Table 3

Mutivariable analysis of 47 studies reporting prevalence of venous thromboembolism

Beta*95% CIIntercept†P value
Proportion of patients in ICU (%)0.00200.0012 to 0.00280.0351<0.0001
Systematic screening (yes/no)0.14230.0590 to 0.22560.03510.0008
  • Number of patients (47 studies, p=0.57) and D-dimer level (28 studies, p=0.26) were not significantly associated with venous thromboembolism when added to the model.

  • As an example, the prevalence of venous thromboembolism in a cohort using systematic screening and including 30% of patients in the ICU is estimated at 0.0351+0.1423+0.0020×30=0.2374 (23.7%).

  • *Beta signification: prevalence increase or decrease for the augmentation of one unit of the variable tested.

  • †Intercept signification: rate of venous thromboembolism for all variables with null value.

  • ICU, intensive care unit.