrent sepsis. Results from the multivariable Cox regression models, Kaplan Meier survival curves, cubic spline analysis, and threshold effect analysis demonstrate a non-linear U-shaped relationship between bicarbonate concentration and mortality. Serum bicarbonate analysed as a continuous variable showed that lower serum bicarbonate concentration significantly increased 28-day and 90-day mortality rates. The finding of reduced short- and long-term mortality with increasing bicarbonate levels, when treated as a continuous variable, was statistically non-significant and likely influenced by the relatively small sample size of patients with elevated serum bicarbonate concentrations. When serum bicarbonate was categorized, patients with levels <22 mEq/L demonstrated significantly higher risks of both short-term and long-term mortality compared to those in the 22–29 mEq/L range. Similarly, levels ≥29 mEq/L were associated with increased mortality risk, though these results were not statistically significant. Threshold effect analysis identified a turning point 25.0 mEq/L for 28-day and 25.9 mEq/L for 90-day mortality, where increasing bicarbonate levels below this point are associated with reduced mortality, and increasing levels above this point are linked to increased mortality. These results align with previous findings regarding the U-shaped relationship between bicarbonate and mortality among critically ill patients. Libório et al. (24), reported a U-shaped relationship between serum bicarbonate levels and in-hospital mortality among critically ill patients. Specifically, their study identified two serum bicarbonate concentration nodes, at 25 mEq/L and 30 mEq/L, that correspond to this U-shaped pattern. These observations are consistent with the findings of the current studies. Notably, no previous studies have explored the specific relationship between serum bicarbonate levels and 28-day or 90-day mortality rates in patients with epilepsy and concurrent sepsis.

The international Sepsis-3 consensus defines sepsis as “life-threatening organ dysfunction caused by a dysregulated host response to infection” (21). This dysfunction leads to hemodynamic failure and abnormal cell metabolism, leading to anaerobic glycolysis in terminal organs and systemic lactic acidosis (25). Sepsis-related organ dysfunction often involves liver and kidney impairment, which exacerbates systemic acidosis (26). Liver damage reduces lactate clearance, while acute kidney injury leads to hydrogen ion retention and bicarbonate loss, intensifying acid–base imbalance (25). Metabolic acidosis, characterized by reduced serum bicarbonate levels, disrupts acid–base homeostasis and contributes to adverse clinical outcomes. Lower serum bicarbonate concentrations have been linked to a higher incidence of acute cardiovascular events and mortality rate within 28 days due to ischemic cardiogenic shock (25, 27). They have also been associated with a higher risk of progression to kidney transplantation or a 50% reduction in glomerular filtration rate in children with acute kidney disease (28). Additionally, they have been significantly correlated with all-cause and cancer mortality rates in patients with type 2 diabetes (19). Low bicarbonate may be related to acidosis, which increases intracellular acidity and releases intracellular protein-bound iron, leading to the formation of hydroxyl radicals that cause oxidative stress and tissue damage (29). Moreover, acidosis activates inflammatory gene expressions in endothelial cells, mediating endothelial plasma leakage, leukocyte recruitment, and exacerbated tissue injury (30). Given these mechanisms, serum bicarbonate is an effective clinical predictor of mortality in sepsis patients, with lower levels strongly associated with increased in-hospital mortality rates compared to those with normal or higher (31). Conversely, elevated serum bicarbonate levels (above 30 mEq/L) have also been associated with increased mortality risks by 21% in critically ill patients (24), Bicarbonate depletion reflects worsening sepsis-related organ dysfunction related to sepsis, whereas bicarbonate excess may suggest metabolic alkalosis, resulting in an increase in pH, a leftward shift of the oxygenated hemoglobin dissociation curve, electrolyte imbalance, and supraventricular or ventricular arrhythmias (32). It thus serves as an effective clinical predictor of mortality in sepsis patients.

Additionally, bicarbonate consumption reflects lactic acid accumulation during seizures, leading to metabolic acidosis, simultaneously activating acid-sensing ion channels (ASIC1a) in both neurons and astrocytes, thereby increasing the frequency of spontaneous seizures. Meanwhile, ASIC1a promotes the activation of the N-methyl-D-aspartate receptor (NMDAR) and exacerbates neuronal death caused by acidosis, leading to calcium ion influx (33, 34). Olaciregui Dague et al. (35), reported significantly lower serum bicarbonate levels (median: 22.6 mmol/L, range 7.1–33.2 mmol/L) in 165 patients with epilepsy following seizures compared with 58 controls with non-epileptic seizure-like events. In the other hand, excessive serum bicarbonate levels can lead to metabolic alkalosis and may strongly predict mortality and adverse functional outcomes in patients with status epilepticus (36). Previous studies have reported that HCO3− plays an important role in neuronal plasticity, as the transmembrane gradient of HCO3− and Cl− determines whether the γ-aminobutyric acid (GABA) ergic signal is depolarizing and excitatory or hyperpolarizing and inhibitory (37, 38). During an epileptic seizure, the consistent depolarization driven by the GABAAR-mediated HCO3− current boosts the activity-dependent Cl− influx into neurons, thereby directly triggering excitation in this state (39). In this study, Kaplan–Meier survival curve analyses revealed that serum bicarbonate levels below 22 mEq/L or above 29 mEq/L were associated with increased mortality at both 28 days and 90 days, corroborating earlier findings. The research demonstrated not only that both depleted and excessive bicarbonate levels are associated with mortality in both sepsis and epilepsy but also reinforced the utility of serum bicarbonate as a biomarker in patients with sepsis and concurrent epilepsy. Individuals with serum bicarbonate levels below 22 mEq/L or above 29 mEq/L can be identified as high-risk at an early stage. This early identification allows for more targeted and timely interventions, such as closer monitoring of vital signs, more frequent laboratory tests, and early adjustment of treatment plans. However, given the complexity of the pathogenesis of epilepsy and concurrent sepsis, a single biomarker may not be adequate for accurately predicting prognosis. Utilizing a combination of serum bicarbonate alongside markers such as procalcitonin, C-reactive protein, and lactate could enhance the precision of risk assessment and prognosis prediction.

This study possesses several advantages. Firstly, it employs data from the well-established MIMIC-IV 3.0 database, which is a rich and reliable source of ICU patient information. Moreover, this research employed appropriate statistical methods to verify the curve relationships of the association among serum bicarbonate concentration and short-term or long-term mortality rates, rather than a simple straight-line relationship. Lastly, we have identified the threshold for the smooth curve, which may contribute to suggesting to ICU physicians a better understanding and management of changes in the serum bicarbonate threshold of critically ill patients. Nevertheless, some limitations need to be taken into account. Firstly, as with other retrospective cohort studies, residual confounding may persist despite comprehensive adjustments for known confounding factors. Secondly, the data were solely sourced from the single-center MIMIC-IV 3.0 database, limiting the generalizability of the findings to broader populations. Thirdly, only the lowest serum bicarbonate levels during the first day of admission to ICU were analyzed. This approach does not account for dynamic changes in serum bicarbonate levels throughout the ICU stay, which may have significant implications for patient outcomes. Specifically, fluctuations in bicarbonate levels over time might reflect different pathophysiological processes effects that were not captured in our analysis. Future studies should consider longitudinal measurements of serum bicarbonate levels to better understand their prognostic significance and potential role in guiding clinical interventions.

There were two separate non-linear U-shaped relationships between serum bicarbonate levels at ICU admission and both short-term and long-term mortality rates in patients with concurrent sepsis and epilepsy rates. Compared to the normal range of serum bicarbonate concentration, both the 28-day and 90-day mortality risks were increased when the serum bicarbonate level deviates from the normal value. The optimal serum bicarbonate values associated with the lowest short-term and long-term mortality risks are 25.0 and 25.9 mEq/L, respectively. These results suggest a useful marker for risk stratification of patients with epilepsy and concurrent sepsis in the ICU, and underscore the importance of maintaining optimal serum bicarbonate levels to improve survival outcomes in these patients.