Clinical limitations of risk assessment models.

Appropriate thromboprophylaxis in hospital patients is effective in preventing clinically important venous thromboembolic events, including deep vein thrombosis (DVT) and fatal pulmonary embolism. Due to the risk of bleeding associated with pharmacological prophylaxis and the cost of administering prophylactic drugs, the clinical benefit and cost-effectiveness of thromboprophylaxis may be optimized by providing prophylaxis only to patients at risk of thrombosis, and tailoring the intensity of prophylaxis to the level of risk. Accurate assessment of patients' thromboembolic risk is therefore highly necessary. Thromboembolic risk is influenced by numerous factors. Several risk factor indices based on clinical risk factors and laboratory variables have been proposed since the 1970s, but these have not been widely adopted due to their complexity and lack of prospective validation. The method of deriving risk data on which these indices are based is questioned, and older prognostic indices excluded recently identified risk factors, particularly molecular factors such as the clotting factor V Leiden mutation, further undermining their clinical value. A number of much simpler risk assessment models (RAMs) have now been developed which stratify patients into low-, moderate- and high-risk categories. However, no RAM currently available provides comprehensive guidance for all patient groups. Use of poorly designed RAMs may fail to identify some patients at risk, leading to omission of prophylaxis and preventable thrombotic events. Certain patient groups develop DVT despite prophylaxis. Current RAMs are not validated to identify these patients. Well-designed and well-validated RAMs, incorporated into standard practice guidelines in hospitals, should contribute to improved clinical outcomes and economic benefits of prophylaxis.