| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Department of Radiology, University of California, San Francisco, China Basin
Landing, 185 Berry St., Ste. 350, Lobby 7, Campus Box 0946, San Francisco, CA
94107.
2 Departments of Epidemiology and Biostatistics, University of California, San
Francisco, San Francisco, CA.
3 Group Health Center for Health Studies, Seattle, WA.
4 Department of Biostatistics, University of Washington, Seattle, WA.
5 Department of Radiology, University of New Mexico, Albuquerque, NM.
6 Applied Research Program, Division of Cancer Control and Population Sciences,
National Cancer Institute, Bethesda, MD.
7 Departments of Family Medicine and Radiology, Vermont Cancer Center,
University of Vermont, Burlington, VT.
8 General Internal Medicine Section, Department of Veterans Affairs, University
of California, San Francisco, San Francisco, CA.
Received May 1, 2007;
accepted after revision September 18, 2007.
Supported by a National Cancer Institute–funded Breast Cancer
Surveillance Consortium cooperative agreement (U01CA63740, U01CA86076,
U01CA86082, U01CA63736, U01CA70013, U01CA69976, U01CA63731, U01CA70040).
Abstract
 Top Abstract IntroductionSubjects and MethodsResultsdiscussionReferences |
|---|
960
mammography examinations within each 2-year period (approximately 480
annually). The purpose of our study was to estimate perphysician annual
volumes of mammography interpretation. SUBJECTS AND METHODS. Our study includes 4.2 million mammography examinations performed at 196 facilities between 1998 and 2004. We calculated the annual interpretive volumes per physician, the proportion of mammography examinations interpreted by radiologists in specified volume categories, and the impact on mammography capacity if annual interpretive volume requirements increased.
RESULTS. The mean annual mammographic interpretive volume was 1,777.
Approximately 31% of radiologists interpreted < 1,000 mammography
examinations annually, yet these low-volume radiologists interpreted only 10%
of all mammograms. The 10% of radiologists who interpreted 3,000
mammography examinations annually interpreted 32% of all examinations. Rural
radiologists interpreted fewer examinations annually compared with urban
radiologists. If the minimum annual volume requirement were increased to 1,000
mammograms per year, only 10% of the overall U.S. mammography capacity would
be affected. If the requirement were increased to 2,000 mammograms annually,
47% of capacity would be eliminated, and a major rearrangement of workload
would be required because most radiologists would no longer interpret enough
examinations to meet the revised standards.
CONCLUSION. Doubling physician annual volume requirements would result in a small impact on overall mammography capacity. Increasing volume requirements to 2,000 mammography examinations annually would require a dramatic increase in the number of mammography examinations interpreted by the higher volume radiologists. Unless previously low-volume radiologists increased their volumes, raising requirements to 2,000 examinations could curtail access to mammography, particularly in rural areas.
Keywords: mammography • physician workforce • volume
|
TopAbstractIntroductionSubjects and MethodsResultsdiscussionReferences |
|---|
960 mammography examinations within each 2-year period to be
certified to interpret mammography in the United States
[1]. This minimum was initially
chosen pragmatically because of a desire to keep mammography widely available
and was not based on actual practice patterns or information regarding the
association between mammography performance and annual volume
[2]. Requirements for annual
volume in several other countries, including the United Kingdom, Canada, and
Australia, are five- to 10-fold higher than in the United States
[3–5].
Recently published European guidelines expect radiologists to interpret
5,000 mammograms per year in a centralized location
[6]. The minimums were
originally chosen because of concerns about the physician workforce, but more
recently they have been guided by the belief that accuracy rises with higher
annual interpretive volumes. Several studies have associated radiologists' annual interpretive volume (along with other factors such as years of practice and receipt of additional training in mammography) with some measures of mammographic performance [3, 7–9]. A result of these studies has been ongoing discussion in the United States about increasing volume requirements as a strategy for improving radiologists' accuracy in interpreting mammograms [2]. The impact of an increase in the annual volume requirement on the capacity of radiologists to provide mammography services in the United States is unknown. Were the annual mammography volume requirement raised, some mammographers would undoubtedly no longer interpret enough mammograms to stay certified. The remaining higher-volume mammographers would need to increase their workload to maintain overall mammographic capacity; or alternatively, strategies would be required to increase the volume of low-volume mammographers.
Increasing the annual volume requirement could jeopardize access to mammography in regions such as rural areas, where radiologists who interpret mammography interpret at lower volumes [2]. This might be particularly important for access to diagnostic (rather than screening) mammography, where patients need to be located near the physician who is interpreting the examination so that the physician can examine the patient and can obtain views tailored to the individual patient's needs. Thus, any change in volume requirements could differentially affect access to screening and diagnostic mammography. It would be wise to estimate the impact of any policy that would reduce the number of radiologists who can interpret mammography before enacting such a policy.
A recent American College of Radiology survey of U.S. radiologists found that two thirds of radiologists in multispecialty and private practice radiology groups interpret mammography as part of their practice, but most interpret mammography as a small part of their total workload [10]. Detailed information is not available on how many mammography examinations most U.S. radiologists interpret annually [10]. Given concern regarding access to and staffing of mammography facilities in the United States [2, 11–13], it is important that accurate estimates of physician volume are available, so that the impact of changing volume requirements on physician staffing can be assessed.
The purpose of this study is to document, in several U.S. locations, the annual volume of mammography examinations that radiologists interpret. Using this information, we also estimate the proportion of mammograms interpreted by radiologists in different volume categories. In addition, we document the annual number of mammography examinations at mammography facilities. This will help policy makers and other interested parties model the potential impact on mammographic capacity of increasing radiologists' annual volume requirements or implementing facility annual volume requirements.
|
TopAbstractIntroductionSubjects and MethodsResultsdiscussionReferences |
|---|
Data were included from the seven BCSC registries, covering more than 4 million mammograms on 1.5 million women: San Francisco Mammography Registry, San Francisco, CA; Group Health, Seattle, WA; Colorado Mammography Project, Denver, CO; Vermont Breast Cancer Surveillance System, Burlington, VT; New Hampshire Mammography Network, Lebanon, NH; Carolina Mammography Registry, Chapel Hill, NC; and New Mexico Mammography Project, Albuquerque, NM. Information on all mammograms obtained at these facilities, including a unique study identifier for the interpreting physician, is sent to the mammography registries and then sent centrally to the Statistical Coordinating Center (SCC) for pooled analysis. Each registry obtains annual approval from its institutional review board to collect mammography-related information.
Subjects
The subjects of this study were radiologists who interpreted mammography
examinations within one or more of the facilities that contributed data to the
BCSC between January 1, 1998, and December 31, 2004. The subjects of the study
were also facilities where the mammograms were interpreted. To estimate
physician mammographic volume, we restricted the analysis to the five
registries where radiologists primarily work at BCSC-affiliated facilities so
that our estimates of physician volume would be accurate. We excluded two
registries when we calculated physician volume based on results from a
separate survey of radiologists' practice patterns because we found that a
large percentage of radiologists at those registries work at BCSC and non-BCSC
facilities.
We included radiologists who provided data for at least 2 full years of the study period, and we included the most recent complete data available to reflect radiologists' current mammography workloads. For each physician, we included 2–7 years of data (median, 4.5 years) to get the most stable estimate of the average annual volume of interpretation. We were also interested in estimating facility volume. Facilities that contribute mammograms to the BCSC contribute data on all mammography examinations performed at their facility. Thus, to estimate facility mammographic volume, we were able to include all seven mammography registries of the BCSC. We included all sites so that the results would reflect all facilities participating in the BCSC. We estimated facility volumes based on the location where mammography examinations were performed (as opposed to where they were interpreted) because we thought this could potentially have a larger impact on patient access to mammography.
Analysis
We calculated descriptive statistics of the annual total volume of
mammograms each physician interpreted and the proportion of mammograms
interpreted by radiologists who interpret in each volume category. Statistics
were also calculated separately by type of mammography (screening vs
diagnostic) and by location (rural vs urban) of the woman who underwent each
mammography. A woman was considered to live in a rural community if 50%
of the area defined by her zip code was deemed rural according to U.S. census
data.
We calculated the average annual number of mammograms interpreted at each
facility and the overall proportion of mammograms interpreted at facilities
within each size category, calculated separately for rural (n = 64)
and urban (n = 132) facilities. We defined a facility as rural if
50% of the mammography examinations were provided to women who lived in a
rural area based on their zip code.
The recent Institute of Medicine [2] report, Improving Breast Imaging Quality Standards, considered the possibility of increasing annual requirements for physician volume. To estimate the impact on physician staffing if annual requirements for physician volume were changed, we estimated the impact on radiologists' daily, weekly, and annual volume if the FDA increased the minimum 2-year volume from 960 to 2,000 or 4,000 mammograms, respectively (corresponding to 1,000 and 2,000 1-year volumes). For this estimate, we assumed radiologists work an average of 5 days per week for 43 weeks per year [10]. We assumed that, if the minimum threshold were changed, radiologists who were below the new threshold would no longer interpret mammograms; and that the smaller pool of eligible radiologists would need to interpret the same total number of mammograms. We divided the additional workload equally among all remaining radiologists. We have used this assumption for illustrative purposes only.
The FDA currently sets no interpretive requirements for facility volume. To estimate the impact on facilities if annual requirements for facility volumes were adopted by the FDA, we estimated the impact of minima set at 1,000 and 2,000 mammograms annually and assumed that facilities were open 52 weeks per year.
|
TopAbstractIntroductionSubjects and MethodsResultsdiscussionReferences |
|---|
Figure 1 shows the
distribution of radiologists' annual mammography workload. The mean annual
volume of mammographic interpretations was 1,777 (median, 1,403; interquartile
range, 797–2,202). Overall, 57% (144/253) of the radiologists
interpreted 750–2,000 mammograms annually, whereas 20% (50/253)
interpreted < 750, and 23% (59/253) interpreted > 2,000. Overall, 10%
(25) of the radiologists interpreted 3,000 mammograms per year.
|
3,000
mammograms annually (the difference between 100%, point c, and 90%,
point d [Fig. 2]). The
10% of radiologists who interpreted 3,000 or more mammograms annually
interpreted 32% of all mammograms; 31% of all screening mammograms (the
difference between 100%, point c, and 69%, point e
[Fig. 2]) and 39% of all
diagnostic mammograms (the difference between 100%, point c, and 61%,
point f [Fig. 2]).
Radiologists in the various volume categories did not differ much in the
proportion of screening and diagnostic mammograms they interpreted, but
higher-volume radiologists tended to interpret a larger proportion of
diagnostic mammograms.
|
To determine whether the provision of mammography services varies by location, we recalculated mammographic interpretive volumes stratified by the location where the mammograms were obtained. Rural radiologists tended to interpret fewer mammograms annually than did urban radiologists. For example, 90% of rural radiologists (point a in Fig. 3) versus 70% of urban radiologists (point b in Fig. 3) interpreted < 2,000 mammograms annually. At lower annual interpretive volumes (for example, < 1,000 mammograms annually), this translated into only small differences in the provision of mammography by location. At higher interpretive volumes (for example, 2,000 mammograms annually), the difference in the proportion of mammograms that urban and rural radiologists interpreted was more pronounced. Thus the impact of a higher annual volume requirement on mammography capacity would be more pronounced in rural locations. In urban locations, a cutoff of 2,000 mammograms would result in 42% of mammograms (point c in Fig. 3) needing to be redistributed to the high-volume radiologists if low-volume radiologists stopped interpreting mammograms. In rural locations, this cutoff would mean that the majority of mammograms (63%, point d in Fig. 3) would have to be redistributed to the high-volume radiologists.
|
Fewer mammography examinations were performed at rural facilities. In 2003,
the mean number of mammography examinations performed at rural facilities was
3,168 compared with 5,512 at urban facilities. In addition, a significant
proportion of rural facilities were small; 23% of rural and 11% of urban
facilities performed < 1,000 mammography examinations annually, although
relatively few mammography examinations were performed at these low-volume
facilities (3% of all rural mammography examinations and 1% of all urban
mammography examinations). At higher-volume facilities, the differences in the
proportions of mammography provided by location became more pronounced. For
example, if facilities were required to obtain 2,000 mammograms annually
to be certified, only half (52%) of rural facilities would currently qualify
(as opposed to 65% of urban facilities), and 16% of rural mammography
examinations are performed at these high-volume facilities (as opposed to 7%
in urban areas). Facilities where mammograms were interpreted (compared with
where the mammography examinations were performed) were larger in both rural
and urban locations, likely reflecting centralization of interpretation.
We estimated what would happen to the average workload of radiologists if
the annual requirements for interpretive volume were increased and required
radiologists to interpret 1,000 (or 2,000) mammograms annually
(Table 1). Currently,
radiologists interpret a mean of eight mammograms per workday and 39
mammograms per week, with relatively small differences between urban and rural
practices (urban weekly volume, 44 mammograms; rural weekly volume, 29
mammograms). If the minimum were raised to 1,000 mammograms per year (a
doubling of the current FDA requirement), approximately a quarter of
radiologists would no longer be certified at their current workload, and if
the low-volume radiologists stopped interpreting mammograms, the workload of
the remaining radiologists would increase by about a third (to 10 mammograms
per day and 51 mammograms per week). If the requirement were increased to
2,000 mammograms annually, 35% of radiologists in urban locations, yet nearly
half (48%) of radiologists in rural locations, would not interpret enough
mammograms to qualify to interpret mammography. Thus the remaining
mammographers would have to increase their volumes substantially to maintain
capacity, and this effect would be most dramatic in rural locations. To
maintain mammographic capacity, urban mammographers would need to interpret 27
mammograms per day and 133 per week, whereas rural mammographers would need to
read 53 mammograms per day and 263 mammograms per week.
|
We estimated what would happen to facility volume if an annual requirement
for facility interpretive volume were adopted
(Table 2). If a volume
requirement per facility were set at 1,000 mammograms annually, the daily
volume would increase approximately 25% (from 18 to 23), with little
difference between urban and rural locations. If a volume requirement per
facility were set at 2,000 mammograms annually, the average daily volume would
more than double in both urban (increasing from 21 to 50) and rural
(increasing from 12 to 31) locations.
|
|
TopAbstractIntroductionSubjects and MethodsResultsdiscussionReferences |
|---|
What are the implications of our work? If the decision were made to double
annual requirements for physician volume (from 480 to 1,000 in 12
months), the impact on overall mammography capacity would be relatively
modest. Low-volume radiologists currently interpret a relatively small
percentage of mammography examinations, and it is likely that other
radiologists in the communities could increase their workload to compensate.
However, the impact could be greater in rural communities, particularly if a
single physician whose volume was below the new 1,000-mammogram threshold
serves an area. Access to mammography could substantially suffer if that
radiologist chose to stop practicing mammography because he or she could not
increase annual volumes to meet new standards.
If the decision were made to increase annual requirements for U.S.
physician volume more substantially (for example, to increase the requirement
to 2,000 in 12 months, a volume consistent with requirements and
guidelines in many other countries), unless many U.S. radiologists chose to
increase their volumes, this would dramatically impact the number of
mammography examinations that the remaining radiologists would have to
interpret. Countries such as the United Kingdom, Canada, and Australia achieve
these higher annual volumes through both specialization and centralization of
mammography interpretation
[3–5].
If volume requirements were increased as described, the actual workload that would be required of the qualifying radiologists would reflect the need for mammography to become more specialized. Currently, most radiologists in private practice each interpret few mammograms; meeting increased volume requirements would require radiologists to specialize in mammography, which only a small proportion of radiologists now do [10]. However, if radiologists were available and willing to interpret these mammograms, the actual workload would be manageable in larger radiology groups because increased mammography work could be traded for other work. This degree of specialization is routine for breast imaging in academic settings but much less common in private practice settings where most mammography occurs and is even less common in rural settings [10]. One reason such specialization has not been as common in private practice is that if only a single physician interpreted mammography examinations within a particular practice, mammography would not be available whenever that physician was not at work.
The impact of physician workload on mammography accuracy remains uncertain, and volume is likely only one factor among many that may impact mammography accuracy. [2, 3, 7–9, 17]. Thus it is not clear how a change in FDA minimum guidelines would affect mammography accuracy. Any impact may depend on how physician staffing changes: whether high-volume radiologists further increase their workload or low-volume radiologists increase their workload to meet a revision to FDA requirements.
Théberge et al. [17]
reported that higher volume for facilities (> 4,000 mammograms annually)
was associated with improved mammographic accuracy. Assessing the impact of
changing volume requirements for facilities is even more complex than that of
changing volume requirements for radiologists. Radiologists in one facility
may shift mammograms among themselves or even send them to a remote facility
for interpretation; however, volume requirements that close a facility may
force patients to travel far for mammography and any follow-up care. Even in
urban areas, insurance arrangements may make this particularly complex.
Table 2 shows a simplified
impact of changing volume requirements for facilities. If facilities were
required to examine 2,000 mammograms annually, we estimate that more than
half of them would need to stop providing mammography, and this would occur
slightly more in rural areas.
Currently many rural counties lack dedicated mammography facilities [2], and this has resulted in some difficulty in women's access to mammography. For example, in some areas in Virginia women need to travel up to 60 miles to get mammography [2]. Several possible solutions have been suggested to improve screening access in rural areas (telemammography, mobile mammography facilities, and centralization of mammography interpretation). All of these are less convenient and generally more expensive than mammography in a local setting.
Even more than access to screening, facility closures could impair access to diagnostic services, such as for women with palpable abnormalities or those needing additional imaging. Any creation of minimum volume requirements for facilities should not negatively affect access to mammography in rural areas. It is important to distinguish between the respective sizes of the facilities where the mammography examinations are performed and where they are interpreted. For interpretation, mammograms can be sent electronically or by mail to areas far from where they were obtained. However, mammography should be available without requiring women to travel several hours to a facility. Thus, any legislation should focus on where mammograms are interpreted, rather than where they are obtained, because this would ensure centralization of reading while preserving access. Digital mammography, for example, would facilitate centralized reading of mammograms at larger (perhaps urban) facilities, even if the facilities where the mammography examinations were performed were small. However, it would be important that mammography technologists obtain sufficient experience to maintain a high level of technical expertise in obtaining the images.
Interest has been expressed in creating specialized regional breast imaging centers of excellence [2] where experienced and high-volume radiologists can potentially provide multidisciplinary and coordinated breast cancer care. One of the impediments has been concern regarding how centralizing care could decrease access to services for some patients; however, it has also been suggested that these centers could alleviate access problems. Our study does not provide data on the desirability of creating centers of excellence. However, our results do suggest that access overall would not be greatly reduced if the interpretation of mammography were limited to radiologists who interpret 1,000 or more mammograms annually, but would be impaired if higher thresholds were used or if volume requirements were extended to facilities. Perhaps all targets for physician or facility volumes should allow exceptions in rural areas, For example, a single radiologist would be acceptable in a small facility performing 500–1,000 mammography examinations annually; whereas two to five readers would be the target for a facility performing 5,000–10,000 mammography examinations a year.
Lewis et al. [10] recently published estimates of physician volume based on self-reporting among a random sample of U.S. radiologists, and our estimates are similar. For example, we found a mean annual volume of 1,673, almost identical to their estimation of 1,670. We found 31% of radiologists interpreted < 1,000 mammograms annually, whereas they estimated 25% of radiologists interpret at these low annual interpretive numbers. The primary difference in the results occurred at the highest annual volumes. On the basis of physician self-reporting, they found that 11% of radiologists interpret > 5,000 mammograms annually, whereas according to actual health service data, we found that only 3% of radiologists interpret at these high annual interpretive volumes. These discrepancies suggest that self-reporting by high-volume radiologists may overestimate the number of mammograms interpreted.
This study has several strengths, including its large size, coverage of seven separate regions of the country, and inclusion of diverse mammography practice types with women who roughly mirror U.S. demographics [16]. This study also has several limitations. We may not have captured all mammographic interpretations for all radiologists. Some doctors interpret at facilities outside the BCSC, and we are not counting these examinations when estimating physician volume. To minimize this problem, we limited our estimation of physician volume to the five sites where we have captured the most mammograms in the communities and where relatively few mammograms are likely to be interpreted outside of the BCSC facilities. In addition, radiologists may double-read mammography examinations (i.e., two radiologists may each read the same case for each to meet MQSA requirements), and we will have only counted each examination for a single physician. Therefore, we are almost certainly undercounting volume for some radiologists, so radiologists may actually interpret more examinations than we have estimated. Given the direction of this bias, any impact on mammography capacity if FDA regulations are changed would likely be smaller than we have estimated.
Second, our estimates were made in seven regions in the United States, and patterns of the provision of mammography may differ in other parts of the country. However, our study is the first to document, in several different types of practices, annual interpretive volumes based on actual data. Third, our analysis of the urban–rural continuum was relatively simple: We considered an area rural if more than half of the mammography examinations were provided to women who lived in zip codes defined as being at least 50% rural by the U.S. Census Bureau. Other definitions exist and would categorize some practices and patients differently. We believe that characterizing facilities by the population they serve will lead to the most accurate reflection of access.
Fourth, when calculating the impact of a change in FDA regulations, we assumed that radiologists who interpret fewer mammograms than any newly imposed threshold would simply stop reading mammograms. In fact, some of these low-volume radiologists would ideally increase their annual averages. The Institute of Medicine concluded that having midvolume radiologists increase their annual volume would be the best way to improve overall capacity [2]. If radiologists with low volumes increased their annual volumes rather than ceasing to perform mammography, the negative impact on access to mammography (in both rural and urban areas) would be smaller than we estimated.
Lastly and importantly, our study does not address the association of outcomes with volumes or the desirability of changing interpretive volume requirements for physicians, nor does our study address the expected impact on mammographic accuracy if these requirements are changed. We are estimating what would happen to mammographic capacity, in the short term, if interpretive volume requirements were changed.
A movement is growing to expect and pay for quality in health care: "pay for performance." Any decision about physician practice requirements should be based on evidence of the impact of those requirements on mammographic accuracy and patient outcomes. If the FDA decides that it is in the best interest of patients to raise interpretive volume requirements for radiologists, facilities, or both, our study outlines the estimated impact of this change on mammographic capacity and raises concerns over access in rural communities if undergoing mammography remains associated with interpretation.
Acknowledgments
We thank the participating mammography facilities and radiologists for the
data they have provided for this study. A list of the BCSC investigators and
procedures for requesting BCSC data for research purposes are provided at
breastscreening.cancer.gov/.
|
TopAbstractIntroductionSubjects and MethodsResultsdiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  B. M. Geller, E. J. A. Bowles, H. Y. Sohng, R. J. Brenner, D. L. Miglioretti, P. A. Carney, and J. G. Elmore Radiologists' Performance and Their Enjoyment of Interpreting Screening Mammograms Am. J. Roentgenol., February 1, 2009; 192(2): 361 - 369. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. S. Baxi, L. Liberman, C. Lee, and E. B. Elkin Breast Imaging Fellowships in the United States: Who, What, and Where? Am. J. Roentgenol., February 1, 2009; 192(2): 403 - 407. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. R. Melton Are Current Mammography Quality Standards Act (MQSA) Physician Guidelines Truly Adequate? Am. J. Roentgenol., August 1, 2008; 191(2): W79 - W79. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
