Association of Intraoperative Transcranial Doppler Monitoring Variables With Stroke From Carotid Endarterectomy
To the Editor:
I read with great interest today the article by Ackerstaff et al1 regarding transcranial Doppler monitoring during carotid endarterectomy (CEA). In particular, the editorial remarks by Drs Babikian and Cantelmo that “a wider use of transcranial Doppler monitoring seems reasonable because it can possibly improve the outcome CEA”2 should be interpreted with caution. I believe that Ackerstaff et al have failed to demonstrate how transcranial Doppler monitoring can improve the stroke rate following CEA. I will not dispute the fact that the authors’ data support the use of transcranial Doppler monitoring as an equivalent means of practicing selective shunting compared with other modalities, namely, EEG monitoring. However, I believe the suggestion that improved outcomes might be obtained by using transcranial Doppler monitoring is potentially dangerous and subject to misinterpretation. The fact that one can detect emboli does not necessarily lead to the ability to prevent them or their consequences. The only remark that I noted to have any potential use as far as prevention of complications relates to early distal clamping of the internal carotid artery in the face of emboli detected during the dissection. This must be balanced, of course, with the potential for this leading to ischemic complications if the artery is not otherwise completely prepared for the surgeons who go on to proceed with a rapid completion of endarterectomy. Moreover, the detection of emboli during wound closure as an indicator of impending occlusion is a problem that can generally be detected by routine inspection of the artery with hand-held Doppler prior to closure. This is a technique that can readily detect intimal flaps and can lead to intraoperative revision.
I do not dispute the fact that transcranial Doppler monitoring may be a good tool for detecting patients at risk for hyperperfusion syndromes. In my experience, clinically significant episodes of hyperperfusion can often be predicted on the basis of a study of the preoperative patient’s angiogram in the context of the patient’s clinical history.
The editorial reviewers2 go on to comment on the issue of making public the morbidity and mortality rates of every medical center. Again, in the United States, I find that this is an issue in which the application of such information may potentially backfire. For instance, if a young, well-trained surgeon, fresh from training, experiences a stroke on his third endarterectomy out of training, should that surgeon’s privileges for performing this procedure be suspended? I think that this is slippery slope, given the fact that we still do not know what would be an adequate N on which to evaluate a single surgeon’s operative outcomes. I would therefore strongly urge professional associations to spearhead maneuvers to continue to educate their members who perform this operation in proven means of reduction the risk of complications.
References
- 1 Ackerstaff RGA, Moons KGM, van de Vlasakker CJW, Moll FL, Vermeulen FEE, Algra A, Spencer MP. Association of intraoperative transcranial Doppler monitoring variables with stroke from carotid endarterectomy. Stroke..2000; 31:1817-1823.CrossrefMedlineGoogle Scholar
- 2 Babikian VL, Cantelmo NL. Cerebrovascular monitoring during carotid endarterectomy. Stroke..2000; 31:1799-1801. Editorial.CrossrefMedlineGoogle Scholar
Recently, we have illustrated that information provided by transcranial Doppler (TCD) monitoring during carotid endarterectomy (CEA) independently predicts postoperative stroke.R1 In their editorial, Babikian and CantelmoR2 welcomed our results but also commented that our study was not randomized and that the efficacy of TCD monitoring on patient outcome cannot yet be considered established. In his letter, Dr Rabb has raised the same concern. Here, we contemplate whether a randomized study is truly necessary to infer the usefulness of TCD during CEA.
In the present era of evidence-based medicine, a medical technology must have been proved to be effective (improve patient outcome) or more efficient (contain costs). This also applies to diagnostic and monitoring tests.R3 Various authors have proposed that, as with therapeutic strategies, medical tests also should undergo consecutive phases of evaluation before implementation in practice.R4R5R6 They more or less propose that after the technical and diagnostic performance of the test (its ability to detect presence or absence of a disorder) has been evaluated, the impact on health (ability to change patient outcome) and cost-effectiveness should be quantified. Obviously, the paradigm for a valid evaluation of therapeutic strategies on patient outcome is the randomized trial. But, even in the realm of evidence based medicine, is this a necessity for diagnostic tests?
TCD monitoring can be considered a diagnostic test to detect cerebral microemboli during CEA. As most diagnostic tests, TCD itself has no impact on patient outcome. Its information directs patient management, which may alter the prognosis of the patient. Hence, especially the therapies chosen on the basis of TCD determine the (cost-)effectiveness of TCD. Diagnostic studies have shown that TCD is capable of detecting microemboli.R7R8R9R10 Therapeutic studies demonstrated that interventions administered to decrease emboli resulted in reduction of perioperative morbidity.R8R11R12
In general, we find that the beneficial effect of a diagnostic test on patient outcome may be considered established if (1) diagnostic studies have shown the test’s ability to detect a particular disorder and (2) therapeutic studies have provided evidence on efficacy of the management of this disorder. Following this reasoning and combining the available evidence of TCD monitoring and emboli treatment, there may be enough evidence that TCD monitoring during CEA improves patient outcome.
In the case of TCD, the limiting factor may be the strength of the evidence from the therapeutic studies: none was randomized. A classic trial would pragmatically randomize patients eligible for CEA to surgery with or without TCD monitoring. A difference in postoperative outcome in favor of the TCD arm would prove the usefulness of TCD during CEA. However, such a study would require large numbers of patients, given the relatively low occurrence of stroke in these patients.R1R2 Second, such a design may be perceived to be unethical, as many surgeons already rely on the use of TCD during CEA. Finally, it would be difficult to avoid the learning effects of the surgeon. The experience with TCD monitoring may make surgeons more careful at the moments they expect emboli when no TCD is used. This learning effect would dilute the effectiveness of TCD. Given these limitations and our reasoning that many tests, including TCD, by themselves have no effect on patient outcome and the results of diagnostic and therapeutic studies ought to be combined, another trial design may be more efficient. Assuming that the value of TCD to detect emboli is established, each patient eligible for CEA will undergo surgery with the use of TCD. Only if emboli occur, patients are randomized to 2 treatment regimens (eg, dextran-40 or platelet glycoprotein IIB/IIIA receptor antagonist). Such a trial design still has the drawback of large patient numbers but lacks the second and third limitations discussed above.
