Comparative Accuracy of Renal Duplex Sonographic Parameters in the Diagnosis of Renal Artery Stenosis: Paired and Unpaired Analysis : American Journal of Roentgenology : Vol. 188, No. 3 (AJR)

Home >
American Journal of Roentgenology >
Volume 188, Issue 3 >
Comparative Accuracy of Renal Duplex Sonographic Parameters in the Diagnosis of Renal Artery Stenosis: Paired and Unpaired Analysis

March 2007, Volume 188, Number 3

Genitourinary Imaging

Original Research

Comparative Accuracy of Renal Duplex Sonographic Parameters in the Diagnosis of Renal Artery Stenosis: Paired and Unpaired Analysis

Gabrielle J. Williams¹ ², Petra Macaskill², Siew F. Chan², Thomas E. Karplus³, Winkle Yung³, Elisabeth M. Hodson¹ and Jonathan C. Craig¹ ²

+ Affiliations:

¹Centre for Kidney Research and Department of Nephrology, Children's Hospital at Westmead, Hawkesbury Rd. (Locked Bag 4001) Westmead, New South Wales 2145, Australia.

²Screening and Test Evaluation Program, School of Public Health, University of Sydney, Camperdown, New South Wales 2006, Australia.

³Department of Vascular Medicine, Concord Hospital, Concord, Sydney, Australia.

Citation: American Journal of Roentgenology. 2007;188: 798-811. 10.2214/AJR.06.0355

ABSTRACT

OBJECTIVE. The purpose of this study was to evaluate the test performance of duplex sonographic parameters in screening for hemodynamically significant renal artery stenosis, which occurs in approximately 5% of persons with hypertension.

MATERIALS AND METHODS. A comprehensive literature search was conducted to find studies on the diagnosis of renal artery stenosis in which duplex sonography and intraarterial angiography were compared and in which sensitivity and specificity were calculated. MEDLINE (1966-2005), EMBASE (1988-2005), and reference lists were searched and the authors contacted. Data were subjected to meta-analysis according to the hierarchical summary receiver operating characteristic curve model. Heterogeneity in test performance relating to population and design features was investigated.

RESULTS. From 1,357 titles, 88 studies involving 9,974 arteries in 8,147 patients were included. The following four parameters were evaluated: peak systolic velocity (21 studies), acceleration time (13 studies), acceleration index (13 studies), and renal-aortic ratio (13 studies). The corresponding diagnostic odds ratios (ORs) were 60.9 (95% CI, 28.3-131.2), 28.9 (95% CI, 7.1-117.2), 16.0 (95% CI, 5.1-50.6), and 29.3 (95% CI, 12.7-67.7). Results based on studies in which parameters were directly compared showed that peak systolic velocity had greater accuracy than renal-aortic ratio (relative diagnostic OR, 1.8; p = 0.03; nine studies) and acceleration index (relative diagnostic OR, 5.3; p < 0.001; five studies). Acceleration time versus acceleration index showed no evidence of a difference in accuracy (relative diagnostic OR, 1.1; p = 0.65; nine studies). Analysis of peak systolic velocity used in combination with other parameters compared with peak systolic velocity alone (seven studies) showed evidence of a shift in test positivity (p < 0.001) but only weak evidence of improvement in accuracy (relative diagnostic OR, 1.6; p =0.09).

CONCLUSION. Sonography is a moderately accurate screening test for renal artery stenosis. The single measurement, peak systolic velocity, has the highest performance characteristics, an expected sensitivity of 85% and specificity of 92%. Additional measurements do not increase accuracy.

Keywords: color Doppler sonography, digital subtraction angiography, renal disease, screening

Introduction

Renal artery stenosis (RAS) is the most common underlying medical condition causing hypertension and is present in 1-5% of all cases of hypertension [1]. RAS is also the indication for renal replacement therapy in 5-15% of cases [2, 3]. Most cases of RAS are due to atherosclerosis [4], the incidence of which increases with age [5]. As the general population ages, it is likely that RAS will be detected more frequently. Identification of patients with RAS-induced hypertension has important clinical implications because correction of RAS with angioplasty and stenting can improve blood pressure control in as many as 64% of patients with hypertension resistant to medical treatment [6]. In addition, improvement or stabilization of renal function occurs in 79% of patients in whom RAS is associated with deteriorating renal function [7]. Intraarterial angiography currently is the reference standard for the diagnosis of RAS. However, the morbidity associated with this technique from bleeding, anaphylaxis, and contrast material-induced nephropathy has largely limited its use for verification of RAS in patients in whom the condition is strongly suspected either on clinical grounds or because of a positive screening test result.

There is no universally accepted screening or triage test for RAS. Clinical assessment to identify patients in whom RAS is a likely cause of hypertension or of deteriorating renal function has not proved sufficiently accurate for routine use; however, clinical prediction algorithms for the detection of RAS have achieved a sensitivity of 65% and a specificity of 87% [8]. Physiologic studies for assessing the renin-angiotensin system have been largely abandoned for screening because of their low predictive accuracy, especially among elderly patients [9, 10]. Radionuclide studies for assessing differential blood flow have several limitations. The results are difficult to interpret in patients with impaired renal function, and the technique cannot be used to identify RAS if the patient has bilateral disease or if only one kidney is present. The accuracy of radionuclide studies in the diagnosis of RAS has ranged from very high [11] to very low [12].

Use of direct imaging of renal vessels for triage in the detection of RAS so that only patients with positive results undergo verification with the reference standard, intraarterial angiography, has attracted much interest in recent years. The techniques used are duplex sonography, CT with administration of contrast material, and MRI [13]. Duplex sonography has several advantages over the other two techniques: it is widely available, noninvasive, and inexpensive. Duplex sonography, however, is not a simple test. Multiple measurements of several indexes of renal blood flow are possible. The absolute and relative accuracy rates of these sonographic parameters, which have been used to diagnose hemodynamically significant RAS, are not well defined. The primary aims of this study were to assess the comparative performance of duplex sonographic parameters in the diagnosis of hemodynamically significant RAS, to determine which test is most discriminating in the detection of disease, to examine threshold effects, and to assess the gain in using tests in combination. Secondary aims were to use population and study design features to explain heterogeneous results among studies.

Materials and Methods

Study Selection

We included studies in which duplex sonography was compared with renal angiography in examinations of patients with hypertension. Articles were excluded if echo-enhancing agents were used or data were insufficient to complete a two-by-two contingency table from which sensitivity and specificity could be calculated. Studies involving echo-enhancing agents were excluded because the aim of the study was to assess noninvasive triage tests, and use of an echo-enhancing agent constitutes an invasive procedure. Pairs of reviewers independently extracted data from each article. Disagreements were resolved by a third reviewer extracting the data and through subsequent discussion. Reviewers were not blinded to details of authorship. Duplicate publication was identified after data extraction, and the article with the most complete data was included in the review.

Literature Search

Studies were identified through MEDLINE (1966-March 2005), EMBASE (1988-March 2005), review of all references in review articles or eligible articles, and contact with investigators. The medical subject headings renal artery obstruction, ultrasonography, Doppler, color, angiography, hypertension renovascular, hypertension renal, sensitivity and specificity, receiver operating characteristic, and hypertension were combined with the text words renal artery stenosis, renal arteriography, sensitivity, specificity and predictive value, renal angiography, and renal arteriography. Details of search strategies appear in Appendix 1. All titles were reviewed online by two of the authors. Abstracts of these articles were reviewed, and those appearing relevant were obtained for full assessment. Articles not in English were translated into the following languages: French, German, Italian, Spanish, Czechoslovakian, Slovak, Portuguese, and Russian.

**APPENDIX 1:** **Electronic Search Strategies**
MEDLINE Search Strategy			EMBASE Search Strategy
No.	Search History	No. of Results	No.	Search History	No. of Results
1	renal artery obstruction/	6,895	1	kidney artery stenosis/	3,310
2	renal artery stenosis.tw	2,738	2	renal artery stenosis.tw	2,093
3	renal artery obstruction.tw	40	3	1 or 2	3,789
4	hypertension renovascular.tw	99	4	kidney artery stenosis/di [Diagnosis]	1,084
5	hypertension renovascular.mp [mp=ti, ab, rw, sh]	5,123	5	Kidney angiography/	747
6	Renovascular hypertension.mp [p=ti, ab, rw, sh]	3,620	6	Kidney arteriography/	1,209
7	Or/1-6	11,798	7	Doppler echography/	6,249
8	Renal angiography.tw	623	8	Duplex Doppler.tw	837
9	Renal arteriography.tw	541	9	Colo?r Doppler.tw	4,951
10	Renal artery/ and angiography/	1,652	10	Doppler pulse.tw	42
11	(dop$adj25ultra$).mp [mp=ti, ab, rw, sh]	13,530	11	Doppler ultrasonography.tw	2,187
12	Ultrasonography Doppler/	3,058	12	Or/4-11	14,769
13	Ultrasonography Doppler color/	4,160	13	3 and 12	1,354
14	Ultrasonography Doppler duplx	1,581	14	Diagnostic accuracy/	51,669
15	Ultrasonography Doppler pulse	522	15	Receiver operating characteristic/	1,395
16	Or/8-15	21,748	16	“sensitivity and specificity”.tw	16,778
17	7 and 16	952	17	(sensitive$ or specific$ or predictive value).tw	440,614
18	17 not animal.mp [mp+ti, ab, rw, sh]	917	18	Or/14-17	475,202
			19	13 and 18	348
			20	Limit 19 to human	334

Methodologic Assessment and Data Extraction

Design and population characteristics were assessed for each study retrieved. Items assessed included participant details (clinical spectrum, prevalence of stenosis), duplex sonographic method (method described or not, proportion of unsuccessful tests, inclusion of failed tests, inclusion of occluded arteries), angiographic method (method described or not, operator specified, views specified, stenosis measurement, results reviewed by two operators, percentage stenosis severity, threshold of positive test result), and study design items (year of publication, prospective or retrospective data collection, consecutive patient enrollment, interpreters of sonographic results blinded to angiographic results, interpreters of angiographic results blinded to sonographic results, interpreters of both types of images blinded to clinical information, number of patients undergoing both sonography and angiography, method of inclusion of occluded arteries, inclusion of accessory arteries in data). Most population and design characteristics were coded yes, no, or not stated. Some characteristics, such as threshold, prevalence of stenosis, percentage of unsuccessful tests, and percentage of patients undergoing both tests, were treated as continuous variables in the metaregression. RAS severity was classified according to angiographic appearance, most commonly 50% or 60%. Sonographic parameters evaluated were peak systolic velocity, renal-aortic ratio (RAR), acceleration time (AT), acceleration index (AI), combinations that included peak systolic velocity, combinations not including peak systolic velocity, and other single parameters. Peak systolic velocity and RAR are usually extrarenal measurements, and AI and AT are intrarenal measurements.

When results were reported for more than one threshold for a given sonographic parameter, the threshold most commonly used by other authors was chosen to make study designs as similar as possible. Threshold values for studies in which peak systolic velocity was evaluated ranged from 100 to 200 cm/s and in which RAR was measured, 1.8-3.5 (ratio), one article not stating the threshold. AT threshold values were > 0.1 to > 0.7 m/s, and AI threshold values were 3-4.5 m/s, one article did not states the threshold used.

In articles that reported results for more than one level of severity of angiographically defined RAS, the data set with the most frequently used threshold in other studies was chosen. This precaution was used to ensure the most appropriate comparison across articles. We attempted to extract data for both patients and arteries, but because only 28 of the 88 eligible articles reported data on patients, we reported only data on individual arteries.

For each study, sensitivity and specificity were calculated at the chosen threshold, and the diagnostic odds ratio (OR) was computed. This value is a single summary measure of test accuracy for each study, taking into account both sensitivity and specificity at a threshold. A high diagnostic OR indicates high test accuracy. A test that performs no better than chance in discriminating presence and absence of disease has a diagnostic OR of 1. In instances in which five or more studies assessed the same Doppler parameter, a summary receiver operator characteristic (SROC) curve was estimated. A preliminary exploratory analysis was conducted with the SROC regression method of Moses et al. [14], Littenberg and Moses [15], and Irwig et al [16]. Regression diagnostics were examined to identify outliers and potentially influential studies.

The hierarchical SROC curve model described by Rutter and Gatsonis [17, 18] was used at the next stage of the analysis. This method has the advantage of taking into account uncertainty in estimates of both sensitivity and specificity within each study and includes a random effect for both test accuracy and positivity criterion (threshold), thereby taking into account unexplained heterogeneity between studies. The model also allows test accuracy to vary with threshold through the inclusion of a scale (shape) parameter that provides for asymmetry in the SROC curve. This shape parameter is assumed to be constant across studies (fixed effect). The model can be referred to as a mixed model because it includes both random and fixed effects. Empiric Bayes estimates of model parameters were obtained with PROC NLMIXED (SAS) [19]. Appendix 2 shows a detailed specification of the model.

View larger version (17K)

Fig. 1 —Flow chart of identified and included studies.

An SROC curve was constructed after selection of a range of values of 1 - specificity (consistent with the observed data) and use of the estimated model parameters to compute the predicted values for sensitivity. A function of the estimated model parameters was used to obtain the expected operating point on the SROC curve [17] that gave summary estimates of sensitivity (1 - specificity) and the corresponding positive and negative likelihood ratios. Covariates were added to the model for assessment of whether test accuracy, the positivity criterion, or the shape of the SROC curve varied with population and design characteristics. The level chosen for statistical significance was 5%. Only significance test results that were robust to the removal of an influential observation were reported. Both the SROC curve regression and hierarchical SROC curve methods gave similar results; therefore, only the results of the hierarchical model are reported.

Test accuracy can vary across studies owing to differences in patient characteristics, disease spectrum, and study design characteristics. Because of missing or poorly reported information, it is difficult to adequately adjust for this confounding effect. We thus focused on paired studies conducted with direct comparisons of two or more parameters against the same reference standard, angiography, within the same study.

Results

Literature Search

Among 1,357 publications reviewed, 88 eligible articles with data on 9,974 arteries were identified and included in this study (Fig. 1). Nine duplicate publications were excluded. The following four Doppler sonographic parameters were assessed: peak systolic velocity (21 studies, 2,785 arteries), acceleration time (AT) (13 studies, 1,927 arteries), acceleration index (AI) (13 studies 1,299 arteries), and renal-aortic ratio (RAR) (13 studies, 1,347 arteries). Fifty-one (58%) of the articles contributed no data to our analyses because they reported on parameters or combinations of parameters that fewer than five authors examined, such as delta resistive index (n =4), peak systolic velocity and delta resistive index (n = 2), RAR or spectral broadening (n = 1), and peak systolic velocity, AT, and AI (n = 1). Seventeen studies contributed one data set to our analyses, three studies contributed two data sets, nine studies contributed three data sets, and five studies provided four or more data sets for the analyses. Dual data extraction reached complete agreement between data extractors in all four cells summarizing test performance (true-positive, true-negative, false-positive, false-negative) in 72% of the included studies. Extracted data in an article that did not agree completely were reextracted by a third author and in the case of one article, a fourth author.

Study Characteristics

The study characteristics are shown in Table 1. Studies were published between 1984 and 2004. The number of patients in each study ranged from 17 to 550. Of the articles that reported sex ratios, all but one [20] included both male and female patients. Only three articles [21-23] reported results for children younger than 12 years, the others reporting a wide range of ages of adults. The clinical spectrum included hypertension alone, hypertension with chronic renal failure, peripheral vascular disease, treatment by kidney transplantation, and potential kidney donation. It was not possible to separate patients with hypertension and chronic kidney failure and patients with hypertension and peripheral vascular disease from the other groups because patients were usually mixed, and separated data were not provided. The prevalence of RAS varied widely, from 6% to 90% with a mean of 37% (median, 33%). The thresholds used for diagnosis of clinically significant stenosis in the analyzed data were almost equal at 50% and 60%.

View Larger Version

TABLE 1: Characteristics of Studies of Duplex Sonography in the Detection of Renal Artery Stenosis (RAS)

Sonographic Parameters

Peak systolic velocity was measured in the main renal artery in all studies in the analysis, and 17 of the 21 studies detailed an angle-adjusted measurement. RAR was analyzed as a ratio between measurement in the main renal artery relative to aortic measurements. Eleven of 13 studies described an angle-adjusted measurement. AI was measured intrarenally in nine of 13 studies, three articles did not state where AI was measured, and in one study, AI was measured in the main renal artery. In two of 14 studies, calculation of AI was adjusted according to ultrasound frequency, but the effect on outcome was inconsistent with highly variable diagnostic ORs (239 and 59). AT was measured intrarenally in 10 of 13 studies, the reports of three studies not stating where measurements were made. Angle adjustment was stated in eight of 13 articles.

Test Accuracy

Figure 2A shows the SROC curve for all four individual duplex sonographic tests: peak systolic velocity, AT, AI, and RAR. The shape parameter reached significance in only one analysis, but this result was due to a single influential study. Hence, symmetry was assumed for all SROC curves. The estimated summary diagnostic OR and corresponding 95% CI based on all studies for each parameter are shown in the insets in Figure 2A, 2B, 2C, 2D, 2E. Peak systolic velocity had the highest accuracy, followed by RAR, AT, and AI. The corresponding areas under the SROC curves were computed from the estimated log diagnostic OR [24] to be 0.95, 0.91, 0.91, and 0.87 for peak systolic velocity, RAR, AT, and AI, respectively. After exclusion of the study identified earlier as influential for AT (Fig. 2A), the diagnostic OR increased from 28.9 to 42.2 (95% CI, 13.1-136.7). The corresponding estimate of the area under the curve was 0.93. Estimates of expected sensitivity, 1 - specificity, positive likelihood ratio, and negative likelihood ratio for each test, based on all studies, are shown in Table 2.

View Larger Version

TABLE 2: Estimated Sensitivity, I - Specificity, and Likelihood Ratios

View larger version (20K)

Fig. 2A —Summary receiver operator characteristics curves. DOR = diagnostic odds ratio, RDOR = relative DOR. Peak systolic velocity (PSV), renal-aortic ratio (RAR), acceleration time (AT), and acceleration index (AI).

View larger version (13K)

Fig. 2B —Summary receiver operator characteristics curves. DOR = diagnostic odds ratio, RDOR = relative DOR. Studies in which both PSV and RAR were examined separately.

View larger version (15K)

Fig. 2C —Summary receiver operator characteristics curves. DOR = diagnostic odds ratio, RDOR = relative DOR. Studies in which AI and AT were examined separately.

View larger version (13K)

Fig. 2D —Summary receiver operator characteristics curves. DOR = diagnostic odds ratio, RDOR = relative DOR. Studies in which PSV was examined alone and in combination with other parameters.

For the nine studies in which peak systolic velocity and RAR were directly compared for the same patients (Fig. 2B), the diagnostic OR was higher for peak systolic velocity than for RAR in almost all studies (relative diagnostic OR, 1.8; p = 0.03). In nine studies, investigators assessed both AT and AI (Fig. 2C). Analysis of paired studies showed no significant difference in accuracy between AT and AI (relative diagnostic OR, 1.1; p = 0.65). Excluding the influential study, the relative diagnostic OR increased to 1.4, which was not significant (p = 0.17), but there was a shift in test positivity (p = 0.01), AI having a higher expected sensitivity but lower expected specificity than AT. Analysis of the seven studies in which peak systolic velocity alone was compared with peak systolic velocity in combination with other parameters (Fig. 2D) showed only weak evidence of improvement in accuracy with the use of combined tests (relative diagnostic OR, 1.6; p = 0.09). Three of the studies, however, showed an increase in sensitivity but a decrease in specificity for the combined tests relative to peak systolic velocity alone. These findings suggest that an “either positive” rule was applied when test results were combined. Three of the other four studies showed negligible difference in either sensitivity or specificity, and one showed no difference. These results suggest that using tests in combination results in a shift in test positivity (p < 0.001) but are only weak evidence of an improvement in discrimination (diagnostic OR). Analysis of the five studies in which peak systolic velocity and AI were assessed (Fig. 2E) showed that peak systolic velocity was more accurate than AI (relative diagnostic OR, 5.3; p < 0.001), but the 95% CI was very wide. The estimated relative diagnostic ORs for the comparisons based on the subset of paired studies were consistent with the results for individual parameters shown in Figure 2A except for AT versus AI, in which the paired studies showed no difference in accuracy.

View larger version (15K)

Fig. 2E —Summary receiver operator characteristics curves. DOR = diagnostic odds ratio, RDOR = relative DOR. Studies in which PSV and AI were examined separately.

Population and Design Characteristics

The population and design characteristics are shown in Table 3. Eighty-seven (99%) of the articles reported the number of patients who underwent both duplex sonography and angiography, the average proportion being 75% (range, 3-100%) and the median being 100% (48 articles reported 100%). Seventy-seven articles reported the proportion of patients in whom sonography failed to give a result, which occurred in 10% of cases on average (median, 6%). The minimum proportion of failures reported was zero (22 studies), and the maximum, 54%.

View Larger Version

TABLE 3: Summary of Population and Design Characteristics of Studies of the Accuracy of Sonography in the Diagnosis of Renal Artery Stenosis (RAS)

To assess whether test accuracy or test positivity cut point was associated with study design or reporting, each population and design characteristic was included as a covariate in the hierarchical SROC curve model. For peak systolic velocity, the approach to failed sonographic examinations was associated with the cut point for test positivity but not with test accuracy. Studies explicitly showing no peak systolic velocity failures had a higher expected sensitivity (0.95) and, hence, a lower expected specificity (0.76) than the following study categories, which had similar expected sensitivity (0.81) and specificity (0.93): peak systolic velocity failures excluded, peak systolic velocity failures included, and no indication of what the investigators did with failed peak systolic velocity (p = 0.004). For AI, test accuracy increased as the test threshold increased. For every 0.5-m/s² increase in test threshold, the diagnostic OR increased an average of 3.8 times (relative diagnostic OR, 3.8; 95% CI, 1.4-10.5; p = 0.01). Other population and study design characteristics had no significant effect on test performance.

Discussion

We found that renal duplex sonography is a moderately accurate test for screening patients with hypertension for hemodynamically significant RAS, but important differences in accuracy exist among the parameters measured. Peak systolic velocity was the most accurate test parameter with values of sensitivity, specificity, and diagnostic OR of 85%, 92%, and 60.9, respectively. A diagnostic OR of 61 means that the odds of a test having a positive result (for peak systolic velocity) are 61 times greater in someone with RAS than in someone without RAS. AI was the least accurate parameter analyzed, the OR being 16.0. Between these extremes were RAR and AT, which had similar accuracy, with diagnostic ORs of 29.3 and 28.9, respectively. Studies in which different parameters were evaluated in the same patient group were useful for comparing the parameters while controlling for variation within patient group. On the basis of the findings of these paired studies, peak systolic velocity had superior discrimination compared with RAR (relative diagnostic OR, 1.8) and AI (relative diagnostic OR, 5.3). No significant difference in accuracy was identified for AI and AT. Analysis of peak systolic velocity alone compared with peak systolic velocity in combination with other parameters showed only weak evidence of improvement in discrimination for the combination over peak systolic velocity alone but suggested a shift in test positivity criterion. Overall, findings from paired analyses were consistent with those of individual, unpaired analyses.

There was considerable heterogeneity in the accuracy of the same test parameters among studies, and this finding was largely unexplained. An operator effect might have lead to variation in test performance, operators with the most experience producing best test performance. It was not possible, however, to analyze this effect because publications did not give details about experience level or number of operators involved in a study, nor were data reported separately for individual operators. The numbers of reported sonographic failures varied widely across publications (0-54%), suggesting substantial variation in study design. Failure rates were not higher in the oldest studies and did not appear to be related to which sonographic machine was used or the country in which the study was performed. A possible explanation for variability is referral filter bias. Few articles provided details on how patients were enrolled in the study; it was therefore impossible to determine how many difficult or likely-to-fail cases were excluded from the reports. Evidence of patient exclusion after study commencement was a discrepancy between reported numbers of patients in the assessment and results tallied. Variation in reported test performance may be a result of authors' selecting favorable patients for the studies. Such selection bias may affect the ability to generalize results to the broader group of patients who may ordinarily be considered for Doppler sonographic screening for RAS. Thus application of these results to an unselected clinical group may result in differing test performances.

In one systematic review [25], investigators evaluated renal duplex sonography in the diagnosis of RAS. The objective was to compare all available noninvasive or minimally invasive techniques currently used for detection of RAS. Thus, the focus was not exclusively on Doppler sonography. The authors used area under the curve to compare diagnostic techniques across studies and concluded that CT angiography and gadolinium-enhanced 3D MRI, which had an area under the curve of 0.99, performed better than other studies, including duplex sonography (area under the curve, 0.93). The authors, however, evaluated renal Doppler sonography as one test rather than determining which of the available measurements was the most accurate. The authors of the review included only 24 of the 88 studies identified in the current study, did not analyze each sonographic test separately, and did not evaluate study population and design characteristics. Given the variety of Doppler parameters that can be measured and our evidence of important differences in accuracy, summary statements about renal duplex sonography in general are probably unhelpful to clinicians and radiologists who need to decide which parameter should be measured.

This study was a comprehensive review of all available data and was conducted with recently developed methods for finding relevant studies, evaluating their quality, and synthesizing results. We depended, however, on the quality of design and reporting of the primary studies, which in many cases was problematic. This finding was not novel, and methods are being developed to improve the design and reporting of studies of diagnostic tests. We hope that the recently published Standards for Reporting of Diagnostic Accuracy [26] will improve future reports of studies of diagnostic tests.

It is accepted that for many diagnostic tests, accuracy relies on operator expertise; however, evidence to support this belief is lacking. None of the 88 studies analyzed provided sufficient details to allow exploration of this theory. Considerable evidence of patient selection bias was apparent in many of the studies. In using peak systolic velocity as a triage test for RAS, it may be worthwhile to consider both who will be performing the test and who will be the subjects.

In the clinical setting, duplex sonography is used primarily as a screening or triage test for RAS to prevent all patients with suspected RAS from undergoing the reference standard test, renal angiography. Of the four possible classifications based on test and target condition status, the proportion of patients with true-positive results (those who would still need angiography) is less important than the proportion with truenegative results (those who avoid angiography) and the proportion of those with false-positive findings (those who undergo unnecessary angiography). Based on the summary sensitivities and specificities determined in this metaanalysis, Table 4 shows the expected number of persons in each of the four test and target condition cells with each duplex sonographic parameter. The data were calculated for two different prevalences of disease (5% and 45%), reflecting the prevalence found in unselected hypertensive populations and in selected populations of patients with hypertension and peripheral vascular disease.

View Larger Version

TABLE 4: Clinical Implications of Use of Renal Duplex Sonography in the Triage of 100 Persons with Hypertension and Possible Renal Artery Stenosis (RAS) with Assumption of Low (5%) and High (45%) Prevalence

Our analysis highlighted and confirmed previous findings that showed studies of diagnostic tests are poorly reported. Further high-quality research that yields data on individual duplex sonographic parameters separately and specifies combinations of parameters is needed. In particular, to explore further the association between expertise and experience and test accuracy, authors should report the experience of the test operators and observers. Further studies are needed to evaluate the comparative accuracy of peak systolic velocity and AT with paired data cross-classified by RAS status.

In conclusion, renal duplex sonography performed with the parameter peak systolic velocity is accurate in triage of many patients with suspected RAS, and unnecessary renal arteriography is avoided. The noninvasive character, ready availability, and low cost of duplex sonography make it an attractive option as a screening test for RAS.

APPENDIX 2: Description of Hierarchical SROC Model and SAS Program for Fitting the Model

In the hierarchical summary receiver operating characteristic model [17, 18], within-study variability for the ith study is modeled as logit , where Y_ij = 1 represents a positive test result for subject j, and D_ij represents that subject's true disease status as defined by the reference standard (coded as -0.5 for nondiseased and 0.5 for diseased). Between-study variability is taken into account by allowing each study (i) to have its own cut point for test positivity (θ_i) and diagnostic accuracy (α_i). The cut point for test positivity is assumed to be on an underlying (latent) continuous scale for test results. The random effects for cut point and accuracy are assumed to be independent (uncorrelated) and normally distributed with and . Hence Λ, gives the expected accuracy (log diagnostic odds ratio) at a given cut point. The modeling takes into account the sampling variability in estimates of sensitivity and specificity (and the correlation between them) within each study in estimation of random effects. The fixed-effect parameter β is used to assess whether test accuracy depends on test threshold. When β = 0, test accuracy is constant across thresholds, resulting in a summary receiver operating characteristic curve that is symmetric about the diagonal line where sensitivity equals specificity, and Λ then is a global estimate of test accuracy (log diagnostic odds ratio) that is constant across thresholds. The model was fitted with PROC NLMIXED (SAS [19]. Functions of the model parameters were used to estimate the expected sensitivity, specificity, likelihood ratios, and confidence intervals.

Address correspondence to G. J. Williams ([email protected]).

Supported by National Health and Medical Research Council Program grant no. 211205 (Australia).

The authors have no financial interests in this article.

References

1. Olin JW. Atherosclerotic renal artery disease. Cardiol Clin 2002; 20:547-562 [Google Scholar]

2. Jacobson HR. Ischemic renal disease: an overlooked clinical entity? Kidney Int 1988; 34:729-743 [Google Scholar]

3. Rimmer JM, Gennari FJ. Atherosclerotic renovascular disease and progressive renal failure. Ann Intern Med 1993; 118:712-719 [Google Scholar]

4. Textor SC. Epidemiology and clinical presentation. Semin Nephrol 2000; 20:426-431 [Google Scholar]

5. Winker MA. The emerging epidemic of atherosclerosis. JAMA 1999; 281:84-85 [Google Scholar]

6. Jensen G, Zachrisson BF, Delin K, Volkmann R, Aurell M. Treatment of renovascular hypertension: one-year results of renal angioplasty. Kidney Int 1995; 48:1936-1945 [Google Scholar]

7. Bonelli FS, McKusick MA, Textor SC, et al. Renal artery angioplasty: technical results and clinical outcome in 320 patients. Mayo Clin Proc 1995; 70:1041-1052 [Google Scholar]

8. Zoccali C, Mallamaci F, Finocchiaro P. Atherosclerotic renal artery stenosis: epidemiology, cardiovascular outcomes, and clinical prediction rules. J Am Soc Nephrol 2002; 13 [suppl 3]:S179-S183 [Google Scholar]

9. Mann SJ, Pickering TG. Detection of renovascular hypertension: state of the art. Ann Intern Med 1992; 117:845-853 [Google Scholar]

10. Safian RD, Textor SC. Renal-artery stenosis. N Engl J Med 2001; 344:431-442 [Google Scholar]

11. Mann SJ, Pickering TG, Sos TA, et al. Captopril renography in the diagnosis of renal artery stenosis: accuracy and limitations. Am J Med 1991; 90:30-40 [Google Scholar]

12. Huot SJ, Hansson JH, Dey H, Concato J. Utility of captopril renal scans for detecting renal artery stenosis. Arch Intern Med 2002; 162:1981-1984 [Google Scholar]

13. Vashist A, Heller EN, Brown EJ, Alhaddad IM. Renal artery stenosis: a cardiovascular perspective. Am Heart J 2002; 143:559-564 [Google Scholar]

14. Moses LE, Shapiro D, Littenberg B. Combining independent studies of a diagnostic test into a summary ROC curve: data analytic approaches and some additional considerations. Stat Med 1993; 12:1293-1316 [Google Scholar]

15. Littenberg B, Moses LE. Estimating diagnostic accuracy from multiple conflicting reports: a new meta-analytic method. Med Decis Making 1993; 13:313-321 [Google Scholar]

16. Irwig L, Macaskill P, Glasziou P, et al. Meta-analytic methods for diagnostic test accuracy. J Clin Epidemiol 1995; 48:119-130 [Google Scholar]

17. Rutter CM, Gatsonis CA. A hierarchical regression approach to meta-analysis of diagnostic test accuracy evaluations. Stat Med 2001; 20:2865-2884 [Google Scholar]

18. Rutter CM, Gatsonis CA. Regression methods for meta-analysis of diagnostic test data. Acad Radiol 1995; 2 [suppl 1]:S48-S56 [Google Scholar]

19. Macaskill P. Empirical Bayes estimates generated in a hierarchical summary ROC analysis agreed closely with those of a full Bayesian analysis. J Clin Epidemiol 2004; 57:925-932 [Google Scholar]

20. Rittgers SE, Norris CS, Barnes RW. Detection of renal artery stenosis: experimental and clinical analysis of velocity waveforms. Ultrasound Med Biol 1985; 11:523-531 [Google Scholar]

21. Lossner F, Ingrisch H, Hepp KD. Color-coded duplex sonography in the diagnosis of renal artery stenosis: a 3-year experience in a large community hospital. Int J Angiol 1998; 7:10-13 [Google Scholar]

22. Snider JF, Hunter DW, Moradian GP, Castaneda-Zuniga WR, Letourneau JG. Transplant renal artery stenosis: evaluation with duplex sonography. Radiology 1989; 172:1027-1030 [Google Scholar]

23. Sievers KW, Lohr E, Werner WR. Duplex Doppler ultrasound in determination of renal artery stenosis. Urol Radiol 1989; 11:142-147 [Google Scholar]

24. Walter SD. Properties of the summary receiver operating characteristic (SROC) curve for diagnostic test data. Stat Med 2002; 21:1237-1256 [Google Scholar]

25. Boudewijn G, Vasbinder C, Nelemans PJ, et al. Diagnostic tests for renal artery stenosis in patients suspected of having renovascular hypertension: a meta-analysis. Ann Intern Med 2001; 135:401-411 [Google Scholar]

26. Bossuyt PM, Reitsma JB, Bruns DE, et al. Standards for reporting of diagnostic accuracy: the STARD statement for reporting studies of diagnostic accuracy—explanation and elaboration. Ann Intern Med 2003; 138:W1-W12 [Google Scholar]

27. Saarinen O, Salmela K, Edgren J. Doppler ultrasound in the diagnosis of renal transplant artery stenosis: value of resistive index. Acta Radiol 1994; 35:586-589 [Google Scholar]

28. Kletter, K, G Mostbeck, R Duczak. Captopril renography and duplex sonography: comparison of two noninvasive methods for the diagnosis and follow-up in renovascular hypertension. Contrib Nephrol 1990; 79:190-195 [Google Scholar]

29. Radermacher J, Chavan A, Schaffer J, et al. Detection of significant renal artery stenosis with color Doppler sonography: combining extrarenal and intrarenal approaches to minimize technical failure. Clin Nephrol 2000; 53:333-343 [Google Scholar]

30. Hua HT, Hood DB, Jensen CC, Hanks SE, Weaver FA. The use of colorflow duplex scanning to detect significant renal artery stenosis. Ann Vasc Surg 2000; 14:118-124 [Google Scholar]

31. Kliewer MA, Tupler RH, Carroll BA, et al. Renal artery stenosis: analysis of Doppler waveform parameters and tardus-parvus pattern. Radiology 1993; 189:779-787 [Google Scholar]

32. Antonica G, Sabba C, Berardi E, et al. Accuracy of echo-Doppler flowmetry for renal artery stenosis. J Hypertens Suppl 1991; 9:S240-S241 [Google Scholar]

33. Argalia G, Cacciamani L, Fazi R, Salera D, Giuseppetti GM. Contrast-enhanced sonography in the diagnosis of renal artery stenosis: comparison with MR-angiography. Radiol Med (Torino) 2004; 107:208-217 [Google Scholar]

34. Avasthi PS, Voyles WF, Greene ER. Noninvasive diagnosis of renal artery stenosis by echo-Doppler velocimetry. Kidney Int 1984; 25:824-829 [Google Scholar]

35. Bardelli M, Jensen G, Volkmann R, Aurell M. Non-invasive ultrasound assessment of renal artery stenosis by means of the Gosling pulsatility index. J Hypertens 1992; 10:985-989 [Google Scholar]

36. Barozzi L, Pavlica P, Sabattini A, et al. Duplex and Doppler color echocardiography for the study of renovascular hypertension: comparison with arteriography. Radiol Med 1991; 81:642-649 [Google Scholar]

37. Baxter GM, Aitchison F, Sheppard D, et al. Colour Doppler ultrasound in renal artery stenosis: intrarenal waveform analysis. Br J Radiol 1996; 69:810-815 [Google Scholar]

38. Blebea J, Zickler R, Volteas N, et al. Duplex imaging of the renal arteries with contrast enhancement. Vasc Endovascular Surg 2003; 37:429-436 [Google Scholar]

39. Breitenseher M, Kainberger F, Hubsch P, et al. The screening of renal artery stenoses: the initial results with the value of color Doppler sonography [in German]. Rofo 1992; 156:228-231 [Google Scholar]

40. Buchet P, Minvielle F, Rainfray M, Mahfouz T, Ancri D, Meyrier A. Diagnostic value of pulsed Doppler in atheromatous renal insufficiency: a study of 32 patients. Arch Mal Coeur Vaiss 1991; 84:1191-1193 [Google Scholar]

41. Burdick L, Airoldi F, Marana I, et al. Superiority of acceleration and acceleration time over pulsatility and resistance indices as screening tests for renal artery stenosis. J Hypertens 1996; 14:1229-1235 [Google Scholar]

42. Claudon M, Plouin PF, Baxter GM, Rohban T, Devos DM. Renal arteries in patients at risk of renal arterial stenosis: multicenter evaluation of the echo-enhancer SH U 508A at color and spectral Doppler US. Radiology 2000; 214:739-746 [Google Scholar]

43. Coen G, Manni M, Giannoni MF, et al. Ischemic nephropathy in an elderly nephrologic and hypertensive population. Am J Nephrol 1998; 18:221-227 [Google Scholar]

44. Combarnous F, Pouteil-Noble C, Lapra C, Poix D, Pinet A, Touraine JL. Diagnosis of stenoses of the renal artery graft by pulsed echo Doppler velocimetry: application to the prospective evaluation of the incidence of stenoses of renal graft artery. Presse Med 1992; 21:1977-1978 [Google Scholar]

45. Conkbayir I, Yucesoy C, Edguer T, Yanik B, Yasar Ayaz U, Hekimoglu B. Doppler sonography in renal artery stenosis: an evaluation of intrarenal and extrarenal imaging parameters. Clin Imaging 2003; 27:256-260 [Google Scholar]

46. De Cobelli F, Venturini M, Vanzulli A, et al. Renal arterial stenosis: prospective comparison of color Doppler US and breath-hold, three-dimensional, dynamic, gadolinium-enhanced MR angiography. Radiology 2000; 214:373-380 [Google Scholar]

47. Deane C, Cairns T, Walters H, et al. Diagnosis of renal transplant artery stenosis by color Doppler ultrasonography. Transplant Proc 1990; 22:1395 [Google Scholar]

48. de Haan MW, Kroon AA, Flobbe K, et al. Renovascular disease in patients with hypertension: detection with duplex ultrasound. J Hum Hypertens 2002; 16:501-507 [Google Scholar]

49. Dondi M, Fanti S, Barozzi L, et al. Evaluation by captopril renal scintigraphy and echo-Doppler flowmetry of hypertensive patients at high risk for renal artery stenosis. J Nucl Biol Med 1992; 36:309-314 [Google Scholar]

50. Engelhorn CA, Engelhorn AL, Pullig R. Vascular color Doppler ultrasound for assessing renovascular hypertension: accuracy of the direct technique for assessing the renal arteries. Arq Bras Cardiol 2004; 82:477-480 [Google Scholar]

51. Equine O, Beregi JP, Mounier-Vehier C, Gautier C, Desmoucelles F, Carre A. Importance of the echo-doppler and helical angioscanner of the renal arteries in the management of renovascular diseases: results of a retrospective study in 113 patients. Arch Mal Coeur Vaiss 1999; 92:1043-1045 [Google Scholar]

52. Ferdinandi A, Pavlica P, Lupattelli L, Barozzi L, Mosca S. Duplex-sonography and color-Doppler evaluation of renal artery stenosis-angiographic correlation. Scand J Urol Nephrol Suppl 1991; 137:67-72 [Google Scholar]

53. Gokhale S, Salgia P. The role of color Doppler sonography in screening for renal artery stenosis. Ultrasound Int 2002; 8:122-126 [Google Scholar]

54. Grataloup-Oriez C, Plainfosse MC, Challande P, et al. Renal artery stenosis detection in 123 hypertensive patients: comparison between Doppler and angiography. Eur J Ultrasound 1997; 5:155-163 [Google Scholar]

55. Grenier N, Douws C, Morel D, et al. Detection of vascular complications in renal allografts with color Doppler flow imaging. Radiology 1991; 178:217-223 [Google Scholar]

56. Halpern EJ, Needleman L, Nack TL, East SA. Renal artery stenosis: should we study the main renal artery or segmental vessels? Radiology 1995; 195:799-804 [Google Scholar]

57. Halpern EJ, Rutter CM, Gardiner GA Jr, et al. Comparison of Doppler US and CT angiography for evaluation of renal artery stenosis. Acad Radiol 1998; 5:524-532 [Google Scholar]

58. Handa N, Fukanaga R, Ogawa S, Matsumoto M, Kimura K, Kamada T. A new accurate and non-invasive screening method for renovascular hypertension: the renal artery Doppler technique. J Hypertens Suppl 1988; 6:S458-S460 [Google Scholar]

59. Hansen KJ, Tribble RW, Reavis SW, et al. Renal duplex sonography: evaluation of clinical utility. J Vasc Surg 1990; 12:227-236 [Google Scholar]

60. Hawkins PG, McKnoulty LM, Gordon RD, Klemm SA, Tunny TJ. Non-invasive renal artery duplex ultrasound and computerized nuclear renography to screen for and follow progress in renal artery stenosis. J Hypertens Suppl 1989; 7:S184-S185 [Google Scholar]

61. Helenon O, el Rody F, Correas JM, et al. Color Doppler US of renovascular disease in native kidneys. RadioGraphics 1995; 15:833-854 [Google Scholar]

62. Hoffmann U, Edwards JM, Carter S, et al. Role of duplex scanning for the detection of atherosclerotic renal artery disease. Kidney Int 1991; 39:1232-1239 [Google Scholar]

63. Hollenbeck M, Kutkuhn B, Grabensee B. Colour Doppler ultrasound in the diagnosis of transplant renal artery stenosis. Bildgebung 1994; 61:248-254 [Google Scholar]

64. Hortling N, Strunk H, Wilhelm K, Hofer U, Schild HH. Visualization of renal arteries and value of color-coded duplex sonography in renal artery stenoses using an ultrasound signal enhancing agent [in German]. Rofo 1998; 169:397-401 [Google Scholar]

65. House MK, Dowling RJ, King P, Gibson RN. Using Doppler sonography to reveal renal artery stenosis: an evaluation of optimal imaging parameters. AJR 1999; 173:761-765 [Abstract] [Google Scholar]

66. Johansson M, Jensen G, Aurell M, et al. Evaluation of duplex ultrasound and captopril renography for detection of renovascular hypertension. Kidney Int 2000; 58:774-782 [Google Scholar]

67. Kaplan-Pavlovcic S, Nadja C. Captopril renography and duplex Doppler sonography in the diagnosis of renovascular hypertension [in German]. Nephrol Dial Transplant 1998; 13:313-317 [Google Scholar]

68. Karasch T, Strauss AL, Grun B, et al. Colourcoded duplex ultrasonography in the diagnosis of renal artery stenosis [in German]. Dtsch Med Wochenschr 1993; 118:1429-1436 [Google Scholar]

69. Kchouk H, Brun P, Sentou Y, Raynaud A, Gaux JC, Loirat C. Renal stenosis in hypertensive children: Doppler/arteriographic correlation. J Mal Vasc 1997; 22:86-90 [Google Scholar]

70. Kohler SM, Muscholl M, Kromer EP, Kramer BK, Riegger GA. Comparison of color coded duplex scanning and intra-arterial angiography for diagnosis of renal artery stenosis in patients with hypertension and coronary artery disease. Nieren Hochdruckkrankh 1994; 23:182-184 [Google Scholar]

71. Krumme B, Blum U, Schwertfeger E, et al. Diagnosis of renovascular disease by intra- and extrarenal Doppler scanning. Kidney Int 1996; 50:1288-1292 [Google Scholar]

72. Lampreave J, Rengel M, Dominguez P, et al. Scintigraphy diagnosis of renovascular hypertension in renal transplanted patients. Rev Esp Med Nucl 1999; 18:340-347 [Google Scholar]

73. Leung DA, Hoffmann U, Pfammatter T, et al. Magnetic resonance angiography versus duplex sonography for diagnosing renovascular disease. Hypertension 1999; 33:726-731 [Google Scholar]

74. Lucas P, Blome S, Roche J. Intra-renal Doppler wave-form analysis as a screening test for renal artery stenosis. Australas Radiol 1996; 40:276-282 [Google Scholar]

75. Magotteaux P, Thille A, Biquet JF, Brisbois D. Comparative assessment of color Doppler sonography and helical CT-scan angiography for exploring renal artery stenosis in hypertensive patients [in French]. J Echogr Med Ultrasons 1997; 18:162-166 [Google Scholar]

76. Malatino LS, Polizzi G, Garozzo M, et al. Diagnosis of renovascular disease by extra- and intrarenal doppler parameters. Angiology 1998; 49:707-721 [Google Scholar]

77. Martin RL, Nanra RS, Wlodarczyk J, De Silva A, Bray AE. Renal hilar doppler analysis in the detection of renal artery stenosis. J Vasc Technol 1991; 15:173-180 [Google Scholar]

78. Materne R, Puttemans T, Goffette P, De Plaen JF, Dardenne AN. Reliability of color Doppler ultrasound for diagnosis of renal artery stenosis [in French]. J Echogr Med Ultrasons 1997; 18:155-157 [Google Scholar]

79. Milon P, Clavier E, Genevois A, Benozio M. Feasibility study of the Doppler exploration of the renal artery. J Radiol 1990; 71:215-220 [Google Scholar]

80. Miralles M, Cairols M, Cotillas J, Gimenez A, Santiso A. Value of Doppler parameters in the diagnosis of renal artery stenosis. J Vasc Surg 1996; 23:428-435 [Google Scholar]

81. Missouris CG, Allen CM, Balen FG, Buckenham T, Lees WR, MacGregor GA. Non-invasive screening for renal artery stenosis with ultrasound contrast enhancement. J Hypertens 1996; 14:519-524 [Google Scholar]

82. Mollo M, Pelet V, Mouawad J, Mathieu JP, Branchereau A. Evaluation of colour duplex ultrasound scanning in diagnosis of renal artery stenosis, compared to angiography: a prospective study on 53 patients. Eur J Vasc Endovasc Surg 1997; 14:305-309 [Google Scholar]

83. Motew SJ, Cherr GS, Craven TE, et al. Renal duplex sonography: main renal artery versus hilar analysis. J Vasc Surg 2000; 32:462-469 [Google Scholar]

84. Moulton-Levy D. Detection of renal artery stenosis by duplex and color flow Doppler. J Diagn Med Sonogr 1990; 6:1-12 [Google Scholar]

85. Napoli V, Palla A, Pinto F, et al. Abdominal Doppler ultrasonography in the diagnosis of renovascular diseases: double-blind prospective study. Radiol Med 1993; 86:496-502 [Google Scholar]

86. Napoli V, Pinto S, Bargellini I, et al. Duplex ultrasonographic study of the renal arteries before and after renal artery stenting. Eur Radiol 2002; 12:796-803 [Google Scholar]

87. Nchimi A, Biquet JF, Brisbois D, et al. Duplex ultrasound as first-line screening test for patients suspected of renal artery stenosis: prospective evaluation in high-risk group. Eur Radiol 2003; 13:1413-1419 [Google Scholar]

88. Norris CS, Pfeiffer JS, Rittgers SE, Barnes RW. Noninvasive evaluation of renal artery stenosis and renovascular resistance: experimental and clinical studies. J Vasc Surg 1984; 1:192-201 [Google Scholar]

89. Olin JW, Piedmonte MR, Young JR, DeAnna S, Grubb M, Childs MB. The utility of duplex ultrasound scanning of the renal arteries for diagnosing significant renal artery stenosis. Ann Intern Med 1995; 122:833-838 [Google Scholar]

90. Oliva VL, Soulez G, Lesage D, et al. Detection of renal artery stenosis with Doppler sonography before and after administration of captopril: value of early systolic rise. AJR 1998; 170:169-175 [Abstract] [Google Scholar]

91. Pedersen EB, Egeblad M, Jorgensen J, Nielsen SS, Spencer ES, Rehling M. Diagnosing renal artery stenosis: a comparison between conventional renography, captopril renography and ultrasound Doppler in a large consecutive series of patients with arterial hypertension. Blood Press 1996; 5:342-348 [Google Scholar]

92. Pedro LM, Freire JP, Machado AS, et al. Assessment of arterial occlusive disease with duplex sonography: prospective study [in Portuguese]. Rev Port Cardiol 1993; 12:905-911 [Google Scholar]

93. Postma CT, van Aalen J, de Boo T, Rosenbusch G, Thien T. Doppler ultrasound scanning in the detection of renal artery stenosis in hypertensive patients. Br J Radiol 1992; 65:857-860 [Google Scholar]

94. Postma CT, Bijlstra PJ, Rosenbusch G, Thien T. Pattern recognition of loss of early systolic peak by Doppler ultrasound has a low sensitivity for the detection of renal artery stenosis. J Hum Hypertens 1996; 10:181-184 [Google Scholar]

95. Rene PC, Oliva VL, Bui BT, et al. Renal artery stenosis: evaluation of Doppler US after inhibition of angiotensin-converting enzyme with captopril. Radiology 1995; 196:675-679 [Google Scholar]

96. Riehl J, Schmitt H, Bongartz D, Bergmann D, Sieberth HG. Renal artery stenosis: evaluation with colour duplex ultrasonography. Nephrol Dial Transplant 1997; 12:1608-1614 [Google Scholar]

97. Ripolles T, Aliaga R, Morote V, et al. Utility of intrarenal Doppler ultrasound in the diagnosis of renal artery stenosis. Eur J Radiol 2001; 40:54-63 [Google Scholar]

98. Ruebben A, Piroth W, Neuerburg J, Wildberger JE, Schmitz-Rode T, Gunther RW. Diagnosis and visualization of renal artery stenosis by color-coded Doppler ultrasonography: comparison of central and peripheral flow patterns. Rofo 1999; 171:319-323 [Google Scholar]

99. Schwerk WB, Restrepo IK, Stellwaag M, Klose KJ, Schade-Brittinger C. Renal artery stenosis: grading with image-directed Doppler US evaluation of renal resistive index. Radiology 1994; 190:785-790 [Google Scholar]

100. Simoni C, Balestra G, Bandini A, Rusticali F. Doppler ultrasound in the diagnosis of renal artery stenosis in hypertensive patients: a prospective study. G Ital Cardiol 1991; 21:249-255 [Google Scholar]

101. Souza de Oliveira IR, Widman A, Molnar LJ, Fukushima JT, Praxedes JN, Cerri GG. Colour Doppler ultrasound: a new index improves the diagnosis of renal artery stenosis. Ultrasound Med Biol 2000; 26:41-47 [Google Scholar]

102. Spies KP, Fobbe F, El-Bedewi M, Wolf KJ, Distler A, Schulte KL. Color-coded duplex sonography for noninvasive diagnosis and grading of renal artery stenosis. Am J Hypertens 1995; 8:1222-1231 [Google Scholar]

103. Stavros AT, Parker SH, Yakes WF, et al. Segmental stenosis of the renal artery: pattern recognition of tardus and parvus abnormalities with duplex sonography. Radiology 1992; 184:487-492 [Google Scholar]

104. Stuhrmann M, Roefke C, Jahnke T, Cramer BM. Duplex sonography of renal artery stenosis: potentials and limits of frequency spectrum analysis of arterial segments compared with angiography. Rofo 1998; 168:258-263 [Google Scholar]

105. Taylor DC, Kettler MD, Moneta GL, et al. Duplex ultrasound scanning in the diagnosis of renal artery stenosis: a prospective evaluation. J Vasc Surg 1988; 7:363-369 [Google Scholar]

106. Vigna C, Pacilli MA, Testa M, et al. Comparison of steerable continuous-wave versus pulsed-wave Doppler ultrasonography to renal artery angiography in diagnosing renal artery stenosis. Am J Cardiol 1998; 81:365-367 [Google Scholar]

107. Voiculescu A, Hofer M, Hetzel GR, et al. Noninvasive investigation for renal artery stenosis: contrast-enhanced magnetic resonance angiography and color Doppler sonography as compared to digital subtraction angiography. Clin Exp Hypertens 2001; 23:521-531 [Google Scholar]

108. Winde G, Buchholz B, Krings W, et al. Duplex sonography in the diagnosis of renal artery stenoses following allogenic kidney transplantation. Langenbecks Arch Chir 1989; 374:284-290 [Google Scholar]

109. Zeller T, Frank U, Spath M, Roskamm H. Color duplex ultrasound imaging of renal arteries and detection of hemodynamically relevant renal artery stenoses [in German]. Ultraschall Med 2001; 22:116-1621 [Google Scholar]

110. Zoller WG, Hermans H, Bogner JR, Hahn D, Middeke M. Duplex sonography in the diagnosis of renovascular hypertension. Klin Wochenschr 1990; 68:830-834 [Google Scholar]

Related Articles

Recommend & Share

March 2007, Volume 188, Number 3

Genitourinary Imaging

Original Research

Comparative Accuracy of Renal Duplex Sonographic Parameters in the Diagnosis of Renal Artery Stenosis: Paired and Unpaired Analysis

Recommended Articles

Comparative Accuracy of Renal Duplex Sonographic Parameters in the Diagnosis of Renal Artery Stenosis: Paired and Unpaired Analysis