Correlation of North American Symptomatic Carotid Endarterectomy Trial (NASCET) angiographic definition of 70% to 99% internal carotid artery stenosis with duplex scanning☆

Article Outline

• Abstract
• Patients and methods
  • Patients
  • Duplex studies
  • Angiograms
  • Data analysis
• Results
• Discussion
• Discussion
• References
• Copyright

Abstract 

Purpose: The North American Symptomatic Carotid Endarterectomy Trial (NASCET) has thus far demonstrated conclusive benefit for carotid endarterectomy for patients with symptomatic 70% to 99% internal carotid artery (ICA) stenosis. In the NASCET, ICA stenosis was classified angiographically: % ICA stenosis = (1—[narrowest ICA diameter/diameter normal distal cervical ICA]) × 100%. However, widely used duplex scan criteria for ICA stenosis correlate with different angiographic categories of high-grade stenosis (50% to 79%, >80%) and were developed on the basis of estimated bulb diameter. We therefore blindly evaluated with separate observers carotid angiograms from 100 patients who also underwent carotid duplex scanning in our vascular laboratory. Methods: “Angiographic stenosis” was calculated as in NASCET. Duplex scan measurements of ICA peak systolic velocity (PSV), ICA end-diastolic velocity, and the ratio of ICA PSV to common carotid artery (CCA) PSV were analyzed for sensitivity, specificity, positive predictive value, negative predictive value, and overall accuracy to identify a 70% to 99% ICA stenosis. Results: Analysis of the data revealed that an ICA PSV/CCA PSV ratio of 4.0 provided the best combination of sensitivity (91%), specificity (87%), positive predictive value (76%), negative predictive value (96%), and overall accuracy (88%) for detection of a 70% to 99% stenosis. Conclusion: We conclude duplex scan determination of 70% to 99% stenosis as defined in the NASCET requires the adoption of duplex criteria modified from those in current use in most vascular laboratories. (J VASC SURG 1993;17:152-9.)

Duplex scanning is well accepted as the most accurate diagnostic modality for noninvasive evaluation of carotid artery stenosis. In most laboratories degrees of arterial narrowing are grouped into categories on the basis of velocity waveform criteria that have been developed by comparing carotid duplex scans to carotid arteriograms obtained in the same patients. It is important to note that the degree of narrowing of the internal carotid artery (ICA) seen on angiography may be determined in different ways. “Angiographic” ICA stenosis may be based on the estimated diameter of the carotid bulb were it to be free of atherosclerosis, or it may be based on the measured diameter of the distal, presumably normal, cervical ICA.

Currently many vascular laboratories use duplex criteria for determining ICA stenosis patterned after criteria developed at the University of Washington.¹ Velocity waveform analysis and spectral criteria are used to classify the ICA as normal, 1% to 15% angiographic stenosis, 16% to 49% angiographic stenosis, 50% to 79% angiographic stenosis, 80% to 99% angiographic stenosis, and occlusion. Percent angiographic stenosis in the development of these criteria was calculated by comparing the site of greatest ICA narrowing seen on angiography to the estimated normal diameter of the bulb. These criteria have been prospectively validated and have proved enormously useful both in clinical practice and in research evaluating the natural history of atherosclerotic carotid artery disease.², ³

In the spring of 1991 interim results of the North American Symptomatic Carotid Endarterectomy Trial (NASCET) were reported. The NASCET investigators reported a clear and striking benefit in terms of stroke reduction for a specific subgroup of patients with symptomatic ICA stenosis who were treated with carotid endarterectomy in addition to medical management versus those treated with medical management alone.⁴ The discussion of the NASCET report repeatedly emphasized that the proven benefits of carotid endarterectomy for symptomatic patients were limited to those with 70% to 99% angiographic ICA stenosis. In the NASCET investigation calculation of angiographic ICA stenosis was based on the measured diameter of the distal cervical ICA.⁵

The University of Washington criteria for determining ICA stenosis with duplex scanning are therefore not applicable to NASCET either in terms of duplex categories of stenosis or in the method used for calculating angiographic stenosis. This, together with the observation that increasing numbers of surgeons recommend performance of carotid endarterectomy based on duplex scanning alone, prompted us to evaluate the potential accuracy of various duplex parameters in determining a 70% to 99% ICA stenosis as defined by NASCET.

Back to Article Outline

Patients and methods 

Patients 

We compared carotid angiograms and carotid duplex scans from 100 patients evaluated at our institution for surgical treatment of carotid artery atherosclerosis from January 1987 to December 1991. Patients were identified through our computer-based vascular registry. The first 100 patients whose angiograms were obtained from the radiology department record room and who also underwent carotid artery duplex scanning in our vascular laboratory within 1 week of their carotid angiographic study were selected for the study.

Duplex studies 

Vascular laboratory carotid artery duplex scanning was performed with use of a color duplex scanner (Acuson 128; Acuson Inc., Mountain View, Calif.) in the manner described by Strandness.⁶ The ultrasonic transducer was maintained at a 60-degree angle of insonation. All the cervical common carotid artery (CCA) and ICA accessible to the transducer were examined. Velocity waveforms were obtained from any location where stenosis was suspected and routinely from the CCA in the base of the neck and just proximal to the bulb, from the proximal, middle, and distal ICA, and from the external carotid artery. The highest peak systolic velocity (PSV) and end-diastolic velocity (EDV) were recorded from each examination site.

Angiograms 

Carotid arteriograms were performed with use of either intraarterial digital subtraction or standard cut-film techniques. Intravenous digital studies were not used. At least two views were obtained of each common carotid bifurcation. Intracranial and aortic arch views were also routinely obtained.

For purposes of this report ICA stenosis was measured as described by NASCET. Therefore the view showing the greatest degree of ICA narrowing was chosen for analysis. The site of greatest ICA stenosis and the diameter of the distal cervical ICA were measured with calipers and loop magnification by a single individual (observer 1) unaware of the results of the duplex studies. The percent angiographic ICA stenosis was then calculated on the basis of the distal cervical ICA (Fig. 1).

• 
  • View Large Image
  • Download to PowerPoint

• Fig. 1. 

  ICA angiographic stenosis may be calculated using as reference vessel either measured diameter of distal cervical ICA: (% ICA )] × 100% or estimated diameter of bulb of atherosclerosis: (% ICA stenosis = [1-(c/b)] × 100%). Former technique is used in calculating ICA stenosis in NASCET,⁵ ACAS study,¹¹ and Veteran's Affairs Cooperative Trial of asymptomatic carotid stenosis.¹⁰ Later method was used in developing University of Washington duplex scanning criteria for ICA stenosis and for calculating ICA stenosis in European Cooperative Trial of Carotid Endarterectomy.¹⁵

When the greatest degree of narrowing seen on angiography involved the distal CCA rather than the ICA, the CCA site of maximal stenosis was used for the calculation of angiographic stenosis. This calculation was, however, still based on the distal cervical ICA as the reference vessel. Extreme narrowing of the proximal ICA, resulting in incomplete filling of the distal cervical ICA, angiographic “string signs,” were designated as 95% ICA stenoses, again in accordance with NASCET.⁵ Angiographically occluded ICAs were excluded from this study because occluded ICAs were also excluded from randomization in NASCET.⁴

To obtain a measure of the reproducibility of the NASCET method of calculating angiographic stenosis, maximal angiographic stenosis was also determined in the same manner by a second observer (observer 2) unaware of the results of the initial measurements. However, because NASCET used only a single observer for final classification of stenosis,⁵ only the determinations of angiographic stenosis by observer no. 1 were used for comparison with the results of duplex scanning.

Data analysis 

In most cases maximal ICA PSV and EDV were used for comparisons with maximal angiographic stenosis. When distal CCA velocities exceeded ICA velocities, maximal distal CCA PSV and EDV were used for comparisons with calculated maximal angiographic stenosis. In addition a comparison of maximal angiographic stenosis was also made with the ratio of the maximal ICA PSV to the PSV of the CCA obtained from a site low in the neck (ICA PSV/CCA PSV ratio). All data were entered into a personal computer, and receiver-operator characteristic (ROC) curves were generated for the sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of different ICA PSVs, ICA EDVs, and ICA PSV/CCA PSV ratios to predict a 70% to 99% angiographic stenosis according to the measurements made by observer 1.

Kappa values were used to compare the angiogram readings of observers 1 and 2 with respect to their classification of ICA or distal CCA angiographic stenosis as greater than or less than 70%.

Back to Article Outline

Results 

One hundred eighty-four ICAs seen as patent on angiography were present in the 100 patients selected for the study. As measured by observer 1, a 70% to 99% angiographic stenosis was present in 58 (32%) of the patent ICAs. Observer 2 also identified 184 patent ICAs in the patients selected for the study. According to this observer, 53 ICAs had an angiographically documented 70% to 99% stenosis when measured by the NASCET method. The kappa value for observers 1 and 2 with respect to classifying the patent ICAs as greater than or equal to 70% stenosis was 0.93. (A kappa of 1.0 indicates perfect agreement, whereas a kappa of 0.0 would indicate complete disagreement between observers.)

The sensitivities, specificities, PPV, and NPV and overall accuracies of selected ICA PSVs, ICA EDVs, and ICA PSV/CCA PSV ratios to predict an angiographically documented 70% to 99% stenosis as measured by observer 1 are shown in Table I. The maximum overall accuracy achieved by any of the duplex variables to predict a 70% to 99% ICA stenosis is 88%. This level of accuracy was achieved with an ICA PSV of 325 cm/sec or an ICA PSV/CCA PSV ratio of 4.0 or 4.2. Of these values, the ICA PSV/CCA PSV of 4.0 provides the greatest sensitivity (91.4%) for the detection of a 70% to 99% stenosis.

Table I. Sensitivities, specificities, PPV, NPV, and accuracies of ICA PSV, ICA EDV, and the ratio of ICA PSV to CCA PSV to identify a 70% to 99% ICA angiographic stenosis

Back to Article Outline

Discussion 

A number of duplex criteria based on either PSV or EDV, or ratios of ICA to CCA velocities have been used for quantitating ICA stenosis.⁷ Among the most widely used are those developed at the University of Washington. Duplex parameters for ICA stenosis were derived from comparisons of ICA velocity waveforms to angiographic stenosis. Angiographic stenosis was calculated by comparing residual ICA diameter to estimated normal bulb diameter. In some instances radiologically visible mural calcifications were used in estimating bulb diameter.⁸ The University of Washington criteria use an ICA PSV of 125 cm/sec as the cutoff for a 50% or greater ICA stenosis.¹, ⁶ If the ICA EDV exceeds 140 cm/sec, the stenosis is judged 80% or greater (94% sensitivity, 70% specificity).⁹

Unfortunately, these criteria, and others based on estimated bulb diameters, are not directly applicable to NASCET, in which ICA angiographic percent stenosis was determined with use of the distal cervical ICA as the reference vessel. To date, NASCET has reported benefit for carotid endarterectomy over medical therapy only for patients whose symptomatic ICA stenosis exceeds 70%. In addition, both the Veterans Affairs Cooperative Trial and the Asymptomatic Carotid Artery Stenosis Study Group (ACAS) trial of surgical versus no surgical therapy for asymptomatic carotid stenosis have determined carotid stenosis based on the diameter of the distal cervical ICA.¹⁰, ¹¹

The obvious importance of duplex scanning in the noninvasive evaluation of carotid stenosis, together with the inapplicability of widely used duplex criteria for ICA stenosis developed from estimated bulb diameters to the major North American trials of carotid surgery, has led us to review the experience in our vascular laboratory to determine which velocity waveform parameters best predict a 70% to 99% ICA stenosis based on a lesion-to-distal ICA diameter calculation. Given the results of the NASCET study presented to date, such information may be relevant and of immediate use in the care of patients with carotid artery occlusive disease. Our data suggest a 70% to 99% stenosis determined by NASCET criteria is best predicted by an ICA PSV/CCA PSV ratio of 4.0 (91% sensitivity, 87% specificity, 88% overall accuracy, Table I).

Clearly no consensus exists as to the optimal method of measuring stenosis of the proximal ICA. The problem can be attributed to disagreement concerning the best method of dealing with the ICA bulb. Because atherosclerosis of the carotid bifurcation preferentially involves the bulb, use of the diameter of the undiseased bulb as the basis for determining proximal ICA stenosis gives a more realistic description of the anatomic extent of proximal ICA atherosclerosis. Such measurements determined angiographically are, however, complicated by the fact that the exact dimensions of the undiseased bulb must be estimated. The arterial wall itself is seldom discernable with conventional angiographic techniques. The use of the distal cervical ICA as the reference vessel for calculating proximal ICA stenosis obviates uncertainty as to the exact angiographic dimensions of the reference vessel. The distal cervical ICA is seldom involved with atherosclerosis, and the luminal diameter is easily measured from standard angiograms. Estimates of angiographic proximal ICA stenosis based on the distal ICA are therefore potentially more precise than those based on bulb diameter.

Calculations of diameter and area reductions based on distal ICA diameter will, however, tend to significantly underestimate carotid bifurcation atherosclerosis, especially with more moderate lesions. For example, as shown in Table II, a proximal ICA stenosis with a 0.6 cm diameter residual lumen, a normal bulb diameter of 1.2 cm, and a distal ICA diameter of 0.6 cm has, with the bulb as the reference site, a 50% diameter reducing lesion and a 81% area reducing lesion. The same lesion referenced to the distal cervical ICA would have 0% diameter and area reductions.

Table II. Calculations of ICA diameter and area reductions with either normal bulb diameter or diameter of distal cervical ICA as reference vessel

D, Diameter in centimeters; A, area in square centimeters.

When stenosis forms in the bulb to the point that the proximal ICA becomes significantly narrower than the distal cervical ICA, differences between measurement techniques become less pronounced. From Table II it can be seen that when the residual proximal ICA lesion narrows to 0.1 cm, diameter and area reductions are 92% and 99% on the basis of normal bulb measurements and 83% and 97% on the basis of distal cervical ICA measurements (Table II).

One may question whether it is realistic, or even medically correct, to attempt to precisely differentiate fine grades of carotid stenosis. Variations in intraobserver and intraobserver interpretations of angiographic data are well known.¹², ¹³ Even in the current study, the ability of two independent observers to independently categorize an angiographic ICA stenosis as simply less than 70% or 70% to 99% was good, but not perfect (kappa: 0.93).

It is interesting to note the European Carotid Surgery Trial also concluded that endarterectomy was superior to medical therapy alone for patients with 70% to 99% symptomatic ICA stenosis.¹⁴ Internal carotid artery stenosis in this trial, however, was determined by comparing maximal ICA stenosis with estimated bulb diameter.¹⁵

From the discussion herein it is obvious that lesions marginally greater than 70% in the European trial would likely be measured as somewhat less than 70% with the NASCET method of calculating ICA stenosis. It is therefore reasonable to conclude that although carotid endarterectomy has now been proved efficacious for symptomatic high-grade carotid stenosis, a precise cutoff and method for determining the minimal degree of ICA stenosis best treated by endarterectomy have yet to be determined. Clinicians caring for patients with carotid artery disease should be aware that the optimal duplex criteria used for the noninvasive estimation of the extent of ICA stenosis may vary. Such criteria depend not only on the method used for calculating angiographic control data, but also significantly on the methods used in the controlled trials on which practicing physicians will undoubtedly base a significant part of their clinical decision making.

Back to Article Outline

Discussion 

Dr. Brian Thiele (Hershey, Pa.). It has been assumed by many that we all know what categories of lesions were being identified in the large trials that the authors have already mentioned. But they have correctly identified that there is a dissociation between the criteria developed at the University of Washington in their arteriographic interpretation of the degree of stenosis and those used by the multicenter studies mentioned. This issue has, in fact, been addressed in part by the Reporting Committee on Standards for Noninvasive Testing under the supervision and authority of Dr. Rutherford and really involves three issues.

Should stenosis in the carotid arteries be reported as diameter or area reduction? If stenosis in fact is to be reported as diameter reduction, what should the denominator in the equation be? What categories of stenosis should be reported?

As published in the JOURNAL OF VASCULAR SURGERY in this past year, the Committee has recommended that percent diameter stenosis be the criterion used and the distal ICA be used as the denominator.

Both of the asymptomatic carotid artery trials—one finished, results not available—one still underway, used 60% diameter reduction as the entry criterion, and in the case of the ACAS, this particular criterion necessitated the development of careful quality control mechanisms to define the individual participating laboratories' duplex criteria for identifying these lesions. Not surprisingly, this resulted in the adoption of frequency or velocity criteria specific for each institution rather than the use of any single criterion for all institutions.

Therefore although the authors have addressed one real source of error in the adoption of Doppler-derived criteria, that of the arteriographic correlation, it is also important to recognize that instrument variability may also preclude the slavish adoption of velocity criteria, and it should therefore be the responsibility of individual laboratories to confirm that the diagnostic criteria that have been presented today are also suitable for use in their institutions.

It should also be recognized by the membership that the quality control for both the arteriographic studies and the duplex studies in the NASCET are not nearly as rigid as those used in the ACAS, nor in the particular study reported today, and may therefore contribute to some confusion in the future, particularly when more data are reported. I therefore urge the membership to carefully review this aspect of the study and subsequent results.

Dr. Moneta and colleagues are to be commended for the careful conduct of this study, including their attention to the recognition of the variability of arteriographic reporting, as well as that encountered during their duplex studies. They are suggesting that we should use a velocity ratio of 4.0 to optimize the statistical correlation as opposed to a single value obtained from the ICA, which the University of Washington group has popularized.

Do you have information regarding the variability involved in obtaining two particular measurements, that is, one from the CCA and one from the ICA in the first place?

Although I believe it is important to be precise, it is also important to recognize the limits of precision, and I also ask whether it is really realistic to use a single cut point of 4 because the ROC showed that if one used values between 3.6 and 4.2, the results were almost identical. This would take into consideration some of the inherent variability of the tests.

The next question relates to the technique used in obtaining the raw PSV information. Can you tell us what the sample volume was from the CCA and ICA because this particular factor may also be responsible for some variability in the velocity data.

Finally, the use of PSV criteria, including ratios, demands that aliasing not be a factor in preocclusive lesions. In this setting, EDV has been used as a secondary diagnostic criterion, and I ask, was aliasing a problem in any of your studies, and do you use EDV criteria in any of your studies?

Dr. Gregory Moneta. Clearly variability of the measurements can be a problem. I think that the criteria proposed today are best used as an asterisk, an addenum if you will, to the criteria commonly used. I would not abandon the criteria that have been used successfully for many years by us and others. We merely suggest reports be amended with a statement as to the likelihood of having a 70% or greater NASCET lesion on the basis of the ratios that we have proposed.

We use a sample volume of 1.5 mm for both the ICA and the CCA. This is the smallest sample volume I believe we can use on our machines.

Aliasing as a technical problem in duplex scanning and as a reason to use EDV criteria is, less of a factor currently, Emphasis on EDV was developed when duplex scanning machines were not as good at recording very high PSVs. We currently have little trouble with aliasing with modern technology.

We do use EDV criteria as our primary determinant of very high-grade, 80% lesions. I think it is important to keep in mind that much of the work emanating from the University of Washington and other institutions on the natural history of ICA disease is based on duplex studies and not on angiograms. My current personal practice is to operate on asymptomatic high-grade lesions in good risk patients when the EDV exceeds 140 cm/sec.

Dr. Robert Hobson (Newark, N.J.). This study has implications for those of us involved in clinical trials regarding the efficacy of carotid endarterectomy in patients with asymptomatic carotid stenosis.

As you know, the threshold for inclusion in the VA Asymptomatic Trial was a 50% stenosis and in the ACAS trial is a 60% stenosis as determined arteriographically by diameter reduction, and yet each of these trials couple these determinations of percentage stenosis with positive non-invasive studies, ocular pneumoplethysmography, and duplex scanning. My question relates to your assessment of these two asymptomatic trials in view of your presentation. The combination of a lesser stenotic luminal diameter reduction on arteriography (50% or 60%) and a positive ocular pneumoplethysmography or duplex scan suggests a lesion that, in fact, is comparable to the University of Washington's 80% stenosis. In reviewing natural history data based on prior determination of percent stenosis by duplex criteria, is it not important to emphasize that the threshold lesion includes the 50% to 60% luminal stenosis when accompanied by the duplex scan data?

Dr. Moneta. This is an important issue. First of all, the concept of microstratification of carotid artery stenosis is a difficult one for me. Unfortunately, I think that we are going to have to live with it for the foreseeable future. Many physicians believe we can subdivide carotid stenoses. However, most surgeons realize you cannot really tell the difference between a 70% or 80% stenosis from an angiogram. My approach is that if there is a disparity between the duplex and angiographic data, and if the accuracy of the duplex studies is well documented, it is reasonable to use noninvasive data and hemodynamic data rather than the angiogram for assessing the severity of the stenosis.

Dr. John Ricotta (Buffalo, N.Y.). The criteria that you develop are going to depend on the prevalence of disease, which, you might expect, would be different in a symptomatic and an asymptomatic population. We found when we started with the ACAS that our velocity criteria were different in part because we were dealing with patients who are asymptomatic. Were all the patients in your study symptomatic or did they have angiograms for symptomatic disease? And, would you comment on whether you think that these criteria need to be developed differently for different clinical categories of patients.

Dr. Moneta. Most of the patients were symptomatic. Our series of patients undergoing carotid endarterectomy, as does most everyone else's series, includes approximately 30% with asymptomatic disease. Whether or not we have to have different categories for asymptomatic or symptomatic, I do not think that that is an issue. However, the issue of the PPV of a noninvasive test relative to the prevalence of the disease in the population that one is wishing to study, is a real one, and clearly patients in our study are selected in the sense that they were having an angiogram and were going to have an operation. They obviously are going to have a greater burden of disease than the patient who just walks in off the street.

So we may find that PPV of these tests is different in a group of patients with less prevalence of disease. However, I believe it will prove to work well in patients with symptomatic disease, most of whom are likely to have more severe stenoses.

Dr. Maria Hunink (Boston, Mass.). In studies such as these, a common problem is that of selection for verification. That is, positive duplex studies or Doppler studies will tend to undergo the gold standard test, the angiogram, whereas the native studies will tend not to undergo the angiogram and, as has been described in the radiologic literature, this can introduce a bias as regards your sensitivity and specificity and the ROC curve analysis. Will you comment on that?

The second issue that I bring up is that as criteria for choosing the cut-off point on your ROC curve you use the highest accuracy. This assumes that you value the false-positive equal to the false-negative outcomes, and, as is probably the case in this situation, that is not true. Depending on your clinical algorithm, your false-positive outcome may be far worse than your false-negative outcome. Which operating point on the ROC curve you choose will depend on the cost and benefits of the various outcomes. I would like to hear your comments on these two issues.

Dr. Moneta. Clearly we identified many lesions before angiography with the duplex scanner. This, like any retrospective study, is therefore hypothesis seeking. This hypothesis should then be evaluated prospectively.

The position of where you want to be on an ROC curve is a dilemma. It depends on what your assessment is of false positives and false negatives. If you practice in an institution where angiography has many complications, then false-positive examination outcomes are bad. You would choose a place on the ROC curve to minimize false positive outcomes. If, however, your institution provides you with very few angiographic problems, and the disease, in your mind, is such that you would rather do a few negative angiograms than miss the lesions, then you would choose a position on the curve that perhaps has more false-positive results but fewer false-negative results. There is no way to avoid this. I think we have presented data for a wide range from the ROC curves. Ideally, individuals will look at the data and try to decide where on the curve they would like to be.

Dr. Sergio Salles Cunha (Burbank, Calif.). A report has been published in the Journal of Vascular Technology that shows that different instruments, even different transducers, give different values for the same velocity. The error can be as high as 40%. The data that you presented suggest that the instrument used in this study overestimates the absolute velocity; so I think that the major contribution of the paper is that we should pay attention to velocity ratios. Would you comment on that?

Dr. Moneta. There are problems with either one, obviously. The PSVs are easy to record.

Ratios obviate the problems of compensatory flow when the other carotid artery is occluded. Therefore, in addition to machine variability, the use of ratios is theoretically attractive.

Dr. Robert Barnes (Little Rock, Ark.). I think your report also demonstrates an interobserver agreement among your radiologists, which I do not think can be generalized in this country right now. In our own state, when we looked at small area analysis of the reasons for variance of carotid endarterectomy, one contributing factor was a difference between various radiologists in the accuracy of their reports of stenoses of the carotid arteries. On average, the original report on the arteriogram overestimated carotid stenosis by an average of 25%, compared with a rereading by several observers with good interobserver agreement. I believe we are seeing this diagnostic disagreement adversely affect NASCET. Because symptomatic patients with carotid stenoses of 70% or greater are candidates for surgery, the prevalence of randomized patients with stenoses between 60% and 69% has dropped by more than 50% in just the past year. This so-called “diagnosis creep” of placing patients into a higher diagnostic category of carotid stenosis than is truly documented by arteriograms is of concern nationally as these randomized trials are being reported.

Could you tell us just very briefly how you were able to get two radiologists to agree on criteria so well that you achieved interobserver agreement with a kappa statistic of 0.93, which is almost perfection? This far surpasses the agreement that I have encountered among radiologists in my experience.

Dr. Moneta. First of all, we did not use radiologists. Vascular surgeons did readings of the angiograms.

What you have pointed out highlights the difference between sitting down in the angiogram suite at the end of the day, looking at the films on the board, and proclaiming the degree of stenosis, versus using loupes and calipers and classifying the stenosis into a single level, yes or no. In that type of situation it is not too surprising that you can do pretty well. If you try to further stratify lesions, there would probably be much more disagreement.

Back to Article Outline

References 

1. Roederer GO, Langlois YE, Chan ATW, et al.  Ultrasonic duplex scanning of the extracranial carotid arteries: improved accuracy using new features from the common carotid artery. J Cardiovasc Ultrasonography. 1982;1:373–380
   • View In Article
2. Roederer GO, Langolis YE, Jager KA, et al.  The natural history of carotid arterial disease in asymptomatic patients with cervical bruits. Stroke. 1984;15:605–613
   • View In Article
   • MEDLINE
   • CrossRef
3. Moneta GL, Taylor DC, Nicholls SC, et al.  Operative versus nonoperative management of asymptomatic high-grade internal carotid artery stenosis: improved results with endarterectomy. Stroke. 1987;18:1005–1010
   • View In Article
   • MEDLINE
   • CrossRef
4. North American Symptomatic Carotid Endarterectomy Trial Collaborators . Beneficial effect of carotid endarterectomy in symptomatic patients with high-grade carotid stenosis. N Engl J Med. 1991;325:445–453
   • View In Article
   • MEDLINE
   • CrossRef
5. North American Symptomatic Carotid Endarterectomy Trial NASCET Steering Committee . North American Symptomatic Carotid Endarterectomy Trial: methods, patient characteristics, and progress. Stroke. 1991;22:711–720
   • View In Article
   • MEDLINE
   • CrossRef
6. Strandness DE. Extracranial arterial disease. In:  Strandness DE editors. Duplex scanning in vascular disorders. New York: Raven Press; 1990;p. 92–120
   • View In Article
7. Zweibel WJ. Spectrum analysis in carotid sonography. Ultrasound Med Biol. 1987;13:625–636
   • View In Article
   • Abstract
   • Full-Text PDF (1032 KB)
   • CrossRef
8. Chikos PM, Fisher LD, Hirsch JH, et al.  Observer variability in evaluating extracranial carotid artery stenosis. Stroke. 1983;14:885–892
   • View In Article
   • MEDLINE
   • CrossRef
9. Roederer GO, Langlois Y, Jager KA, et al.  A simple spectral parameter for accurate classification of severe carotid artery disease. Bruit. 1984;8:174–178
   • View In Article
10. Veterans Administration Cooperative Trialists . Role of carotid endarterectomy in asymptomatic carotid stenosis. Stroke. 1986;17:534–538
    • View In Article
    • MEDLINE
    • CrossRef
11. Asymptomatic Carotid Atherosclerosis Study Group . Study design for randomized prospective trial of carotid endarterectomy of asymptomatic atherosclerosis. Stroke. 1989;20:844–849
    • View In Article
    • MEDLINE
    • CrossRef
12. Slot HB, Strijbosch L, Greep JM. Intraobserver variability in single plane angiography. Surgery. 1981;90:497–503
    • View In Article
    • MEDLINE
13. Thiele BL, Strandness DE. Accuracy of angiographic quantification of peripheral atherosclerosis. Prog Cardiovasc Dis. 1983;26:223–236
    • View In Article
    • MEDLINE
    • CrossRef
14. European Carotid Surgery Trialists' Collaborative Group . MRC European Carotid Surgery Trial: interim results for symptomatic patients with severe (70-99%) or with mild (0-29%) carotid stenosis. Lancet. 1991;337:1235–1243
    • View In Article
    • Abstract
    • CrossRef
15. Hankey GJ, Warlow CP. Symptomatic carotid ischaemic events: safest and most cost effective way of selecting patients for angiography, before carotid endarterectomy. BMJ. 1990;300:1485–1491
    • View In Article
    • MEDLINE
    • CrossRef