Meta-Analysis to Assess the Quality of Warfarin Control in Atrial Fibrillation Patients in the United States

BACKGROUND: Atrial fibrillation (AF) affects a significant proportion of the American population and increases ischemic stroke risk by 4- to 5-fold. Oral vitamin K antagonists, such as warfarin, can significantly reduce this stroke risk but can be difficult to dose and monitor. Previous research on the effects of setting (e.g., randomized controlled trials, anticoagulation management by specialty clinics, usual care by community physicians) on the proportion of time spent within therapeutic range for the international normalized ratio (INR) has not specifically examined anticoagulation in AF patients. OBJECTIVES: Use traditional meta-analytic and meta-regressive techniques to evaluate the effect of specialty clinic versus usual care by community physicians on anticoagulation control, measured as the proportion of time spent in therapeutic INR range, for AF patients that received warfarin anticoagulation in the United States. METHODS: Studies included in a previously published meta-analysis (van Walraven et al., 2006), which systematically searched reports between 1987 and 2005, were also screened for inclusion in our analysis. A subsequent systematic literature search of MEDLINE, EMBASE, and the Cochrane Central Register of Clinical Trials from January 2005 through February 2008 was conducted. Studies were included if they (a) contained at least 1 warfarin-treated group including more than 25 patients for whom INR control was monitored for at least 3 weeks; (b) included patients treated for AF in the United States; (c) used a patient-time approach (patient-year) to report outcomes; and (d) reported data on the proportion of time spent in traditional therapeutic INR ranges (i.e., a lower limit INR between 1.8 and 2.0 and an upper limit INR between 3.0 and 3.5. Studies with INR goals outside this range were excluded). The proportion of time spent within the therapeutic INR range for each study group was expressed as an incidence density using a person-time approach (in years). All studies were pooled using a random effects model and were weighted by the inverse of the variance of proportion of time spent in the therapeutic range. In order to determine how study setting influenced the proportion of time spent within a therapeutic INR range, both subgroup and meta-regression analyses were conducted. RESULTS: This analysis included 8 studies and a total of 14 unique warfarin- treated groups; 3 of the 8 studies and 4 of the warfarin groups were not included in the previous meta-analysis (van Walraven et al., 2006). Overall, patients spent a mean 55% (95% CI = 51%-58%) of their time in the therapeutic INR range. Meta-regression suggested that AF patients treated in a community usual care setting compared with an anticoagulation clinic spent 11% (95% CI = 2%-20%, n = 6 studies with 9 study groups) less time in range. CONCLUSIONS: In the United States, AF patients spend only about one-half the time within therapeutic INR. Anticoagulation clinic services are associated with somewhat better INR control compared with standard community care.