Prospective studies of consecutive patients with objectively confirmed symptomatic deep venous thrombosis (DVT) or pulmonary embolism (as defined in the paper) and objectively confirmed recurrent events were eligible for inclusion in the review.
Patients had to receive five days of initial treatment with unfractionated heparin (intravenous, adjusted-dose or weight-based subcutaneous), low-molecular-weight heparin (LMWH: intravenous or weight-based subcutaneous), fondaparinux, idraparinux or ximelagatran. Patients also had at least three months of anticoagulant treatment that comprised vitamin K antagonist (target international normalised ratio 2.0 to 3.0), unfractionated heparin (weight-based or adjusted-dose), LMWH (subcutaneous weight-adjusted or fixed-dose), idraparinux or ximelagatran.
Studies had to include at least one primary outcome: fatal recurrent VTE or fatal major bleeding event; or secondary outcomes (probable fatal recurrent VTE, recurrent nonfatal VTE and major bleeding events). Definitions were given in the paper.
Most of the included studies administered intravenous unfractionated heparin or subcutaneous LMWH combined with a vitamin K antagonist. Fatal recurrent VTE was poorly defined and not objectively assessed in a large proportion of studies (although independent adjudication for the latter was employed in many cases).
Two reviewers independently selected studies for inclusion. Discrepancies were resolved by reference to a third reviewer.