New-Onset Atrial Fibrillation in Patients Hospitalized With COVID-19: Results From the American Heart Association COVID-19 Cardiovascular Registry.

BACKGROUND: New-onset atrial fibrillation (AF) in patients hospitalized with COVID-19 has been reported and associated with poor clinical outcomes. We aimed to understand the incidence of and outcomes associated with new-onset AF in a diverse and representative US cohort of patients hospitalized with COVID-19. METHODS: We used data from the American Heart Association COVID-19 Cardiovascular Disease Registry. Patients were stratified by the presence versus absence of new-onset AF. The primary and secondary outcomes were in-hospital mortality and major adverse cardiovascular events (MACE; cardiovascular death, myocardial infarction, stroke, cardiogenic shock, and heart failure). The association of new-onset AF and the primary and secondary outcomes was evaluated using Cox proportional-hazards models for the primary time to event analyses. RESULTS: Of the first 30 999 patients from 120 institutions across the United States hospitalized with COVID-19, 27 851 had no history of AF. One thousand five hundred seventeen (5.4%) developed new-onset AF during their index hospitalization. New-onset AF was associated with higher rates of death (45.2% versus 11.9%) and MACE (23.8% versus 6.5%). The unadjusted hazard ratio for mortality was 1.99 (95% CI, 1.81-2.18) and for MACE was 2.23 (95% CI, 1.98-2.53) for patients with versus without new-onset AF. After adjusting for demographics, clinical comorbidities, and severity of disease, the associations with death (hazard ratio, 1.10 [95% CI, 0.99-1.23]) fully attenuated and MACE (hazard ratio, 1.31 [95% CI, 1.14-1.50]) partially attenuated. CONCLUSIONS: New-onset AF was common (5.4%) among patients hospitalized with COVID-19. Almost half of patients with new-onset AF died during their index hospitalization. After multivariable adjustment for comorbidities and disease severity, new-onset AF was not statistically significantly associated with death, suggesting that new-onset AF in these patients may primarily be a marker of other adverse clinical factors rather than an independent driver of mortality. Causality between the MACE composites and AF needs to be further evaluated.

Development of a Portable Tool to Identify Patients With Atrial Fibrillation Using Clinical Notes From the Electronic Medical Record.

BACKGROUND: The electronic medical record contains a wealth of information buried in free text. We created a natural language processing algorithm to identify patients with atrial fibrillation (AF) using text alone. METHODS AND RESULTS: We created 3 data sets from patients with at least one AF billing code from 2010 to 2017: a training set (n=886), an internal validation set from site no. 1 (n=285), and an external validation set from site no. 2 (n=276). A team of clinicians reviewed and adjudicated patients as AF present or absent, which served as the reference standard. We trained 54 algorithms to classify each patient, varying the model, number of features, number of stop words, and the method used to create the feature set. The algorithm with the highest F-score (the harmonic mean of sensitivity and positive predictive value) in the training set was applied to the validation sets. F-scores and area under the receiver operating characteristic curves were compared between site no. 1 and site no. 2 using bootstrapping. Adjudicated AF prevalence was 75.1% at site no. 1 and 86.2% at site no. 2. Among 54 algorithms, the best performing model was logistic regression, using 1000 features, 100 stop words, and term frequency-inverse document frequency method to create the feature set, with sensitivity 92.8%, specificity 93.9%, and an area under the receiver operating characteristic curve of 0.93 in the training set. The performance at site no. 1 was sensitivity 92.5%, specificity 88.7%, with an area under the receiver operating characteristic curve of 0.91. The performance at site no. 2 was sensitivity 89.5%, specificity 71.1%, with an area under the receiver operating characteristic curve of 0.80. The F-score was lower at site no. 2 compared with site no. 1 (92.5% [SD, 1.1%] versus 94.2% [SD, 1.1%]; P<0.001). CONCLUSIONS: We developed a natural language processing algorithm to identify patients with AF using text alone, with >90% F-score at 2 separate sites. This approach allows better use of the clinical narrative and creates an opportunity for precise, high-throughput cohort identification.