Safety and efficiency of D-dimer testing in combination with clinical decision rules to exclude pulmonary embolism in patients with cancer: individual patient data meta-analysis

BACKGROUND: Failure rates of clinical decision rules (CDRs) combined with D-dimer to exclude pulmonary embolism (PE) are higher in patients with cancer compared with noncancer patients, raising concerns about their use in this patient group. OBJECTIVES: To compare the failure rates of the Wells score, revised Geneva score, and YEARS algorithm in patients with cancer who underwent standard imaging with those in whom imaging was withheld, and report the diagnostic yield of these algorithms. METHODS: We used data from an individual-patient level meta-analysis of prospective diagnostic management studies of patients with suspected PE. The primary outcome was the 3-month incidence of venous thromboembolism, excluding PE (failure), using fixed and age-adjusted D-dimer results across different patient management categories for all investigated CDRs. The secondary outcome included the proportion of patients for whom PE could be ruled out without further imaging (diagnostic yield). RESULTS: A total of 2258 (7.6%) patients with cancer from 17 studies were included in the analysis. The 3-month incidence of venous thromboembolism in patients after excluding PE ranged from 0.52% (95% CI, 0.06%-4.35%) to 2.83% (95% CI, 0.96%-8.34%) and was comparable across all management categories. The highest diagnostic yield of 26% was found for the revised Geneva score, with an age-adjusted D-dimer cutoff, and the YEARS algorithm. CONCLUSION: The failure rates of patients with cancer for whom PE was excluded using CDRs and withholding imaging were similar to those observed after standard imaging. Current diagnostic algorithms for suspected PE are applicable to patients with cancer.