Short- and medium-term complications after emergency department treatment of patients with atrial fibrillation: design and internal validation of the CoSTuM risk scale

To design and validate a clinical risk scale for 90-day adverse events after emergency department (ED) discharge of patients with atrial fibrillation (AF). Prospective, observational CoSTuM (Spanish acronym for comorbidity, symptoms, treatments, and poor heart rate control) study of patients with AF. We recorded all adverse events occurring during follow-up and used multivariate logistic regression analysis to identify variables independently associated with the occurrence of any such event. A risk model was developed with data from a random sample of 66% of the patients. Data for the remaining patients were used for internal validation of the model. We analyzed data for 645 patients (50.2% women; median [interquartile range] age, 76 (67-81 years). At 90 days, at least 1 adverse event had occurred in 165 patients (25.6%); 114 were in the development sample and 51 were in the validation sample. Independent variables associated with risk for adverse events were age 70 years or older; vascular disease; heart failure; chronic obstructive pulmonary disease; kidney disease; obesity; poorly controlled AF symptoms (European Heart Rhythm Association score, III-IV); use of oxygen therapy, digoxin, or other antiarrhythmics in the ED; and poor control of heart rate. The CoSTuM scale based on these variables demonstrated optimal predictive value in both the development sample (c statistic, 0.901; 95% CI, 0.869-0.928; P .001) and the validation sample (c statistic, 0.850, 95% CI, 0.796-0.894; P .001). The CoSTuM risk score was a good predictor of 90-day adverse events after ED discharge of patients treated for AF.