PURPOSE: To synthesize the available evidence regarding the diagnostic accuracy of color duplex ultrasonography (CDU) versus the accepted gold-standard of contrast-enhanced computed tomography (CT) for the detection and classification of endoleaks after aortic endografting. METHODS: A systematic search of the literature was conducted using electronic bibliographical databases and other means to gather articles published between 1991 and 2004. Articles were scrutinized against inclusion/exclusion criteria that broadly followed the QUADAS quality assessment guidelines. The results of diagnostic CDU were expressed for each study as a 2x2 contingency table, and summary statistics (sensitivity/specificity with 95% confidence intervals [CI]) were calculated. Pooled and random effects meta-analyses were performed. RESULTS: Eight published studies and 2 unpublished studies from Charing Cross and St. George's Hospitals (711 patients, 1355 paired scans performed > or = 1 month after endografting) were eligible for inclusion. From meta-analyses, the pooled sensitivity of CDU (versus CT as the gold standard) was 69% (95% CI 52% to 87%) and the specificity of CDU was 91% (95% CI 87% to 95%). These parameters did not appear to vary over time when a smaller dataset of 117 patients with 239 paired scans was used to compare CT and CDU specifically at 3, 12, and 24 months after endografting. Endoleak classification data, which was derived from only 5 small studies, indicated that CDU appeared to have better diagnostic accuracy in detecting type I or type III endoleaks compared with type II endoleaks; however, the data were insufficient for statistical analysis. CONCLUSIONS: CDU currently does not have sufficient diagnostic accuracy for the detection of all endoleaks in routine clinical practice. The diagnostic accuracy of CDU may improve if type II endoleaks are ignored.
PURPOSE: To synthesize the available evidence regarding the diagnostic accuracy of color duplex ultrasonography (CDU) versus the accepted gold-standard of contrast-enhanced computed tomography (CT) for the detection and classification of endoleaks after aortic endografting. METHODS: A systematic search of the literature was conducted using electronic bibliographical databases and other means to gather articles published between 1991 and 2004. Articles were scrutinized against inclusion/exclusion criteria that broadly followed the QUADAS quality assessment guidelines. The results of diagnostic CDU were expressed for each study as a 2x2 contingency table, and summary statistics (sensitivity/specificity with 95% confidence intervals [CI]) were calculated. Pooled and random effects meta-analyses were performed. RESULTS: Eight published studies and 2 unpublished studies from Charing Cross and St. George's Hospitals (711 patients, 1355 paired scans performed > or = 1 month after endografting) were eligible for inclusion. From meta-analyses, the pooled sensitivity of CDU (versus CT as the gold standard) was 69% (95% CI 52% to 87%) and the specificity of CDU was 91% (95% CI 87% to 95%). These parameters did not appear to vary over time when a smaller dataset of 117 patients with 239 paired scans was used to compare CT and CDU specifically at 3, 12, and 24 months after endografting. Endoleak classification data, which was derived from only 5 small studies, indicated that CDU appeared to have better diagnostic accuracy in detecting type I or type III endoleaks compared with type II endoleaks; however, the data were insufficient for statistical analysis. CONCLUSIONS:CDU currently does not have sufficient diagnostic accuracy for the detection of all endoleaks in routine clinical practice. The diagnostic accuracy of CDU may improve if type II endoleaks are ignored.