Accuracy of Duplex Ultrasonography for Renal Artery Stenosis: A Meta Analysis of Diagnostic Test Studies

G.Williams¹,² T.Karplus³, E.Hodson¹, W.Yung³, P.Macaskill², S.Chan², J Craig¹,²
1. The Centre for Kidney Research, The Children’s Hospital, Westmead.
2. Screening and Test Evaluation Program (STEP), University of Sydney
3. Dept of Vascular Medicine, Concord Hospital.

Although duplex ultrasonography (DUS) is a widely utilised triage test for renal artery stenosis (RAS), the overall accuracy of the test and the individual parameters is uncertain. 1251 studies were identified through a comprehensive search of Medline (1966-2001), Embase (1980-2001), reference lists and contact with investigators.

Studies were eligible if they compared DUS parameter/s with the reference standard, renal angiography for diagnosing RAS. 75 eligible studies were found. 14 papers assessed peak systolic velocity (PSV), 9 examined acceleration time (AT), 8 acceleration index (AI), 7 looked at renal aortic arch (RAR), 4 assessed PSV as well as RAR and 24 studies analysed PSV in combination with other parameters. The summary estimates of DOR for the individual parameters were: PSV 59.6 (95% CI 26.6-133.5), AT 65.3 (22.1-193.2), AI 38.2 (10.0-145.4) and RAR 21.8 (8.6-55.7). The more robust analysis of paired data showed that PSV was more accurate than RAR (relative DOR 1.20 95% CI 0.7-2.1, PSV/RAR). Combining parameters did not improve test performance beyond that of PSV alone (relative DOR 1.23, 95% CI 0.4-3.9, PSV/PSV+other parameters).

Overall study quality was sub-optimal, blinding to clinical information was reported in only 8%, prospective design 32% and angiography method reported in 44%. Multiple regression analysis of 13 quality items identified PSV threshold as a significant determinant of DOR in those studies (P=0.026). Severity of RAS as determined by angiography was a significant factor influencing the DOR in studies assessing AT and AI (P=0.068, 0.023 respectively).

In conclusion, PSV at >200m/s is probably the most accurate test for RAS. In 1000 people with suspected RAS, assuming 100 have disease, PSV would correctly diagnose 89, miss 11, lead to 108 unnecessary angiograms and avoid 792 unnecessary angiograms.