Objective
Electrocardiogram (EKG) abnormalities are predictive of subsequent cardiovascular events. Cardiac involvement is common in systemic lupus erythematosus (SLE). We aimed to determine the prevalence of EKG abnormalities in SLE patients and to examine the factors associated with EKG abnormalities with machine learning approaches.
Methods
Consecutive SLE patients’ records were retrieved from the database of the hospital for the cross‐sectional study. Abnormal EKGs with clinical significance were grouped by the presence of tachyarrhythmias, atrioventricular block, nonspecific ST segment changes, T wave abnormalities, ventricular hypertrophy, axis deviation, bundle branch block, and QT interval prolongation. Associated factors of the most common EKG abnormalities were assessed by comparing logistic regression and 4 other machine learning approaches.
Results
In the present study, 299 patients were enrolled, with 128 showing clinically significant abnormalities on EKG. T wave changes (52.3%), nonspecific ST segment–T wave (ST‐T) changes (26.6%), and prolonged QT interval (8.6%) were the most prevalent abnormalities among patients with abnormal findings on EKG. Random forests models had the best performance in the discovery of associated factors. Age, disease duration, antinuclear antibody titer, disease activity (as measured by the Systemic Lupus Erythematosus Disease Activity Index 2000) were associated with nonspecific ST‐T changes, prolonged QT interval, and T wave changes. Hypertension, positivity for anti‐SSA antibodies, and secondary Sjögren's syndrome were influential factors for nonspecific ST‐T changes, prolonged QT interval, and T wave changes, specifically.
Conclusion
ST‐T and T wave changes were the most common abnormalities seen on EKGs of SLE patients. Our finding suggests that age, longer disease duration, higher disease activity, hypertension, anti‐SSA antibody positivity, and secondary Sjögren's syndrome are important and influential factors in these EKG abnormalities.