The diagnosis of venous thromboembolism is difficult in the postoperative setting because signs such as hypoxemia, leg pain, and swelling are so common. CTPA can also detect subsegmental PE (SSPE), of which the clinical significance has been widely debated. Clinical decision rules (CDR), such as the Wells and PISA 2, have been developed to identify symptomatic patients at low risk for PE who could forgo imaging. We performed this study in order to (1) compare the performance of the Wells and PISA 2 CDR in orthopedic patients; (2) compare CDR scores in patients with subsegmental PE (SSPE) versus larger clots; and (3) identify variables that improve performance of the Wells in orthopedic patients. This retrospective cohort study included all orthopedic surgery patients that underwent computerized tomographic pulmonary angiography at a single institution from 1/1/13 to 12/31/14 and had data to calculate both Wells and PISA 2 scores. CDR sensitivity, specificity and c-statistics were calculated. Multivariable logistic regression was used to identify variables that improved CDR performance. 402 patients were included in the study. The Wells rule (cutoff > 4) had sensitivity 74% and specificity 45%. PISA 2 (cutoff 0.6) had sensitivity 90% and specificity 11%. The Wells performed better than PISA 2: c-statistic 0.60 vs. 0.50; p = 0.007. The mean Wells score was 5.20 ± 1.68 for patients with SSPE and 5.41 ± 1.86 for patients with larger clots. Adding the variables prior smoking and varicose veins improved the performance of the Wells rule (c-statistic 0.66 vs. 0.60, p = 0.008). The Wells rule (cutoff > 4) performs better than PISA 2 in orthopedic patients. Neither can distinguish patients with SSPE from those with larger clots. Although adding past smoking and varicose veins to the Wells improves its performance, this requires validation in other populations.
Keywords: Arthroplasty; Clinical decision support; Orthopedic; Pulmonary embolism; Wells rule.