Background: Several randomized clinical trials have shown that among patients with atrial fibrillation, warfarin sodium use protects against stroke. Recently, experts have voiced concern about possible underuse of warfarin by practicing physicians. Few studies, however, have quantitated the amount of warfarin underuse.
Methods: We randomly sampled 65 Medicare beneficiaries discharged alive from each of 5 small Pennsylvania hospitals between July 1, 1993, and June 30, 1994, with a discharge diagnosis code for atrial fibrillation. Trained abstractors verified that atrial fibrillation was present at some time during the hospitalization, determined the presence of contraindications to anticoagulation, and identified warfarin or aspirin use at discharge for each patient. An internist used implicit criteria to identify the reason for warfarin nonuse in patients who had none of the explicit contraindications to warfarin and did not receive it.
Results: Of 322 charts reviewed, 48 patients were not in atrial fibrillation during the hospitalization, 79 had contraindications to warfarin use, 21 either died or were transferred to another hospital, and 2 were admitted with a complication of warfarin. Of the 172 remaining patients, 76 (44%) underwent anticoagulation. On implicit review of the 96 patients who did not undergo anticoagulation, the internist judged that warfarin would not have been appropriate in 54. After excluding those patients, 76 (64%) of the remaining 118 patients underwent anticoagulation. Patients not receiving warfarin were slightly older (81.6 vs 78.3 years old), but this was not statistically significant after stratifying by hospital. Rates of warfarin use at the 5 hospitals varied widely (32%, 57%, 79%, 82%, 94%; P < .001, chi2 with 4 df). Patients with newly diagnosed atrial fibrillation were not more likely to undergo anticoagulation, nor were patients treated by internal medicine or cardiology specialists.
Conclusions: There may be significant warfarin underuse in some hospitals. Overall, approximately one third of patients with atrial fibrillation for whom it appeared appropriate were not anticoagulated with warfarin. Although the fact that data were not available to or were missed by our review surely justifies some of the underuse, one should recall that even a small amount of underuse may affect a large number of people with this common condition.