Several diagnostic studies have indicated that certain risk factors and clinical signs increase or decrease the likelihood of deep vein thrombosis (DVT) in individual people. In particular, signs and symptoms taken together as the Wells Clinical Prediction Rule provide a relatively sensitive and specific test for DVT, but these need to be combined with the D-dimer test to safely exclude DVT. The combination approach of Wells Rule and D-dimer test has also been shown to be cost-effective.

• A systematic review (search date: July 2004) identified 14 studies (n > 8000) that used a clinical prediction rule to diagnose DVT (of which 11 studies incorporated the D-dimer test as part of the algorithm) [Wells et al, 2006].
• The clinical prediction rules (in most cases the Wells Rule) stratified people into low, moderate, or high risk of DVT:
• The actual prevalence of DVT was found to be 5% (95% CI 4 to 8), 17% (95% CI 13 to 23), and 53% (95% CI 44 to 61) respectively in people deemed to be at low, moderate, or high clinical risk.
• The likelihood ratios of a high-sensitivity D-dimer test was 0.20 (95% CI 0.12 to 0.31), 0.23 (95% CI 0.13 to 0.39), and 0.15 (95% CI 0.10 to 0.38) respectively in people deemed to be at low, moderate, or high clinical risk.
• The authors concluded that in people with a low clinical probability of DVT, a negative D-dimer test can effectively rule out DVT. However, in people with a high pre-test probability of DVT, the D-dimer test should not influence clinical decisions and these people require further assessment with ultrasound.
• A systematic review and meta-analysis was commissioned by the NHS Technology Assessment R&D Programme to investigate the cost-effectiveness of techniques used to diagnose DVT [Goodacre et al, 2006b]. The authors concluded that the 'optimal strategy for DVT diagnosis is to use ultrasound selectively in patients with a high clinical risk or positive D-dimer. Radiological testing for all patients does not appear to be a cost-effective use of health service resources'.
• A systematic review (search date: January 2005) identified 51 cohort studies suitable for inclusion in a meta-analysis that investigated the accuracy of physician's empirical judgements, clinical findings, or clinical score in diagnosing DVT with a reference standard test (venography, ultrasonography, or plethysmography) [Goodacre et al, 2005].
• The authors calculated the diagnostic usefulness of a range of risk factors, signs, and symptoms using likelihood ratios (LRs), and found:
• Three risk factors and one clinical sign were found to be useful predictors for ruling in DVT. These were malignancy (LR 2.71), previous DVT (LR 2.25), previous surgery (LR 1.76), and difference in calf diameter (LR 1.76).
• Two clinical signs were described as being useful for excluding DVT; these were absence of calf swelling (LR 0.67) and similarity in calf diameter (LR 0.57).
• The Wells score was found to be more accurate at diagnosing DVT than any other single measure, with a positive LR of 6.2 (95% CI 1.0 to 40.0, p < 0.001) and a negative LR of 0.18 (95% CI 0.13 to 0.26, p > 0.2).
• The authors concluded that individual clinical features were of limited value in predicting DVT, but overall the Wells score was a more useful measurement.
• A systematic review (search date: December 2003) identified 12 studies (n > 5000) suitable for inclusion that studied the accuracy of combining clinical prediction rules (Wells test) with a D-dimer assay in the diagnosis of DVT [Fancher et al, 2004].
• It was found that when a highly sensitive D-dimer test was employed, the 3 month incidence rate of thromboembolism in people predicted to be of low or moderate clinical probability of DVT was 0.4% (95% CI 0.04 to 1.10).
• The authors concluded 'A normal result from a highly sensitive D-dimer test effectively rules out deep vein thrombosis among patients classified as having either low or moderate clinical probability of deep vein thrombosis'.
• A systematic review (search date: January 2003) identified 23 eligible studies that investigated some aspect of the diagnosis of DVT; 17 studies investigated clinical prediction rules and 15 of these specifically evaluated the Wells Clinical Prediction Rule [Tamariz et al, 2004].
• The authors investigated the accuracy of the Wells Rule and found that:
• People who had a high pre-test probability according to the rule had a prevalence of DVT of 17–85%. Those with a moderate pre-test probability had a prevalence of DVT of 0–38%. Those with a low pre-test probability had a prevalence of DVT of 0–13%.
• Overall, the negative predictive values were generally high in people in the low pre-test probability range (87–96%), but rarely exceeded 75% with people in the high probability range, indicating the Well's rules were not useful for identifying people who did not have DVT.
• The authors concluded that the Wells Clinical Prediction Rule is useful in identifying people at low risk of being diagnosed with venous thromboembolism, and the addition of a latex D-dimer assay improved the overall performance of the prediction rule.