A systematic review 1 including 16 studies with a total of 8306 subjects was abstracted in DARE.
Clinical gestalt. The proportion of patients in each pre-test category that actually had PE varied from 8 to 19% (low), 26 to 47% (moderate) and 46 to 91% (high) across studies. The LR was 0.25 (95% CI: 0.14, 0.45) for low pre- test probability, 0.92 (95% CI: 0.71, 1.2) for moderate pre-test probability and 4.7 (95% CI: 2.3, 9.7) for high pre-test probability.
Clinical prediction rules. The proportion of patients in each pre-test category that actually had PE varied from 3 to 28% (low), 16 to 46% (moderate) and 38 to 98% (high) across studies. The LR was 0.12 (95% CI: 0.05, 0.31) for low pre-test probability, 1.1 (95% CI: 0.76, 1.6) for moderate pre-test probability and 23 (95% CI: 7.6, 69) for high pre-test probability.
Comment: The quality of evidence is downgraded by inconsistency (heterogeneity in interventions and outcomes). There was no direct comparison of the accuracy of clinical prediction rules and clinical gestalt. The authors recommend that clinical prediction rules should be used because they can be used accurately by less-experienced clinicians.
Primary/Secondary Keywords