Development and validation of a predictive model for new-onset atrial fibrillation in sepsis based on clinical risk factors

Abstract

Objective: New-onset atrial fibrillation (NOAF) is a common complication and one of the primary causes of increased mortality in critically ill adults. Since early assessment of the risk of developing NOAF is difficult, it is critical to establish predictive tools to identify the risk of NOAF.

Methods: We retrospectively enrolled 1,568 septic patients treated at Wuhan Union Hospital (Wuhan, China) as a training cohort. For external validation of the model, 924 patients with sepsis were recruited as a validation cohort at the First Affiliated Hospital of Xinjiang Medical University (Urumqi, China). Least absolute shrinkage and selection operator (LASSO) regression and multivariate logistic regression analyses were used to screen predictors. The area under the ROC curve (AUC), calibration curve, and decision curve were used to assess the value of the predictive model in NOAF.

Results: A total of 2,492 patients with sepsis (1,592 (63.88%) male; mean [SD] age, 59.47 [16.42] years) were enrolled in this study. Age (OR: 1.022, 1.009-1.035), international normalized ratio (OR: 1.837, 1.270-2.656), fibrinogen (OR: 1.535, 1.232-1.914), C-reaction protein (OR: 1.011, 1.008-1.014), sequential organ failure assessment score (OR: 1.306, 1.247-1.368), congestive heart failure (OR: 1.714, 1.126-2.608), and dopamine use (OR: 1.876, 1.227-2.874) were used as risk variables to develop the nomogram model. The AUCs of the nomogram model were 0.861 (95% CI, 0.830-0.892) and 0.845 (95% CI, 0.804-0.886) in the internal and external validation, respectively. The clinical prediction model showed excellent calibration and higher net clinical benefit. Moreover, the predictive performance of the model correlated with the severity of sepsis, with higher predictive performance for patients in septic shock than for other patients.

Conclusion: The nomogram model can be used as a reliable and simple predictive tool for the early identification of NOAF in patients with sepsis, which will provide practical information for individualized treatment decisions.

Keywords: SOFA score; new-onset atrial fibrillation; nomogram; predictive model; sepsis.

FIGURE 1

The flow diagram of developing and validating the prediction model.

FIGURE 2

Variable selection using the least absolute shrinkage and selection operator (LASSO) binary logistic regression model. (A) The tuning parameter (λ) in the LASSO model was selected for 10-fold cross-validation by the minimum criteria. The dotted vertical lines were drawn at the best values using the minimum criteria and 1 standard error of the minimum criteria (the 1-SE criteria). A λ-value of 0.021, with log (λ), –3.855 was chosen (1-SE criteria) according to 10-fold cross-validation. (B) LASSO coefficient curves of the 48 variables. A coefficient profile plot was produced against the log (λ) sequence. Vertical line was drawn at the value selected using 10-fold cross-validation, where optimal λ resulted in 7 non-zero coefficients.

FIGURE 3

Forest plot showing the relationship between risk factors and the development of new-onset atrial fibrillation in patients with sepsis.

FIGURE 4

Nomogram for predicting the risk of new-onset atrial fibrillation in patients with sepsis. A 70-year-old patient with sepsis and no history of congestive heart failure. During hospitalization INR was 0.83, fibrinogen was 4.87 g/L, C-reactive protein was 108 mg/L, SOFA score was 11, and dopamine was not used during treatment. This patient had a total score of 163 and a 33.0% risk of developing new-onset atrial fibrillation.

FIGURE 5

Discrimination and calibration of nomogram prediction models in the training and validation cohorts. (A) Calibration plot in the training cohort. (B) Calibration plot in the validation cohort. (C) ROC curves in both the training and validation cohorts.

FIGURE 6

Evaluation of clinical utility of nomogram prediction models in the training and validation cohorts. (A) Decision curves in both the training and validation cohorts. (B) Clinical impact curve in the training cohort. (C) Clinical impact curve in the validation cohort.

FIGURE 7

Cumulative mortality in patients with sepsis based on kaplan-meier curves. (A) Cumulative mortality in all patients with sepsis. (B) Comparison of cumulative mortality between new-onset atrial fibrillation and non-new-onset atrial fibrillation.