The value and economic analysis of routine postoperative carotid duplex ultrasound surveillance after carotid endarterectomy

Abstract

Background

Several studies have reported on the role of postoperative duplex ultrasound surveillance after carotid endarterectomy (CEA) with varying results. Most of these studies had a small sample size or did not analyze cost-effectiveness.

Methods

We analyzed 489 of 501 CEA patients with patch closure. All patients had immediate postoperative duplex ultrasound examination and were routinely followed up both clinically and with duplex ultrasound at regular intervals of 1 month, 6 months, 12 months, and every 12 months thereafter. A Kaplan-Meier analysis was used to estimate the rate of ≥50% and ≥80% post-CEA restenosis over time and the time frame of progression from normal to ≥50% or ≥80% restenosis. The cost of post-CEA duplex surveillance was also estimated.

Results

Overall, 489 patients with a mean age of 68.5 years were analyzed. Ten of these had residual postoperative ≥50% stenosis, and 37 did not undergo a second duplex ultrasound examination and therefore were not included in the final analysis. The mean follow-up was 20.4 months (range, 1-63 months), with a mean number of duplex ultrasound examinations of 3.6 (range, 1-7). Eleven of 397 patients (2.8%) with a normal finding on immediate postoperative duplex ultrasound vs 4 of 45 (8.9%) with mild stenosis on immediate postoperative duplex ultrasound progressed to ≥50% restenosis (P = .055). Overall, 15 patients (3.1%) had ≥50% restenosis, 9 with 50% to <80% and 4 with 80% to 99% (2 of these had carotid artery stenting reintervention), and 2 had late carotid occlusion. All of these were asymptomatic, except for one who had a transient ischemic attack. The mean time to ≥50% to <80% restenosis was 14.7 months vs 19.8 months for ≥80% restenosis after the CEA. Freedom from restenosis rates were 98%, 96%, 94%, 94%, and 94% for ≥50% restenosis and 99%, 98%, 97%, 97%, and 97% for ≥80% restenosis at 1 year, 2 years, 3 years, 4 years, and 5 years, respectively. Freedom from myocardial infarction, stroke, and deaths was not significantly different between patients with and without restenosis (100%, 93%, 83%, and 83% vs 94%, 91%, 86%, and 79% at 1 year, 2 years, 3 years, and 4 years, respectively; P = .951). The estimated charge of this surveillance was 3.6 × 489 (number of CEAs) × $800 (charge for carotid duplex ultrasound), which equals $1,408,320, to detect only four patients with ≥80% to 99% restenosis who may have been potential candidates for reintervention.

Conclusions

This study shows that the value of routine postoperative duplex ultrasound surveillance after CEA with patch closure may be limited, particularly if the finding on immediate postoperative duplex ultrasound is normal or shows minimal disease.

A few nonrandomized studies have reported mixed results regarding the timing and value of postoperative carotid duplex ultrasound (CDUS) surveillance after carotid endarterectomy (CEA).^1-5 Advancements in medical treatment and the current emphasis on cost-effectiveness have questioned the value of performing routine carotid ultrasound scanning after CEA.

This study is an attempt to address the evolving clinical dilemma concerning the utilization of CDUS for surveillance after CEA with patch closure. The argument will continue on the clinical equipoise of using such a diagnostic tool routinely, especially with the current constraints from the Accountable Care Act. The main controversies revolve around the natural history of asymptomatic carotid stenosis in light of advanced medical treatment and the need, if any, for post-CEA CDUS surveillance at appropriate intervals. This study also analyzed the economic implication of such surveillance.

METHODS

This is a retrospective analysis of prospectively collected data of all patients who unde1went CEA during a recent period (September 2008-0ctober 2011) by six full-time board-certified academic vascular surgeons of the Vascular Center of Excellence at Charleston Area Medical Center/West Virginia University, Charleston, West Virginia. The study was approved by the Institutional Review Board,and informed consent of the patients was not necessary. Demographics and clinical characteristics of the patients were collected, including age, race, gender, and other cardiovascular comorbidities (hypertension, diabetes mellitus, coronary artery disease, hyperlipidemia, smoking, and chronic renal insufficiency). All data were collected using electronic medical records and all progress notes at the Vascular Center of Excellence during follow-up. CEA indications were classified into symptomatic (transient ischemic attack [TIA] or stroke) or asymptomatic (carotid bruit and nonhemispheric TIA). All patients underwent preoperative duplex scanning of both extracranial carotid arteries in our accredited vascular laboratory (Intersocietal Commission for the Accreditation of Vascular Laboratories) by registered vascular technologists. Patients with combined CEA and coronary artery bypass grafts, repeated CEA, and complex brachiocephalic reconstruction with CEA were excluded.

All procedures were performed under general anesthesia with intravenous systemic heparin and routine shunting and patching. All patients were administered aspirin therapy (325 mg or 81 mg daily) within a few days before the operation, and it was continued postoperatively indefinitely, except in patients with a history of intolerance, who were given clopidogrel 75 mg daily after CEA.

Post-CEA surveillance protocol

All patients underwent immediate postoperative duplex ultrasound (within 24 hours of CEA while in the hospital), which was repeated at 1 month, 6 months, 12 months, and every 12 months thereafter with an iU22 System (Philips Healthcare, Bothell, Wash). The color duplex scanning was used to assess the presence of postoperative residual stenosis or recurrent restenosis at a later follow-up. Restenosis was considered to be present only if the abnormality detected with duplex ultrasound was not detected on the first immediate post operative carotid ultrasound examination. We used our previously validated CDUS criteria to define ≥50% and ≥80% post-CEA restenosis (peak systolic velocity of ≥274 cm/s or end-diastolic velocity of 100 cm/s).⁶

The early (30-day postoperative) and late complications were recorded and analyzed. The primary end points were early and late ipsilateral strokes, myocardial infarction (MI), and deaths and the incidence of ≥50% and ≥80% post-CEA restenosis. In the event of a new neurologic event (TIA or stroke), patients were assessed by a neurologist, and all strokes were confirmed with computed tomography scans or magnetic resonance imaging. An MI was defined as an abnormal electrocardiogram associated with chest pain or elevated plasma troponin with a value above the 99th perceentile of the upper reference limit.⁷ The Social Security Death Index was also used to verify all deaths.

Economic analysis of post-CEA duplex surveillance

Hospital charges for CDUS were captured during this period to estimate the charges for long-term postoperative CDUS surveillance. The mean number of postoperative duplex ultrasound scans was used to calculate the mean cost (charge) for long-term surveillance per patient. Meanwhile, a similar calculation was made using the actual global reimbursement to the hospital, presuming a payer mix of practice consisting of 60% Medicare patients, 30% other insurance, 5% Medicaid, and 5% that were self-insured or with no insurance. The estimated reimbursement using this calculation was $399 × the mean number of ultrasound scans × the number of CEAs.

Statistical analysis

A statistical analysis was performed using SAS 9.2 software. Categorical variables were compared by contingency table analysis with χ² or Fisher exact test to determine statistical significance. A Kaplan-Meier analysis was used to estimate the rate of ≥50% and ≥80% post-CEA restenosis and the time frame of progression from normal to ≥50% or ≥80% post-CEA restenosis. Statistical comparisons were made with the log-rank test.

RESULTS

Of 501 CEAs performed during this period, 12 patients were discharged without having an immediate postoperative CDUS because of technical reasons or because the patient wanted to leave early; therefore, 489 patients were analyzed. The mean age was 68.5 years (range, 43-93 years). The Table summarizes the demographic and clinical characteristics of the group. The indications for CEA were 38% for symptomatic and 62% for asymptomatic indications. The perioperative stroke rate was 1.6% (8 of 489), the MI rate was 2.3% (11 of 489), the death rate was 1.4% (7 of 489), and the combined perioperative stroke and death rate was 2.7% (13 of 489). Of these 489 patients, 10 (2%) had residual postoperative stenosis and 37 did not undergo a second duplex ultra-sound examination and therefore were not included in the final analysis. Of the 10 patients with residual disease, one had perioperative carotid occlusion (asymptomatic) and two had ≥80 stenosis (which were thought to be higher carotid lesions, one of which was treated with stenting and the other by reoperation; both were done during the same hospitalization). The remaining seven had ≥50 to <80% stenosis (one of which progressed to 80% to 99% stenosis, which was treated later with stenting). The mean follow-up of these seven patients was 22.1 months (0-48 months).

Table.

Demographics and clinical characteristics

	No. (%) (N = 489)
Gender, male	258 (53)
Hypertension	404 (83)
Coronary artery disease	197 (40)
Hyperlipidemia	336 (69)
Diabetes mellitus	186 (38)
Chronic renal insufficiency	53 (11)
History of stroke	73 (15)
Chronic obstructive pulmonary disease	106 (22)
Current smoking	167 (34)
Previous smoking	114 (23)
All smokers	281 (57)

Late follow-up and post-CEA restenosis

We observed 479 patients (excluding 10 residual patients) at a mean follow-up of 20.4 months (range, 1-63 months). The mean number of postoperative duplex ultrasound scans was 3.6 (range, 1-7). The mean time to post-CEA restenosis was 14.7 months for ≥50% to <80% restenosis (range, 1-35 months), with a mean of 19.8 months (range, 1-56 months) for ≥80% restenosis (P= .5236). The mean time for restenosis for the whole series was 19.3 months (range, 1-56 months).

Overall, 15 patients of 442 (3.4%; these are of 479 patients who had a second duplex ultrasound examination or more on follow-up) had ≥50% restenosis: 9 (2%) with 50% to <80% restenosis and 6 (1.4%) with ≥80% to 99% restenosis (2 of these were late carotid occlusions). If we include the 10 patients with residual stenosis, 9 of 448 (2%) had ≥50% stenosis and 7 of 448 (1.6%) had ≥80% restenosis. In regard to symptomatic vs asymptomatic patients, overall, 9 of 170 symptomatic patients (5.3%) progressed to ≥50% restenosis vs 6 of 272 (2.2%) asymptomatic patients (nonsignificant, P = .08), and only 3 of 9 stenoses of ≥50% in the symptomatic category were ≥80% restenoses. The mean age at the time of CEA for patients with restenosis was 71.9 years (range, 53-86 years) vs 68.4 years for patients without restenosis (range, 43-98 years; P = .178). There were no statistically significant differences between patients with normal findings and patients with mild stenosis on immediate postoperative duplex ultrasound on the rate of ≥50% restenosis (11 of 397 [2.8%] vs 4 of 45 [8.9%]; P= .055).

Post-CEA restenosis and symptoms

All 15 patients with restenosis were asymptomatic, except for one with ≥80% restenosis who had TIA. Of the four patients with ≥80% restenosis, two underwent carotid artery stenting (including the one with TIA), and the other two were observed.

Kaplan-Meier analysis

Freedom from ≥50% and ≥80% post-CEA restenosis is seen in Fig 1, A. As noted, freedom from ≥50% restenosis rates were 98%, 96%, 95%, and 95% at 1 year, 2 years, 3 years, and 4 years, respectively; for ≥80% restenosis, rates were 99%, 99%, 98%, and 98%. Fig 1, B includes residual stenosis and freedom from restenosis rates: for ≥50% restenosis, 96%, 94%, 93%, and 93% at 1 year, 2 years, 3 years, and 4 years, respectively; and for ≥80% restenosis, 98%, 98%, 97%, and 97%. Freedom from MI, stroke, and death was not significantly different between patients with and without restenosis: 100%, 93%, 83%, and 83% vs 94%, 91%, 86%, and 79% at 1 year, 2 years, 3 years, and 4 years, respectively (Fig 2, A; P = .951). Fig 2, B includes residual stenosis and freedom from MI, stroke, and death in patients with and without restenosis: 100%, 93%, 83%, and 83% vs 94%, 91%, 86%, and 79% at 1 year, 2 years, 3 years, and 4 years, respectively (P= .973).

Figure 1.

A, Freedom from ≥50% and ≥80% restenosis rates at 1 year, 2 years, 3 years, and 4 years. B, Freedom from ≥50% and ≥80% restenosis rates (including 10 patients with residual stenosis) at 1 year, 2 years, 3 years, and 4 years. SE, Standard error.

Figure 2.

A, Freedom from myocardial infarction (MI), stroke, deaths at 1 year, 2 years, 3 years, and 4 years. B, Freedom from MI, stroke, and deaths (including 10 patients with residual stenosis) at I year, 2 years, 3 years, and 4 years. SE, Standard error.

Economic analysis of post-CEA surveillance

The overall hospital charge of this surveillance was 3.6 multiplied by $800 multiplied by 489, for a total of $1,408,320, and the overall reimbursement was 3.6 × 339 × 489 = $702,400 to detect only four patiente with ≥80% to 99% restenosis who may have been potential candidates for reintervention. To be noted, only two of these four patients underwent carotid artery stenting in our series. This analysis did not count patients with residual stenosis on immediate postoperatie duplex ultrasound imaging.

DISCUSSION

Since its inception in early 1950, CEA has contributed to a significant decrease in the stroke rate as the third most common cause of death in the United States.^7-9 Carotid patching has also been adopted by most authorities as a valid adjunct to decrease the chance of restenosis and thus has resulted in increased primary patency after CEA.^9,10 On the other hand, during the last two and a half decades, CDUS has became an accurate noninvasive diagnostic tool to estimate the degree of post-CEA restenosis.^6,11-13 Bandyk et al¹² reported that compared with angiography, the overall accuracy of duplex scanning in predicting recurrent carotid bifurcation disease was 83% However, that review was published before that an internal angioplasty were revised.⁶ It was found that an internal carotid artery peak systolic velocity of 274 cm/s (or end diastolic velocity of ≥100 cm/s) was optimal for detecting ≥80% restenosis with sensitivity, specificity, positive predictive value, negative predictive value, and overall accuracy of 100%, 91%, 99%, 100%, and 99%, respectively. This value may not be applicable to other laboratories. With these revised velocity numbers, only 4 of 479 (0.8%) of our patients were found to have 80% to 99% post-CEA restenosis on CDUS without ipsilateral neurologic events. This directly reflects that restenosis is not only uncommon in occurrence but also benign in nature. The mean time to ≥80% restenosis in this study was 19 months. This raises anot11er valid concern about the appropriate timing for repeating the CDUS if the findings of the first one are normal. It is conceivable that the value of any imaging modality is fulfilled only if it contributes to a meaningful change in care of the patient. The current study did not show any benefit in performing CDUS at 6 or 12 months and suggests that if it is warranted, CDUS should be considered at an 18-month interval. This recommendation may be controversial because the most common time for restenosis secondary to intimal hyperplasia is between 6 and 18 months; therefore, some may be reluctant not to obtain a duplex ultrasound examination at 6- or 12-month periods. In addition, this study found that an early postoperative CDUS, performed within the first week or before discharge, identifies >50% residual stenosis, mild residual (<50%) stenosis, and small carotid dimensions, which are good predictors of >50% restenosis at 6 months.¹³ Still, some authors have reported that the mild spectral broadening and flow turbulence that appeared immediately after CEA usually disappeared or remained stable during the follow-up CDUS surveillance.^14-16 Aldoori and Baird¹⁶reported that 36% of the arteries with spectral broadening had returned to normal at the 6-month CDUS follow-up.

In addition to CDUS performed after CEA, the need for repeated CDUS to detect the progression of post-CEA restenosis is decreasing. Constantinou et al¹⁷ reported on the best evidence of medical therapy for carotid disease during last two decades and found that the annual ipsilateral stroke rate for patients with underlying asymptomatic carotid disease dropped from 4% to <1%. This is directly related to lifestyle modifications, better control of blood pressure, and the institution of antiplatelet and statin therapy. Other authorities have echoed this opinion.^18-20 Therefore, compliance with medical treatment after CEA is imperative. With the ongoing advances in medical therapy, along with better compliance of patients, investigators of the Carotid Revascularization Endarterectomy vs Stenting Trial (CREST)²¹ reported a low incidence of restenosis after intervention (only 13 patients [0.75%]). In our study, the rates of freedom from restenosis at 1 year, 2 years, 3 years, 4 years, and 5 years, respectively, were 98%, 96%, 94%, 94%, and 94% for ≥50% restenosis and 99%, 98%, 97%, 97%, and 97% for ≥80% restenosis. The overall incidence of ≥80% post-CEA restenosis in our sn1dy was only 0.8%. Our conclusions may not be applicable to other groups who may have a higher incidence of restenosis.

Others concluded that 80% of patients with normal site repair (by CDUS) after CEA with <50% contralateral internal carotid artery diameter reduction stenosis can be safely observed with clinical examination with CDUS every 1 to 2 years.² Golledge et al¹ recommended restriction of follow-up to those with ≥50% stenosis, as this protocol will reduce CDUS surveillance by 78%. Another review by Golledge et al indicated a very low yield of CDUS surveillance after CEA, with only 13 patients (5.9%) with severe disease warranting additional intervention.^1,3 To noted is that our current study evaluated CDUS results only after an ipsilateral CEA; therefore, patients with severe disease of the contralateral carotid should be observed appropriately and objectively, according to the current standard of care.^22-25

The present study also estimated the hospital charge of CDUS surveillance, which was 3.6 (mean number for CDUS) × 489 (number of CEAs) × $800, for a total of $1,408,320 with an overall reimbursement of 3.6 × $399 × 489 for a total of $702,400 to detect only four patients with ≥80% to 99% post-CEA restenosis that may need reintervention. The observations of our current study are consistent with a recent retrospective study by Cull et al.⁵ They reviewed a 9-year vascular surgical database to identify patients enrolled in a CDUS surveillance program for asymptomatic carotid stenosis or after CEA. The reimbursement data were evaluated to calculate the average cost of each CDUS and the cost of the CDUS surveillance program for each stroke prevented. In that study, 1l,531 CDUS scans were performed on 3003 patients who were enrolled in the CDUS surveillance program. CEA for asymptomatic carotid stenosis was performed on 225 patients (7.5%). The CDUS surveillance program prevented 13 strokes (871 CDUS examinations for each stroke prevented). The mean cost of each duplex scan was $332 ± $170. The total cost of the CDUS surveillance was $3,830,000 or $290,000 for each stroke prevented. The authors concluded that a CDUS surveillance program is costly and inefficient. They also thought that consideration should be given to eliminating routine post-CEA CDUS surveillance in the absence of contralateral disease.

The results of the current study raise concerns about the unrealistic overutilization of CDUS after CEA and underscore the ultimate need for interdisciplinary collaborative experts to work together for a quality improvement consensus to standardize routine CDUS surveillance after CEA and to improve outcomes.

There are still many inherent aspects to CDUS, such as differences in the diagnostic criteria used at different institutions to interpret CDUS, that may result in significant variations in the classification of carotid artery stenosis and may lead to differences in the number and subsequent costs of revascularizations.²⁶ It is noteworthy that although ongoing utilization of CDUS is increasing, its benefit and clinical implications are still in question. One study reported an increase in utilization by diagnostic radiology at a compound annual growth rate of 1% during 2000 to 2007. In that study, both interventional radiologists and vascular surgeons experienced higher compound annual growth rates of 3% and 6%, respectively. Utilization by cardiologists increased at a rate of 11 times that of diagnostic radiologists, with an increase in Medicare beneficiaries being tested for carotid artery stenosis, especially by specialties that perform revascularization for carotid stenosis. The health benefits of this practice are uncertain.²⁷

Our present study has some limitations; most important, being a retrospective study of existing data, it thus carries all of the inherent bias related to selecting rather than assigning patients to treatment groups. Next, despite our best efforts to standardize follow-up, some patients were still lost to follow-up and did not complete the planned CDUS surveillance at 6 months or 1 year. Finally, these data are limited to CEA patients with patching and not primary closure.

CONCLUSIONS

Our current study dissects the validity of performing CDUS after CEA in a large contemporary practice with a realistic analysis of all related outcomes and underscores this important issue in relation to current health care financial constraints. We did not find any significant value for repeating routine CDUS after CEA with patching, particularly in patients with normal findings on immediate postoperative duplex ultrasound examination. Also, our study suggests that this practice may not be cost-effective, especially if the immediate findings on postoperative CDUS are normal or there is only minimal disease.

DISCUSSION

Dr David L. Cull (Greenville, SC). Dr AbuRahma has presented a provocative study that challenges the value of duplex surveillance after carotid endarterectomy (CEA). In his personal series of nearly 500 patients over a 3-year period, only 3.1% of patients developed a stenosis >50% and only one patient developed an ipsilateral symptom. I have the following questions.

The mean time to restcnosis in your study was just over 19 months. The mean follow-up time of your study was 20 months. These data suggest that for a substantial number of patients in your study, the surveillance period may not have been long enough for a stcnosis to develop. Short follow-up may explain why carotid restenosis in your study was half that reported by the Carotid Revascularization Endarterectomy vs Stcnting Trial (CREST). How do you reconcile the difference in restenosis between your study and CREST? Had your restenosis rate been closer to the CREST restenosis rate of 6.2% at 4 years, would your conclusions regarding the value of surveillance change?

CREST also found that female gender, continued smoking, diabetes, and dyslipidemia were independent risk factors for restenosis. Were you able to identify any factors associated with restenosis that we could use to identify those patients at higher risk and thus require closer follow-up?

Although your study evaluates the value of duplex surveillance in detecting carotid restenosis, there are two other benefits of surveillance not considered by your study. Duplex can also be used to follow the status of the contralateral carotid artery. Furthermore, my senior partner, Bruce Snyder, has taught me that duplex surveillance provides a third benefit. It encourages patient follow-up. Patients prefer to leave the doctor's office with a more tangible measure of their health than simply the reassurance from the physician that they are doing OK. The duplex scan provides a result they can understand. This is what draws them back to our office so we can determine if they have had any neurologic symptoms; so we can reiterate which symptoms they should be watching for; so we can make sure they are on appropriate medications for cardiovascular risk reduction; so we can ensure they have no significant disease in other vascular territories. Considering all three benefits, can the cost of duplex surveillance be justified? Given your experience, what is your surveillance algorithm for patients following CEA?

I would like to thank the program committee for giving me the opportunity to discuss this paper.

Dr Ali F. AbuRahma. Thank you very much, David, for your kind comments.

In regard to your question as to how we reconcile the difference in restenosis between our study and CREST, in our present study and our previously published studies of several randomized prospective carotid trials, the restenosis rate after CEA with patch closure was around 5% over a mean follow-up of 3 to 4 years, which is somewhat similar to CREST. With this in mind, my recommendation in regard to the value of surveillance would not change, especially since most of these restenoses have been asymptomatic. I happen to believe that most asymptomatic lesions, particularly in the first few years, have a benign course; therefore, surveillance would not be cost-effective, as reflected in this study.

As to whether there are any specific risk factors that may put patients at higher risk for restenosis, multiple randomized prospective trials published over the past decade have shown that when the cndartcrectomy was closed using patching, the only factor that may be a contributor to restenosis was female gender. Therefore, if you are closing the endarterectomy site using primary closure or if you have a female patient, the risk of restenosis may be somewhat higher. However, if these restenoses were asymptomatic, then I would not justify the frequency of routine duplex surveillance in these patients.

In regard to the other two benefits of surveillance, I totally agree with you that duplex can be used to follow the status of the contralateral carotid artery; however, our present study did not analyze that effect. Therefore, our recommendation would not have changed, ie, the contralateral carotid artery can be followed using duplex surveillance, and its frequency depends on the severity of the contralateral stenosis.

As to the potential benefit of encouraging patients for follow-up, again, this study did not take this into account, but I do concur with you in this regard.

Again, thank you very much, David, for your comments and for discussing this paper.