In teaching and practicing the art of physical diagnosis, we must continually ask: What parts of the physical examination should we keep? And what parts should we put on the shelf ? Some time-honored elements of the periodic physical examination, such as auscultation of the lungs or measurement of temperature, are of little diagnostic value when used for screening asymptomatic patients.1, 2 Other elements of the physical examination also are of little value when we apply them imperfectly—for example, when we do not identify a cardiac murmur accurately or fail to detect an enlarged spleen.3, 4 Time spent teaching or using skills of little value is time we do not spend teaching, improving, or practicing skills of greater value. In this era of corporate medicine, shortened patient visits, and competing demands, our time is precious.
Auscultation of the carotid arteries has conventionally been part of the physical examination, especially for older patients because of their increased risk for cerebrovascular disease. Should we listen to their carotids routinely?
Routine auscultation of the carotids has a compelling rationale. In the asymptomatic patient, a bruit may indicate occult carotid artery stenosis that can be repaired surgically before it causes stroke, thus preventing the unsuspecting patient from death or crippling stroke. This rationale is now supported by evidence from an excellent clinical trial, the Asymptomatic Carotid Artherosclerosis Study (ACAS), which found that endarterectomy reduced the risk over 5 years for ipsilateral stroke, perioperative stroke, or death from 11.0% to 5.1%.5 Nonetheless, despite the allure of finding a carotid bruit, some authorities have weighed in against screening for carotid disease in asymptomatic patients.6–8
In this issue of JGIM, Shorr et al. contribute to our thinking about the value of searching for asymptomatic carotid bruits.9 They used data from the landmark Systolic Hypertension in the Elderly Program to study 4,442 community-dwelling persons who were aged 60 years or older and had no history of stroke, transient ischemic attack, or myocardial infarction. The annual rates of stroke were 1.86% in persons who had carotid bruits noted on enrollment and 1.21% in those without carotid bruits. Therefore, the absolute risk of stroke was 0.61% higher per year in those with carotid bruits compared with those without carotid bruits, and the relative risk was 1.53 (95% confidence interval [CI] 0.98, 2.40). Also, two subgroups of patients, those with unilateral bruit and those aged 60 to 69 years, had even higher absolute risks (approximately 1% per year) and relative risks (approximately 2). Although these results did not quite reach the conventional standards of statistical significance, they are consistent with earlier studies. Prospective, population-based studies in Framingham, Massachusetts, and Evans County, Georgia, found higher risks of stroke in asymptomatic persons with carotid bruit,10, 11 and the risk associated with carotid bruit may decrease with age, especially in persons older than 70 years.12
Nonetheless, Shorr et al. concluded that carotid bruit was not a useful indicator of increased stroke risk in their study population. Several arguments support this conclusion. First, many strokes in patients with a unilateral carotid bruit occurred on the contralateral side and thus were unlikely to be related directly to the lesion causing the bruit. Previous epidemiologic studies have reported similar findings.10, 11 These findings are consistent with other findings that the presence of a bruit does not accurately reflect the presence of significant stenosis and that the absence of a bruit does not rule out carotid disease.8 Second, the association between carotid bruit and stroke is confounded by other factors. Figure 1 in the article by Shorr et al. provides a classic demonstration of confounding 13: the relative risk of stroke in patients with carotid bruit fell when underlying risk of stroke was taken into account in a stratified analysis. Finally, the absolute risk of stroke associated with carotid bruit was small in their patients, and it was not statistically significant in most analyses. Thus, only a slight benefit could be gained by repairing carotid stenoses that were found in asymptomatic patients because carotid bruits were detected.
We agree with Shorr et al. Moreover, the conclusion that time should not be wasted listening for carotid bruits most likely is generalizable to other asymptomatic patients in the broader population. A recent cost-effectiveness analysis used the results of the ACAS, which provides the most favorable information to date on the potential benefits of screening for asymptomatic carotid disease.7 This analysis found that ultrasonographic screening of 65-year-old men would on average extend their quality-adjusted life span by approximately 5 days. The costs associated with this strategy were so high that routine screening was estimated to cost $120,000 per quality-adjusted life-year, which is substantially more than is usually considered acceptable. This conclusion would not be changed substantially by a less expensive screening method, such as auscultation, even if it were as accurate as ultrasonography.
In summary, given current data, the time limits of clinical practice, and the greater importance of other risk factors such as hypertension and smoking, we cannot recommend routine carotid auscultation in asymptomatic patients. Rather, we should concentrate our efforts on what we have learned about the efficacy and appropriateness of other medical and surgical interventions to decrease the risk of stroke.14 It is now well established that aspirin and other platelet inhibitors decrease stroke rates modestly in high-risk patients, that warfarin prevents most strokes in persons over age 60 who have atrial fibrillation (and that warfarin is more effective than aspirin), that antihypertensive drugs lower stroke rates even in older adults with isolated systolic hypertension, that lowering the cholesterol level with statins in patients with coronary heart disease lowers stroke risk as well, and that carotid endarterectomy reduces stroke risk in symptomatic persons with more than 70% stenosis. Until we learn more, we should make greater use of these management strategies and not listen for carotid bruits in asymptomatic patients.—Louise Aronson, MD, andC. Seth Landefeld, MD,The Division of Geriatrics and the Center on Aging, University of California San Francisco, and the San Francisco Veterans Affairs Medical Center.