Subspecialization in Radiology and Radiation Oncology

Article Outline

• Abstract
• Introduction
• Methods and Materials
• Tabulated Highlights of Survey Results
  • Practice Type and Location of Respondents in Practice and Interests of Those in Training
  • Current Subspecialization and Workload of Respondents in Practice
  • Expanding Subspecialization in Existing Practice
  • Subspecialty Plans of Radiation Oncologists and Radiologists in Training
• Discussion
• Conclusions
• Acknowledgments
• References
• Copyright

Practicing radiologists, radiation oncologists, and trainees were surveyed regarding the current state of subspecialization in practice and in training curricula. The authors present the results of these surveys, establish trends compared with previous survey data, and compare the plans of trainees with current postgraduate practice. Subspecialization is increasing in both radiology and radiation oncology. There remain substantial numbers of practitioners who perform work they deem “general” in nature. The authors also present a method to more accurately measure subspecialization and workload.

Key Words: Subspecialization, radiology, radiation oncology, residents and fellows, workload, surveys

Back to Article Outline

Introduction 

During the past several decades, the complexity of diagnostic and therapeutic radiology has increased dramatically. Fueled by an increasing fund of medical knowledge and technical innovation, what was once a single specialty was split into two separate training and certification programs now known as radiology and radiation oncology [1]. The drive toward subspecialization has continued within both areas, and what was once on the margin of the profession has moved to the mainstream [2].

As the professional organization representing radiology, radiation oncology, and their related subspecialties, it is critical that the American College of Radiology (ACR) understand trends in subspecialization. These data are important for the support of membership, for rational recommendations from ACR leadership, to properly allocate ACR resources, and, most important, to foster high-quality patient care in the context of contemporary medical practice.

With an understanding of these concepts and goals, in January 2008, the chair of the ACR Board of Chancellors, Arl Van Moore Jr, MD, commissioned a study of trends in subspecialization in radiology and radiation oncology. Residents and fellows, as well as postgraduate, professionally active radiologists and radiation oncologists, were included.

Back to Article Outline

Methods and Materials 

The data for this study are the result of a group of surveys conducted from April through October 2008 by ACR economics and health policy staff and independent agencies.

Questions regarding subspecialization were included in a previously planned 2008 survey of radiologists. The ACR provided a random sample of 6,000 postgraduate, professionally active radiologists from their membership records. The term radiology used in the remainder of this report is understood to include diagnostic radiology, interventional radiology, and nuclear medicine. A pretest was conducted to assess the flow and interpretation of the questions. A letter from the ACR president, a practice manager worksheet, and a link to the Web survey were electronically conveyed to radiologists with e-mail addresses (n = 4,893). The remaining radiologists received materials by regular mail (n = 1,029). Interviews were conducted between April 24 and June 13, 2008. Responses were obtained from 1,095 currently practicing radiologists; these physicians were interviewed via a Web survey (n = 575) or computer-assisted telephone interviewing (n = 520). To ensure the confidentiality, the survey was conducted by an outside contractor, dmrkynetec (St Louis, Missouri). The data set delivered to the ACR was stripped of all individual identifiers. The response rate was 26.8% (1,095 complete responses/[5,205 contacted − 911 disconnected or wrong phone numbers − 189 retired or deceased − 18 not qualified]). The margin of error was approximately ±3% at the 95% level.

The data were weighted to be representative of all radiologists in the United States. The characteristics of survey respondents do not necessarily match those of the population about which one would like to make inferences which was in this case all postgraduate, professionally active radiologists in the United States. Accordingly, as is the custom, weights were constructed to be applied to the survey response data from ACR members so that the data would be representative of all US radiologists. Experience has shown that the most important criteria for rendering the survey response pool of ACR members to be representative of all radiologists in the country are age, gender, and census division (ie, the nation divided into 9 geographic areas). These criteria were used to create the weights, which were applied to each survey respondent. For example, if there were half as many women in the response pool as there are among all US radiologists, the data for each woman received a weight of 2.

Surveys of radiation oncologists, radiology residents and fellows, and radiation oncology residents and fellows were conducted between July and October 2008. This portion of our data was collected separately from the 2008 survey of radiologists. E-mail invitations were delivered to potential respondents, asking them to take a short survey using an electronic instrument (SurveyMonkey.com; Portland, Oregon). The results reported for these 3 groups are not weighted and therefore do not necessarily represent all members these populations.

Information about radiation oncologists was collected via a non-random-sample, Internet-based survey conducted by the ACR. All ACR members who are practicing radiation oncologists with e-mail addresses were sent e-mail invitations to take the survey. The American Society for Therapeutic Radiation and Oncology notified its members of the existence of the ACR survey and provided a link to the survey Web site. Responses were obtained from 308 radiation oncologists. The margin of error was approximately ±6%.

Information about radiology residents and fellows was collected via a census-like, Internet-based survey. The names and e-mail addresses of all ACR members who were either residents or fellows were drawn from ACR membership files. Those who indicated that they did not want to be contacted by e-mail were excluded. The remainder were sent an e-mail invitation with a link to the survey Web site. Responses were obtained from 758 radiology residents and fellows. The margin of error was approximately ±4%.

Information about radiation oncology residents and fellows was collected via a census-like, Internet-based survey. The Association of Residents in Radiation Oncology notified its members of the existence of the ACR survey and provided a link to the survey Web site. Responses were obtained from 96 radiation oncology residents and fellows. The margin of error was approximately ±10%.

All survey results were analyzed by ACR staff members and member volunteers.

Back to Article Outline

Tabulated Highlights of Survey Results 

Practice Type and Location of Respondents in Practice and Interests of Those in Training 

Although not the primary focus of our work, data regarding demographic characteristics of respondents are relevant to populations they serve, their ability to aggregate service demand, and therefore their ability to subspecialize. In addition, physicians-in-training were polled regarding their demographic interests in future practice.

Survey data were collected on the distribution of practicing (ie, postgraduate, professionally active) radiation oncologists and radiologists across several demographic categories in 2008. The mean number of radiation oncologists (full-time and part-time combined) in a practice was about 9, and for radiologists, it was 23. Slightly more than half (55.5%) of radiologists were in private radiology practices, followed by primarily academic (18.2%) and multispecialty (16.2%) practices (Supplemental Table S1). Radiologists were generally located in the main cities of metropolitan areas (32.7% large and 26.7% small). Radiation oncologists who responded were more likely affiliated with academic practices in more urban settings; nearly all (79.9%) practiced in large metropolitan areas or main cities of small metropolitan areas. About one-quarter of radiologists were employed by hospitals or health care systems, whereas roughly one-third of radiation oncologists were so employed.

Supplemental Table S1. Practice type and location of radiation oncologists and radiologists

Characteristic	Radiation Oncology	Radiology
Number of practitioners
1	13.0%	3.6%
2–9	59.4%	34.1%
10–20	14.6%	25.6%
≥21	13.0%	36.6%
Practice type
Solo	8.8%	3.6%
Locum tenens	0.0%	1.2%
Primarily academic	32.1%	18.2%
Government	0.6%	2.7%
Private, multispecialty, not primarily academic	12.7%	16.2%
Private radiology, not primarily academic	38.3%	55.5%
Private radiation oncology and radiology combined, not primarily academic	4.5%	N/A
Other	2.9%	2.6%
Employed by hospital or health care system
Yes	32.8%	25.4%
No	67.2%	74.6%
Practice location
Main city of a large metropolitan area (population > 1 million)	33.1%	32.7%
Suburb of a large metropolitan area (population > 1 million)	20.5%	17.2%
Main city of a small metropolitan area (population 50,000 to 1 million)	26.3%	26.7%
Suburb of a small metropolitan area (population 50,000 to 1 million)	5.5%	7.3%
Nonmetropolitan or rural (population < 50,000)	11.4%	12.3%
Varied locations	3.2%	3.8%

Sources: ACR 2008 survey of postgraduate, professionally active radiation oncologists. ACR 2008 survey of postgraduate, professionally active radiologist ACR members.

Note: Results for radiation oncologists are not weighted; the percentage for “solo” is less than the percentage for one radiation oncologist in the practice because several practices with one radiation oncologist were in multispecialty practices with physicians in other specialties. Results for radiologists are weighted to be representative of all US radiologists. Number of practitioners includes full-time and part-time. N/A = not available.

Radiology and radiation oncology residents and fellows were questioned regarding the most desirable practice type and practice location. Forty-one percent of radiology trainees preferred private, multispecialty practices (Supplemental Table S2). Private, radiology-only practices and academic practices were each preferred by one-quarter of these residents and fellows. The remaining practice types received only very small-scale interest. Large metropolitan areas and main cities of smaller metropolitan areas dominated the most desired practice locations. About 61% of survey respondents expected to complete their residency in 2009 or after.

Supplemental Table S2. Desired practice types and locations of radiology residents and fellows

Characteristic	Radiation Oncology	Radiology
Most desired practice type
Solo	3.1%	2.0%
Locum tenens	1.0%	0.4%
Primarily academic	37.5%	26.1%
Government	1.0%	2.3%
Private, multispecialty, not primarily academic	11.5%	40.9%
Private radiology, not primarily academic	36.5%	25.7%
Other	N/A	2.9%
Private radiation oncology and radiology combined practice, not primarily academic	5.2%	N/A
Most desired practice location
Main city of a large metropolitian area (population > 1 million)	45.8%	31.2%
Suburb of a large metropolitian area (population > 1 million)	20.8%	28.1%
Main city of a small metropolitian area (population 50,000 to 1 million)	24.0%	25.3%
Suburb of a small metropolitian area (population 50,000 to 1 million)	6.3%	11.6%
Nonmetropolitian or rural (population < 50,000)	1.0%	4.1%
Varied locations	1.0%	2.4%
Year finished/will finish residency
2006 or before	2.1%	3.6%
2007	7.3%	13.5%
2008	41.7%	22.3%
2009	34.4%	31.3%
2010	14.6%	21.5%
2011 or after	3.1%	7.9%

Sources: ACR 2008 survey of radiation oncology residents and fellows. ACR 2008 survey of radiology residents and fellows who are ACR members.

Note: Results for radiation oncology are not weighted. Results for radiologists are weighted to be representative of all radiologists in the United States. N/A = not available.

Of radiation oncologists in training, 38% said that their most desired practice type was primarily academic. Another 37% preferred private, radiation oncology-only practices. Slightly more than 10% preferred private, multispecialty practices. No other practice type received more than 5% primary interest. Large metropolitan areas and main cities of small metropolitan areas were by far the most preferred practice locations. About three-fourths of radiation oncology residents who took the survey anticipated finishing residency in 2008 or 2009.

Current Subspecialization and Workload of Respondents in Practice 

The surveys of practicing radiologists and radiation oncologists asked how many times respondents had changed the main focus of their work, for example, from one specialty to another, or from subspecializing to being a generalist, or vice versa. About one-fourth of radiologists had changed the main concentration of their work at least once (Supplemental Table S3). In contrast, nearly half of radiation oncologists had changed their main focus at least once.

Supplemental Table S3. Number of times changed main focus of work

Source: ACR 2008 survey of postgraduate, professionally active radiologist ACR members.

Note: Results for radiology are weighted to be representative of all radiologists in the United States.

Twenty-four percent of practicing radiologists did not have main subspecialties. The remainder reported being at least somewhat subspecialized. Table 1 shows percentages for main subspecialties as well as percentages of work time in subspecialties. Neuroradiology (17.1%) was the most frequently cited main subspecialty; other subspecialties with double-digit responses were breast imaging or mammography, interventional or vascular radiology, and body imaging or cross-sectional imaging. As shown in the second column of Table 1, respondents also provided information regarding the amount of time they spent working in particular subspecialty areas. For most subspecialties, the distribution of work time for all respondents correlated well with the distribution of subspecialists. The most conspicuous exception pertained to “general radiology,” claimed by only 1.5% of respondents as a subspecialty, but accounting for 18.9% of overall work time. This outlier may be explained by the nearly one-quarter of respondents who indicated that they were “not specialized” and could therefore be considered “general radiologists.”

Table 1. Radiologists' main subspecialties and work times in subspecialties

Source: ACR 2008 survey of postgraduate, professionally active radiologist ACR members.

Note: Results are weighted to be representative of all radiologists in the United States.

Table 2 shows subspecialization and workload among responding radiation oncologists. The survey asked about radiation oncologists' main subspecialties and the proportions of work time in several subspecialty areas. Subspecialization by radiation oncologists can be categorized by modality and body part or organ system. About 70% of radiation oncologists considered their main subspecialties to be in body parts or organ systems. Prostate (18.9%), breast (17.7%), and head and neck (17.1%) and were the most prevalent; “other” also received relatively numerous responses (6.9%). Among therapeutic modalities, radiosurgery (12.0%) was the most frequent choice. Brachytherapy and intensity-modulated radiation therapy were the only other modalities that were indicated as main subspecialties.

Table 2. Radiation oncologists' main subspecialties and percentages of work time in subspecialties

Subspecialty	Main Subspecialty	Work Time
Body part/organ system
Head and neck	17.1%	14.2%
Prostate	18.9%	21.9%
Gynecologic	4.0%	7.3%
Breast	17.7%	23.6%
Lung	3.4%	14.8%
Pediatric	4.6%	2.2%
Other (thyroid, GI, CNS)	6.9%	16.0%
Modality
Brachytherapy	9.1%	8.5%
Radiosurgery	12.0%	6.3%
Conventional radiotherapy	0.0%	17.3%
3-D conformal radiotherapy	0.0%	32.0%
Intensity-modulated radiation therapy	6.3%	34.6%
Coronary artery brachytherapy	0.0%	0.0%
Other		1.3%

Source: ACR 2008 survey of postgraduate, professionally active radiation oncologists.

Note: Results are not weighted. CNS = central nervous system; GI = gastrointestinal.

In the case of work time spent in subspecialties, an expert panel of survey questionnaire developers believed it best to separate body part or organ system from modality. The questionnaire instructed respondents to provide the percentage of work time such that the total was 100% for body part or organ system and also 100% separately for modality. For body part or organ system, the percentages of time spent in prostate, breast, head and neck, and lung were all in the range of 14% to 24%. Among modalities, Intensity Modulated Radiation Therapy (IMRT) and 3-D conformal radiotherapy both averaged in the 32% to 35% range.

Survey data showed that more than three-fourths (76.6%) of radiologists and nearly as many (69.4%) radiation oncologists were satisfied with the extent of their subspecialization (Supplemental Table S4). Among the remaining physicians, those who would like more subspecialization outnumbered those who would like less by a 5-to-1 margin, with a slightly smaller predominance for radiation oncology. Most radiologists covered absences with practitioners in the same subspecialty. Radiation oncologists were more likely to substitute generalists or practitioners from different subspecialties. Fewer than one-third of radiation oncologists felt pressure to subspecialize, as opposed to half of radiologists. Both groups felt significant motivation to subspecialize on the basis of quality of care and demand from referring physicians. Hospital expectations were mentioned by more than one-third of radiologists but by only 15% of radiation oncologists.

Supplemental Table S4. Subspecialization satisfaction, cross-coverage, and motivations

Response	Radiation Oncology	Radiology
Extent of subspecialization in own work compared with desired work
Would like a lot more subspecialization	4.5%	4.9%
Would like somewhat more subspecialization	20.5%	14.6%
About right	69.4%	76.6%
Would like somewhat less subspecialization	4.9%	2.7%
Would like a lot less subspecialization	0.7%	1.2%
Subspecialty coverage in respondent's absence
A practitioner from a different subspecialty from within practice	19.8%	3.3%
A practitioner from the same subspecialty from within practice	28.4%	64.2%
A generalist from within practice	20.9%	14.8%
A practitioner from a different subspecialty or a generalist from outside practice	3.0%	0.8%
A practitioner from the same subspecialty from outside practice	1.1%	2.0%
A practitioner from outside practice, field not known	0.0%	0.7%
Other	2.6%	N/A
Don't subspecialize	24.3%	14.1%
Practitioner's perception of pressure to subspecialize
Much pressure	3.0%	10.5%
Moderate pressure	28.0%	39.3%
Little or no pressure	69.0%	50.2%
Main sources of motivation to subspecialize (multiple responses allowed)
To improve quality of service to patients	71.5%	57.7%
Demand by referring physicians	39.8%	70.8%
Demand by hospitals	14.6%	34.7%
Demand by payers	1.6%	13.2%
New diagnostic imaging technology	N/A	45.4%
Teleradiology	N/A	12.5%
Other	13.0%	N/A

Sources: ACR 2008 survey of postgraduate, professionally active radiation oncologists. ACR 2008 survey of postgraduate, professionally active radiologist ACR members.

Note: Results for radiation oncologists are not weighted. Results for radiologists are weighted to be representative of all radiologists in the United States. N/A = not available.

Expanding Subspecialization in Existing Practice 

Most radiologists (62.9%) and a minority of radiation oncologists (29.9%) reported recent expansions of subspecialization within their practices (Supplemental Table S5). For radiologists, the addition of new members was the almost universal means to augment subspecialization, compared with 60% of radiation oncologists who reported new practice membership. To a similar degree, both radiologists and radiation oncologists undertook additional training, used other local practices, or had undergone practice consolidation. By virtue of the digital nature of imaging, three-quarters of radiologists moved studies within their groups, and about one-quarter electronically outsourced some studies for second opinions or primary interpretations.

Supplemental Table S5. Subspecialization expansion of radiation oncologists and radiologists

Response	Radiation Oncology	Radiology
Practice has recently increased subspecialization through new members or other means	29.9%	62.9%
Method of increasing subspecialization (multiple choices allowed)
Existing practitioners undertook subspecialty training	56.2%	64.0%
Hired additional practitioners	59.4%	95.2%
Electronic intrapractice image distribution	N/A	74.7%
Electronic extrapractice image distribution (outsourcing)	N/A	24.6%
Consolidated with other practices	27.9%	29.1%
Used other local practices to do some subspecialty work	22.2%	24.6%
Other	40.0%	N/A
Subspecialty training for existing practitioners
Courses, on-job training, not involving multiweek absences from practice	84.7%	76.1%
Fellowship or minifellowship away from practice	27.7%	61.2%

Sources: ACR 2008 survey of postgraduate, professionally active radiation oncologists. ACR 2008 survey of postgraduate, professionally active radiologist ACR members.

Note: Results for radiation oncologists are not weighted. Results for radiologists are weighted to be representative of all radiologists in the United States. N/A = not available.

Subspecialty Plans of Radiation Oncologists and Radiologists in Training 

Table 3 lists residents and fellows' plans regarding subspecialization. The differences here between the specialties are striking. Over 90% of radiologist trainees planned on fellowship training, whereas fewer than 1 in 10 radiation oncology residents pursued this pathway. Despite the lack of additional training, many radiation oncology residents (42.7%) still wished to subspecialize. About 67% of radiology trainees wished to work in 2 or more areas once in practice. About three-quarters of radiology trainees felt moderate or much pressure to subspecialize. Conversely, three-quarters of radiation oncology trainees felt little or none of this stress.

Table 3. Subspecialty plans and views of radiation oncology and radiology residents and fellows

Response	Radiation Oncology	Radiology
Intend to pursue a fellowship	7.3%	91.5%
Choice of whether to do a fellowship (multiple choices allowed)
Do not want to, but likely necessary to get a job	6.3%	14.1%
Always wanted and planned to do a fellowship	6.3%	80.1%
Compelled to by recent proposed changes to ABR boards	N/A	8.0%
Do not want to, but encouraged by residency program	0%	8.9%
Plan to subspecialize
Yes	42.7%	89.9%
No	26.0%	5.9%
Too early in residency to know	31.3%	4.2%
Desired degree of subspecialization to practice
One area (ie, modality, organ system)	N/A	26.6%
Two or 3 areas	N/A	52.7%
More than 3 areas	N/A	14.8%
Desired percentage of clinical work time at most prominent subspecialty	100.0%	61.5%
Amount of pressure felt to subspecialize
Much pressure	1.0%	27.1%
Moderate pressure	22.9%	51.9%
Little or no pressure	76.0%	21.0%
Main sources of pressure to subspecialize (multiple mentions allowed)
Intended or prospective employers	21.9%	62.0%
Improve quality of service to patients	41.7%	60.3%
Faculty encouragement	25.0%	44.6%
New radiation oncology technology	30.2%	41.5%
Referring physicians	2.1%	22.3%
Hospitals	0.0%	13.9%
Payers	1.0%	9.9%
Subspecialization desirability in light of competition from other specialties
Likely beneficial to profession	47.9%	51.7%
Subspecialization not necessary, makes no difference	31.3%	5.0%
Imperative for continuation of the profession	10.4%	36.2%
Likely to harm the profession	7.3%	5.3%
Will ultimately result in demise of the medical specialty	3.1%	1.8%

Sources: ACR 2008 survey of radiation oncology residents and fellows. ACR 2008 survey of radiology residents and fellows who are ACR members.

Note: Results are not weighted. N/A = not available.

Radiology residents and fellows were asked to indicate the fields in which they planned to subspecialize after the completion of training. The top 6 subspecialties among residents and fellows included the top 5 main subspecialties of postgraduate radiologists (Supplemental Table S6). No single subspecialty garnered more than 23% of choices. Four percent of residents and fellows said that they did not plan to subspecialize.

Supplemental Table S6. Preferred subspecialty areas of radiology residents and fellows (multiple mentions allowed)

Source: ACR 2008 survey of radiology residents and fellows who are ACR members.

Note: Results are not weighted.

Radiation oncology residents and fellows were surveyed regarding their preferred subspecialty areas. Fifty-six percent said that they planned to subspecialize in particular fields (Supplemental Table S7). (Forty-four percent said that they did not plan to subspecialize or that it was too early in residency to make a decision.) The most frequently cited subspecialties were radiosurgery (24%) and head and neck (19%) (multiple selections were allowed). The least frequently mentioned were 3-D conformal radiotherapy (2%) and coronary artery brachytherapy (0%).

Supplemental Table S7. Preferred subspecialty areas of radiation oncology residents and fellows (multiple mentions allowed)

Source: ACR 2008 survey of radiation oncology residents and fellows.

Note: Results are not weighted.

We compared the plans of residents and fellows with the main subspecialties of practicing radiologists and the percentage of work time in subspecialties of practicing radiologists. These are presented in Table 4. The subspecialty preferences and practice percentages were adjusted proportionally from those presented in previous tables to add to 100% and facilitate comparisons. The plans of residents and fellows matched reasonably well with the main subspecialties of practicing radiologists. Musculoskeletal radiology and abdominal, body, chest, or cross-sectional imaging were the main outliers, in that residents and fellows had a greater preference for these specialties compared with the main subspecialties of practicing radiologists. Substantially more time was spent in general radiology by practicing radiologists compared with the preferences of residents and fellows. However, the definition of “general radiology” was not subject to the same degree of precision and agreement across radiologists as other subspecialties, perhaps in part because the survey instruments did not define the term general radiology.

Table 4. Radiologist subspecialization and plans of residents and fellows

Sources: ACR 2008 survey of radiology residents and fellows who are ACR members. ACR 2008 survey of posttraining, professionally active radiologists who are ACR members.

Note: Results for radiologists are weighted to be representative of all radiologists in the United States. Results for residents and fellows are not weighted. Responses of “no subspecialty” were deleted and the percentages for the named subspecialties rescaled to add to 100%. Some subspecialties are combined that are presented separately in other tables.

Using trainees' preferred subspecialties as a surrogate for “future supply,” the current distribution of subspecialists as “current supply,” and the work time in particular areas as “demand,” we juxtaposed the data, as shown in Table 5. For abdominal, body, and cross-sectional imaging, cardiac, emergency, and general radiology, demand is greater than supply. In these subspecialties, the market seems to be reacting as expected by showing an increasing future supply of providers. When supply is greater than demand, an economic market model would predict a weakening in future supply. This is the case for interventional or vascular radiology, mammography or women's imaging, neuroradiology, and nuclear medicine. These simple market relationships are not present in the subspecialties of magnetic resonance imaging, musculoskeletal, neurointerventional, and pediatrics. In these fields, current supply is greater than demand at the same time as there is increased interest in undergoing a fellowship. In the subspecialty of ultrasound, current demand exceeds both current and future supply.

Table 5. Radiologist subspecialization supply and demand

Sources: ACR 2008 survey of radiology residents and fellows who are ACR members. ACR 2008 survey of postgraduate, professionally active radiologists who are ACR members.

Note: Results for radiologists are weighted to be representative of all radiologists in the United States. Results for residents and fellows are not weighted. Responses of “no subspecialty,” “not involved in any clinical activity,” and “other specialty” were deleted and the percentages for the named subspecialties rescaled to add to 100%. Some subspecialties are combined that are presented separately in other tables.

Table 6 presents subspecialty comparisons between practicing radiation oncologists and radiation oncology residents and fellows. The subspecialty preferences and practice percentages were adjusted proportionally from those presented in previous tables to add to 100%. The greatest excesses of the preferences of residents and fellows over the main subspecialties of practicing radiation oncologists were for radiosurgery, IMRT, and lung. The opposite relationship exists for prostate and breast. Comparing residents' and fellows' preferences with those of practicing radiation oncologists for work time, the largest mismatch was for radiosurgery. At the opposite end of the spectrum, practicing radiation oncologists spent far more work time in 3-D conformal radiotherapy than was preferred by residents and fellows.

Table 6. Radiation oncologist subspecialization and plans of residents and fellows

Sources: ACR 2008 survey of radiation oncology residents and fellows. ACR 2008 survey of postgraduate, professionally active radiation oncologists.

Note: Results are not weighted. Responses of “no subspecialty” were deleted and the percentages for the named subspecialties rescaled to add to 100%. Some subspecialties may be combined that are presented separately in other tables.

We examined the linkage between main subspecialty and work time for the 5 most frequently cited main subspecialties of the radiology respondents. The results are presented in Table 7.

Table 7. Five most common main radiology subspecialties and work time spent in subspecialty areas by radiologists

Source: ACR 2008 survey of posttraining, professionally active radiologist ACR members.

Note: Results are weighted to be representative of all posttraining, professionally active radiologists in the United States.

For each of the 5 most frequently cited main subspecialties, the same subspecialty was the one in which the radiologists spent most of their work time. There was considerable variation, however, in the percentage of work time spent in that subspecialty. Slightly more than half of work time was spent in neuroradiology and breast imaging or mammography by radiologists with the matching main subspecialties. Those who indicated that magnetic resonance imaging was their main subspecialty, however, spent only 29% of their work time doing magnetic resonance imaging. Body imaging and cross-sectional imaging fell between the two extremes, with 40% of work time being spent in body imaging or cross-sectional imaging by those for whom that was their main subspecialty.

For all 5 of the top main subspecialties, general radiology was the area in which the second most work time was devoted. Similarly, body imaging or cross-sectional imaging was the third most frequently reported work area (with the exception of those who indicated that body imaging or cross-sectional imaging was the main subspecialty). Musculoskeletal radiology, abdominal imaging, and ultrasound tended to be cited among the fourth and fifth most frequent subspecialty work areas, even though none of these was among the top 5 in terms of main subspecialty.

Table 8 shows the distribution by size (ie, number of radiologists in the practice) of radiologists with various characteristics. Only the main results of interest are highlighted here. With respect to practice type, smaller practices involved mainly radiologists in private, radiology-only groups. Relatively few radiologists were in small academic practices. As practice size increased, the percentage of radiologists in private, radiology-only practices dropped, while the percentage of radiologists who were in academic practices rose. Larger practices tended to be located in metropolitan areas or the main cities of smaller metropolitan areas; smaller practices were in nonmetropolitan or rural areas. Radiologists were much more likely to subspecialize if they were in larger practices. Radiologists' work time in general radiology decreased as practice size increased.

Table 8. Characteristics of radiologists by practice size

Characteristic	Number of Radiologists in the Practice (%)
	1	2 to 10	11 to 20	≥21
Practice type
Solo	100.0	0.0	0.0	0.0
Locum tenens	0.0	0.7	0.4	0.3
Primarily academic	0.0	4.0	10.6	36.5
Government	0.0	4.3	3.7	1.0
Private multispecialty	0.0	16.5	17.9	17.2
Private radiology	0.0	70.7	65.2	43.4
Other practice type	0.0	3.8	2.3	1.6
Location
Main city of large metropolitian area	11.9	18.2	27.9	51.1
Suburb of large metropolitian area	9.2	17.5	20.3	16.4
Main city of smaller metropolitian area	25.7	24.8	40.0	25.0
Suburb of smaller metropolitian area	2.5	11.6	9.8	2.3
Nonmetropolitian or rural	43.5	24.7	5.8	2.5
Varied locations	7.2	3.2	2.2	2.7
Region
Northeast	16.7	19.3	28.0	22.7
Midwest	10.8	19.5	16.1	28.1
South	48.2	38.9	31.8	29.8
West	24.3	22.3	24.1	19.3
% who subspecialize	39.3	62.3	81.8	89.1
% of work time in general radiology
0	23.3	38.3	32.1	52.3
1 to 20	29.8	32.9	40.9	30.5
21 to 40	6.5	6.3	12.5	7.3
41 to 60	5.9	7.4	5.8	3.9
61 to 80	5.5	3.6	4.3	2.0
81 to 100	26.0	11.5	4.5	4.0

Source: 2008 ACR survey of posttraining, professionally active radiologist ACR members.

Note: Results are weighted to be representative of all post-training, professionally active radiologists in the United States. Number of radiologists includes full-time and part-time.

The subspecialty results from the 2008 survey of radiologists may be compared with a similar survey conducted in 2003 [3]. Table 9 shows the percentage who reported particular subspecialties in each of the 2 years, as well as relative percentage changes from the earlier to the later year. Most subspecialties grew, with musculoskeletal imaging and women's imaging having the largest relative increases. Five subspecialties had no measurable changes. The subspecialties of ultrasound and pediatric imaging were the only two for which there were decreases.

Table 9. Main subspecialty of radiologists in 2003 and 2008

Sources: ACR 2003 survey of radiologists. ACR 2008 survey of postgraduate, professionally active radiologists who are ACR members.

Note: Entries exclude radiologists who indicated that they did not have main subspecialties. Data for 2008 are weighted to be representative of all radiologists in the United States. Rows are arranged in descending order of 2003 to 2008 difference. For calculating the 2003 to 2008 difference, values ≤1.0% were set to 1.0%. N/A = not available.

Back to Article Outline

Discussion 

Discussion, debates, and disagreements about subspecialization in radiology are as old the specialty. A number of articles have been published regarding the imperatives for subspecialization, while others have argued for the importance of general medical practice [4, 5, 6].

As the practice of medicine evolves, the demands on individual radiologists, radiation oncologists, and the medical profession as a whole will continue to evolve. As reflected in the survey data, the practice of radiology is in a watershed period. The specialty and its environment vary by size, patient and referring physician expectations, referral patterns, and to some degree geographic constraints. On the basis of survey results over the long term, there has been a distinct trend toward the practice of and training for subspecialization. The marketplace continues to make greater demands for subspecialty expertise. Although not measured by our study, demand seems to vary according to credentialing, geography, practice size, and hospital location.

Motivations for subspecialization may be categorized as external or internal. External factors include demand by referring physicians and hospitals, turf battles, and competition for patients. Internal motivations include personal satisfaction, academic interests, and the desire to have areas of expertise. It is important for radiologists and radiation oncologists to recognize where they are on this spectrum and respond accordingly. Small groups may be pushed beyond their internal capacity to respond and may need to develop referral relationships with larger practices or other centers. Our collective experience indicates that this process, in the absence of external edicts or third parties, is happening in an organic way. From this standpoint, group size may not be a barrier to offering subspecialty-level expertise. Radiologists' groups continue to transform to facilitate subspecialization by expanding existing members' talents, retraining members, targeting new hires, and electronic image transmission.

The mean number of radiation oncologists in a practice is about 9, compared with 23 for radiologists. This inhibits, to some degree, subspecialization by radiation oncologists; most members of a group need to remain generalists. For group sizes of 2 or 3 radiation oncologists, this limitation is important. For intermediate-sized groups of 4 to 8 members, subspecialization may begin to occur, but members will still cover multiple areas (and all deal with palliation, metastatic disease, etc). One radiation oncologist may cover gynecology and breast, another central nervous system and lymphoma, and so on. The greatest degree of subspecialization in these groups relates to technology, such as an individual doing primarily brachytherapy or radiosurgery (the utilization of 3-D conformal radiotherapy and IMRT is assumed universal).

Most large groups of radiation oncologists cover multiple facilities, a circumstance that also limits specialization (except for brachytherapy and radiosurgery). In general, large academic practices house radiation oncologists who deal with only one disease site (eg, lung, central nervous system, lymphoma). Radiation oncologists in practice may be more likely to expand their groups with an emphasis on the utilization of new technologies, such as radiosurgery or novel brachytherapy procedures.

A distinction between radiation oncology and radiology is that there are no certificates of added qualifications or Accreditation Council for Graduate Medical Education–accredited fellowships in the former. Even nonaccredited fellowships are uncommon. As a result, fewer than 1 in 10 radiation oncology trainees plan to undertake fellowships. In contrast, radiology residents and fellows nearly all pursue some type of additional subspecialty training. Many of the pressures radiology trainees experience parallel those of their colleagues in practice. These residents may also be reacting to the increasing complexity of the field, as more than three-quarters indicated an independent intent to pursue fellowship.

There is also a new external stress prompting radiology residents to consider additional subspecialty training. In October 2007, the American Board of Radiology announced a requirement that residents would be able to spend their fourth year in general rotations or up to 3 subspecialty areas of training. These in turn will determine the content of their final certifying examinations [7]. At this point in time, radiation oncology does not have a similar examination structure.

Academic radiology practices are increasingly being pushed to have specific experts in each of the 9 subspecialty areas. As of July 1, 2008, the Accreditation Council for Graduate Medical Education Residency Review Committee (RRC) and new program requirements require that 9 separate subspecialty faculty oversee each of these areas [8].

Private radiology practices are increasingly expecting radiology residents to have completed fellowships to provide new expertise for the practice. Radiation oncology also uses new members to supply subspecialty expertise, but not to the same degree.

For trainees in both radiology and radiation oncology, there are occasional imbalances between the expectations of residents and fellows (trainees) and their counterparts in practice. If one uses fellowship training as a proxy for a desired area of subspecialization, a potentially serious mismatch might occur in what subspecialties are desired by trainees and what radiologists actually practice. For instance, as shown in Table 4, for abdominal, body, chest, and cross-sectional imaging added together, 23.4% of trainees desire these subspecialties, whereas only 17.1% of practicing radiologists currently consider these as their subspecialty. Similarly, 10.5% of trainees desire magnetic resonance compared with 7.5% practicing radiologists, 13.1% of trainees desire musculoskeletal compared with 6.3% practicing radiologists, and 4.4% of trainees desire pediatric radiology compared with only 2.7% of practicing radiologists considering that as their subspecialty. Are these apparent differences only a reflection of a marketplace supply and demand, as suggested by the data in Table 4, Table 5? Could these disparities cause incongruities between the needs of the medical market and the skills supplied by upcoming trainees? If the latter is true, some trainees may need to redirect or retrain. Although 79% of radiology residents feel pressure to subspecialize and more than 90% do fellowships, it is noteworthy that 73.4% do not seek to limit themselves to a single subspecialty.

To be valuable to a practice, residents or fellows must bring new expertise in an area of marketplace demand. Thus, those trained in new areas such as musculoskeletal magnetic resonance imaging might find an advantage in the job market. Ultimately, the job market for new radiologists will determine which fellowships make residents most competitive for new positions.

In many, if not most practice settings, both radiologists and radiation oncologists should anticipate variety. Despite subspecialty interests, radiation oncologists currently treat across a spectrum of modalities and organ systems. About one-quarter of practicing radiologists do not consider themselves subspecialists. Of those who do consider themselves specialized, only about half of their professional time is spent in their primary subspecialties. Ten percent to 20% of their time is in general radiology, and the remainder in other non-primary-subspecialty areas. Even if smaller groups can practically and economically outsource their specialty work and call responsibilities electronically, “general radiology,” however it is defined, will probably remain a viable and necessary part of radiology in the future. It is here that opportunities may exist for larger practices and academic radiology departments to develop beneficial subspecialty referral relationships with smaller practices.

In considering the topic of subspecialization, definitions pose a significant limitation. Our survey suffers from this pitfall, as the details of credentialing for a subspecialty were not specified. In addition, the term general radiology or generalist may have been perceived as a category of exclusion. Just what is the definition of general radiology? Did our respondents infer these words to mean “everything outside of my primary areas of expertise”? The ambiguity and imprecision of these words may well reflect the confusion that we as a profession have regarding the role and future of “general radiology.”

The definition of subspecialization is a variable for which we did not control. There is certainly a trend that more recent radiology residents have elected to invest additional training in fellowships for a variety of reasons. Although identification for those who continue to practice in the fields of their fellowships is better defined, whether or not they have certificates of added qualifications, those who have either varied their areas of interest or graduated to given fields during their courses of practice without formal credentialing remain less clear. For instance, many “general radiologists” who do not do interventions or perform breast imaging may now identify themselves as “cross-sectional imaging” specialists rather than generalists. Does satisfying the Mammography Quality Standards Act requirements for breast imagers certify that one is a specialist? And what of those who have achieved certificates of added qualifications but have not practiced their subspecialties?

Another point of confusion is the designation of proportions for subspecialty work. Some responders to our survey may have relied on time as the basis of measuring work, whereas others may have used effort or relative value units. Furthermore, there is probably significant crossover of studies that can be categorized as falling either within several subspecialties or within general radiology. Instead of objective criteria, or even definitions yet to be determined by arbiters such as credentialing bodies, the survey depended on respondents' interpretations and “perceptions” of their work distribution. By their very nature, surveys are victim to the subjective judgment of respondents [9].

Although the ideal of subspecialization can be debated, the practical applications are left to the variety of practices across the country. Practice size is a critical determinant. Although larger practices may be better equipped to accommodate the internal or external demands of subspecialization, the critical number of radiologists needed for adequate subspecialty coverage has yet to be determined. For both radiologists and radiation oncologists, the personal expectations or those of the institutions in which they practice will determine the degree of subspecialization manifest. These data indicate that a more important contributor for radiologists is the expectation of referring physicians. Unfortunately, these expectations were not measured directly. These demands may vary depending on the time of day; as groups expand and potentially acquire very focused subspecialists, they may also have trouble in staffing on-call duties. Practices may have to change traditional compensation models to provide incentives for skill maintenance. The distribution of vacation and continuing education time may be altered to provide time for skill acquisition.

Radiology and radiation oncology practices are also challenged to address the balance of compensation versus productivity. Historically, many groups, at least in private practice, equated time with compensation: those who worked an equivalent number of hours would realize an equivalent share of the profits of the practice. However, as subspecialization becomes more pervasive, along with the acceptance of the varying lifestyle demands of those in practice, the compensation formulas may need to become more complex. For example, survey data from the Society of Chairmen of Academic Radiology Departments clearly indicate that radiologists who predominantly interpret cross-sectional imaging studies are more productive than those reviewing plain films [10]. Groups that look to relative value units as the basis for compensation will need to consider these inherent productivity differences. Radiation oncology practices may find similar differences on the basis of patient diagnosis, condition, and coexisting disease.

As professional disciplines, radiology and radiation oncology may not want to deem practitioners who have attended weekend courses, or even minifellowships, as having subspecialty expertise. The categories established by credentialing boards may provide clearer guidelines regarding “credentialing” for those who require maintenance of certification. These categories are unlikely to be exhaustive. Should organized radiology and radiation oncology attempt to provide a fair definition of situations in which a non-fellowship-trained practitioner might be legitimately termed a subspecialist?

The current state of the “marketplace” and the reactions of practitioners and trainees are important indicators of the current and future trends of radiology and radiation oncology. There are no stronger variables shaping medical practices and organizations than the demands of the marketplace. Currently, horizontal integration, including mergers and consolidation, seem to be popular in certain markets. Nonetheless, the data suggest that smaller practices and multispecialty opportunities remain in strong demand by candidates seeking jobs.

We acknowledge shortcomings in this survey and work. From its inception in January 2008, data collection extended over a 9-month period. Fortunately, data collection for practicing radiologists dovetailed with a larger and previously planned survey under way by the ACR economics and health policy staff. This portion of the data was carefully selected and controlled and carries with it the greatest statistical validity.

Radiology and radiation oncology trainees responded in significant numbers to e-mail invitations to a separate poll. Data collection for trainees was occasionally hampered by access to correct electronic contact information. The ACR has an ongoing need to communicate with its membership, and to this end, we applaud ongoing efforts to maintain close contact with individual members and their practices. Recently, the college instituted My Profile and the Practice of Radiology Environment Database (PRED) to better serve the discipline [11].

Practicing radiation oncologists were challenging to contact, in part because of the migration of younger practitioners to other professional organizations. The ACR may be further served by proactive alliances with other professional societies, with the aim of improving communications among constituents with common interests.

As reflected by the data, radiologists, radiation oncologists, and trainees all see quality improvement as an important driver for subspecialization. Credentialing boards and professional organizations, including the ACR, share similar perceptions. Unfortunately, their collective ability to measure marketplace reality is limited. Objective data reflecting the value added to patients and practitioners, much less the work done by a given practitioner or group, is also limited.

Organizational leadership acts, and memberships react, on the basis of personal or limited collective experience rather than more objective and broad measures of service delivery and patient outcome. Does the ACR have the ability to better measure subspecialization among its membership? Certainly. The College already hosts a robust informatics infrastructure. These facilities are used in conjunction with a variety of registries, such as the National Radiology Data Registry [12]. The voluntary collection of aggregate Current Procedural Terminology^® code submissions by members would be relatively easy in comparison with other ongoing projects and would leave the ACR with objective data regarding practitioners' actual care delivery. Using this information—the actual care delivered—would immediately eliminate the variable of definitions, concepts of work effort, and many of the inaccuracies inherent in data collected through a polling process.

Back to Article Outline

Conclusions 

Subspecialization in the field of radiology is increasing for a majority of respondents, but only a minority of radiation oncologists are so affected. Both specialties are using similar mechanisms to accommodate this trend. Radiologists identify significantly more stress from referring physicians to subspecialize, and in response, radiology trainees far outnumber radiation oncologists in their pursuit of fellowship training.

Despite the growth of subspecialty care, generalists remain a substantial portion of practitioners in both fields. A significant amount of subspecialists' time is devoted to work they describe as “general.” In addition to studies such as this work, the ACR has additional opportunities to support its membership and their patients with improved data collection techniques.

Back to Article Outline

Acknowledgments 

We wish to thank Jim Moser and Jonathan Sunshine for their excellent work, commentary, and contributions. We wish to thank Christie Ji for assistance with the statistical analysis and tables and Stephanie Le for coordinating workgroup meetings. We thank all those who replied to the surveys; their responses have provided valuable information for the entire radiology and radiation oncology professions.

Back to Article Outline

References 

1. Linton OW. The American College of Radiology: the first 75 years. Reston, Va: American College of Radiology; 1997;
   • View In Article
2. Yee KM. Radiology subspecialization: From margins to center. Available at: http://www.auntminnie.com/index.asp?Sec=sup&Sub=imc&Pag=dis&ItemId=81988&wf=1236. Accessed August 7, 2008.
   • View In Article
3. Meghea C, Sunshine JH. How much do radiologists and radiation oncologists specialize?. J Am Coll Radiol. 2005;2:906–913
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (94 KB)
   • CrossRef
4. Atlas SW. Embracing subspecialization: the key to the survival of radiology. J Am Coll Radiol. 2007;11:752–753
   • View In Article
5. Williams CR. Re: “Embracing subspecialization: the key to the survival of radiology.”. J Am Coll Radiol. 2008;5:229–230
   • View In Article
   • Full Text
   • Full-Text PDF (54 KB)
   • CrossRef
6. Luttenton CR. Re: “Embracing subspecialization: the key to the survival of radiology.”. J Am Coll Radiol. 2008;5:230–231
   • View In Article
   • Full Text
   • Full-Text PDF (47 KB)
   • CrossRef
7. American Board of Radiology. Overview of ABR changes: the exam of the future. Available at: http://theabr.org/Images/overview_changes.pdf. Accessed October 7, 2008.
   • View In Article
8. Amis ES. New program requirements for diagnostic radiology: update and discussion of the more complex requirements. AJR Am J Roentgenol. 2008;190:2–4
   • View In Article
   • CrossRef
9. Groves RM, Fowler FJ, Couper MP, Lepkowski JM, Singer E, Tourangeau R. Survey methodology. Hoboken, NJ: Wiley-Interscience; 2004;
   • View In Article
10. Ying L, Shao S, Chu P, Arenson RL. An update survey of academic radiologists' clinical productivity. J Am Coll Radiol. 2008;5:817–826
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (247 KB)
    • CrossRef
11. Moore AV. Recognizing the ACR's silent service. J Am Coll Radiol. 2008;5:153–154
    • View In Article
    • Full Text
    • Full-Text PDF (82 KB)
    • CrossRef
12. American College of Radiology. National Radiology Data Registry (NRDR). Available at: http://www.acr.org/SecondaryMainMenuCategories/quality_safety/NRDR.aspx. Accessed October 8, 2008.
    • View In Article