Impact Of Clinical Trial Design And Execution-Related Factors On Incidence Of Thromboembolic Events In Cancer Patients: A Systematic Review And Metaanalysis

Objective: The objective of this study was to quantify the incidence of thromboembolic events (specifically, deep vein thromboses [DVT] and pulmonary embolism [PE]) in patients with cancer, and to examine the effects of a major clinical trial design and execution factors on those incidence rates.Research design and methods: The study included a systematic review of Medline, Current Contents, and accepted study bibliographies, as well as an analysis of studies. Studies included both longitudinal studies (prospective and retrospective) published in the English language between January 1990 and October 2005. Studies of patients with cancer that reported the incidence of thromboembolic events (DVT, PE, and total venous thromboembolic events [VTE]) were eligible for inclusion. Incidence of these events was calculated by study design, surveillance type (active or passive), length of follow-up, and other treatment risk factors. Incidence rates were estimated by random effects Poisson meta-regression modeling.Results: One hundred and eighty-three studies met all inclusion criteria. Incidence rates of all outcomes (DVT PE, and total VTE) were 3-55 times higher for active surveillance than for passive surveillance. Studies with a follow-up time ! 6 months reported thromboembolic event rates that were 3-26 times higher than study groups with a follow-up time > 6 months. Additionally, the incidence rates for all outcome events when using passive surveillance were 3-12 times higher in non-randomized clinical trials (non-RCTs) than in RCTs.Conclusions: These results provide a benchmark for the incidence of thromboembolic events in patients with cancer. Factors such as study design, length of follow-up, and method of case ascertainment (type of surveillance - active or passive) must be considered when interpreting thromboembolic incidence rates. This review is comprehensive in its inclusion of all studies with a scientific objective of examining the risk of thromboembolic events in patients with cancer from 1990 to 2005. However, other studies published prior to 1990, more recently than 2005, or with other scientific objectives in their research may also provide supportive information to these risk estimates.