Abstract
Objective
To investigate the influencing factors of new-onset atrial fibrillation (AF) in patients with sepsis and to construct a nomogram prediction model.
Methods
A retrospective analysis of 245 sepsis patients admitted to our hospital from March 2021 to March 2024 was used as the training set. An additional 107 sepsis patients admitted to our hospital from April 2024 to April 2025 were included as the validation set. The training and validation sets were divided into an AF group and a non-AF group based on the occurrence of new-onset AF.
Results
In the training set, there were significant differences between the two groups in terms of age, mechanical ventilation, APACHE II score, acute kidney injury, metabolic disorders, theophylline medication, TNF-α, E/e′, and NT-proBNP (P < 0.05). LASSO regression analysis was used to screen for 7 predictive factors. Logistic regression analysis identified age, mechanical ventilation, APACHE II score, theophylline medication, TNF-α, E/e′, and NT-proBNP as risk factors for new-onset AF in sepsis patients (P < 0.05). The area under the curve (AUC) of the ROC curve for the training set was 0.869, and the Hosmer-Lemeshow test yielded χ2=7.346 (P=0.713). The decision curve analysis (DCA) showed that the model has high clinical application value when the threshold probability is between 0.10 and 0.89. For external validation, the AUC of the ROC curve was 0.875, the Hosmer-Lemeshow test yielded χ2=6.992 (P=0.703), and the DCA curve showed that the model has high clinical application value when the threshold probability is between 0.12 and 0.83.
Conclusion
Age, mechanical ventilation, APACHE II score, theophylline medication, TNF-α, E/e′, and NT-proBNP are influencing factors for new-onset AF in patients with sepsis. The nomogram prediction model constructed on the basis of these factors has good clinical applicability.
Keywords: sepsis, atrial fibrillation, influencing factors, nomogram model
Introduction
Sepsis is a life-threatening systemic inflammatory response syndrome caused by infection. After pyogenic bacteria invade the bloodstream, they multiply extensively, causing a relatively intense inflammatory response in the patient, which can increase the risk of death in severe cases.1 Studies have shown that approximately 18 million people worldwide develop sepsis each year, with an annual increase of 1.5%–8%, and a mortality rate of about 30–70%.2 The heart, as the core of the circulatory system, is a key target organ affected during the progression of sepsis, with nearly half of patients experiencing varying degrees of myocardial injury.3 New-onset atrial fibrillation (AF) in sepsis patients is a serious complication that induces abnormal cardiac function. Studies have shown that new-onset AF occurs in about 40% of sepsis patients, which leads to unstable hemodynamics, increases the cost of medical treatment and hospitalization, and severely affects the patient’s prognosis.4,5 Therefore, early clinical identification of the influencing factors of new-onset AF can effectively improve patient prognosis. The nomogram model, as a visual graphical prediction tool, is a tool for clinical prognosis assessment. It is simple and fast to operate, can predict the risk value of a certain event, provides a reference for stratified management, and can assist clinicians in formulating corresponding measures to reduce the risk of morbidity.6,7 Studies have shown that a nomogram for individualized prediction of delirium in patients with sepsis was constructed, identifying four independent predictive factors. The calibration curves of the predictive model were close to the ideal curve in both the training and validation cohorts, and decision curve analysis (DCA) indicated that the nomogram had clinical value. Its performance and clinical utility were satisfactory, helping clinicians to identify patients at risk of delirium in a timely manner, implement early interventions, and improve neurological outcomes.6 Currently, there are few reports on new-onset AF in sepsis patients using nomogram prediction models. Therefore, this study aims to investigate the influencing factors of new-onset AF in sepsis patients and to construct a nomogram prediction model.
Materials and Methods
General Information
A retrospective collection of 245 sepsis patients admitted to our hospital from March 2021 to March 2024 was used as the training set. An additional 107 sepsis patients admitted to our hospital from April 2024 to April 2025 were collected as the validation set (for external validation). The training and validation sets were divided into an AF group (no prior history of atrial fibrillation, transferred to the ICU after the first detection of AF, or developed AF after entering the ICU) and a non-AF group. The case collection flowchart is shown in Figure 1. Inclusion criteria: (1) met the diagnostic criteria for sepsis;8 (2) met the diagnostic criteria for AF,9 with a duration >1 min, disappearance of P waves on the electrocardiogram, and irregular R-R intervals with normal atrioventricular conduction; (3) survival time >24 h after admission; (4) age >18 years; (5) complete clinical data. Exclusion criteria: (1) congenital heart disease; (2) pregnancy or lactation; (3) chronic or permanent atrial fibrillation; (4) immune dysfunction; (5) cognitive dysfunction. This study was approved by our hospital’s ethics committee.
Figure 1.
Case flow collection diagram.
Clinical Data Collection
Data were collected from the electronic systems of hospitalized patients in our hospital, mainly including: age, gender, smoking history, alcohol consumption history, chronic obstructive pulmonary disease (COPD), heart failure, ischemic heart disease, cerebrovascular disease, malignant tumors, acute kidney injury, new-onset heart failure, metabolic disorders, coagulation dysfunction, respiratory failure, liver dysfunction, mechanical ventilation, APACHE II score, continuous renal replacement therapy (CRRT), cedilanide, anticholinergic drugs, theophylline drugs, amiodarone, white blood cell count (WBC), hemoglobin (HGB), fibrinogen (FIB), interleukin-6 (IL-6), tumor necrosis factor-α (TNF-α), procalcitonin (PCT), creatine kinase-MB (CK-MB), cardiac troponin I (cTnI), potassium ion (K+), left ventricular ejection fraction (LVEF), E/e′, and N-terminal pro-B-type natriuretic peptide (NT-proBNP).
Statistical Analysis
SPSS 25.0 was used to process the data. Measurement data were analyzed using the t-test, expressed as (mean ± SD), while count data were analyzed using the χ2 test, expressed as n (%). LASSO regression was used to screen variables. Logistic regression analysis was used to identify the influencing factors of new-onset AF in sepsis patients in the training set. R software was used to construct the nomogram prediction model. Bootstrap was used for internal validation (1000 resamples). The ROC and calibration curves were used to evaluate the discrimination and consistency of the nomogram model. DCA was used to assess the clinical application value. A P-value <0.05 was considered statistically significant.
Results
Comparison of Clinical Data Between AF and Non-AF Groups in the Training Set
Among the 245 patients in the training set, 72 developed AF, with an incidence rate of 29.39%. There were significant differences between the two groups in terms of age, mechanical ventilation, APACHE II score, acute kidney injury, metabolic disorders, theophylline medication, TNF-α, E/e′, and NT-proBNP (P < 0.05). There were no significant differences in other clinical data (P>0.05). See Table 1.
Table 1.
Comparison of Clinical Data Between the AF Group and the Non-AF Group in the Training Set
| Factor | AF Group (n=72) | Non-AF Group (n=173) | t/χ2 | P |
|---|---|---|---|---|
| Age (years old) | 69.35±5.81 | 61.10±5.64 | 10.338 | <0.001 |
| Gender | 0.046 | 0.829 | ||
| Male | 46 (63.89) | 108 (62.43) | ||
| Female | 26 (36.11) | 65 (37.57) | ||
| History of smoking | 0.097 | 0.756 | ||
| Yes | 29 (40.28) | 66 (38.15) | ||
| No | 43 (59.73) | 107 (61.85) | ||
| History of alcohol consumption | 0.032 | 0.857 | ||
| Yes | 25 (34.72) | 58 (33.53) | ||
| No | 47 (65.28) | 115 (66.47) | ||
| COPD | 0.355 | 0.551 | ||
| Yes | 24 (33.33) | 51 (29.48) | ||
| No | 48 (66.67) | 122 (70.52) | ||
| Heart failure | 0.062 | 0.804 | ||
| Yes | 10 (13.89) | 22 (12.72) | ||
| No | 62 (86.11) | 151 (87.28) | ||
| Hypertension | 0.002 | 0.968 | ||
| Yes | 31 (43.06) | 74 (42.77) | ||
| No | 41 (56.94) | 99 (57.23) | ||
| Diabetes | 0.474 | 0.491 | ||
| Yes | 24 (33.33) | 50 (28.90) | ||
| No | 48 (66.67) | 123 (71.10) | ||
| Ischemic heart disease | 0.698 | 0.404 | ||
| Yes | 26 (36.11) | 53 (30.64) | ||
| No | 46 (63.89) | 120 (69.36) | ||
| Cerebrovascular disease | 0.044 | 0.833 | ||
| Yes | 15 (20.83) | 34 (19.65) | ||
| No | 57 (79.17) | 139 (80.35) | ||
| Malignancy | 0.543 | 0.461 | ||
| Yes | 6 (8.33) | 10 (5.78) | ||
| No | 66 (91.67) | 163 (94.22) | ||
| Acute kidney injury | 5.744 | 0.017 | ||
| Yes | 35 (48.61) | 56 (32.37) | ||
| No | 37 (51.39) | 117 (67.63) | ||
| New-onset heart failure | 2.225 | 0.136 | ||
| Yes | 18 (25.00) | 29 (16.76) | ||
| No | 54 (75.00) | 144 (83.24) | ||
| Metabolic disorders | 5.411 | 0.020 | ||
| Yes | 36 (50.00) | 59 (34.10) | ||
| No | 36 (50.00) | 114 (65.90) | ||
| Coagulopathy | 1.593 | 0.207 | ||
| Yes | 18 (25.00) | 31 (17.92) | ||
| No | 54 (75.00) | 142 (82.08) | ||
| Respiratory failure | 3.436 | 0.064 | ||
| Yes | 38 (52.78) | 69 (39.88) | ||
| No | 34 (47.22) | 104 (60.12) | ||
| Liver dysfunction | 0.894 | 0.344 | ||
| Yes | 12 (16.67) | 21 (12.14) | ||
| No | 60 (83.33) | 152 (87.86) | ||
| Mechanical ventilation | 12.657 | <0.001 | ||
| Yes | 39 (54.17) | 52 (30.06) | ||
| No | 33 (45.83) | 121 (69.94) | ||
| APACHE IIscore (points) | 25.26±3.41 | 23.41±2.74 | 4.469 | <0.001 |
| CRRT treatment | 2.612 | 0.106 | ||
| Yes | 15 (20.83) | 22 (12.72) | ||
| No | 57 (79.17) | 151 (87.28) | ||
| Cediran | 14 (19.44) | 32 (18.50) | 0.030 | 0.863 |
| Anticholinergic drugs | 13 (18.06) | 30 (17.34) | 0.018 | 0.893 |
| Theophylline | 38 (52.78) | 34 (32.37) | 8.954 | 0.003 |
| Amiodarone | 9 (12.50) | 16 (9.25) | 0.587 | 0.444 |
| WBC (×109/L) | 18.96±5.02 | 18.02±5.01 | 1.337 | 0.182 |
| HGB (g/L) | 108.65±20.41 | 109.24±21.04 | 0.202 | 0.840 |
| FIB (mg/dL) | 4.92±1.03 | 4.83±1.08 | 0.602 | 0.548 |
| IL-6 (ng/L) | 2531.26±156.24 | 2498.62±146.28 | 1.559 | 0.120 |
| TNF-α | 132.54±21.02 | 89.65±16.24 | 17.210 | <0.001 |
| PCT (μg/L) | 32.05±6.54 | 30.68±6.41 | 1.515 | 0.131 |
| CK-MB (μg/L) | 23.64±6.25 | 22.05±6.12 | 1.841 | 0.067 |
| cTnI (μg/L) | 5.94±1.21 | 6.13±1.24 | 1.100 | 0.272 |
| K+ (mmol/L) | 4.31±0.97 | 4.26±0.98 | 0.365 | 0.716 |
| LVEF (%) | 51.43±10.32 | 51.67±10.26 | 0.167 | 0.868 |
| E/e′ | 15.89±1.24 | 13.14±1.31 | 15.201 | <0.001 |
| NT-proBNP (ng/L) | 3046.28±165.23 | 2491.68±156.26 | 24.881 | <0.001 |
LASSO Regression Analysis
In the training set, with the occurrence of new-onset AF in sepsis patients as the dependent variable (Yes=1, No=0), and the aforementioned differing indicators—mechanical ventilation (Yes=1, No=0), acute kidney injury (Yes=1, No=0), metabolic disorders (Yes=1, No=0), theophylline medication (Yes=1, No=0), age, APACHE II score, TNF-α, E/e′, and NT-proBNP (all as measured values)—as independent variables, LASSO analysis was performed using R software (with 10-fold cross-validation to obtain the optimal λ value). The results showed that when the penalty coefficient λ=0.03671194, the model performance was optimal with the fewest influencing factors. Ultimately, 7 predictive factors were selected: age, mechanical ventilation, APACHE II score, theophylline medication, TNF-α, E/e′, and NT-proBNP. See Figure 2.
Figure 2.
LASSO Regression Analysis LASSO. (A) Relationship plot of regression coefficients. (B) 10-fold cross-validation results of LASSO regression.
Multivariate Analysis of New-Onset AF in Sepsis Patients
The 7 independent variables selected by LASSO regression were included in a Logistic regression analysis (forward stepwise method). The results showed that age, mechanical ventilation, APACHE II score, theophylline medication, TNF-α, E/e′, and NT-proBNP were risk factors for new-onset AF in sepsis patients (P < 0.05). See Table 2.
Table 2.
Multivariate Analysis of New-Onset AF in Patients with Sepsis
| Variable | Β value | SE value | Wald χ2 value | P value | OR value | 95% CI |
|---|---|---|---|---|---|---|
| Age | 0.199 | 0.053 | 13.901 | <0.001 | 1.220 | 1.099~1.355 |
| Mechanical ventilation | 2.056 | 0.641 | 10.280 | 0.001 | 7.815 | 2.224~27.464 |
| APACHE II score | 0.111 | 0.039 | 8.092 | 0.004 | 1.117 | 1.035~1.206 |
| Theophylline | 1.278 | 0.345 | 13.718 | <0.001 | 3.590 | 1.825~7.061 |
| TNF-α | 0.023 | 0.010 | 6.021 | 0.014 | 1.024 | 1.005~1.043 |
| E/e′ | 1.329 | 0.267 | 24.698 | <0.001 | 3.778 | 2.237~6.381 |
| NT-proBNP | 1.103 | 0.326 | 11.453 | 0.001 | 3.014 | 1.591~5.710 |
| Constant | −44.700 | 7.191 | 38.644 | <0.001 | <0.001 | – |
Construction of the Nomogram Model for New-Onset AF in Sepsis Patients
A nomogram model was constructed using the risk factors identified above, with P=ex/(1+ex), where x = −44.700 + 0.199 × Age + 2.056 × Mechanical Ventilation + 0.111 × APACHE II Score + 1.278 × Theophylline Medication + 0.023 × TNF-α + 1.329 × E/e′ + 1.103 × NT-proBNP. In this model, E/e′ was identified as the strongest predictive factor. See Figure 3.
Figure 3.
Construction of a nomogram model for new-onset AF in patients with sepsis.
Internal Validation of the Nomogram Model for New-Onset AF in Sepsis Patients
The ROC curve was used to assess discrimination, yielding an AUC of 0.869 (95% CI = 0.823–0.915). The Hosmer-Lemeshow (H-L) test was used to validate consistency, yielding χ2=7.346 (P=0.713). See Figure 4.
Figure 4.
Internal validation of the nomogram model for new-onset atrial fibrillation (AF) in sepsis patients in the training set. (A) ROC curves — the larger the area under the curve (AUC), the better the model’s discriminative ability. (B) Calibration curves — the closer the curve is to the 45° diagonal line, the better the agreement between predicted and observed outcomes.
Internal Validation of the Nomogram Model’s DCA Curve in the Training Set
The results of the decision curve analysis (DCA) showed that when the threshold probability was between 0.10 and 0.89, the model had high clinical application value for assessing new-onset AF in sepsis patients. See Figure 5.
Figure 5.
Internal validation of the nomogram model in the training set using the decision curve analysis (DCA). The horizontal axis represents the threshold probability, ranging from 0 to 1, corresponding to the probability threshold used in clinical practice to determine whether a patient will experience a disease or an adverse outcome. The vertical axis represents net benefit, with higher values indicating greater net clinical benefit at that threshold.
Comparison of Clinical Data Between AF and Non-AF Groups in the Validation Set
Among the 107 patients in the validation set, 32 developed AF, with an incidence rate of 29.91%. There were significant differences between the two groups in age, mechanical ventilation, APACHE II score, theophylline medication, TNF-α, E/e′, and NT-proBNP (P < 0.05). There were no significant differences in other variables (P>0.05). See Table 3.
Table 3.
The Clinical Data of the AF Group and the Non-AF Group in the Validation Set
| Factor | AF Group (n=32) | Non-AF Group (n=75) | t/χ2 | P |
|---|---|---|---|---|
| Age (years old) | 68.65±5.24 | 60.89±5.13 | 7.119 | <0.001 |
| Gender | 0.247 | 0.619 | ||
| Male | 20 (62.50) | 43 (57.33) | ||
| Female | 12 (37.50) | 32 (42.67) | ||
| History of smoking | 0.036 | 0.849 | ||
| Yes | 13 (40.63) | 29 (38.67) | ||
| No | 19 (59.38) | 46 (61.33) | ||
| History of alcohol consumption | 0.011 | 0.917 | ||
| Yes | 11 (34.38) | 25 (33.33) | ||
| No | 21 (65.63) | 50 (66.67) | ||
| COPD | 0.039 | 0.843 | ||
| Yes | 10 (31.25) | 22 (29.33) | ||
| No | 22 (68.75) | 53 (70.67) | ||
| Heart failure | 0.076 | 0.783 | ||
| Yes | 4 (12.50) | 8 (10.67) | ||
| No | 28 (87.50) | 67 (89.33) | ||
| Hypertension | 0.791 | 0.374 | ||
| Yes | 14 (43.75) | 26 (34.67) | ||
| No | 18 (56.25) | 49 (65.33) | ||
| Diabetes | 0.267 | 0.605 | ||
| Yes | 11 (34.38) | 22 (29.33) | ||
| No | 21 (65.63) | 53 (70.67) | ||
| Ischemic heart disease | 0.172 | 0.678 | ||
| Yes | 12 (37.50) | 25 (33.33) | ||
| No | 20 (62.50) | 50 (66.67) | ||
| Cerebrovascular disease | 0.048 | 0.826 | ||
| Yes | 7 (21.08) | 15 (20.00) | ||
| No | 25 (78.13) | 60 (80.00) | ||
| Malignancy | 0.238 | 0.626 | ||
| Yes | 3 (9.38) | 5 (6.67) | ||
| No | 29 (90.63) | 70 (93.33) | ||
| Acute kidney injury | 1.414 | 0.234 | ||
| Yes | 15 (46.08) | 26 (34.67) | ||
| No | 17 (53.13) | 49 (65.33) | ||
| New-onset heart failure | 0.173 | 0.677 | ||
| Yes | 8 (25.00) | 16 (21.33) | ||
| No | 24 (75.00) | 59 (78.67) | ||
| Metabolic disorders | 3.020 | 0.082 | ||
| Yes | 14 (43.75) | 20 (26.67) | ||
| No | 18 (56.25) | 55 (73.33) | ||
| Coagulopathy | 0.146 | 0.702 | ||
| Yes | 7 (21.88) | 14 (18.67) | ||
| No | 25 (78.13) | 61 (81.33) | ||
| Respiratory failure | 0.850 | 0.357 | ||
| Yes | 15 (46.88) | 28 (37.33) | ||
| No | 17 (53.13) | 47 (62.67) | ||
| Liver dysfunction | 0.130 | 0.718 | ||
| Yes | 14 (43.75) | 30 (40.00) | ||
| No | 18 (56.25) | 45 (60.00) | ||
| Mechanical ventilation | 10.914 | 0.001 | ||
| Yes | 16 (50.00) | 14 (18.67) | ||
| No | 16 (50.00) | 61 (81.33) | ||
| APACHE IIscore (points) | 24.98±3.21 | 22.10±2.34 | 5.192 | <0.001 |
| CRRT treatment | 0.280 | 0.597 | ||
| Yes | 6 (18.75) | 11 (14.67) | ||
| No | 26 (81.25) | 64 (85.33) | ||
| Cediran | 5 (15.63) | 10 (13.33) | 0.098 | 0.755 |
| Anticholinergic drugs | 6 (18.75) | 12 (16.00) | 0.121 | 0.728 |
| Theophylline | 17 (53.13) | 18 (24.00) | 8.644 | 0.003 |
| Amiodarone | 4 (12.50) | 6 (8.00) | 0.536 | 0.464 |
| WBC (×109/L) | 17.59±4.15 | 17.11±4.06 | 0.556 | 0.579 |
| HGB (g/L) | 106.38±19.86 | 107.69±20.24 | 0.308 | 0.759 |
| FIB (mg/dL) | 4.89±0.99 | 4.85±1.01 | 0.189 | 0.851 |
| IL-6 (ng/L) | 2530.68±143.25 | 2510.68±136.58 | 0.683 | 0.496 |
| TNF-α | 128.62±20.11 | 96.58±17.21 | 8.377 | <0.001 |
| PCT (μg/L) | 31.09±6.12 | 30.72±6.21 | 0.283 | 0.777 |
| CK-MB (μg/L) | 22.68±6.11 | 22.34±6.04 | 0.266 | 0.791 |
| cTnI (μg/L) | 5.96±1.06 | 6.02±1.03 | 0.274 | 0.785 |
| K+ (mmol/L) | 4.29±0.99 | 4.27±0.85 | 0.106 | 0.916 |
| LVEF (%) | 50.96±9.86 | 50.72±9.62 | 0.117 | 0.907 |
| E/e′ | 15.43±1.11 | 12.34±1.05 | 13.702 | <0.001 |
| NT-proBNP (ng/L) | 3001.69±156.38 | 2476.58±154.65 | 16.028 | <0.001 |
External Validation of the Nomogram Model for New-Onset AF in Sepsis Patients in the Validation Set
The AUC of the ROC curve for external validation was 0.875 (95% CI=0.811–0.939). The Hosmer-Lemeshow (H-L) test was used to validate consistency, yielding χ2=6.992 (P=0.703). See Figure 6.
Figure 6.
External validation of the nomogram model for new-onset atrial fibrillation (AF) in septic patients in the validation cohort. (A) ROC curves — the larger the area under the curve (AUC), the better the model’s discriminative ability. (B) Calibration curves — the closer the curve is to the 45° diagonal line, the better the agreement between predicted and observed outcomes.
External Validation of the Nomogram Model’s DCA Curve in the Validation Set
The results of the decision curve analysis (DCA) showed that when the threshold probability was between 0.12 and 0.83, the model had high clinical application value for assessing new-onset AF in sepsis patients. See Figure 7.
Figure 7.
External validation of the nomogram model using decision curve analysis (DCA) in the validation set. The horizontal axis represents the threshold probability, ranging from 0 to 1, corresponding to the probability threshold used in clinical practice to determine whether a patient will experience a disease or an adverse outcome. The vertical axis represents net benefit, with higher values indicating greater net clinical benefit at that threshold.
Discussion
The pathogenesis of sepsis is related to the pathophysiological changes in multiple organs and systems of the body, including systemic inflammation, tissue injury, etc. Clinically, most patients exhibit a systemic inflammatory response, increased respiratory rate, and decreased blood oxygen, which can also lead to shock, threatening human life.10 Previous research has found that sepsis-induced cardiac dysfunction increases patient mortality; patients experience changes in cardiac structure, which ultimately induce cardiovascular events such as AF, impair cardiac function, and in severe cases, lead to death.11 Some studies have confirmed that patients undergoing sepsis-related treatment are prone to tachycardia and AF. Under these circumstances, treatment for rapid arrhythmias is generally ineffective, leading to a poor prognosis.12 Therefore, clinically identifying the factors that can influence AF for early prevention is crucial to maximally improve patient outcomes.
This study found that in the training set, there were differences between the two groups in age, mechanical ventilation, APACHE II score, acute kidney injury, metabolic disorders, theophylline medication, TNF-α, E/e′, and NT-proBNP. After screening with LASSO regression, 7 predictive factors were identified. Logistic analysis confirmed that age, mechanical ventilation, APACHE II score, theophylline medication, TNF-α, E/e′, and NT-proBNP are risk factors for new-onset AF in sepsis patients. The reasons are analyzed as follows: (1) For older patients, due to lower muscle mass, their immune function is correspondingly reduced, and the function of various organs also declines. The inflammatory response produced by sepsis can rapidly invade myocardial tissue, causing myocardial injury and inducing AF. Furthermore, elderly individuals have a high prevalence of cardiovascular diseases and poorer physical constitution, which reduces myocardial fiber compliance and cardiac reserve function. When sepsis occurs, the severity of the condition increases the risk of new-onset AF.13,14 Studies have shown that sepsis is more common in the elderly population mainly due to advanced age and reduced physical fitness, which predisposes them to AF,15 consistent with the findings of this study. (2) Mechanical ventilation can improve a patient’s ventilation and oxygenation, but during treatment, improper operation by medical staff can cause the patient to resist violently, leading to a rapid increase in heart rate. Additionally, excessive ventilation pressure during treatment can cause excessive intrathoracic pressure, reducing the patient’s cardiac output, stimulating the sympathetic nervous system and increasing myocardial excitability, which can easily trigger AF, thereby lowering blood pressure, increasing heart rate, and impairing cardiac function, thus increasing the risk of AF.16,17 (3) One study found that the APACHE II score is a factor for AF in ICU sepsis patients, which is similar to our study’s findings, indicating that the more severe the patient’s condition, the higher the likelihood of developing AF.18 Therefore, regular clinical monitoring of changes in the APACHE II score can be used to assess disease progression and thus predict the occurrence of AF. (4) Theophylline drugs can increase myocardial blood supply and cardiac contractility when treating sepsis patients. However, due to the significant stimulatory nature of these drugs, studies have found that infusing theophylline in sepsis patients may cause AF,19 which is consistent with the results of this study. Therefore, close monitoring of the patient’s heart rate is necessary when using this drug clinically, as this class of drugs may also impair cardiac function while improving the patient’s condition, and attention should be paid to the dosage used. (5) The pathogenesis of AF is closely related to cardiac structural remodeling caused by the inflammatory response. TNF-α plays a role in the progression of inflammation. Studies have found that TNF-α is significantly elevated in patients with atrial fibrillation, and the use of tumor necrosis factor inhibitors can alleviate atrial remodeling, suggesting that it may induce AF by participating in the structural remodeling of the atria, mainly because inflammatory factors can impair cardiac function, leading to an acute increase in left ventricular end-diastolic pressure and excessive stretching of the left atrium, thereby promoting the occurrence of AF.20–22 (6) E/e′ is an indicator for assessing left ventricular diastolic function. Sepsis-induced cardiomyopathy mainly manifests as left atrial enlargement and diastolic dysfunction. An elevated E/e′ indicates possible diastolic dysfunction, which can easily lead to AF. This is because when the body’s inflammation damages cardiac function, it causes an acute increase in left ventricular end-diastolic pressure, leading to a subsequent rise in E/e′, which causes excessive stretching of the left atrium and increases the risk of AF.23,24 (7) NT-proBNP is a hormone secreted by ventricular myocytes in response to heart failure and increased ventricular wall tension. When ventricular volume load and ventricular wall tension increase, the secretion of NT-proBNP is promoted. Studies have found that inflammatory factors and endotoxins released in sepsis patients can cause AF, leading to elevated NT-proBNP levels,25,26 which is consistent with this study, indicating that its elevated level can be used as an indicator to assess AF in patients.
Nomograms can transform complex regression equations into graphical representations, making the results of predictive models more intuitive and facilitating the assessment of patients by clinicians. Based on LASSO regression and Logistic analysis, this study constructed a nomogram model. The AUCs for the training set and external validation were 0.869 and 0.875, respectively, and the H-L test showed good consistency. The DCA curve showed high clinical application value when the threshold probability was 0.10–0.89 for the training set and 0.12–0.83 for the external validation set. This indicates that the nomogram model has high clinical utility and can assist medical personnel in predicting the risk of AF and implementing timely prevention, thereby effectively improving patient prognosis. Therefore, in clinical practice, high-risk patients should be closely monitored, and patients with inflammatory responses should also receive heightened attention, in order to minimize the occurrence of AF as much as possible. This study has certain limitations. As a retrospective study with a relatively small sample size, there is a potential risk of model overfitting. Moreover, being a single-center study, it may be subject to bias. Future studies with larger sample sizes are needed for further validation.
Conclusion
In summary, age, mechanical ventilation, APACHE II score, theophylline medication, TNF-α, E/e′, and NT-proBNP are influencing factors for new-onset AF in sepsis patients. The nomogram prediction model constructed based on these factors has good clinical applicability and can serve as an assessment tool for screening patients at risk of new-onset AF.
Data Sharing Statement
The datasets used during the present study are available from the corresponding author upon reasonable request.
Ethics Approval and Consent to Participate
This study involving human participants was in accordance with the ethical standards of The Guangdong Medical University Affiliated Hospital Research Committee (No. PJ2025134) and with the 1964 Helsinki Declaration. Written informed consent was obtained from all participants. For patients who were unable to provide informed consent themselves due to severe sepsis or impaired consciousness, consent was obtained from their legal guardians in accordance with ethical requirements.
Disclosure
The authors declare no conflicts of interest in this work.
References
- 1.Liu D, Huang S-Y, Sun J-H, et al. Sepsis-induced immunosuppression: mechanisms, diagnosis and current treatment options. Mil Med Res. 2022;9(1):56. doi: 10.1186/s40779-022-00422-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Fleischmann-Struzek C, Mellhammar L, Rose N, et al. Incidence and mortality of hospital- and ICU-treated sepsis: results from an updated and expanded systematic review and meta-analysis. Intensive Care Med. 2020;46(8):1552–13. doi: 10.1007/s00134-020-06151-x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Habimana R, Choi I, Cho HJ, et al. Sepsis-induced cardiac dysfunction: a review of pathophysiology. Acute Crit Care. 2020;35(2):57–66. doi: 10.4266/acc.2020.00248 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Yao T, Chen Y, Ma J, et al. Clinical characteristics and risk factors for death in patients with stress cardiomyopathy in the ICU. Ann Palliat Med. 2021;10(12):12420–12430. doi: 10.21037/apm-21-3223 [DOI] [PubMed] [Google Scholar]
- 5.Angriman F, Rosella LC, Lawler PR, et al. Risk factors for major cardiovascular events in adult sepsis survivors: a population-based cohort study. Crit Care Med. 2023;51(4):471–483. doi: 10.1097/CCM.0000000000005793 [DOI] [PubMed] [Google Scholar]
- 6.Gu Q, Yang S, Fei D, et al. A nomogram for predicting sepsis-associated delirium: a retrospective study in MIMIC III. BMC Med Inform Decis Mak. 2023;23(1):184. doi: 10.1186/s12911-023-02282-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Ren Y, Zhang L, Xu F, et al. Risk factor analysis and nomogram for predicting in-hospital mortality in ICU patients with sepsis and lung infection. BMC Pulm Med. 2022;22(1):17. doi: 10.1186/s12890-021-01809-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Evans L, Rhodes A, Alhazzani W, et al. Surviving sepsis campaign: international guidelines for management of sepsis and septic shock 2021. Crit Care Med. 2021;49(11):e1063–e143. doi: 10.1097/CCM.0000000000005337 [DOI] [PubMed] [Google Scholar]
- 9.Sagris M, Vardas EP, Theofilis P, et al. Atrial fibrillation: pathogenesis, predisposing factors, and genetics. Int J Mol Sci. 2021;23(1). doi: 10.3390/ijms23010006 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Rababa M, Bani Hamad D, Hayajneh AA. Sepsis assessment and management in critically Ill adults: a systematic review. PLoS One. 2022;17(7):e0270711. doi: 10.1371/journal.pone.0270711 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Honorato MO, Sousa Filho JTD, Honorato Junior LFB, et al. Atrial fibrillation and sepsis in elderly patients and their associaton with in-hospital mortality. Arq Bras Cardiol. 2023;120(3):e20220295. doi: 10.36660/abc.20220295 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Ruslan MA, Baharuddin KA, Noor NM, et al. Norepinephrine in septic shock: a systematic review and meta-analysis. West J Emerg Med. 2021;22(2):196–203. doi: 10.5811/westjem.2020.10.47825 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Law AC, Bosch NA, Peterson D, et al. Comparison of heart rate after phenylephrine vs norepinephrine initiation in patients with septic shock and atrial fibrillation. Chest. 2022;162(4):796–803. doi: 10.1016/j.chest.2022.04.147 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Guan C, Gong A, Zhao Y, et al. Interpretable machine learning model for new-onset atrial fibrillation prediction in critically ill patients: a multi-center study. Crit Care. 2024;28(1):349. doi: 10.1186/s13054-024-05138-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Rudd KE, Johnson SC, Agesa KM, et al. Global, regional, and national sepsis incidence and mortality, 1990–2017: analysis for the Global Burden of Disease Study. Lancet. 2020;395(10219):200–211. doi: 10.1016/S0140-6736(19)32989-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Lin Z, Han H, Guo W, et al. Atrial fibrillation in critically ill patients who received prolonged mechanical ventilation: a nationwide inpatient report. Korean J Intern Med. 2021;36(6):1389–1401. doi: 10.3904/kjim.2020.142 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Liu Y-W, Wang Y-F, Chen Y, et al. A nationwide study on new onset atrial fibrillation risk factors and its association with hospital mortality in sepsis patients. Sci Rep. 2024;14(1):12206. doi: 10.1038/s41598-024-62630-x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Chen B, Tang WB, Li XJ, et al. [Analysis of the clinical characteristics and risk factors of postoperative atrial fibrillation in patients with critical burns]. Zhonghua Shao Shang Yu Chuang Mian Xiu Fu Za Zhi. 2022;38(5):408–414. Danish. doi: 10.3760/cma.j.cn501225-20220214-00026 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Ramalho NJD, Švecová O, Kula R, et al. Aminophylline at clinically relevant concentrations affects inward rectifier potassium current in a dual way. Pflugers Arch. 2022;474(3):303–313. doi: 10.1007/s00424-021-02646-8 [DOI] [PubMed] [Google Scholar]
- 20.Dai H, Ye J, Wang S, et al. Myeloperoxidase and its derivative hypochlorous acid combined clinical indicators predict new-onset atrial fibrillation in sepsis: a case-control study. BMC Cardiovasc Disord. 2024;24(1):377. doi: 10.1186/s12872-024-04034-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Yang H, Feng L, Jiang Z, et al. Amlexanox reduces new-onset atrial fibrillation risk in sepsis by downregulating S100A12: a Mendelian randomization study. Front Cardiovasc Med. 2024;11:1401314. doi: 10.3389/fcvm.2024.1401314 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Eyiol A. The relationship of pan-immune-inflammation value (PIV) and HALP score with prognosis in patients with atrial fibrillation. Medicine. 2024;103(36):e39643. doi: 10.1097/MD.0000000000039643 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Lv X, Shi X, Maihemuti M, et al. Correlation of HMGB1, TLR2 and TLR4 with left ventricular diastolic dysfunction in sepsis patients. Scand J Immunol. 2023;97(4):e13260. doi: 10.1111/sji.13260 [DOI] [PubMed] [Google Scholar]
- 24.Pujara AC, Koprivanac M, Stembal F, et al. Atrial fibrillation after descending aorta repair: occurrence, risk factors, and impact on outcomes. Aorta. 2023;11(3):116–124. doi: 10.1055/s-0043-1770960 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Nishikimi T, Nakagawa Y. Potential pitfalls when interpreting plasma BNP levels in heart failure practice. J Cardiol. 2021;78(4):269–274. doi: 10.1016/j.jjcc.2021.05.003 [DOI] [PubMed] [Google Scholar]
- 26.Lörstad S, Wang Y, Tehrani S, et al. Development of an extended cardiovascular SOFA score component reflecting cardiac dysfunction with improved survival prediction in sepsis: an exploratory analysis in the Sepsis and Elevated Troponin (SET) Study. J Intensive Care Med. 2025;40(3):320–330. doi: 10.1177/08850666241282294 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The datasets used during the present study are available from the corresponding author upon reasonable request.