Nine studies (3,225 participants) were included: five cohorts from RCTs (1,306 participants) and four cohort studies (1,919 participants). Four studies scored 6 for quality, four scored 4 and one scored 3.
For increasing durations of treatment (by month) there was no significant change in rates of deep vein thrombosis, pulmonary embolism or venous thromboembolism and no statistically significant differences between groups for three to six months, six to 12 months and more than 12 months of treatment. For venous thromboembolism there was significant heterogeneity between studies in the six- to 12-month analysis (Ι²=92.3%).
Once treatment had ceased there was a decrease in mortality in the more than 12 months treatment group compared to the three to six month group but no difference with six to 12 month group.
Very few bleeds were recorded after cessation of treatment. There was no significant difference in percentage bleeds per patient year according to each additional month of anticoagulation regardless of whether on or off treatment. The combined venous thromboembolism and bleeding outcome was similar to the venous thromboembolism outcome as bleeds were few.
Re-analyses showed no statistically significant effects of study quality, study design, cohort age or percentage of men.