Eighteen studies providing 25 treatment arms were included (n=3,186 patients). There were 16 RCTs, one study that was a combination of an RCT and a prospective cohort, and one prospective cohort study.
For all three categories of treatment duration, the monthly incidence rate of recurrent venous thromboembolism declined over time after discontinuation of vitamin K antagonist treatment. The monthly incidence rates of recurrent venous thromboembolism appeared to stabilise 9 months after the index event, regardless of the duration of treatment: 0.50 to 0.71% at 9 to 12 months, and 0.42 to 0.59% at 12 to 18 months.
The monthly incidence rates with short-term treatment were 3.33% (95% CI: 2.34, 4.59) at 1 to 3 months, decreasing to 1.23% (95% CI: 0.79, 1.83) at 3 to 6 months.
The monthly incidence rates with medium-term treatment were 1.19% (95% CI: 0.91, 1.53) at 3 to 6 months, decreasing to 0.51% (95% CI: 0.31, 0.78) at 6 to 9 months.
The monthly incidence rates with long-term treatment were 0.99% (95% CI: 0.62, 1.49) at 6 to 9 months, decreasing to 0.71% (95% CI: 0.40, 1.18) at 9 to 12 months.
Heterogeneity was statistically significant for incidence rates at 3 to 6 months with short-term treatment (P=0.005). The authors stated that heterogeneity was mainly due to one study in which 28% of patients had cancer and there was a relatively high number of recurrent thromboembolic events (statistical heterogeneity and results without this study were not reported). No significant statistical heterogeneity was found for the other analyses (P>0.20).
For medium-term treatment in patients without cancer (data not available for short- or long-term treatment), the monthly incidence rates of venous thrombombolism were 1.59% at 3 to 6 months, reducing to 0.58% at 6 to 12 months and to 0.34% at 12 to 18 months. These results were based on 5 studies with 672 patients.