Retrospective cohort study of new-onset atrial fibrillation in acute pulmonary embolism on prognosis

Machine learning-based model to predict composite thromboembolic events among Chinese elderly patients with atrial fibrillation

OBJECTIVE

Accurate prediction of survival prognosis is helpful to guide clinical decision-making. The aim of this study was to develop a model using machine learning techniques to predict the occurrence of composite thromboembolic events (CTEs) in elderly patients with atrial fibrillation(AF). These events encompass newly diagnosed cerebral ischemia events, cardiovascular events, pulmonary embolism, and lower extremity arterial embolism.

METHODS

This retrospective study included 6,079 elderly hospitalized patients (≥ 75 years old) with AF admitted to the People's Liberation Army General Hospital in China from January 2010 to June 2022. Random forest imputation was used for handling missing data. In the descriptive statistics section, patients were divided into two groups based on the occurrence of CTEs, and differences between the two groups were analyzed using chi-square tests for categorical variables and rank-sum tests for continuous variables. In the machine learning section, the patients were randomly divided into a training dataset (n = 4,225) and a validation dataset (n = 1,824) in a 7:3 ratio. Four machine learning models (logistic regression, decision tree, random forest, XGBoost) were trained on the training dataset and validated on the validation dataset.

RESULTS

The incidence of composite thromboembolic events was 19.53%. The Least Absolute Shrinkage and Selection Operator (LASSO) method, using 5-fold cross-validation, was applied to the training dataset and identified a total of 18 features that exhibited a significant association with the occurrence of CTEs. The random forest model outperformed other models in terms of area under the curve (ACC: 0.9144, SEN: 0.7725, SPE: 0.9489, AUC: 0.927, 95% CI: 0.9105-0.9443). The random forest model also showed good clinical validity based on the clinical decision curve. The Shapley Additive exPlanations (SHAP) showed that the top five features associated with the model were history of ischemic stroke, high triglyceride (TG), high total cholesterol (TC), high plasma D-dimer, age.

CONCLUSIONS

This study proposes an accurate model to stratify patients with a high risk of CTEs. The random forest model has good performance. History of ischemic stroke, age, high TG, high TC and high plasma D-Dimer may be correlated with CTEs.

Ischemic Stroke Temporally Associated With New-Onset Atrial Fibrillation: A Population-Based Registry-Linkage Study

BACKGROUND

Limited data exist on the temporal relationship between new-onset atrial fibrillation (AF) and ischemic stroke and its impact on patients' clinical characteristics and mortality.

METHODS

A population-based registry-linkage database includes all patients with new-onset AF in Finland from 2007 to 2018. Ischemic stroke temporally associated with AF (ISTAF) was defined as an ischemic stroke occurring within ±30 days from the first AF diagnosis. Clinical factors associated with ISTAF were studied with logistic regression and 90-day survival with Cox proportional hazards analysis.

RESULTS

Among 229 565 patients with new-onset AF (mean age, 72.7 years; 50% female), 204 774 (89.2%) experienced no ischemic stroke, 12 209 (5.3%) had past ischemic stroke >30 days before AF, and 12 582 (5.8%) had ISTAF. The annual proportion of ISTAF among patients with AF decreased from 6.0% to 4.8% from 2007 to 2018. Factors associated positively with ISTAF were higher age, lower education level, and alcohol use disorder, whereas vascular disease, heart failure, chronic kidney disease cancer, and psychiatric disorders were less probable with ISTAF. Compared with patients without ischemic stroke and those with past ischemic stroke, ISTAF was associated with ≈3-fold and 1.5-fold risks of death (adjusted hazard ratios, 2.90 [95% CI, 2.76-3.04] and 1.47 [95% CI, 1.39-1.57], respectively). The 90-day survival probability of patients with ISTAF increased from 0.79 (95% CI, 0.76-0.81) in 2007 to 0.89 (95% CI, 0.87-0.91) in 2018.

CONCLUSIONS

ISTAF depicts the prominent temporal clustering of ischemic strokes surrounding AF diagnosis. Despite having fewer comorbidities, patients with ISTAF had worse, albeit improving, survival than patients with a history of or no ischemic stroke.

REGISTRATION

URL: https://www.clinicaltrials.gov; Unique identifier: NCT04645537. URL: https://www.encepp.eu; Unique identifier: EUPAS29845.

New-Onset Paroxysmal Atrial Fibrillation in the Setting of Acute Pulmonary Embolism Is Associated with All-Cause Hospital Mortality in Women but Not in Men

BACKGROUND

Patients with acute pulmonary embolism (PE) may have various types of atrial fibrillation (AF). The role of AF in hemodynamic states and outcomes may differ between men and women.

METHODS

In total, 1600 patients (743 males and 857 females) with acute PE were enrolled in this study. The severity of PE was assessed using the European Society of Cardiology (ESC) mortality risk model. Patients were allocated into three groups according to their electrocardiography recordings taken during hospitalization: sinus rhythm, new-onset paroxysmal AF, and persistent/permanent AF. The association between the types of AF and all-cause hospital mortality was tested using regression models and net reclassification index (NRI) and integrated discrimination index (IDI) statistics with respect to sex.

RESULTS

There were no differences between the frequencies of the types of AF between men and women: 8.1% vs. 9.1% and 7.5% vs. 7.5% (p = 0.766) for paroxysmal and persistent/permanent AF, respectively. We found that the rates of paroxysmal AF significantly increased across the mortality risk strata in both sexes. Among the types of AF, the presence of paroxysmal AF had a predictive value for all-cause hospital mortality independent of mortality risk and age in women only (adjusted HR, 2.072; 95% CI, 1.274-3.371; p = 0.003). Adding paroxysmal AF to the ESC risk model did not improve the reclassification of patient risk for the prediction of all-cause mortality, but instead enhanced the discriminative power of the existing model in women only (NRI, not significant; IDI, 0.022 (95% CI, 0.004-0.063); p = 0.013).

CONCLUSION

The occurrence of paroxysmal AF in female patients with acute PE has predictive value for all-cause hospital mortality independent of age and mortality risk.