| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1
Research Department, American College of Radiology, 1891 Preston White Dr.,
Reston, VA 20191.
2
Committee on Radiologist Resources, American College of Radiology, Reston, VA
20191.
3
Department of Diagnostic Imaging, Brown University Medical School, Box G,
Providence, RI 02912.
4
Department of Diagnostic Imaging, Rhode Island Hospital, 593 Eddy St.,
Providence, RI 02903.
Received August 10, 2001;
accepted after revision August 27, 2001.
Address correspondence to J. H. Sunshine.
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. In spring 2000, we surveyed 3,027 randomly selected radiologists by mail, of whom 74% responded. We weighted responses to make answers representative of all radiologists and compared findings with a similar 1995 survey.
RESULTS. Thirty percent of radiologists age 65-69 were working full-time; 21% were working part-time. Overall, the full-time equivalency of radiologists age 55-74 was 67.6%, not significantly different from 1995. If current patterns of retirement and production of graduates continue, the workforce will grow at a rate of approximately 2% annually. Fifty-one percent of radiologists said that recognizing that income depends largely on work done, they had "much too much work" or "somewhat too much work"; 5% reported "somewhat too little work" or "much too little work." Six percent of posttraining professionally active radiologists were in solo practice, down from 8% in 1995; 15% were in two-to-four-radiologists groups, down from 17%; and 38% were in groups of 15 or more, up from 30%. Sixteen percent of posttraining professionally active radiologists were women. The percentage was highest (29%) for those younger than age 35 but was lower (22%) among trainees.
CONCLUSION. The findings of excess work are further evidence of a radiologist shortage. However, contrary to surveys of groups that are hiring, we found minimal evidence of earlier retirement. Nonetheless, workload currently is increasing faster than the workforce is likely to grow, so the shortage will probably intensify. The typical number of radiologists in a practice is increasing, but slowly.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
First, we report characteristics of radiologists and the practices in which they work, including findings on changes in practice size. Then, we report information on early retirement and on how actual workload compares with that which radiologists desire.
Particularly in addressing the issues, changes over time should be analyzed. We, therefore, make extensive reference to information from the ACR's 1995 Survey of Radiologists (hereafter called the "1995 Survey") [13, 14, 16, 17] comparing information that we report herein from the 2000 Survey of Diagnostic Radiologists and Radiation Oncologists (hereafter called the "2000 Survey") with information from the 1995 Survey.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
A total sample of 3,545 was received. We reduced the list to our desired sample size of approximately 3,000 by random deletions from the list. In late March 2000, we mailed the survey, which consisted of one page with six questions, to these 3,027 diagnostic radiologists and nuclear medicine physicians. Nonrespondents were contacted with up to three remailings, as necessary, at monthly intervals. The last remailing took place in June and was conducted by United States Postal Service Priority Mail, which uses a large red, white, and blue envelope. A total of 2,241 valid responses was received for a response rate of 74%. We also received information that eight of the physicians in the sample were deceased, with the deaths concentrated among those age 70 and older. Survey operations were conducted for the ACR by The Blackstone Group (Chicago, IL).
Responses were weighted so that weighted statistics would be representative
of the answers that would have been received if all diagnostic radiologists
and nuclear medicine physicians in the United States had been surveyed and had
responded. To achieve this representation, diagnostic radiologists were
divided into 48 strata: six age groups (<35 years, 35-44, 45-54, 55-64,
65-69, and 
70 years) each separated by the four census regions (Northeast,
Midwest, South, and West), and these categories further divided by whether the
surveyed physician was an ACR member or not. In previous surveys, we found
differential response rates according to characteristics of age, region, and
ACR membership, and a logistic regression analysis of who had responded to
this survey also showed statistically significant variation according to these
three characteristics. A different weight was assigned to each of the 48
strata. The weight for each was calculated as (number in the stratum in the
sample provided by Medical Marketing Service x 10) ÷ (number of
responses from the stratum). For example, if there were eight in a given
stratum (e.g., radiologists age 65-69 years in the South who were not ACR
members) in the Medical Marketing Service sample and five responses, the
weight would be (8 x 10) ÷ 5 = 16. The logic is that if there
were eight radiologists in this stratum in the sample, then because it was a
one-in-10 sample, there were 80 radiologists with these characteristics in the
United States. With five responses, weighting each response as 80 ÷ 5
makes the responses added together total the 80 such radiologists estimated to
be in the United States.
Because the number of nuclear medicine physicians is only approximately 5% the number of diagnostic radiologists, and the number in the sample was, therefore, only approximately 5% as large, we divided the nuclear medicine physicians into only four strata—those certified by the American Board of Radiology (ABR) and those not so certified, each certification status then divided into ACR members and nonmembers—and then followed the same process of weighting each of the four strata separately.
In reporting and analyzing data, we included only nuclear medicine physicians who had a major connection to diagnostic radiology. We defined "major connection" in this context as either being certified by the ABR (regardless of certification by the American Board of Nuclear Medicine), being a member of the ACR, or both. We eliminated nuclear medicine physicians who did not have a major connection to diagnostic radiology, as so defined. This process eliminated approximately 60% of the nuclear medicine physicians. Note that we included all physicians whose self-described primary specialty was nuclear radiology; these were part of the diagnostic radiology sample. Except as otherwise explicitly indicated, we will use the terms "diagnostic radiologists" and "radiologists" to refer to both diagnostic radiologists per se and those nuclear medicine physicians who were retained in the analysis.
All responses from outside the 50 states and the District of Columbia were deleted. The logistic regression analyzing the factors associated with response and nonresponse showed a small, not statistically significant, difference in response rates between men and women. To adjust for this, we multiplied the previously described weights by 1.0135 for women and by 0.997 for men. Because the Physician Masterfile is apparently slow to record deaths, we multiplied the weights for respondents age 70 and older by 0.99 to approximately compensate for the lag in recording deaths. Because the Masterfile is slow to record board certification, we restricted analyses involving board certification to radiologists 45 years and older.
For analyses involving practices, we divided the final weight for each radiologist, as described in the preceding paragraph, by the size of the respondent's practice. For example, if a radiologist was a member of a 10-radiologist group, the individual weight was divided by 10. These practice weights made the data set representative of all radiology practices in the United States. We have described this practice weighting process previously [4]. Practice size was defined as the number (not full-time equivalence) of posttraining diagnostic radiologists (residents and fellows excluded) in the practice. In the rare cases in which radiation oncologists were members of the same group, or same department of a multispecialty group, they were included.
The item response rate—that is, the percentage of respondents answering any given question—was high, 97% or 98%. If we report numbers, rather than proportions or percentages, they have been augmented to compensate for item nonresponse.
Analysis of Issues Related to the Radiologist Shortage
To study whether diagnostic radiologists were retiring earlier than
previously, we measured the fulltime equivalency of radiologists age 55-74
years, comparing the finding for 2000 with that for 1995. Age 55-74 is the
appropriate age range to study because it is the age range in which reduction
of work and retirement largely take place. It forms the entire age range in
which the fraction of radiologists not working full-time is substantial, but
the fraction not retired is also substantial. Our 1995 Survey showed that the
average hours worked by a self-described part-time radiologist were
approximately 70% of those worked by a self-described full-time radiologist
[13]. Therefore, to measure
full-time equivalency, we counted each full-time radiologist as 1.0, each
part-time radiologist as 0.7, and each retired (or otherwise not working in
radiology) radiologist as zero. We compared this full-time equivalency in 2000
with the situation in 1995, as ascertained by the 1995 Survey. Because changes
could be caused by change in the age and sex mix of radiologists within the
age 55-74 range as well as by genuine changes in work behavior, we
standardized the 1995 observations to the 2000 population characteristics
using the method epidemiologists call "direct standardization"
[21,
22], creating eight cells;
these consisted of four 5-year age categories (e.g., 55-59 years, 60-64) for
each sex. We measured the full-time equivalency in 1995 for each of these
eight cells and applied that full-time equivalency to the actual 2000
population to obtain a measure of what full-time equivalency would have been
if the 2000 population had been at work with the work characteristics of the
1995 population.
As part of our analysis, we needed to estimate the number of retirements
that would have taken place in 1995 and in 2000, assuming the age- and
sex-specific pattern of work versus retirement remained constant. We divided
the population of radiologists in 1995 and in 2000 into 5-year age categories,
beginning with 45-49 years and ending with 75-79; 35-44 years and 80 years and
older were also used as categories. We measured the difference in the
percentage of radiologists who were not working between each age category and
assumed that this percentage of radiologists would retire when they crossed
the boundary from one age group to the next. We estimated the number of
radiologists who crossed a boundary each year as one tenth of the sum of the
number in the 5-year age group below that boundary plus the 5-year age group
above the boundary. For example, we estimated the number turning age 50 as one
tenth of the sum of the number of radiologists age 45-49 years plus the number
age 50-54 years. (For the 35- to 44-year-old category and 80 category, we
made appropriate adjustments for the wider age span in the category.) We then
multiplied the computed percentage retiring at each boundary by the computed
number crossing the boundary and summed across all the boundaries. (Note that,
strictly speaking, both the survey and the analysis refer to persons not
working in radiology, rather than to retirement per se. In addition to
retirees, persons not working in radiology can include those who have taken up
some other activity after retiring from radiology and persons temporarily not
working in radiology.)
Individuals depart from the radiology workforce not only by retirement, but also by death, and we needed an estimate of the number of deaths among working radiologists. Estimating the number of deaths among radiologists requires recognizing the fact that among an upper middle class highly educated population, such as radiologists, death rates are considerably lower than among the general population. Analysis of 1997 survivorship tables (the latest data available) show that in the age ranges for which significant numbers of deaths among nonretired radiologists are expected, the death rate for white females is approximately 61% of that for white males [23]. A study of life expectancy by income and education shows that for white males, life expectancy in the highest income or education category exceeds the whole-population average by half to two thirds of the approximately 5-year gap between the average for all males and the average for all females [24]. On the basis of this finding and the percentage of females among working radiologists, we estimated that the death rate of working radiologists is approximately 70% of that for white males in general. To estimate deaths, we applied this age-specific death rate to the number of working radiologists in each of the age categories described in the preceding paragraph and summed the expected number of deaths across all the categories.
To investigate how radiologists' workload compared with what they would like, our survey included the question, "How would you rate your individual workload relative to what you would like, recognizing that income largely depends on work done?" To obtain a thoughtful answer, we judged that it was important to stress that income depends on workload. The survey offered five answer categories for the question. As part of the analysis of responses, we combined the two categories, "much too much work" and "somewhat too much work." The text and tables refer to these two categories combined as "too much work." In addition to tabular analysis, we carried out logistic regression analysis of the factors associated with having too much work. Logistic regression analysis, a standard multivariate technique used when the dependent variable has two values (in our analysis, having or not having too much work), indicates the independent effect of each of several explanatory variables, controlling for the simultaneous effect of all other explanatory variables included in the analysis.
Statistical Significance and Sampling Variability
In many tables we present standard errors. Because our data come from a
sample, rather than from a survey that reached every diagnostic radiologist in
the United States, our data, like all sample survey data, are subject to
sampling variability. Sampling variability is the name given to the fact that,
in general, a statistic—such as a percentage—from a sample will
deviate somewhat from the same statistic for the underlying population from
which the sample is drawn. The standard error is the usual measure of sampling
variability. There is a 95% probability that the true value of a statistic for
a population lies within approximately two standard errors of the
corresponding statistic for a sample drawn from that population. Standard
errors are also used in calculating the statistical significance of a
difference between two statistics—for example, the statistical
significance of the difference between the percentage of radiologists
reporting too much work in one age group and the percentage reporting too much
work in a different age group. We use p 
0.05 as the measure of
statistical significance, and in the text, only mention differences that are
statistically significant, except where explicitly otherwise stated.
Data analysis was conducted with release 8.01 software (SAS Institute, Cary, NC).
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Work status.—Overall, 73% of posttraining radiologists were working full-time in radiology, 10% were working part-time in radiology, and 17% were not working in radiology (Table 2). With those not working in radiology added in, the total number of posttraining radiologists was approximately 30,700.
|
Overwhelmingly, the 17% not working in radiology was retired, and in the remainder of this report, we, therefore, generally call this category "retired" or "not working." However, the 17% includes approximately 0.6% of posttraining radiologists who were disabled, approximately half of these permanently: 0.2% not working because they were taking care of family members, mostly children; 0.2% voluntarily taking time off between jobs; 0.2% involuntarily unemployed; 0.3% working as administrators or in other ways working, but not as radiologists. (The true percentage working, but not in radiology, may be larger, for the survey did not explicitly ask about it.) Note that because each nonretirement category is small, the uncertainty regarding its size, arising from sampling variability, is relatively large.
The percentage of posttraining radiologists working part-time or not working varies by age and sex, as Table 2 shows. The percentage not working is, of course, highest in the oldest age categories. Nonetheless, half of all radiologists age 65-69 years were working, most of them full-time; and 40% of those age 70-74 were working.
There does not appear to be a difference between men and women in the pattern of completely retiring from work, although the small number of women age 55 years or older in the profession makes this somewhat uncertain. In contrast, there is clearly a sex-linked difference in working part-time. Working part-time was not unusual among women younger than age 55, but was much rarer among men in that age range. Conversely, a substantial fraction of men age 55-74 were working part-time, whereas this practice was rare among women. Another difference between men and women was that there was a small percentage of women aged 35-54 not working, whereas this situation was much rarer among men.
Sex.—Overall, 16% of posttraining professionally active radiologists were women (Table 3). In general, we found that the younger the age group in question, the higher the percentage of women. However, this pattern has reversed recently, with women comprising 29% of posttraining radiologists younger than 35 but only 22% of current trainees.
|
Board certification.—(Note that all data we report on certification by the ABR are based solely on radiologists age 45 years and older; the findings may not hold for younger radiologists.) Using American Medical Association Physician Masterfile data on certification, our survey found that 90% of posttraining professionally active diagnostic radiologists age 45 or older were certified by the ABR. The percentage of radiologist certified varied by radiologists' characteristics, as Table 4 shows. Among radiologists age 45 and older, the percentage was higher among men than among women; relatively low among radiologists in solo practices, in groups consisting of two radiologists, and in nonacademic government groups; and relatively high among radiologists in groups of 15 or larger.
|
Every radiologist has multiple characteristics, making it impossible to tell from a table that shows only one characteristic at a time, such as Table 4, which relationships are significant when multiple characteristics are considered simultaneously. We, therefore, used logistic regression to identify which characteristics had a statistically significant relationship with certification by the ABR above and beyond the relationship arising from all other characteristics being considered. The logistic regression showed that the following relationships among radiologists 45 and older were statistically significant when we simultaneously controlled for the effects of all other characteristics considered: women were less likely to be board certified than men. Radiologists in solo practice and in groups consisting of two radiologists were less likely to be board certified than those in groups consisting of five to seven radiologists (our reference category for size), whereas those in groups of 30 or larger were marginally (p = 0.06) more likely to be board certified. Radiologists in private nonacademic multispecialty groups and nonacademic government groups were less likely to be board certified than those in private nonacademic radiology groups (our reference category for group type). The logistic regression showed no statistically significant independent relationships for geographic region, age, or part-time versus full-time status. Also, other things equal, radiologists in academic groups were no more likely to be board certified than those in private nonacademic radiology groups.
Radiology Practices and the Characteristics of Radiologists in
Different Types of Practices
Number and size of practices.—In 2000, there were
approximately 4,566 diagnostic radiology practices in the United States;
approximately 1,556, or 34%, were one-radiologist (solo) practices; the
remaining 3,010 (66%) were multiradiologist practices
(Table 5). Of the
multiradiologist practices, 45% had two-to-four radiologists, and these
practices plus solo practices constituted almost two thirds of all practices
in the country. In general, there were not prominent regional differences from
the national pattern of practice size. The one exception was that in the West,
40% of all practices were solo practices.
|
It is important to recognize that the distribution of practices by size differs from the distribution of radiologists by practice size. This difference occurs because a large practice and a small practice are each one practice and so count equally when practices are counted, but the large practice may contain many times as many radiologists as the small practice and hence contributes far more to a count of radiologists. In keeping with this phenomenon, in 2000, solo practices were 34% of all practices, but only 6% of posttraining professionally active radiologists were in solo practice (Fig. 1). In contrast, practices with 15 or more radiologists were only 9% of all practices, but 38% of posttraining professionally active radiologists were in these practices. Practices consisting of four or fewer radiologists (including solo practices) constituted 64% of all practices but included only 21% of posttraining professionally active radiologists.
|
Trends in numbers and size of practices.—Tabulation of the 1995 Survey shows that in 1995, there were approximately 4,439 diagnostic radiology practices in the United States; of these, approximately 1,728, or 39%, were one-radiologist (solo) practices. The remaining 2,711 (61%) were multiradiologist practices. Thus, between 1995 and 2000, the number of solo practices decreased by 10%, and the number of multiradiologist practices increased by 11%.
In 1995, 39% of radiology practices were solo practices, 27% were composed of two-to-four radiologists, and 6% were composed of 15 or more radiologists (Fig. 1). In 2000, the corresponding statistics were that 34% of practices were solo, 29% were composed of two-to-four radiologists, and 9% were composed of 15 or more radiologists. For radiologists (as opposed to practices), the statistics were that in 1995, 8% of posttraining professionally active radiologists were in solo practice, 17% were in practices with two-to-four radiologists, and 30% were in groups with 15 or more. In 2000, the corresponding statistics were, respectively, 6%, 15%, and 38%. Clearly, solo practices are declining in relative importance, whereas large groups are increasing in relative importance. However, we found no strong trend with respect to the relative importance of small groups.
Types of multiradiologist groups.—Nonacademic private radiology groups were by far the most common type of multiradiologist group (Table 5). They numbered approximately 2,105 and constituted 70% of the 3,010 multiradiologist groups in the United States (calculated from Table 5). The radiology components of nonacademic multispecialty groups constituted 12% of all multiradiologist groups; academic radiology groups were 8% of all multiradiologist groups; and the radiology components of nonacademic government groups were 6% of multiradiologist groups. (Four percent are missing because of anomalies in the data on group type.) Of the 3,010 multiradiologist groups, 3% were reported to include radiation oncologists. For the remaining 97% of groups, radiation oncology was a separate department (in a multispecialty entity) or a separate practice.
The types of multiradiologist groups differed in size. Academic groups were relatively large. Nine percent of them were composed of two-to-four members, and 47% were composed of 15 or more members (Table 5) compared with 45% and 13%, respectively, for all multiradiologist groups. Nonacademic government groups deviated from the overall pattern in the opposite direction; 65% were composed of two-to-four members, whereas only 4% were composed of 15 or more members.
Characteristics of radiologists in different types of practices.—The different types of practices also varied in the demographic characteristics of the radiologists who worked in them. Compared with the regional distribution of all posttraining professionally active radiologists, radiologists in academic groups were concentrated in the Northeast and were relatively few in the South and West (Table 1). In contrast, radiologists in nonacademic private multispecialty groups were concentrated in the Midwest and West but were relatively few in the South and Northeast. Those in nonacademic government groups were concentrated in the South and were relatively few in the Northeast and Midwest.
There were also differences in age patterns. Nonacademic private multispecialty groups had a relatively large proportion of radiologists age 65 years or older, and this feature was even truer of government groups, Eight percent of all posttraining professionally active radiologists were 65 or older, but 21% of those in government groups were in this age category. Solo practices also appeared to have a relatively large proportion (16%) of radiologists 65 or older. But because there are few solo practitioners, sampling variability is large and this difference was not statistically significant. Unambiguously, however, radiologists in solo practice were concentrated in the older age categories and infrequently were young: 54% of them were age 55 years or older compared with 31% of all posttraining professionally active radiologists. Only 12% of solo practitioners were younger than age 45, whereas 37% of all posttraining professionally active radiologists were in this age range.
The most prominent differences in the male-female composition of practices were the relatively large percentage of radiologists in academic groups who were women—23% compared with 16% for all posttraining professionally active radiologists—and the small percentage (8%) of solo practitioners who were women (Table 1).
Table 6 presents information on age, sex, and type of practice from a different perspective. Rather than asking the question, "What is the age (or sex) composition of a given type of practice, and how does that compare with the average for all practice types?" which Table 1 answers, Table 6 asks the question, "What is the distribution across practice types of radiologists in a given age and sex category, and how does that compare with the distribution of all radiologists?"
|
Overall, Table 6 provides these answers: 6% of posttraining professionally active radiologists were in solo practice; 16% were in academic groups; 61%, in nonacademic private radiology groups; 12%, in nonacademic private multispecialty groups; 4%, in nonacademic government groups; 1%, in other situations. Statistically significant differences from this overall pattern were the following: radiologists age 65 years and older were more likely to be in nonacademic government or "other" practice and less likely to be in nonacademic private radiology groups—only 39% of them were in nonacademic private radiology groups. Radiologists age 55-64 years had an increased likelihood of being in solo practices. Radiologists younger than 45 years old had an atypically low likelihood of being in solo practice and an atypically high likelihood of being in a nonacademic private radiology group.
Only 3% of women were in solo practice compared with 7% of men. However, this finding was a function of women's generally younger age, not a true difference between men and women, for there was no statistically significant difference between men and women in any age category in the percentage in solo practice. Women were more likely than men to be in an academic group (23% vs 15%), principally because of the high percentage (35%) of women age 45-54 years old in academia. There was no statistically significant excess of women over men in academia in other age categories, and among radiologists younger than 35, women were less likely than men to be in academic groups (5% vs 22%). Women were less likely than men to be in nonacademic private radiology practices. This difference was true for all ages for which we had a sufficient sample size to have reliable data except for those younger than 35. Women age 45-54 were more likely than men of the same age to be in a nonacademic private multispecialty group, but there were no statistically significant differences between men and women in the percentage in multispecialty groups in other age categories or overall. Men 45-54 were less likely to be in nonacademic government or "other" practices than radiologists overall.
Issues Related to the Radiologist Shortage
Early retirement.—Annual surveys of the hiring activity that
radiology groups conducted in 1996 through 1998 showed elevated numbers of
radiologists leaving the workforce
[18,19,20,
25]. To study the effect of
this apparently much increased level of retirement, we conducted an analysis
of changes in the work patterns of radiologists age 55-74 years. We studied
this age range because it encompasses almost the entire process of retirement.
Few radiologists in their early 50s do anything other than work full-time, and
few 75 years old or older work at all
(Table 2).
In 2000, the full-time equivalency of radiologists 55-74 years old was 67.6%. If radiologists in each of the eight categories in this age range defined by a combination of sex and 5-year age group (e.g., women age 55-59 years old or men age 70-74 years old) had had the pattern of work that the same category had in 1995, rather than the 2000 pattern shown in Table 2, the full-time equivalency of radiologists 55-74 years old would have been 70.3%. The 2.7 percentage-point decline is not statistically significant and corresponds to a difference of approximately 280, or 1.1%, in the total number of radiologists of all ages at work (rather than retired). The only statistically significant difference found between the 1995 work patterns and the 2000 patterns was among men age 65 to 69 years old. In 1995, 47% of men in this age group were working full-time and 20% were working part-time, whereas in 2000, 30% were working full-time and 22% parttime.
Given the number of radiologists in 2000 in each category defined by a combination of sex and 5-year age group and assuming the percentage in each such category that was working (rather than retired) remained constant at the 2000 level, approximately 390 retirements and 140 deaths of professionally active radiologists would occur each year, for a total of 530 radiologists leaving the workforce each year. (Note that this analysis encompasses posttraining radiologists of all ages, not merely those 55-74 years old, although most retirements occur among those 55-74 years old.) Given the 1995 number of radiologists in each category (which was generally smaller, especially in the older age categories) and the 1995 pattern of working and assuming the 1995 pattern of working continuing, approximately 320 retirements and 115 deaths of professionally active radiologists would have occurred each year, for a total of 435 radiologists leaving the workforce annually.
Excessive workload.—In 2000, 10% of posttraining professionally active radiologists said that, recognizing that income depends largely on work done, they had "much too much work"; 41% said they had "somewhat too much work"; 45% said they had an "about-right" workload; 4% said they had "somewhat too little work"; and 1% said they had "much too little work."
Whereas the percentage of posttraining professionally active radiologists saying they had somewhat or much too much work was 51% overall, this percentage differed according to characteristics of radiologists and their practices, as Table 7 shows. Statistically significant differences from the 51% overall average found when, as in the table, only one or two variables at a time were considered were the following: only 35% of male radiologists younger than age 35 years old and 28% of male radiologists age 65 years and older reported too much work. In contrast, 57% of male radiologists age 45-54 said they had too much work. Only 34% of radiologists working part-time said they had too much work. Only 38% of solo practitioners and 33% of radiologists in practices consisting of two radiologists reported too much work, whereas 60% of radiologists in private nonacademic multispecialty practices did so.
|
Again, a logistic regression serves as our method to ascertain which relationships are statistically significant when all variables are considered jointly, rather than examined one at a time. A logistic regression examining the determinants of having too much work showed the following relationships were statistically significant when we simultaneously controlled for the effects of all other characteristics considered: part-timers were less likely to report too much work than those working full-time. Solo practitioners and radiologists in groups consisting of two radiologists were less likely to report too much work than those in groups consisting of five-to-seven radiologists (the reference category against which we measured others). Radiologists younger than 35 years, 35-44 years old, and 65 years old and older were less likely to report too much work than those age 45-54 years, which was the reference category. Female radiologists were more likely to report too much work than men. Radiologists in private nonacademic multispecialty groups were more likely to report too much work than those in private nonacademic radiology groups (the reference category).
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Our analysis shows that if current patterns of work versus retirement remain constant, approximately 530 radiologists will leave the workforce each year because of retirement and death. This analysis implies that the radiologist workforce will grow at approximately 2% annually in the near future. (Annually, there are approximately 1,000-1,100 radiologists completing training and, therefore, entering the workforce each year [26]. Subtract the number leaving the work-force each year, and the annual increase is approximately 500-550, approximately 2% of the current workforce.) The 2% annual increase in the number of radiologists is less than the rate of growth of workload, which we recently estimated at 3.5% annually measured in procedures and 5.5% annually measured in relative value units [27]. Thus, the shortage of radiologists seems likely, if anything, to become worse in the near future.
However, predictions of radiologist surplus or shortage—and, more generally, of physician surplus or shortage—have been notably fallible [28, 29], and we have no reason to think this prediction is any less subject to eventually being proved wrong.
Voiced dissatisfaction with managed care [16] and the large number of retirements reported in our recent surveys of hiring by radiology practices [18,19,20, 25] could easily lead to an impression that early retirement has become the general practice among radiologists. In contrast, we found that the striking feature of radiologists' retirement pattern in 2000, as in 1995 [13], was the large percentage still at work beyond age 65 years old.
The small percentage of radiologists younger than age 65 who are retired and the substantial percentage of those 65-74 still at work implies that bringing radiologists out of retirement is not promising as a large-scale remedy to the radiologist shortage. The current pattern of work among those age 55-74 years means there simply are not a great many retirees who are of an age at which coming back to work is highly plausible. However, returning retirees might possibly make an important, if not large, contribution. As indicated in the "Results" section, if 1995 patterns of retirement had held, the total workforce would probably be approximately 1% larger. Thus, it seems that drawing radiologists out of retirement might perhaps increase the workforce by at most a relatively small amount.
Unlike the situation in some other professions, such as nursing, there is no substantial reservoir of persons trained as radiologists but not working in the field, although not yet retired. The percentages of radiologists unemployed, taking time off between jobs, caring for family members, or working at a nonradiology job are minuscule—approximately 1% for all these categories combined.
Some of the variation according to practice type and radiologists' characteristics in the percentage reporting too much work seems easily explicable. For example, the relatively low percentage of part-time radiologists and of radiologists 65 years old and older reporting too much work probably reflects a relatively large percentage of these radiologists who could adjust their work hours to suit their preferences as to how much they work. However, the relatively low percentage of solo practitioners and radiologists in groups of two radiologists reporting too much work then seems puzzling, for one would expect that radiologists in these small practices would be least able to adjust their workload. The reason that relatively few posttraining radiologists younger than age 35 years report excessive work may be that they are too new to the profession to expect anything other than current patterns of work or that they are inured to the heavy workload experienced during the training that they just completed [8].
The finding of this study that there has been no statistically significant increase in retirement among radiologists in the age range at which retirement predominantly takes place (age 55-74 years) contradicts findings of the ACR's recent annual surveys of hiring by radiology practices, which have shown elevated numbers of retirements [18,19,20, 25]. Specifically, these latter surveys found approximately 800, 800, and 700 radiologists leaving the workforce in 1996, 1997, and 1998, respectively, and these numbers omit radiologists leaving solo practice (as opposed to learning multiradiologist groups). In contrast, interpolating the number of departures from the workforce expected in 1995 and 2000 if the pattern of retirement had remained fixed (see "Results") indicates approximately 450-500 radiologists in total (including those leaving solo practice) would have been expected to leave the workforce in each of these years. Thus, the hiring surveys report in total at least 900 more radiologists leaving the workforce than steady-retirement-pattern models predict. In contrast, the best estimate from the 2000 Survey (see Results) is that the excess was only approximately 280.
We are unable to account for this discrepancy. Possible explanations are that when a radiologist leaves a group, the group often thinks he or she is retiring when he or she, in fact, goes to work somewhere else or that many radiologists come out of retirement after a few years or less. Neither of these explanations seems particularly plausible.
Although the 2000 Survey was a one-page survey, the response rate was no higher than that for the 1995 Survey, which was eight pages long. Both surveys involved the same number and type of remailings. Normally, a much shorter survey will produce a substantially higher response rate. Thus, there clearly seems to have been a decline in responsiveness to surveys between 1995 and 2000. This decline may reflect diminished time available to radiologists because of their having too much work. But it also may reflect reaction to the increasing communication flow that bombards most Americans.
Radiology practices and radiologists.—Our study shows a decrease in solo practices and an increase in large practices between 1995 and 2000. However, as in the first half of the 1990s [15], the shift has been gradual, not the dramatic transformation some commentators predicted. For example, with approximately 2% of all practicing radiologists retiring or dying annually and solo practitioners concentrated among the older age groups, the 10% decline in the number of solo practices that took place between 1995 and 2000 is much more in keeping with solo practitioners having retired at a normal—or even unexpectedly slow—pace than with predictions of radiologists abandoning solo practice in droves because it is outmoded. Similarly, at the large-group end of the spectrum, the increase has been gradual, not cataclysmic, with the fraction of radiologists in groups of 15 or larger increasing from 30% to 38%.
The concentration in the Northeast of radiologists in academic practice reflects the distribution of the overall population of the United States many decades ago when most of America's medical schools were established.
The difference between men and women in patterns of part-time work and of not working probably reflects family responsibilities, principally child rearing. A substantial percentage of women radiologists younger than age 55 years are not working full-time, whereas this practice is rare among men.
However, whereas part-time work among men is fairly common from age 55 years onward and clearly serves as a route for easing into retirement, part-time work among women in this age range is rare. The reason for this difference is not apparent.
Men and women radiologists also differ in the higher percentage of women who are in academic groups. Currently, this finding is primarily characteristic of women in the 45-54 year age range, rather than of younger women. This suggests that the difference will not continue into the future as new graduates of training programs join the profession.
The relatively low percentage of women among physicians entering diagnostic radiology is a matter for concern. In the late 1990s, the percentage of each year's class of graduates of medical schools in the United States who were women was in the low-to-mid 40s and rising. In contrast, our survey shows the percentage of radiology residents and fellows who were women was only in the low 20s in 2000 and apparently was lower than the percentage among radiologists who had recently graduated from training, suggesting a downward trend.
Although nonacademic private radiology groups are the locus of practice of most radiologists younger than 65 years, radiologists age 65 years and older are concentrated in other types of practice. It may be that radiologists in nonacademic private radiology groups tend to retire at a younger age than those in solo practice or in nonacademic government and "other" practices (including locum tenens). Alternatively, some radiologists in nonacademic private radiology groups may shift to these other types of practices as they age, finding them more suitable to the pattern of work they wish to perform. This trend seems particularly likely with respect to locum tenens work.
Study Limitations
Our data are from a large, carefully conducted survey that achieved a high
response rate. The detailed weighting described in the "Materials and
Methods" section adjusts for nonresponse bias—that is, differences
between respondents and nonrespondents—in the characteristics used in
the weighting (age, ACR membership, region, and sex).
Agreement of findings from this survey with data from independent information sources is generally good. For example, most findings are similar to those from the 1995 Survey, and where there are differences—for example, the smaller number of solo practices in 2000—they are generally in the expected direction. Also, given the existence of sampling variability in all survey findings, the survey's estimate of 232 academic groups in the United States agrees well with the 204 diagnostic radiologist residency training programs in the United States known to the ACR's education department (after deletion of obviously osteopathic residency programs). The difference is less than twice the standard error of the estimate from the 2000 Survey, which is the difference usually regarded as quite possibly due to sampling variability. Similarly, the radiation oncologist portion of the 2000 Survey yielded an independent estimate of the number of groups in the United States that include both diagnostic radiologists and radiation oncologists, and the estimates from the two sources agree within expected sampling variability.
Nonetheless, the study has noteworthy limitations. It omits osteopathic radiologists, who are approximately 2% of the total workforce. There may be nonresponse bias with respect to characteristics not included in the weighting. For example, those with too much work may be underrepresented because time pressures made them less likely to respond than other radiologists. Or those with too much may be overrepresented because their unhappiness with overwork energized them to respond to our survey because it asked about having too much work. In any case, having too much work is a subjective matter.
The sample—and therefore the number of respondents—was small for categories of radiologists that are small (i.e., older women radiologists), and sampling variability is, therefore, relatively large for these categories. Also, the survey instrument did not give a definition of an academic group or other practice types it listed.
Despite careful and extensive data cleaning, some anomalies remain. Approximately 1% of respondents said they were in solo practice but reported the number of radiologists in their practice as other than one, and approximately 1% of respondents reported the number of radiologists in their practice was one but reported a practice type other than solo.
The American Medical Assosication Physician Masterfile is notably slow to register changes in physicians' status, such as the beginning of residency training, certification by a specialty board, completion of training, retirement, and death. Our survey asked about respondents' work status, so information about completion of training and retirement should be sound. However, deaths are probably underreported, because relatives or colleagues of decedents were probably less likely than living radiologists to respond to our survey. This underreporting should have only a minor effect on our statistics. In contrast, trainees are massively underreported in the Masterfile; our survey shows a weighted count of 3,600, whereas the correct total of residents and fellows is probably in excess of 5,000. The Physician Masterfile apparently averages a 1- to 2- year delay in recording a physician's entry into residency training, and the characteristics we report for trainees thus reflect mostly physicians in the subsequent years of training.
We took board certification status from the Physician Masterfile rather than ascertaining it in the survey. The percentage of radiologists that the Masterfile records as being board certified is clearly too low for those younger than 45 years. Although we, therefore, analyzed board certification only for those 45 years or older, the reporting of certification may also be slightly incomplete for these radiologists. More important, patterns of board certification among radiologists younger than 45 years old may be different.
Despite its limitations, we believe our survey is a valid and valuable source of information. It provides new evidence that there is a shortage of radiologists, and it suggests the shortage is, if anything, likely to intensify even if departures from the workforce are not swelled by early retirements. However, past attempts to forecast surplus or shortage have been so unsuccessful that one should not rely on current information to predict the future. In any case, the survey clearly shows solo practices are gradually diminishing in number and large groups gradually increasing in importance, with small groups holding their own in relative importance.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  M. Bhargavan, A. H. Kaye, H. P. Forman, and J. H. Sunshine Workload of Radiologists in United States in 2006-2007 and Trends Since 1991-1992 Radiology, August 1, 2009; 252(2): 458 - 467. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. L. Weiss and C. P. Langlotz Structured Reporting: Patient Care Enhancement or Productivity Nightmare? Radiology, December 1, 2008; 249(3): 739 - 747. [Full Text] [PDF]
|
|
|
|
 |
|
 N. R. Nayak, C. Meghea, M. Bhargavan, H. P. Forman, and J. H. Sunshine Prevalence of Productivity-Enhancing Technologies in Radiology Am. J. Roentgenol., June 1, 2008; 190(6): 1445 - 1452. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. R. Fielding, N. M. Major, B. F. Mullan, J. A. Neutze, K. Shaffer, C. B. Wilcox, L. Paynter, and E. D. Pisano Choosing a Specialty in Medicine: Female Medical Students and Radiology Am. J. Roentgenol., April 1, 2007; 188(4): 897 - 900. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. S. Lewis, M. Bhargavan, and J. H. Sunshine Women Radiologists in the United States: Results from the American College of Radiology's 2003 Survey Radiology, March 1, 2007; 242(3): 802 - 810. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. L. Ebbert, C. Meghea, S. Iturbe, H. P. Forman, M. Bhargavan, and J. H. Sunshine The State of Teleradiology in 2003 and Changes Since 1999 Am. J. Roentgenol., February 1, 2007; 188(2): W103 - W112. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. Meghea and J. H. Sunshine Retirement patterns and plans of radiologists. Am. J. Roentgenol., December 1, 2006; 187(6): 1405 - 1411. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. S. Lewis, J. H. Sunshine,, and M. Bhargavan A portrait of breast imaging specialists and of the interpretation of mammography in the United States. Am. J. Roentgenol., November 1, 2006; 187(5): W456 - W468. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Merewitz and J. H. Sunshine A Portrait of Pediatric Radiologists in the United States Am. J. Roentgenol., January 1, 2006; 186(1): 12 - 22. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. D. Saket, C. C. Nwanze, C. D. Maynard, J. H. Sunshine, and H. P. Forman Update on the Diagnostic Radiologist Employment Market: Findings Through 2004 Am. J. Roentgenol., December 1, 2005; 185(6): 1408 - 1415. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
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]
|
|
|
|
|
|
M. Bhargavan and J. H. Sunshine Workload of Radiologists in the United States in 2002-2003 and Trends Since 1991-1992 Radiology, September 1, 2005; 236(3): 920 - 931. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. B. Larson, Y. S. Cypel, H. P. Forman, and J. H. Sunshine A Comprehensive Portrait of Teleradiology in Radiology Practices: Results from the American College of Radiology's 1999 Survey Am. J. Roentgenol., July 1, 2005; 185(1): 24 - 35. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. T. Bramson and G. A. Taylor SOS: Can We Save Pediatric Radiology? Radiology, June 1, 2005; 235(3): 719 - 722. [Full Text] [PDF]
|
|
|
|
|
|
J. H. Sunshine and Y. S. Cypel Types of Procedures Performed by Diagnostic Radiology Practices: Past Patterns and Future Directions Am. J. Roentgenol., November 1, 2004; 183(5): 1193 - 1201. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. D. Smith, J. Temte, J. W. Beasley, and M. Mundt Radiographs in the Office: Is a Second Reading Always Needed? J Am Board Fam Med, July 1, 2004; 17(4): 256 - 263. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. Hogan and J. H. Sunshine Financial Ratios in Diagnostic Radiology Practices: Variability and Trends Radiology, March 1, 2004; 230(3): 774 - 782. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. H. Sunshine, C. D. Maynard, J. Paros, and H. P. Forman Update on the Diagnostic Radiologist Shortage Am. J. Roentgenol., February 1, 2004; 182(2): 301 - 305. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. H. Deitch and J. H. Sunshine The Relationship of Managed Care To Business, Professional, and Organizational Aspects of Radiology Practices Am. J. Roentgenol., January 1, 2004; 182(1): 29 - 38. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. Z. Wang Editorial Comment-- Telemedicine: The Solution to Provide Rural Stroke Coverage and the Answer to the Shortage of Stroke Neurologists and Radiologists Stroke, December 1, 2003; 34(12): 2957 - 2957. [Full Text] [PDF]
|
|
|
|
|
|
Y. S. Cypel and J. H. Sunshine Characteristics of Mixed Diagnostic Radiology-Radiation Oncology Practices Am. J. Roentgenol., September 1, 2003; 181(3): 643 - 647. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Y. S. Cypel and J. H. Sunshine Basic Characteristics of Radiology Practices: Results from the American College of Radiology's 1999 Survey Am. J. Roentgenol., August 1, 2003; 181(2): 341 - 349. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. D. Saketkhoo, J. H. Sunshine, A. M. Covey, and H. P. Forman Findings in 2002 from a Help Wanted Index of Job Advertisements: Is the Job-Market Shortage of Diagnostic Radiologists Easing? Am. J. Roentgenol., August 1, 2003; 181(2): 351 - 357. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N. R. Dunnick Report of the 2002 Intersociety Commission Meeting: Radiology 2002--Today's Research Is Tomorrow's Practice Am. J. Roentgenol., April 1, 2003; 180(4): 925 - 928. [Full Text] [PDF]
|
|
|
|
|
|
M. Bhargavan and J. H. Sunshine Workload of Radiologists in the United States in 1998-1999 and Trends Since 1995-1996 Am. J. Roentgenol., November 1, 2002; 179(5): 1123 - 1128. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. D. Saketkhoo, A. M. Covey, J. Sunshine, and H. P. Forman Updated Findings from a Help Wanted Index of Job Advertisements and an Analysis of the Policy Implications: Is the Job-Market Shortage for Diagnostic Radiologists Stabilizing? Am. J. Roentgenol., October 1, 2002; 179(4): 851 - 858. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. H. Sunshine, R. S. Lewis, B. Schepps, and H. P. Forman Data from a Professional Society Placement Service as a Measure of the Employment Market for Physicians Radiology, July 1, 2002; 224(1): 193 - 198. [Abstract] [Full Text]
|
|
|
|
|
|
M. Bhargavan, J. H. Sunshine, and B. Schepps Too Few Radiologists? Am. J. Roentgenol., May 1, 2002; 178(5): 1075 - 1082. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
