Under-reporting of venous and arterial thrombotic events in randomized clinical trials: a meta-analysis.

For the detection of unwanted outcomes of new interventions, physicians rely on adverse event reporting. We attempt to quantify the reported incidence of venous thromboembolism (VTE) and arterial thrombosis (AT) in randomized clinical trials (RCTs), and evaluate the extent of under-reporting. We selected all therapeutic RCTs published in the four highest-impact general medicine journals between January 2011 and July 2011. Patients were categorized according to VTE risk. The occurrences of VTE and AT, either as predefined outcome or adverse event, were assessed. We identified 131 RCTs. VTE and AT were not reported in 89 and 70 % of these studies, respectively. The raw-unweighted reported incidence in the 3 studies with predefined outcomes for VTE was 8.4 (7.8-9.1) per 1,000 person-years. In the 128 studies without predefined outcomes for VTE, (consisting of 322,029 individuals, including patients with cancer, inflammatory disease, cardiovascular disease, surgery, adding up to a follow-up >500,000 person-years), an incidence of 0.4 (0.4-0.5) per 1,000 person-years was found. The reported incidence of AT in 18 studies in which AT was part of predefined outcomes was 25.6 (24.9-26.3) per 1,000 person-years. In 92 studies without predefined outcomes for AT (231,638 individuals, follow-up >200,000 person-years,), the incidence was 2.5 (2.3-2.7) per 1,000 person-years. The incidence of VTE and AT in RCTs is highly under-reported. Uniform registration of adverse events, even when unlikely to be related to the intervention, is necessary to be able to inform physicians about the potential toxicities of new therapeutic strategies.