US National Survey of Physician Practices for the Secondary and Tertiary Prevention of Ischemic Stroke
Background and Purpose Data from several randomized clinical trials concerning the efficacy of carotid endarterectomy (CE) in patients with symptomatic and asymptomatic stenoses of the extracranial carotid artery are now available. Yet, there are few data concerning the patterns of use of CE by physicians for their patients at risk for stroke. These data are critical for the rational allocation of resources and targeting of educational efforts.
Methods Between August 1993 and February 1994, we surveyed the stroke prevention practices of a stratified random sample of 2000 US physicians. The survey queried the perceived availability and use of diagnostic studies and surgery for specific types of patients who might be considered candidates for CE.
Results Of eligible physicians, 67% (n=1006) completed the survey. Seventy percent reported that they always or often obtain carotid ultrasonography for evaluation of patients with asymptomatic bruits; 89% do so in patients with recent transient ischemic attack or minor stroke (P<.001). For asymptomatic patients, 13% always or often obtain a cerebral angiogram if carotid ultrasonography indicates 50% to 70% stenosis versus 33% if carotid ultrasonography indicates >70% stenosis (P<.001). For asymptomatic patients with >70% stenosis, a cerebral angiogram was reported as seldom or never used by 42% of physicians who viewed the test as readily available versus 67% if cerebral angiography was perceived as not readily available (P=.005). Multinomial multiple logistic regression analysis showed that symptom status, the degree of stenosis, perceived availability of CE, and physician specialty independently contributed to the explained variance in the reported use of CE (P<.001). The odds of performing CE were approximately four times greater in patients with recent symptoms compared with asymptomatic patients (P<.001) and four times greater in patients with >70% stenosis compared with patients with 50% to 70% stenosis (P<.001). Physicians who perceived CE as not being readily available were one third as likely to report using the procedure compared with physicians who reported having ready access (P=.004). CE was reported as being always or often used by more than 80% of neurologists and surgeons but by only about half of internists and noninternist primary care physicians for patients with newly symptomatic high-grade stenosis (P<.001). Almost one in four noninternist primary care physicians responded that they would seldom or never use CE for these patients.
Conclusions These data show that (1) symptom status and degree of carotid artery stenosis strongly influence the reported frequency with which CE is used by practicing physicians; (2) the perceived availability of cerebral angiography and CE significantly affects their reported frequency of use; and (3) physician specialty significantly influences the reported frequency of use of CE.
The efficacy of CE in selected patients with TIA or nondisabling stroke ipsilateral to a high-grade stenosis of the extracranial CA has been established through three randomized clinical trials.1 2 3 A review of CEs performed in US hospitals based on the National Hospital Discharge Survey found that the rate at which the procedure was performed per 100 000 population increased substantially after these results became available, a finding consistent with an impact of these clinical trial data on medical practice.4 Clinical trial data concerning the efficacy of CE in patients with asymptomatic stenosis of the extracranial CA have also become available,5 6 7 but the effect of these data on the frequency of this surgery has not yet been assessed.
Despite the wealth of clinical information concerning the use of CE, there are few data available examining how practicing physicians evaluate their patients and use the procedure. The US National Survey of Physician Practices for the Secondary and Tertiary Prevention of Ischemic Stroke was designed to address these questions.8 An initial report was devoted to a description of the survey methodology, physician and practice characteristics, the perceived availability of basic and advanced stroke prevention services, and the reported use of basic stroke prevention strategies. Carotid ultrasonography, a key test in the diagnostic evaluation of patients being considered for CE, was reported as being readily available by at least 90% of physicians. MR angiography, a relatively new technique, was perceived as being readily available by 68% of respondents. Surprisingly, 12% of physicians reported cerebral arteriography and 10% reported CE as not being readily available. Multiple logistic regression analyses showed that the perceived availability of services varied with physician specialty (noninternist primary care, internal medicine, neurology, or surgery), practice location (nonmetropolitan, small metropolitan, or large metropolitan area), and for CE, region of the country (South, North Central, Northeast, or West). The odds of CE being reported as readily available were approximately 2.5 to 3.5 times greater for physicians practicing in the North Central, Northeast, and Western regions of the United States compared with those practicing in the South, independent of practice setting and specialty.
In addition to these general questions, the survey queried the use of diagnostic tests and CE through a series of specific patient scenarios. The present analyses were designed to determine both how practicing physicians approach the diagnosis and surgical treatment of potential CE candidates and how variables such as physician specialty, practice location, region of the country, and perceived availability of services influence these decisions.
Details of the survey methodology and analytic procedures have previously been presented.8 Briefly, the study population consisted of all physicians, excluding residents and fellows, who were practicing in the United States, actively involved in patient care, and listed in the 1993 American Medical Association’s Physician Masterfile under one of the following primary specialties: family practice, general practice, general internal medicine, neurosurgery, neurology, vascular surgery, or a group of internal medicine subspecialties (including diabetology, hematology, geriatrics, nephrology, rheumatology, and endocrinology). A stratified random sample was then chosen by selecting physicians from a list ordered by region for each specialty (each specialty formed a stratum, with the exception of the group of internal medicine subspecialties, which were combined to form a single stratum). Survey instruments were mailed to approximately equal numbers of the randomly selected physicians from each specialty (n=2000). Survey data were collected between August 1993 and February 1994.
The self-administered questionnaire included 23 items and was divided into two distinct parts. The first part of the questionnaire consisted of items that described physician and practice characteristics and the availability of selected procedures and tests. The second part of the questionnaire consisted of 12 patient scenarios. These included several vignettes that probed the reported use of diagnostic modalities and surgery in potential CE candidates (eg, patients without contraindications for CE). Physicians were asked how frequently they would obtain carotid ultrasonography or cerebral angiography in specific clinical settings and whether they would either perform CE or refer for surgery asymptomatic and symptomatic patients subsequently found to have moderate (50% to 70%) or high-grade (>70%) stenosis of the extracranial CA.
To simplify the analyses, the responses of family practice specialists and general practice physicians (noninternist primary care physicians) were combined as “primary care,” general internal medicine physicians and internal medicine subspecialists were combined as “internal medicine,” and neurosurgeons and vascular surgeons were combined as “surgery.” The physicians reported using diagnostic procedures or CE always, often, sometimes, seldom, or never in a given clinical setting. Because the categorizations do not represent a meaningful difference in clinical practice and to further simplify the analyses, the responses “always” and “often” and “seldom” and “never” were combined (“always/often” and “seldom/never”). The perceived availability of services was scored by the physicians as either being provided by the physicians themselves, readily available, or not readily available. The first two categories were combined as “readily available.” To determine whether geographic variation influenced the reported patterns of care, the data were analyzed with regard to the region of the country in which the physician practiced (South, North Central, Northeast, or West). It was also possible to determine whether a physician practiced within a large metropolitan area (more than 1 million persons), a small metropolitan area (less than 1 million persons), or a nonmetropolitan area (area not included in a metropolitan statistical area).
Nonresponse-adjusted weighted data were used in the statistical analyses.8 This was accomplished with the SUDAAN software package (Research Triangle Institute).9 SUDAAN analyses include estimates of basic statistical parameters such as frequency distribution, means, percentages, regression coefficients, test statistics (χ2, Fisher’s F, and Student’s t tests), standard errors, and multinomial multiple logistic regression analysis–based odds ratios that take into account the complex nature of the sample design when estimating variances.
Overall, 67% of eligible physicians fully responded to the survey (n=1006), either by mail or telephone.8 All physician specialties and geographic regions were well represented among the respondents.8
Overall, 89% of physicians reported that they always or often obtain a carotid ultrasound study in patients with recent (eg, within the previous month) carotid-distribution TIA or completed minor stroke. Cerebral angiography was reported as being always or often used by 19% of physicians in these newly symptomatic patients. Seventy percent of physicians responded that they always or often obtain a carotid ultrasound study in asymptomatic patients with a carotid bruit. In asymptomatic patients without surgical contraindications, 13% of physicians reported that they always or often obtain a cerebral angiogram if a carotid duplex ultrasonography indicates 50% to 70% stenosis (Fig 1⇓, top). The proportion doing so increased to 33% if the noninvasive test indicated >70% stenosis of the extracranial CA (Fig 1⇓, middle). Approximately 42% of physicians who reported that they had cerebral angiography readily available responded that they seldom or never obtain the study in asymptomatic patients with >70% CA stenosis demonstrated by ultrasonography; this proportion increased to approximately 67% if angiography was viewed as not readily available (Fig 1⇓, middle; P=.005). The perceived availability of cerebral angiography also influenced its reported use in newly symptomatic patients. Forty-two percent of physicians who viewed the test as being readily available responded that they seldom or never use it for these patients, whereas 85% of those who viewed cerebral angiography as not being readily available reported that they seldom or never refer this type of patient for the procedure (Fig 1⇓, bottom; P<.001).
The perceived availability of CE varies with physician specialty, practice location, and region of the country.8 Multinomial multiple logistic regression analysis was used to evaluate the independent contributions of these variables and the clinical characteristics of patients (eg, degree of stenosis and symptom status) on the reported frequency of use (always or often versus sometimes versus seldom or never) of CE (Table⇓). The overall model was significant (P<.001). Symptom status, the degree of stenosis, perceived availability of CE, and physician specialty independently contributed to the explained variance in the reported use of CE. The odds of performing CE were approximately four times greater in patients with recent symptoms compared with asymptomatic patients (P<.001). The odds of performing the surgery were approximately four times greater in patients with >70% stenosis compared with patients with 50% to 70% stenosis (P<.001). Overall, noninternist primary care physicians, internists, and neurologists were 50% to 60% less likely to use CE than surgeons (P<.001). There were no independent effects of practice location or region after the other variables (eg, perceived availability) were considered.
Fig 2⇓ shows the impact of perceived availability of CE on its reported frequency of use in four types of patients without contraindications to surgery. For physicians perceiving CE as being readily available, the reported frequency of use increases with both the presence of recent symptoms and with increasing degrees of stenosis (the procedure was reported as being always or often recommended by 7% of physicians for asymptomatic patients with 50% to 70% stenosis, by 26% for asymptomatic patients with >70% stenosis, by 23% for newly symptomatic patients with 50% to 70% stenosis, and by 60% for newly symptomatic patients with >70% stenosis). The perceived availability of CE influenced the reported frequency of its use for all groups of patients except for those who were asymptomatic and who had moderate-grade (50% to 70%) stenosis. This impact of perceived availability is present even in newly symptomatic patients with high-grade (>70%) stenosis. If CE was perceived as being readily available, it was reported as being seldom or never used by 15% of physicians; if the procedure was viewed as not being readily available, it was reported as being seldom or never used by 47% of physicians (Fig 2⇓, bottom right).
The multinomial multiple logistic regression analysis indicated that the reported frequency of use of CE also varied by physician specialty, independent of the effect of the perceived availability of the procedure (Table⇑). Fig 3⇓ compares the reported frequency of use of CE according to physician specialty in the same four types of patients without contraindications to surgery. There is not a significant difference among the physician specialties in the reported use of endarterectomy in asymptomatic patients with moderate (50% to 70%) degrees of stenosis of the extracranial CA (Fig 3⇓, top left; P=.126). For asymptomatic patients with higher degrees of stenosis, surgeons reported using CE more frequently than it was recommended by other groups of physicians (Fig 3⇓, top right; P<.001). Similarly, surgeons responded that they use CE more frequently than other groups of physicians for newly symptomatic patients (TIA or minor stroke within the previous month) with moderate ipsilateral CA stenosis (Fig 3⇓, bottom left; P<.001). The reported use of CE also varied by physician specialty for patients with recent TIA or minor stroke and >70% stenosis of the ipsilateral CA (Fig 3⇓, bottom right; P<.001). More than 80% of both surgeons and neurologists responded that they always or often use CE for newly symptomatic patients with high-grade CA stenosis, whereas approximately 55% of noninternist primary care physicians and internists reported that they would do so. About 47% of noninternist primary care physicians and 54% of internists reported that they always or often refer this type of patient to a neurologist. Approximately 23% of noninternist primary care physicians, 16% of internists, 5% of neurologists, and 9% of surgeons reported that they seldom or never use CE for patients with high-grade CA stenosis and a TIA or minor stroke within the previous month.
Considerable data derived from randomized clinical trials are accumulating concerning the impact of CE on the prevention of stroke in defined high-risk patient populations. However, physicians in practice are often faced with patients who do not have characteristics that precisely conform to those of patients for whom the procedure has been demonstrated to be of benefit. In addition, physicians must decide whether to recommend the surgery for patients in whom the role of CE is uncertain given the available data. At the time this survey was conducted, the results from three randomized trials indicating that CE was efficacious in selected symptomatic patients with high-grade stenosis of the CA had been available for about 2 years.1 2 3 These data provided a clear basis for performing the operation in that group of patients, provided that it could be carried out safely, and likely had an impact on the overall use of the procedure in the United States.4 Clinical trial data definitively supporting or refuting the use of CE in symptomatic patients with more moderate degrees of stenosis were not (and still are not) available. A small, inconclusive randomized trial of CE in patients with asymptomatic stenosis was published 1 year before our survey was performed.5 A report from a larger trial of CE use in patients with asymptomatic stenosis was published several months before our survey was conducted6 ; however, the implications of that study for clinical practice were uncertain.10 The results of the Asymptomatic Carotid Atherosclerosis Study have only recently become available.7 Therefore, on the basis of data widely available at the time, patients without surgical contraindications with newly symptomatic high-grade CA stenosis would have been expected to be referred for CE. Decisions with regard to other patient groups were not made with the benefit of unambiguous clinical trial data. This survey systematically probed the reported use of CE and preoperative diagnostic testing in a nationwide sample of practicing physicians.8 The central findings of the present analyses are that the reported use of diagnostic testing and CE vary independently with the perceived availability of the tests and the procedure, with the specialty of the physician, and, as expected, with variables representing clinical characteristics of patients (symptom status and degree of CA stenosis).
Carotid duplex ultrasonography and cerebral angiography are important tests used in the evaluation of patients with cerebrovascular disease. In the setting of newly symptomatic patients, these tests may be used both as an aid in establishing the cause of the symptoms and as part of the presurgical evaluation of potential CE candidates. Carotid ultrasonography was reported as being readily available by 98% of surveyed physicians,8 and almost 90% of physicians reported that they always or often use the test in newly symptomatic patients. Only about 20% always or often obtain cerebral angiography in newly symptomatic patients, suggesting that this invasive test is being used selectively by practicing physicians.
In asymptomatic individuals, carotid duplex ultrasonography is primarily used as a screening tool in candidates for prophylactic CE. The subsequent use of conventional angiography would be expected to vary with the degree of CA stenosis revealed by the screening test. In fact, the proportion of physicians reporting that they always or often obtain a cerebral angiogram in asymptomatic patients increases nearly threefold (13% to 33%) as the degree of stenosis demonstrated by the noninvasive test increases from 50% to 70% to >70%. In view of the fact that 70% of physicians reported that they always or often obtain a carotid ultrasound study in patients with an asymptomatic bruit, the reasons why this difference is not greater are not entirely clear. Despite differences in opinion on the advisability of performing CE solely on the basis of the results of carotid ultrasonography,11 12 it is unlikely that most physicians would proceed without conventional arteriography. Performing a screening test but not acting on the result suggests that the test is being used inappropriately. However, physicians were not asked whether they would perform angiography if the carotid ultrasound indicated a “critical” asymptomatic stenosis. It is possible that physicians were using the screening test to identify this subgroup. The perceived availability of a test may also influence the frequency of its use, and cerebral angiography was viewed as relatively unavailable by 12% of surveyed physicians.8 The reported availability of cerebral angiography did not influence its use for asymptomatic patients with moderate degrees of stenosis (Fig 1⇑, top). In contrast, the frequency of use of cerebral angiography in asymptomatic patients did vary with the availability of the test if the ultrasound study indicated higher-grade stenosis (Fig 1⇑, middle). Thus, part of the disparity between the use of carotid ultrasonography as a screening tool in asymptomatic patients and the reported frequency of use of angiography if the ultrasound study indicated high-grade stenosis may be due to the perceived availability of angiography.
In accord with the data available at the time, the proportions of physicians indicating that they would often or always recommend CE varied with symptom status and degree of stenosis (Table⇑) and were highest for newly symptomatic patients with high-grade stenosis (Fig 2⇑, bottom right), intermediate in the setting of symptomatic moderate-grade stenosis (Fig 2⇑, bottom left) or asymptomatic high-grade stenosis (Fig 2⇑, top right), and lowest for asymptomatic moderate-grade stenosis (Fig 2⇑, top left). However, even in the group of patients most likely to benefit from the surgery given the available data (those with newly symptomatic high-grade stenosis), the perceived availability of CE has a significant impact on its reported frequency of use (Fig 2⇑, bottom right). After considering the impact of perceived availability, there is no significant effect of practice location or region of the country on the frequency of CE use (Table⇑). However, CE was reported to be least available by physicians practicing in the South and in nonmetropolitan settings.8 The reasons for the perceived unavailability of CE to physicians practicing in the South remain uncertain, but it is of particular importance because the South represents the country’s “stroke belt,” with the highest incidence of the disease.13 If the perceived unavailability of CE to physicians practicing in the South is due to actual unavailability, then the problem could be addressed through the allocation of additional resources. If this is not the case, then the problem could be addressed through clarification of referral mechanisms and targeted provider education. Clarification and streamlining of referral mechanisms might also partially address the perceived unavailability of cerebral angiography and CE to physicians practicing in nonmetropolitan areas.
The reported frequency of use of CE also varied by physician specialty (Table⇑). The effect of physician specialty is independent of the perceived availability of the procedure (Table⇑). The frequency of use of CE was consistent with the data for diagnostic testing and uniformly low across specialties for patients with asymptomatic moderate-grade stenosis (Fig 3⇑, top left). In patients with newly symptomatic high-grade stenosis, CE was reported as being always or often used by more than 80% of neurologists and surgeons but only by about half of internists and noninternist primary care physicians (Fig 3⇑, bottom right). Almost one in four noninternist primary care physicians responded that they would seldom or never use CE for these patients (Fig 3⇑, bottom right). It is possible that these groups of physicians were not as aware of the findings of the randomized trials that were available or had not yet incorporated the findings into their practices. Possibly reflecting this uncertainty with regard to optimal management strategies, about half of internists and noninternist primary care physicians responded that they frequently consult neurologists for this type of patient. Physicians in these specialties are particularly important targets for educational efforts because they are responsible for the majority of the care of patients at elevated stroke risk.8
There were also significant differences by specialty in the use of CE for patients with less clear indications for the surgery based on the data available at the time of the survey. CE was used significantly more frequently by surgeons than by other groups of physicians for these patients (eg, asymptomatic patients with high-grade stenosis [Fig 3⇑, top right] and symptomatic patients with moderate-grade stenosis [Fig 3⇑, bottom left]), possibly representing a practice bias.
In summary, the perceived availability of cerebral angiography and CE influences their reported frequency of use by physicians, even for the group of patients most likely to benefit from the procedure. The reasons for this perceived unavailability are not apparent. The use of CE also varies with physician specialty training. The relative lack of use of CE for newly symptomatic patients with high-grade stenosis by noninternist primary care physicians and internists could be partially addressed through targeted educational efforts.
Selected Abbreviations and Acronyms
|SUDAAN||=||SUrvey DAta ANalysis|
|TIA||=||transient ischemic attack|
This work was performed as part of the Stroke Prevention Patient Outcomes Research Team (PORT) and was funded through contract #282-91-0028 from the US Agency for Health Care Policy Research. The authors wish to acknowledge Gordon H. DeFriese, PhD, Eugene Z. Oddone, MD, and John E. Paul, PhD, who contributed to the design of the survey instrument.
- Received December 18, 1995.
- Revision received January 25, 1996.
- Accepted January 25, 1996.
- Copyright © 1996 by American Heart Association
Gillum RF. Epidemiology of carotid endarterectomy and cerebral arteriography in the United States. Stroke. 1995;26:1724-1728.
Goldstein LB, Bonito AJ, Matchar DB, Duncan PW, DeFriese GH, Oddone EZ, Paul JE, Akin DR, Samsa GP. US National Survey of Physician Practices for the Secondary and Tertiary Prevention of Stroke: design, service availability, and common practices. Stroke. 1995;26:1607-1615.
Shah BV, Barnwell BG, Hunt PN, LaVange LM. SUDAAN User’s Manual, Release 6.00. Research Triangle Park, NC: Research Triangle Institute; 1992.