Our results for major thromboembolism and bleeding
events seem to be robust with negligible heterogeneity, and
similar findings have been reported in other reviews (6, 43,
45). For mortality, however, there was evidence of heterogeneity
among studies that was probably attributable to the
VA study. That study, the largest included in this review,
enrolled 2922 persons receiving warfarin therapy for either
mechanical heart valves or atrial fibrillation, randomly assigned
them to receive care with PST or at a high-quality
anticoagulation clinic, and followed them for an average of
3 years. The patients were slightly older than the average
participants in this review (67 vs. 65 years) and were more
likely to be men (98% vs. 75%). There was no significantdifference in the primary outcome, time to first major
event (death, stroke, or major bleeding), among patients
randomly assigned to PST compared with patients randomly
assigned to anticoagulation clinic. Rates of loss to
follow-up and warfarin discontinuation did not differ between
groups, and although there was substantial crossover,
this should not have affected the intention-to-treat
analysis.