Study objective: We determine the prevalence of nontherapeutic and coagulopathic international normalized ratios (INRs) among patients receiving warfarin and presenting to an emergency department (ED). As a secondary goal, we aim to determine whether a simple decision aid composed of physical examination and historical features could be predictive of INR greater than 5.
Methods: This was a prospective, observational study at 2 associated urban academic centers from February 2003 through May 2004, using a convenience sample of patients identified by direct questioning and contemporaneous medical record review in the ED as receiving long-term warfarin therapy. Inclusion criteria were warfarin therapy and self-reported compliance. Patients were enrolled by trained researchers. The primary outcome measure was the percentage of patients within appropriate therapeutic range for their condition according to accepted national guidelines. Descriptive statistics were used, and multivariate regression analysis was performed to identify associations.
Results: One thousand nineteen patients were enrolled. INR values were obtained in 77% (782/1019) of patients. Of these patients, 72% (95% confidence interval 67% to 76%) were outside the desired range. Values were less than 2 in 43% of patients and greater than 3 in 29% of patients. INR greater than 5 was present in 11% (86/782) of patients, and 40% (34/86) of these patients exhibited gross bleeding. Emergency therapy was administered in 12% (96/782) of patients: fresh frozen plasma in 7% of patients, heparin in 5% of patients. Intracranial hemorrhage was found in 12 patients, 5 with INR greater than 3. Ischemic stroke or venous thromboembolism occurred in 51 patients known to be receiving warfarin specifically for prevention of the event that occurred. Of these patients, 49% (25/51) had INR less than 2. Regression analysis indicated no sensitive or specific constellation of features, though 2 factors were associated with INR greater than 5: gross hemorrhage (P=.006) and increasing duration of therapy (P=.047).
Conclusion: The prevalence of undesirable INR in the ED is higher than in warfarin populations previously studied, and a significant number of nontherapeutic levels were associated with thromboembolism, stroke, or hemorrhage. Given the prevalence and established danger of subtherapeutic and supratherapeutic levels, a low threshold should be maintained for testing and addressing INR levels in patients receiving warfarin in the ED.