BACKGROUND

Little is known about the relationship between circulating electrolyte concentrations and paroxysmal atrial fibrillation in the emergency department. We aimed to characterize circulating electrolyte concentrations in patients with paroxysmal atrial fibrillation compared with those of nonspecific control patients admitted to the emergency department.

METHODS

In total, data from 520 individuals with paroxysmal atrial fibrillation and 1,040 randomly selected 1040 patients without atrial fibrillation (1:2 ratio), all admitted to the emergency department (January 2010-December 2015), were analyzed. A classification model was developed using a tree-based machine learning algorithm, and the importance of variables was measured.

RESULTS

Patient age, serum glucose, sodium, potassium, calcium, phosphate, and sex were significantly associated with paroxysmal atrial fibrillation (all p < 0.001). For serum magnesium, the difference approached significance (p = 0.096). The model had a moderate performance with a 10-fold cross-validation accuracy of 0.728 and a sensitivity, specificity, area under the curve, and likelihood ratio of 0.613, 0.770, 0.692, and 2.67, respectively. Overall, age and glucose were the most important variables followed by serum sodium, potassium, and calcium. Male sex, older age, and a higher serum sodium, calcium, potassium, and magnesium, and a lower serum glucose and phosphate were associated with a higher likelihood of paroxysmal atrial fibrillation in the emergency department.

CONCLUSION

Serum electrolyte imbalances, particularly in sodium, potassium, and magnesium, are significantly associated with paroxysmal atrial fibrillation in emergency settings. Emergency physicians should monitor and correct these electrolytes to improve early PAF management and potentially prevent adverse outcomes.