Summary receiver operating characteristic curves as a technique for meta-analysis of the diagnostic performance of duplex ultrasonography in peripheral arterial disease
被引:80
作者:
deVries, SO
论文数: 0引用数: 0
h-index: 0
机构:UNIV GRONINGEN, FAC MED, DEPT HLTH SCI MED DECIS MAKING, 9713 AV GRONINGEN, NETHERLANDS
deVries, SO
Hunink, MGM
论文数: 0引用数: 0
h-index: 0
机构:UNIV GRONINGEN, FAC MED, DEPT HLTH SCI MED DECIS MAKING, 9713 AV GRONINGEN, NETHERLANDS
Hunink, MGM
Polak, JF
论文数: 0引用数: 0
h-index: 0
机构:UNIV GRONINGEN, FAC MED, DEPT HLTH SCI MED DECIS MAKING, 9713 AV GRONINGEN, NETHERLANDS
Polak, JF
机构:
[1] UNIV GRONINGEN, FAC MED, DEPT HLTH SCI MED DECIS MAKING, 9713 AV GRONINGEN, NETHERLANDS
[2] HARVARD UNIV, SCH PUBL HLTH, BOSTON, MA 02115 USA
[3] HARVARD UNIV, BRIGHAM & WOMENS HOSP, SCH MED, DEPT RADIOL, BOSTON, MA 02115 USA
Rationale and Objectives. We summarized and compared the diagnostic performance of duplex and color-guided duplex ultrasonography in the evaluation of peripheral arterial disease. We present our research as an example of the use of summary receiver operating characteristic (ROC) curves in a meta-analysis of diagnostic test data. Methods. A search of the English-language medical literature published between 1984 and 1994 retrieved 48 reports, 14 of which met the inclusion criteria. The analysis was limited to the aortoiliac and femoropopliteal segments because only two studies reported results of infrapopliteal arteries. Diagnostic performance of duplex and color-guided duplex, defined as the ability to detect a stenosis of 50-99% or an occlusion, was compared using summary ROC curve methodology. This method takes into account heterogeneity across studies attributable to differences in the threshold values used. Results. The summary ROC curves demonstrated a high level of diagnostic performance for both types of duplex imaging, with color-guided duplex scanning being superior (P = .022). For example, at a false-positive rate of .05 (specificity = .95), the analysis predicted a true-positive rate (sensitivity) of .83 for duplex alone and .93 for color-guided duplex. Differences in the case mix of the study population and study design did not affect the results. Furthermore, sensitivity analysis did not reveal a strong effect of any single study on the results. Conclusion. For aortoiliac and femoropopliteal arteries, the addition of color flow imaging to duplex scanning improves diagnostic performance in evaluating peripheral arterial disease.