Predication of Wells and Revised Geneva Scores for Pulmonary Embolism / 中山大学学报(医学科学版)

ABSTRACT

[Objective] To reduce misdiagnosis and underdiagnosis rate of pulmonary embolism,the prediction of the revised Geneva score and Wells score for pulmonary embolism were compared and analyzed by receiver operating characteristic curves.[Methods] Sixty-five cases with suspected pulmonary embolism (PE) were collected in the Third Affiliated Hospital of SUN Yat-sen University from January 1998 to October 2008.Of which 44 cases with PE were clinically confirmed.Relevant clinical data were recorded,summarized and the analysis variables were input to SPSS11.0 for statistical analysis.ROC curves was used to evaluate the probability of PE predicted by the Wells and the revised Geneva scores.[Results] Twenty-four patients had a low clinical probability of PE (Wells score ＜ 2 points ),of which 8 (33.3%) had proven PE.The prevalence of PE was 87.2% in the 39 patients with intermediate probability (2-6 points) and 100% in the 2 patients with high probability (＞ 6 points) (P ＝ 0.000).The confirmed PE was 22.2% in the 18 patients with a low probability (Geneva score 0-3 points),82.1% (32/39) in intermediate probability (4-10 points),100% (8/8) in high probability (score ≥11 points) (P ＝ 0.000).The area under curve (AUC) of the ROC curve in the Wells and Geneva scores was 0.785 ± 0.060 and 0.900 ± 0.038,respectively (P ＝ 0.000).The optimal cutoff value was 2 points in the Wells score and 6.5 points in the Geneva score.The Wells score more than 2 points predicted PE with a sensitivity of 81.8% and specificity of 76.2%.The Geneva score more than 6 points predicted PE with a sensitivity of 72.7% and specificity of 100%.The comparison of the area under curve between the Wells and the Geneva score had a significant difference statistically (P ＜ 0.05).[Conclusion] The Wells score and the revised Geneva score are beneficial to predict pulmonary embolism.The revised Geneva score is roughly superior to the Wells score both in sensitivity and specificity.