| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Research Department, American College of Radiology, 1891 Preston White Dr.,
Reston, VA 20191.
2 Department of Diagnostic Radiology, Yale University, New Haven, CT.
Received April 2, 2004;
accepted after revision April 12, 2004.
Address correspondence to J. H. Sunshine
(jonathans{at}acr.org).
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. The American College of Radiology (ACR) surveyed 970 practices by mail, using a 65-item questionnaire, in 1999. A response rate of 66% was achieved. Weighting was used to make responses representative of all radiology practices in the United States. Trends were explored by making comparisons with data from a 1991-1992 ACR study.
RESULTS. Among the types of procedures studied, the highest percentage of multiradiologist diagnostic-radiology-only practices performed mammography (95%) and sonography (94%). (General conventional radiography and fluoroscopy were not studied.) The lowest percentage of these practices performed interventional procedures (69%) and MRI (77%). Solo practices showed less diversity in types of procedures performed than did multiradiologist practices and were a good deal less likely to perform each type of procedure except mammography and sonography. Generally, higher percentages of practices in nonmetropolitan cities or towns and rural practices performed various types of procedures than practices in metropolitan areas. Practice size, types of settings served (hospital or nonhospital), and practice type also influenced the number of types of procedures performed by a practice. The fraction of practices performing CT decreased from 91% to 83% between 1991-1992 and 1999. Percentages for other types of procedures were generally stable over time.
CONCLUSION. Certain practice characteristics play a role in determining the types of imaging procedures a diagnostic radiology practice performs. The decline in the percentage of practices providing CT and the failure during the 1990s of percentages for MRI and interventional radiology to increase from a relatively low base is worrisome. Future analyses based on a subsequent ACR survey will provide further insights into trends.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Have interventional radiology and MRI become more widespread, thus making the benefits of these relatively new technologies available to more patients—but simultaneously demanding more varied skills of radiologists at a time when a serious shortage of radiologists exists [2-4]? Has mammography become less widely available because of problems of relatively low payment rates and a high incidence of malpractice suits, thereby potentially generating problems for patients in accessing this important preventive service?
This article answers these and similar questions using data from the American College of Radiology's (ACR's) 1999 Survey of Practices [5]. We first present results on seven broad imaging categories, such as MRI and sonography. This is followed by information about specific subcategories of these procedures, such as neurologic MRI or obstetric sonography. Findings are compared with those obtained from the ACR's 1991-1992 Census of Groups.
Exact numbers, such as the exact percentage of groups providing a given subcategory of procedures, change from year to year, if not more often. Therefore, we concentrate on broad, longer-term trends and on differences between various categories of groups, which tend to be more stable.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Statistical and Data Analysis Methods
Responses to the 1999 survey were weighted so that weighted statistics
would be representative of the answers that would have been received if all
practices in the United States had been surveyed and responded. Detailed
information about how responses were weighted has been presented previously
[5]. In this article, only
responses from diagnostic-radiology-only practices (either solo or
multiradiologist) were included. Mixed diagnostic radiology and radiation
oncology practices were omitted because they are very few (
3% of all
multiradiologist practices [6])
and because they are analyzed in another paper
[7]. Data for 1999 from almost
600 responding practices are included.
We computed basic descriptive statistics and conducted regression analyses. Depending on whether a dependent variable (outcome variable) was dichotomous or scalar, logistic regression or ordinary least squares multiple regression analysis, respectively, was performed to examine the relationship between the outcome variable and a set of explanatory variables (independent variables). The explanatory variables used were chosen because earlier published research [4, 8] had found them to be important. In all regression models, we included the square of the practice size (measured by number of radiologists) as an independent variable to explore whether the association between practice size and the outcome variable is curvilinear [9]. For categoric independent variables, the most common value was used as the reference (omitted) value in regressions.
We used a p value of less than or equal to 0.05 (two-tailed) as the measure of statistical significance. All findings reported in the text are statistically significant unless otherwise stated.
All data analyses were conducted with SAS System software (release 8.01,
SAS Institute). Readers may approximate the SE of percentages shown in the
tables by using the formula
 |
in which p = proportion of practices with a characteristic and n = unweighted number of practices, which is shown in the first column of Table 2.
|
For the 1991-1992 census, data are taken from Table 3 of the study by Bansal and Sunshine [1] and recomputed to be based on all practices providing diagnostic radiology services, rather than, as the table was published, on all practices including radiation-oncology-only practices. In these 1991-1992 data, mixed practices are included because the available information did not make possible their exclusion.
|
The information presented in the analyses was based on the practice's interpretations of the survey questions and was self-reported. For example, a practice is classified as "academic" because it self-reported "primarily academic" as its practice type. When information about practice location is discussed, the types of locations are center city or suburb of a large metropolitan area (area population > 1 million), center city or suburb of a smaller metropolitan area (area population, 50,000-1 million), nonmetropolitan city or town or rural area (area population < 50,000), and varied locations (no one type of location is principal).
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Practice Type
In 1999, for each of the seven types of procedures, nonacademic private
radiology practices generally were the type of practice most likely to perform
the procedures (Table 2). The
pattern in 1991-1992 was quite different, with academic practices most likely
to perform four of the six types of procedures studied
(Table 1). By 1999, academic
practices led only in the likelihood of performing interventional procedures
(Table 2). In 1999, the
percentage of solo practices performing each of the seven types of procedures
was usually lower than for all other types of practices. For example, only 59%
of solo practices reported performing CT and 20% of solo practices reported
performing interventional procedures.
After controlling for other practice characteristics, including practice size, logistic regression of the 1999 data indicated that being a solo practice, relative to private, nonacademic radiology practices, significantly decreased the likelihood of a practice performing MRI, mammography, or interventional procedures (Table 3). On average, solo practices performed fewer than four of the seven types of procedures studied, whereas all other practice types performed more than five (Table 2).
Practice Size
Larger multiradiologist practices in 1999 were more likely than smaller
ones to perform each of the seven types of procedures
(Table 2). For example, for
practices with 15 or more members, the percentage performing a procedure type
ranged from 84% for nuclear medicine and MRI to 100% for mammography. In
comparison, of practices with 2-4 members, 69% performed nuclear medicine, 70%
performed MRI, and 93% performed mammography. A similar pattern of the
likelihood of performing any given type of procedure increasing with practice
size was seen in 1991-1992 (Table
1). On average in 1999, 2-4 radiologist practices performed 5.2 of
the seven types of procedures, whereas the average for larger practices was
more than 6.0 (Table 2).
Logistic regression indicated that in 1999, after controlling for the effects of all other practice characteristics studied, increasing practice size significantly increased the likelihood that CT, interventional studies, imaging-guided breast biopsy, and nuclear medicine procedures would be performed (Table 3).
Geographic Region
Logistic regression indicated that in 1999, after controlling for all other
practice characteristics studied, being in the Northeast significantly
decreased the likelihood that a practice would perform MRI relative to
practices located in the South (the reference region), but significantly
increased the likelihood that a practice would perform mammography and
imaging-guided breast biopsy (Table
3). After controlling for other practice characteristics, being a
practice in the West (relative to those in the South) significantly decreased
the likelihood that MRI and nuclear medicine procedures would be performed.
Multivariate analyses are not available for 1991-1992, but simple tabulations
(Table 1) suggest most of these
regional differences existed at that time.
General Practice Settings Served
The percentage of multiradiologist practices serving only nonhospital
settings that performed each of the seven types of procedures was, in 1999,
generally much lower than the percentage of practices serving hospitals only
or serving both types of settings (Table
2). For example, 58% of practices serving only non-hospital
settings performed CT compared with 85-87% of practices serving only hospitals
or serving both types of settings. (Nonhospital settings were predominantly
offices, clinics, or imaging centers or nonhospital facilities of
multispecialty practices). Practices serving only nonhospital settings
provided, on average, 3.6 of the seven types of services compared with
approximately six types provided by other practices.
Logistic regression indicated that after controlling for other practice characteristics, being a practice that served only nonhospital settings significantly decreased the likelihood that the practice would perform MRI, nuclear medicine, interventional procedures, and imaging-guided breast biopsy relative to practices serving both hospitals and nonhospital settings (Table 3).
Location
Generally, in 1999, higher percentages of multiradiologist practices
located in nonmetropolitan cities or towns or in rural areas performed various
types of procedures than those in metropolitan areas
(Table 2). For instance,
approximately 89% of practices in nonmetropolitan cities and rural areas
reported performing CT, whereas approximately 78% of practices in larger
metropolitan areas performed CT; and similar differences were seen for most
other types of procedures except imaging-guided breast biopsy. In 1991-1992,
similar patterns were seen for some types of procedures, but interventional
radiology and MRI were more likely to be performed by practices in
metropolitan areas than by those in nonmetropolitan or rural areas
(Table 1). In 1999, the total
number of types of procedures performed was approximately 5.5 for practices in
large metropolitan areas compared with approximately 6 for practices located
elsewhere (Table 2). Logistic
regression of the 1999 data shows that, after controlling for all other
practice characteristics studied, practices that were located in
nonmetropolitan cities or towns were significantly more likely to perform CT,
sonography, and nuclear medicine than those practices located in a center city
of smaller metropolitan areas, the reference type of location
(Table 3). Similarly, rural
practices were significantly more likely to perform nuclear medicine and
mammography than otherwise similar practices in the center city of smaller
metropolitan areas. Some reduction was seen in the likelihood of performing CT
or MRI for practices in large metropolitan areas relative to similar practices
in center cities of smaller metropolitan areas.
Number of Types of Procedures Performed
We used multiple regression analysis to examine the relationship between
practice characteristics and the total number of types of procedures a
practice performed. We observed the following statistically significant
relationships, each controlling for the effect of all other variables studied
(not shown in tables).
The total number of types of procedures performed by a practice increased with practice size. For small practices, one type of procedure was added for approximately every eight additional radiologists in the practice, but the increase was less for practices of larger size. Moreover, solo practices performed approximately 0.8 fewer types of procedures than otherwise similar private nonacademic practices, even after taking into account differences associated with practice size. Practices that were located in either rural areas or in nonmetropolitan cities or towns performed on average approximately 0.8 more types of procedures than otherwise similar practices located in the center city of smaller metropolitan areas (the reference type of location). Practices that served only nonhospital settings performed on average 1.5 fewer types of procedures than otherwise similar practices that served both hospitals and nonhospital settings.
Specific Subcategories of Procedures
Virtually all (> 98%) multiradiologist practices and solo radiology
practices performing CT in 1999 reported performing neurologic CT. The same
was true for body CT (not shown in tables).
Virtually all (> 97%) multiradiologist and solo practices performing MRI in 1999 reported performing neurologic MRI, and the same was true for body MRI (not shown in tables). In contrast, only 32% of multiradiologist practices performing MRI performed cardiovascular MRI (Table 4). However, the proportion was much higher for practices with 15 or more members (63%) and for academic practices (54%). Logistic regression (Table 5) showed that, all other characteristics equal, practices performing MRI were more likely to perform cardiovascular MRI the larger they were and if they functioned in nonhospital settings only. They were less likely to perform cardiovascular MRI if they were private multispecialty groups compared with private nonacademic multiradiologist practices (the reference practice type); in the West compared with the South (the reference region); and located in nonmetropolitan, rural, or varied locations.
|
|
Of all multiradiologist practices performing sonography, 92% performed obstetric sonography (Table 4). For solo practices, the corresponding statistic was 87%. Logistic regression analysis (Table 5) showed that, after statistically controlling for the effects of all other factors considered, practices performing sonography were more likely to perform obstetric sonography if they were in the West compared with the South (the reference region) and if they were in the suburbs of a smaller metropolitan area or in a nonmetropolitan or rural area compared with those in the central city of a smaller metropolitan area (the reference location). Conversely, practices were less likely to perform obstetric sonography if they were virtually any type other than private, nonacademic multiradiologist; the larger they were (but this effect weakened with increasing size); and if they served nonhospital settings only.
Of all multiradiologist practices performing sonography, 85% performed vascular sonography (Table 4). For solo practices, the corresponding statistic was 75%. Logistic regression (Table 5) showed many practice characteristics had a statistically significant effect on whether sonography-performing practices provided vascular sonography services, with the pattern of effects often like that for providing obstetric sonography services. Of practices performing sonography, 73% of multiradiologist practices and 70% of solo practices provided other sonography services (Table 4) with, again, various practice characteristics affecting the likelihood of doing so (Table 5).
Ninety-six percent of multiradiologist practices performing interventional procedures performed abdominal interventional procedures (not shown in tables). The percentage was lower for the six other types of interventional procedures investigated in the survey (Table 4), and a number of practice characteristics affected the likelihood of an interventional-procedure-performing practice offering these six other types of interventional services (Table 5). However, there was rarely a unidirectional, statistically significant effect of any characteristic across the six other types of interventional procedures. The closest to an exception was functioning in nonhospital settings only. All other things equal, relative to functioning in both hospital and nonhospital settings, functioning in nonhospital settings only decreased the likelihood of offering four of the six types of interventional services but increased the likelihood of offering cardiac interventional services.
Ninety-seven percent of multiradiologist practices that performed nuclear medicine procedures performed SPECT, but for solo radiologist practices the corresponding statistic was only 74% (not shown in tables). For non-SPECT nuclear medicine procedures, these figures were 50% and 63%, respectively (Table 4), and a variety of practice characteristics affected the likelihood that practices that performed nuclear medicine procedures performed these procedures (Table 5).
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Perhaps the decrease and the failures to increase are part of the toll taken by managed care. From 1992 to 1999, the prices paid for radiologists' services decreased 4% in dollar terms and 19% in inflation-adjusted terms [10]. This decrease may have caused some practices to cease offering services requiring relatively expensive equipment or facilities. Certainly, a period existed at the height of the managed care pressures in the mid 1990s when annual sales of MRI units decreased radically. To be specific, annual sales in the United States were approximately 500 early in the decade, but declined to a nadir of less than 350 in 1994-1995 and did not recover to their beginning-of-decade level until 1997 (Bell R, personal communication).
Whatever the cause, what happened in the 1990s quite likely resulted in a loss of access—or a loss of expected access—to services that are valuable to patients. At the least, there was a loss of convenience if the services were available from some other, but less conveniently located, practice.
Of even greater concern is the fact that, unlike the situation in the early 1990s, by the decade's end academic practices were not the type of practice most likely to perform most types of procedures. Academic practices are where radiologists learn their profession, and if any major category of procedures is not provided by a substantial fraction of academic practices, it is hard to see how residents can receive good training in that aspect of radiology. In fact, residents are already seriously concerned about inadequate training in MRI, as evidenced by the ACR's Resident Section focusing on this problem as one of the section's principal concerns in 2002-2003.
On the positive side, it is reassuring that mammography remained widely available, with 95% of multiradiologist practices offering it in 1999. It is similarly reassuring that practices in nonmetropolitan cities and towns and rural areas provide a greater variety of types of procedures than would otherwise be expected. They are far more likely than practices in metropolitan areas to be the only practice in their area, and with their patients not having the option of obtaining care from another practice, it is particularly important that they offer a wide range of services.
The finding that nonhospital-only practices offer a relatively limited range of services is not surprising; this is how such practices are designed. But this finding should sound an important cautionary note to those who would reduce the cost of health care in the United States by moving care away from hospitals. It indicates that hospitals offer clinically important types of imaging services that are not so readily available elsewhere.
Study Strengths and Limitations
General strengths and limitations of the 1999 survey have been described in
detail elsewhere [5]. In brief,
despite the known difficulty associated with collecting information from
physicians and other professionals
[11] and the growing workload
of radiologists [2], response
rates were achieved that were comparable to earlier ACR surveys of this type
[12] and other surveys of
physicians [11]. Weighting
addressed differences in response rates related to region and practice size.
Extensive pretesting and careful data cleaning were conducted to minimize
errors in responses. The number of responses is relatively large. However,
data are self-reported, with no auditing conducted. Definitions generally were
not provided on the survey questionnaire. The list of practices from which the
sample was drawn generally omits practices that have no ACR members.
Differential response associated with factors other than region and group size
is not accounted for. Strengths and limitations of the 1991-1992 census are
similar [1].
The 1991-1992 data in Table 1 are not entirely comparable to the 1999 data in Table 2 because the former includes mixed diagnostic radiology and radiation oncology groups. This should have little effect, given how few mixed groups there are. Other minor differences between the 1991-1992 census and the 1999 survey may also have small effects, although we know of none likely to do so. One might wonder whether the decline in the percentage of multiradiologist practices performing CT is an artifact of differences in the two data sources rather than a real phenomenon. Consistency of results for other techniques, particularly the widely available techniques of sonography and mammography, make this unlikely.
Responses were sometimes inconsistent, with a respondent not checking the box that indicated his or her practice performed a given category of procedures (e.g., MRI), but then checking a subordinate box (e.g., body MRI) indicating the practice performed a subcategory of procedures within the category. In such instances, we assumed the category markings were correct. We expected this would provide most comparability with the 1991-1992 census, which inquired only about categories. Taking the opposite view—namely, that subcategory markings were correct and category markings had to be amended—would have led to higher estimates of the percentage of practices performing the various categories of procedures.
Other limitations are also noted. Standard errors for some categories of practices are sizable—for example, 6-8% for the proportion of academic practices providing each of the seven broad types of services. Believing essentially all radiology practices perform conventional radiography or fluoroscopy, we have gathered no data on these forms of imaging. The ideal method of measuring patient access to care would be somewhat different from the analysis of practices we have conducted—for example, it would probably involve factors like travel time.
Conclusion
Methods to improve access and overcome the disturbing findings noted in
this article should be considered, because the financial pressures present in
the heath care system today that probably contribute to these findings are
unlikely to disappear. Given the success of mobile mammography units and
mobile MRI units, similar mobile equipment for other imaging technologies may
be desirable. Also, academic practices may need to create more innovative
arrangements to assure residents a strong training experience in all types of
procedures, and possibly to forge more links with other practices.
Monitoring the situation remains important, not only because of concern about the problems identified, but also because there are ongoing changes in the health care system. In the last few years, spending on health care has again been rising rapidly after a hiatus during much of the 1990s. Also, the "managed care backlash" has resulted in the dismantling of some of the more onerous features of managed care [13]. These two changes may have led to a greater percentage of practices now providing some of the major types of procedures. On the other hand, although mammography saw a major increase in the Medicare payment rate in 2002, which would be expected to improve access, a malpractice crisis in which "missed" breast cancers are by far the most common complaint may cause some practices to cease offering mammography services. Indeed, the Food and Drug Administration's registry of accredited mammography facilities in the United States shows a 7% decline in their number between the beginning of 2001 and mid 2003.
More recent data, coming from a subsequent ACR survey, should be available in approximately a year. These should provide enlightening information on whether trends described here have continued.
Acknowledgments
We thank Rebecca Lewis, MPH, for her contribution to the analyses of the
data from the 1999 Survey of Practices.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  J. H. Sunshine, R. S. Lewis, and M. Bhargavan A Portrait of Interventional Radiologists in the United States Am. J. Roentgenol., November 1, 2005; 185(5): 1103 - 1112. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
