We assessed the influence of the following potential risk factors for VTE recurrence: age; sex; inpatient or outpatient status; anticoagulant treatment at inclusion (prophylactic anticoagulant treatment was defined as an anticoagulant treatment with a dosage recommended to prevent VTE in acutely ill medical patients); and the SVT characteristics (unilateral or bilateral SVT; saphenofemoral/popliteal Protease Inhibitor Library
involvement, if the distance from the clot to the saphenofemoral/popliteal junction was ��?3?cm). Chronic risk factors were as follows: a personal or a family history of VTE; cancer (previous and/or active); chronic reduced mobility; known biological thrombophilia at the time of diagnosis; estrogen therapy within the last 2?months; obesity (body mass index of ��?30?kg?m?2); and varicose veins (defined according to the clinical component of the CEAP classification, C?��?2). Transient risk factors were as follows: bed confinement; recent travel; surgery within the previous 45?days; congestive heart failure (NYHA class?III or IV) or respiratory check details
insufficiency (acute respiratory failure or exacerbation of chronic obstructive pulmonary disease); infectious disease; and pregnancy or postpartum period of <?6?weeks. Both research teams provided their original datasets electronically, and statisticians from both teams (C.G., J.L.B., and E.P.) independently performed all analyses. First, preliminary analyses were conducted to confirm the appropriate transfer of data. In all cases, we were able to reproduce the results published in both reports [10,11]. Data were <a href="http://en.wikipedia.org/wiki/Unoprostone
">Unoprostone then combined into a single database. Categorical variables were expressed as frequencies and percentages. Continuous variables were expressed as medians and interquartile ranges. The predictive values for VTE or DVT/PE at 3?months were tested for each variable with a random effect model (Cox model) adjusted to the variable ��study��. For each of these variables, heterogeneity between studies was tested. We then performed multivariate analyses with Cox proportional hazard ratio (HR) models, including predefined predictors for VTE (e.g. personal history of VTE, cancer, and advanced age) and potential confounders (e.g. anticoagulant prescribed at inclusion) as dependent variables . We also included, in each multivariate model, variables achieving a P-value of ��?0.15 in univariate analyses, and interaction terms if appropriate. The results of univariate analyses were graphed as forest plots for the variables retained in the multivariate analysis models. Cumulative rates of 3-month adverse events (e.g. symptomatic VTE and death) were estimated by use of the Kaplan�CMeier method. As the date of the major bleeding event was unknown in the POST study, the rates of 3-month major bleeding episodes were presented as crude percentages and 95% confidence intervals (CIs).