The Practice of Radiolog y Original Research

Downloaded from www.ajronline.org by 24.239.202.23 on 01/25/16 from IP address 24.239.202.23. Copyright ARRS. For personal use only; all rights reserved

Meghea and Sunshine

Retirement Patterns and
Plans of Radiologists

C E N T U
R Y

MEDICAL

O F

IMAGING

Cristian Meghea1
Jonathan H. Sunshine2
Meghea C, Sunshine JH

Retirement Patterns and

Plans of Radiologists
OBJECTIVE. The objectives of our study were to describe radiologists recent retirement
plans and patterns and to assess whether changes in radiologists retirement patterns over the
period of 19952003 explain the recent easing of the radiologist shortage.
MATERIALS AND METHODS. We present detailed information from 2003 about the
planned retirement age of radiologists, their labor force participation late in their careers, and their
actual retirement pattern based on data from the American College of Radiologys (ACR) 2003 Survey of Radiologists. To analyze changes over time, we compare these data with information from
the ACRs 1995 and 2000 Surveys of Radiologists. Multivariate regression analysis was also used
to identify the effects of radiologist and practice characteristics on radiologists retirement plans.
RESULTS. The percentage of radiologists fully retired and the average retirement age were
the same in 1995 and 2003. Overall, labor force participation rates were decreasing over the
period 19952003 for both women and men. Standardized labor force participation rates for
radiologists age 5574 years appeared to decrease from 1995 to 2000 and remained at a lower
level in 2003, but the changes were not statistically significant. As of 2003, radiologists retired
at 64, approximately 2 years older than the average U.S. worker.
CONCLUSION. Radiologists remain active in their profession longer than the typical
U.S. worker. There was no change in radiologists pattern of gradually moving into retirement.
If anything, radiologists were retiring earlier in 2003 than in the past. A delay in retirement is
not an explanation of the recent easing of the radiologist shortage.
nly a few years ago, as the 21st
century began, there was a severe shortage of radiologists,
and that shortage was projected
to worsen [19]. Recently, the situation had
changed substantially, with a few sources
pointing to an easing of the shortage and
some suggesting its disappearance [1012].
A number of indicators document these
findings: first, overall, radiologists workload in 2003 was what was desired; second,
there have been decreasing vacancies in academic radiology departments in recent
years; third, there has been a decreasing ratio of job listings to job seekers at a major
placement service, from approximately 4
jobs per job seeker in 2000 to 1.1 in 2004;
and, fourth, there have been a decreasing
number of positions advertised in Radiology
and the American Journal of Roentgenology. Paradoxically, given the easing of the
shortage, Americans are using progressively more radiologic services and radiologists workloads are increasing [1315].

Keywords: career planning, radiologists, retirement,

workload, work trends
DOI:10.2214/AJR.05.1857
Received October 21, 2005; accepted without revision
November 3, 2005.
1Institute for Health

Care Studies, Michigan State

University, A134 East Fee Hall, East Lansing, MI 48824-1316.
Address correspondence to C. Meghea
(cristian.meghea@hc.msu.edu).

2Research Department, American College of Radiology,

Reston, VA 20191.
AJR 2006; 187:14051411
0361803X/06/18761405
American Roentgen Ray Society

AJR:187, December 2006

There are several possible explanations that

might reconcile increasing utilization with the
easing of the shortage [10]. One is that radiologists now retire later or work more late in their
careerthat is, they switch to part-time work
later or not at all. The belief that retirement and
switching to part-time work have been delayed
has been supported by the major decrease in
the value of personal assets, including retirement savings, triggered by the decline in the
stock market since its 2000 peak.
Radiologists want to know their peers path
to retirement and trends in the recent years.
This article provides a service to the radiology
profession, especially because the most recent
article about retirement patterns and plans was
published approximately 3 years ago [16].
In this study, we present detailed information from 2003 about the planned retirement
age of radiologists, their labor force participation (amount of work) late in their careers, their
actual retirement pattern, and the extent to
which they work part-time before retirement.
To analyze changes over time that might partly

1405

Meghea and Sunshine

TABLE 1: Work Status of Posttraining Diagnostic Radiologists in 2003,
by Age and Sex
Men

All Respondents

FT

PT

96.4

2.0

1.6

55.8

38.9

5.3

86.4

11.1

2.5

SE

1.2

0.9

0.8

5.4

5.3

2.6

1.9

1.7

0.9

92.8

5.1

2.1

55.3

34.6

10.1

83.9

12.1

4.0

SE

1.5

1.3

0.8

5.0

4.8

3.2

1.8

1.6

1.0

Age (y)

Downloaded from www.ajronline.org by 24.239.202.23 on 01/25/16 from IP address 24.239.202.23. Copyright ARRS. For personal use only; all rights reserved

Women
Not in
Radiology

FT

PT

Not in
Radiology

FT

PT

Not in
Radiology

3544

4554

Analysis Methods

5559
%

81.3

12.1

6.6

SE

3.3

2.7

2.1

41.0a

40.7a

53.8

25.5

20.7

8.3a

8.3a

SE

3.9

3.4

3.2

29.9

32.3

37.9

SE

4.4

4.4

4.6

19.6

23.2

57.2

SE

4.3

4.8

5.4

5.3

16.8

77.9

SE

2.0

3.5

3.9

68.2

12.5

19.4

54.5

34.0

SE

1.3

0.9

1.1

3.2

3.1

77.2

15.7

7.1

18.3a

3.3

2.9

2.1

6.9a

52.1

26.6

21.3

3.7

3.3

3.1

33.6

29.8

36.7

4.3

4.1

4.4

20.5

22.5

57.1

4.3

4.7

5.3

6064

6569

7074

75

5.8

16.2

78.0

2.0

3.4

3.8

11.5

65.9

16.1

18.1

2.1

1.2

1.0

1.0

All ages

NoteDash () indicates data are not reported because the agesex cell size was too small to provide reliable
statistics (fewer than 20 responses). FT = full-time, PT = part-time.
a Statistics for women 5564 years old.

explain the easing of the shortage, we compare

these data with information from the American
College of Radiologys (ACR) 1995 and 2000
Surveys of Radiologists.
Materials and Methods
The Data
Data are from the ACRs 2003 Survey of Radiologists, a nationally representative stratified random sample of all radiologists in the United States
that has been described in detail elsewhere [15, 17].
In brief, the 2003 Survey was conducted by mail
between March and July 2003 and covered a broad
range of topics. With up to four remailings to nonrespondents and other reminders, it achieved an
overall 63% response rate, including 1,676 responses from posttraining radiologists.
Of main interest for this study were the work status of the survey participants (full-time, part-time,
or not working in radiology) and, if professionally

1406

The 2003 Survey also asked radiologists

whether they subspecialize; their weekly work
hours; whether they feel overworked; whether they
are an owner of their main practice; and other information about their main practice, such as practice
type and practice location.
We also used data from two other nationally representative surveys, the ACRs 1995 Survey of Radiologists and the ACRs 2000 Survey of Diagnostic Radiologists and Radiation Oncologists, both of
which are described in detail elsewhere [16, 18].

active, the radiologists retirement plans. The professionally inactive radiologists were asked for the
main reason they were temporarily or permanently
not working in radiology. Their most frequent responses were that they were retired, disabled, or
raising children. When active radiologists were
asked about their future retirement plans, the response options were keep working full-time in radiology, never retire; change careers; go part-time
before retiring; and fully retire. These responses
were not mutually exclusive, and there were often
more than one response per radiologist.
The radiologists were classified in seven age
groups. Two groups span 10 years: 3544 and
4554 years old. In most cases, retirement occurs
between the ages of 55 and 74 years. To better capture radiologist retirement behavior, we classified
these physicians in 5-year age groups: 5559,
6064, 6569, and 7074 years. Radiologists 75
years old or older constitute the last group.

Our study excludes radiologists in training (fellows and residents). Survey responses were
weighted for different sampling rates and response
rates to make the data representative of what answers would have been if all radiologists in the
United States had been surveyed and had responded. Reported means, SEs, tests of statistical
significance, and regression coefficients are calculated taking into account not only the weighted nature of the data but also the complex survey designthat is, the fact that there are three distinct
physician strata in our data set (osteopathic radiologists, radiologists self-designated as vascular or
interventional radiologists in the American Medical Associations [AMA] Physician Masterfile, and
all other allopathic radiologists).
The work status of radiologists in 2003 (fulltime, part-time, or not working in radiology) is
presented by sex and age groups in Table 1. We
excluded the radiologists younger than 35 from
this analysis because there are few posttraining
radiologists at this age. Having considerably
fewer women than men in our data, we aggregated women in the age group 5564 to obtain a
usable sample size for that age category. We did
not report results for women over 65 because the
sample size was too small to obtain reliable results. The 2003 statistics on work status are
compared with similarly derived statistics based
on the ACRs 1995 Survey reported in this article and with 2000 data already published elsewhere [16].
To investigate the labor force participation of
both men and women, we measured the full-time
equivalency of radiologists in 1995, 2000, and
2003, as reported in Table 2 and Figure 1. We recognized from analyzing the 2003 Survey data that
there were problems with the previous surveys
questions to part-timers about the hours they
worked. On the basis of the more reliable information from the 2003 Survey, we realized part-timers hours average about 50% of full-timers. We
therefore computed the 2003 labor force participation and adjusted the previously published 1995
and 2000 numbers counting a full-time radiologist

AJR:187, December 2006

Retirement Patterns and Plans of Radiologists

TABLE 2: Labor Force Participation Rates of Posttraining Radiologists, by Age
and Sex, from the 1995, 2000, and 2003 Surveys of Radiologists
1995
Age (y)

2000

2003

Men

Women

Men

Women

Men

Women

99.4

85.7

99.3

81.9

97.4

75.3

SE

0.3

3.0

0.3

2.5

0.9

3.3

96.5

89.2

96.8

81.8

95.3

72.6

SE

0.8

4.5

0.7

3.3

1.0

3.5

Downloaded from www.ajronline.org by 24.239.202.23 on 01/25/16 from IP address 24.239.202.23. Copyright ARRS. For personal use only; all rights reserved

3544

4554

5559
%

91.2

SE

2.0

76.0a

87.5
1.8

76.0a

87.4
2.4

61.3a

73.1

13.0a

71.6

8.0a

66.6

6.4a

SE

3.7

6064
3.4

3.2

6569
%

57.8

SE

4.3

41.1

4.0

45.9

3.9

7074
%

19.3

SE

3.8

27.3

3.9

31.2

4.2

75
%

11.5

SE

3.4

11.9

2.4

13.6

2.5

Total
%

81.3

84.1

78.1

79.3

74.5

71.3

SE

1.0

2.0

1.0

2.1

1.1

2.2

NoteDash () indicates data are not reported because the agesex cell size was too small to provide reliable
statistics (fewer than 20 responses).
a Statistics for women 5564 years old.

as 1, a part-time radiologist as 0.5, and a radiologist not working in radiology as zero.

To study whether radiologists were retiring earlier than in the past, we present overall labor force
participation rates in 1995, 2000, and 2003 in
Figure 2. The rates are for radiologists 5574 years
old because that age range is the one in which the
overwhelming majority of radiologists reduce their
workload or retire. Changes over time in labor force
participation could be caused not only by genuine
changes in work behavior, but also by changes in
the age-and-sex mix of radiologists. To account for
the latter potential changes, we standardized the
1995 and 2000 labor force participation rates to the
2003 population characteristics using the method
called by epidemiologists direct standardization
[19, 20]. We created eight cells consisting of four 5year age categories for each sex, measured the fulltime equivalency for each cell, and applied the fulltime equivalencies from 1995 and 2000 to the 2003
population distribution by age and sex. As a result,

AJR:187, December 2006

we obtained a measure of what labor force participation would have been if the 2003 population had
been at work with the work characteristics of the
2000 and 1995 populations.
Retired radiologists in 2003 were asked about
their retirement age, whether they worked part-time
before retirement, whether they would consider returning to work, and whether the state of the economy
influenced their retirement. We present this information in Table 3, along with data about the future plans,
including retirement plans, of professionally active
radiologists. Similar information based on the 1995
Survey was published elsewhere, allowing us to measure changes over time in retirement behavior and retirement plans of radiologists [18].
To measure the effects of radiologist and practice
characteristics on radiologists retirement plans, we
perform logistic and ordinary least-squares (OLS) regressions. Logistic regression is used when the variable analyzed is dichotomous, such as plan to never
retire: yes/no, and OLS is used when the analyzed

variable is continuous, such as planned retirement

age. Regression analysis identifies the independent
effect of each factor considered, such as the radiologists age or the practice type, statistically controlling
for the effects of all other variables included. The explanatory variables we considered were the sex and
the age of the radiologist; whether the radiologist subspecialized or not, worked full-time, or felt overworked; the extent the radiologist liked the profession; and the location and the type of practice where
the radiologist works. Logistic regression results are
reported in terms of odds ratios (the ratio between the
probability that an event does and the probability that
the event does not take place). We use odds ratios to
compare whether the probability of a certain event is
different between pairs of radiologist subgroups.

Results
Work Status of Posttraining Radiologists
The Survey found that in 2003 there were
approximately 32,700 posttraining radiologists in the United States. Overall, 66% of
posttraining radiologists were working fulltime in radiology, 16% were working part-time
in radiology, and 18% were not working in radiology (Table 1). Excluding those not working in radiology, the total number of posttraining radiologists was approximately 26,800.
Most (83%) of those not working in radiology were retired. However, another 7% of the
inactive posttraining radiologists, approximately 1.2% of all posttraining radiologists,
said they were disabled, being on average 63
years old, and 3%, approximately 0.5% of all
posttraining radiologists, were not working in
radiology because they were raising children.
All six respondents not working in radiology
because they were raising children were
women. As in the previous 2000 Survey, because each nonretirement category is small,
the uncertainty regarding its size is relatively
large because of sampling variability.
Overall, 11.5% of women and 19.4% of
men were retired from radiology in 2003 or
otherwise not working in the field. These percentages are significantly different (p 0.05).
At younger ages ( 54 years), women were
more likely than men not to be working in radiology: For example, 10.1% of women age
4554 were not working in radiology versus
2.1% of men in that same age range. The percentages were not significantly different between men and women age 5564. The small
number of female radiologists age 65 years or
older makes any comparison between mens
and womens retirement patterns uncertain.
Thirty-four percent of all women radiologists versus 12.5% of all men worked part-

1407

Meghea and Sunshine

100
90
Men '95

80

Men '03

100

Women '03

60

Percent

90

50

80

40

Percent

Downloaded from www.ajronline.org by 24.239.202.23 on 01/25/16 from IP address 24.239.202.23. Copyright ARRS. For personal use only; all rights reserved

Women '95

70

30

70

65.0

61.6

61.5

2000

2003

60

20

50
10

40
0
3544

4554

5559

6064

6569

7074

Fig. 1Bar graph shows labor force participation by age and sex. Light gray bars
show results for men from American College of Radiology (ACR) Survey 1995; dark
gray bars, men from ACR Survey 2003; white bars, women from ACR Survey 1995;
striped bars, women from ACR Survey 2003.

time; this difference is significant (p 0.05).

More than one third of women younger than
55 years worked part-time, whereas the fraction for men was only approximately 25% at
similar ages (Table 1). However, working
part-time was quite common for older men:
for example, 32.3% of men age 6569 years
worked part-time.
Statistics on the work status of posttraining
radiologists in 2000 by age group and sex
have been published before [16]. As an additional basis of comparison, in 1995, 93.5% of
radiologists age 3544 were working fulltime and 6.5% were working part-time. In
1995, the full-time versus part-time distribution at other ages was 92.8% full-time versus
5.8% part-time at age 4554 years; 87.2%
full-time versus 8.8% part-time at age 5559;
65.2% full-time versus 13.3% part-time at
age 6064; and 46.7% full-time versus 18.7%
part-time radiologists age 6569.
Overall labor force participation rates in
2003 were lower than in the past both for
men81% in 1995, 78% in 2000, and 75% in
2003and for women84% in 1995, 79% in
2000, and 71% in 2003 (Table 2). Labor force
participation rates decreased slightly between
1995 and 2003 for men younger than age 65
years (Fig. 1): For example, the labor force participation rate decreased from 91% in 1995 for
men age 5559 to 87% in 2003. Conversely, at

1408

30
1995

Age (y)

Fig. 2Bar graph shows data standardized to 2003 population distribution by age
and sex.

ages 7074 the labor force participation rates of

men were substantively higher in 2003 than in
1995: 31% in 2003 versus 19% in 1995. Also,
there were relatively large decreases in labor
force participation rates between 1995 and
2003 for women younger than 55.
To study the trends in early retirement, we
performed an analysis of changes in the work
patterns of radiologists age 5574 in the 1995,
2000, and 2003 Surveys. This age range covers
almost the entire process of retirement: few radiologists younger than 55 do anything other
than work full-time, and few radiologists age
75 or older work at all (Table 1). The full-time
equivalency of radiologists (labor force participation rate) 5574 years old in 2003 was 61.5%
(Fig. 2). If radiologists in each agesex category (such as men 5559 or women 7074) had
had the work pattern that the same category had
in 2000, the standardized full-time equivalency
of radiologists age 5574 years would have
been 61.6%. Derived in a similar fashion, the
standardized labor force participation rate in
1995 would have been 65% (Fig. 2). The decreases in labor force participation from 1995
to 2000, from 2000 to 2003, and from 1995 to
2003 were not statistically significant.
Retirement Patterns and Plans
Thirteen percent of posttraining radiologists were completely retired at the time of the

2003 Survey (Table 3). Only 2% of radiologists age 5559 were retired. Fifty-one percent of those age 65 or older remained professionally active, including 71% of radiologists
age 6569, 53% of those age 7074, and 30%
of radiologists age 70 or older.
The average retirement age of those who
had already retired was 64 years in our 2003
Survey. Forty-two percent of retired radiologists had worked part-time before full retirement, on average for the last 4 years of their
professional career. Twenty-eight percent of
retired radiologists age 6064 said they
would consider returning to radiology parttime, whereas the corresponding statistic for
those age 65 years or older was only 17%.
The economy reportedly influenced the full
retirement of 15% of retired radiologists, including 25% of those who retired between
1999 and 2003 and only 9% of those who retired in 19941998.
As for future plans, only 4% of those working (full-time or part-time) reported a plan to
change careers in the future; 13% of all professionally active posttraining radiologists indicated they hoped to keep working full-time,
never retiring from radiology; 57% of fulltimers said they planned to go part-time before fully retiring; and 39% had plans to fully
retire at some date in the future (more than
one response could be given).

AJR:187, December 2006

Retirement Patterns and Plans of Radiologists

TABLE 3: Posttraining Radiologists: Retirement Patterns and Plans, by Age
Variables

Age 65 (y)

Age (y)

All
Respondents

3544

13

64

42
4

5559

6064

All 65

6569

7074

75

17

49

29

47

70

57

66

62

65

67

36

42

32

37

48

4554

Completely retired from radiology

Downloaded from www.ajronline.org by 24.239.202.23 on 01/25/16 from IP address 24.239.202.23. Copyright ARRS. For personal use only; all rights reserved

Completely retired (%)

0.2

Age (y) at retirement

Mean
SE
Worked PT before full retirement (%)

0.5

0.9

0.4

0.5

0.6

0.6

Years worked PT
Mean
SE

0.5

0.4

0.8

0.4

0.6

Would consider returning to PT radiology (%)

18

28

17

23

27

Current economy influenced full retirement (%)

15

13

15

29

11

13

15

10

13

23

19

27

29

55

52

56

Professionally active radiologists: future plans and

postretirement work
Keep working FT in radiology, never retire (%)
Change careers (%)
If yes, at what age (y)?
Mean
SE
Work PT before retiringa (%)

0.9

1.4

0.8

57

51

60

63

69

45

50

48

12

59

55

58

62

65

70

69

If yes, at what age (y)?

Mean
SE
Fully retire (%)

0.3

0.5

0.3

0.4

0.3

1.0

0.6

39

38

37

40

42

45

48

42

41

63

61

62

64

66

72

69

74

If yes, at what age (y)?

Mean
SE
Current economy affecting future work plans (%)
Been retired and returned to FT or PT work (%)

0.3
42
7

0.5
29
0.4

0.4

0.4

0.4

0.5

0.4

0.5

47

56

51

44

49

41

32

33

27

35

47

NoteDash () indicates too few observations were obtained for reliable statistics (fewer than 10 responses). PT = part-time, FT = full-time.
a Excluding the radiologists already working PT.

Present age influenced the responses to the

questions about future retirement plans.
Twenty-three percent of all professionally active posttraining radiologists age 65 or older
planned never to retire from radiology,
whereas only 10% in the age range 4554 had
similar plans. Among those planning to retire,
the percentage of full-timers who intended to
work part-time before retiring increased with
age up to age 65, from 51% for those age
3544 years to 69% for radiologists age
6064 years. However, it dropped to 45% for
full-time radiologists age 65 years or older.
Forty-two percent of professionally active
posttraining radiologists said the economy had
influenced their future work plans. The percentage of radiologists influenced by the economy was lower at younger ages29% among

AJR:187, December 2006

those age 3544 yearsincreased to 56%

among radiologists age 5559 years and
dropped to 44% for those age 65 years or older.
Overall, 7% of all professionally active radiologists have been retired and have returned
to work. There were very few (2%) among
those age 4554 years, and the percentage increased with age to 47% among those age 75
years or older.
Effects of Radiologist and Practice Characteristics
on Retirement Plans: Regression Findings
Logistic regression showed that, other factors being equal, younger radiologists, those
who enjoyed their profession very much, and
those who wanted their workload increased
had higher odds of reporting they plan never
to retire. Conversely, the odds of planning

eventually to retire for radiologists who feel

overworked were more than twice as high as
the odds for those who reported that their
workload was appropriate.
As for planned retirement age, OLS regression showed that, other factors equal, radiologists age 3544 planned to completely retire
1 year younger than radiologists age 4554
years (the reference group). Those planning
to go part-time before retirement intended to
completely retire from radiology approximately 2 years older than radiologists who do
not plan to go part-time. Radiologists who enjoy their profession very much planned to retire 2 years older than others; those working
in a government practice planned to retire 5
years older than radiologists working in a
multispecialty practice (the reference cate-

1409

Downloaded from www.ajronline.org by 24.239.202.23 on 01/25/16 from IP address 24.239.202.23. Copyright ARRS. For personal use only; all rights reserved

Meghea and Sunshine

gory). Radiologists who wanted a lower
workload planned to retire almost 2 years
younger than those feeling their workload
was about right.
Regarding plans to change careers, others
things equal, radiologists who were owners of
their practice had half the odds of planning to
change careers as did nonowners. In addition,
radiologists who sought a decreased workload had odds of planning to change careers
that were twice as large as for radiologists
who reported their workload was about right.
In terms of going part-time before retirement, logistic regression findings were that,
other factors equal, the odds that radiologists
who subspecialized at least to some extent
planned to work part-time before retirement
were 0.7 of the odds of nonsubspecialists. The
odds that radiologists practicing in the Northeast census region planned to work part-time
before retiring were 1.5 the odds of those practicing in the Midwest. Finally, the odds of planning to go part-time before retirement of radiologists who sought a decrease in their workload
were more than twice the odds for radiologists
who reported their workload was about right.
Discussion
One of the possible explanations for the recent easing of the radiologist shortage is that
fewer radiologists are retiring and that they are
retiring at a later age. This is not, in fact, the
case. The crude evidence is that the overall percentage of radiologists fully retired and the average retirement age were the same in 1995
and 2003. In addition, overall labor force participation rates were decreasing over the period
19952003 for both women and men.
To investigate trends in retirement in a more
sophisticated fashion, we analyzed standardized labor force participation rates for radiologists age 5574. The standardized rates appeared to decrease from 1995 to 2000 and
remain at that lower level in 2003, although the
change was not statistically significant. If anything, radiologists are retiring earlier. Therefore, a delay in retirement is not an explanation
of the recent easing of the radiologist shortage.
Women were more likely than men to work
part-time, more so in 2003 than in 1995, and
the difference is more dramatic at younger
ages. This is a sign of women taking more advantage than men of the flexible hours possible in the radiology profession. The large difference at younger ages is most likely
indicative of social role distinctions, with
women taking a more active role than men in
raising children. The finding that radiologists

1410

not working at all due to child-rearing responsibilities were predominantly women supports this conclusion.
Approximately half of radiologists age 65
years or older still remained active in the profession in 2003. In the general U.S. population,
in comparison, fewer than 15% of the individuals 65 years and older were still actively
working [21]. The mean retirement age across
the entire U.S. population remained constant at
approximately 62 over the period 19952002
[22, 23]. Radiologists retire, on average, 2
years later than the average U.S. worker, and
we found no change over the past decade in the
retirement age of radiologists either. The much
higher percentage of elderly radiologists still
active in the workforce and the later retirement
age probably occur because radiology is both
more remunerative than the average job and
more interesting than many other occupations
in the United States.
We cannot make reliable comparisons between the retirement age of women and that
of men because we have too few retired
women in our data set. As more women from
the increased number who joined the radiology profession a decade or two ago retire, we
will know whether women and men tend to
retire at different ages and whether the average retirement age will change from what it is
now. If the working behavior at ages approaching retirement is an indication of retirement behavior, it is useful to note that
there was a decreasing trend in labor force
participation rates for women in the 5564
age range over the period 19952003, a similar trend as for men in that same age group.
Four in 10 radiologists who were fully retired as of 2003 had worked part-time before retiring, a fraction similar to that in 1995, indicating no change in radiologists pattern of
gradually moving into retirement. In contrast,
almost six in 10 of the active radiologists in
2003 said they planned to go part-time before
fully retiring. This may be an indication of a future increase in the percentage of radiologists
who will work part-time before retiring. However, the 1995 Survey also showed plans for
about six in 10 full-time radiologists to work
part-time before retiring, and those plans did
not, in fact, materialize by 2003. As a result, the
future change apparently implied by 2003 plans
may simply reflect that radiologists do not predict very well their behavior around retirement
years, more so at younger ages further from retirement. Overall, the full pattern of findings
from the 1995 and 2003 Surveys makes this latter explanation of the data seem likely.

One third of active radiologists age 65

years or older in 2003 had been retired and
had returned to work, most of them to parttime. A possible reason for the quite large percentage having returned to work is the major
stock market decline after 2000 and the resulting likely decline in the value of radiologists retirement assets. Support for the hypothesis that the stock market was important
is that one in four radiologists who retired in
the years 19992003 said the economy had
affected their retirement decision compared
with one in 10 who retired in 19941998.
This higher percentage saying the economy
affected their retirement in 19992003 could
have been due to both the rising stock prices
up to 2000 encouraging retirement and the
subsequent sharp decline generating a return
to work or a delay in retirement. In other
words, findings can be explained by the following two points: First, the apparent (but not
statistically significant) decline in labor force
participation rate for those age 5574 years
shows that the stock market slump did not
outweigh other forces enough to, overall, delay retirement. Second, the large percentage
of recent retirees who say the economy influenced their retirement and the large portion of
active radiologists 65 and older who were
fully retired for a period suggest the economy
had substantial effects on some radiologists.
However, alternative explanations should be
considered. The working radiologists age 65
years and older who had been retired for a time
could be individuals who found that in retirement they missed the mental stimulation of radiology or whose retirement income was what
they had anticipated, but who found they
wanted the extra income that working parttime brings. Similarly, the fact that radiologists
who retired before 1999 relatively infrequently
reported that the economy influenced their decision to retire may merely reflect that radiologists decreasingly remember all the factors that
influenced retirement as the retirement event
becomes further and further in the past.
Limitations
The 2003 Survey, the source of our data, has
strengthsmost notably, its high response
rate, careful weighting to adjust for differences
among categories of radiologists in response
rate, relatively large number of responses, and
careful data cleaning and editing to improve
data quality. Nonetheless, like surveys generally, it has important limitations. These include
imprecision due to sampling (measured by the
SEs presented in Tables 13 for most statis-

AJR:187, December 2006

Downloaded from www.ajronline.org by 24.239.202.23 on 01/25/16 from IP address 24.239.202.23. Copyright ARRS. For personal use only; all rights reserved

Retirement Patterns and Plans of Radiologists

tics); possible response bias with respect to
characteristics not considered in the weighting;
and some residual implausible and probably
erroneous data, despite the careful data cleaning. Also, because few women surveyed in
2003 were over the age of 55, we have little information about older female radiologists.
There is a possibility of selection bias in
our regression analysis of decisions regarding
the future work-related plans of radiologists.
For example, elderly radiologists still in the
workforce are more likely to be the ones planning to never retire from radiology, and their
odds of ever retiring are then artificially low.
Elderly, not retired radiologists are the ones
who tend to work longer, and their planned retirement age is artificially high. To limit the
extent of this bias, we restricted the regression
analyses of retirement plans to radiologists
age 55 or younger.
The survey responses presumably were accurate as to whether radiologists were retired
or not. Responses as to when they retired and
the reasons that influenced retirement are
probably less accurate, especially for radiologists who left the profession far in the past.
Finally, as our data suggest, plans seem not
to be a reliable guide to what will happen in
the future, especially for younger radiologists
who are not close to retirement and have not
thought much about it.
Conclusions
Radiologists remain active in their profession longer than do U.S. workers overall, retiring approximately 2 years older than the
average U.S. worker. The fraction of retired
radiologists who had worked part-time before
full retirement was similar in 2003 to that in
1995, indicating no change in radiologists
pattern of gradually moving into retirement.
Similar percentages of radiologists were retiring, and at similar ages, in 1995 and in 2003.
Labor force participation rates appeared to

AJR:187, December 2006

decrease from 1995 to 2000 and remain at that

lower level in 2003, although the change was
not statistically significant. A delay in retirement is thus not an explanation of the recent
easing of the radiologist shortage; if anything,
radiologists are retiring earlier.

References
1. Bhargavan M, Sunshine JH, Schepps B. Too few radiologists? AJR 2002; 178:10751082
2. Brogdon BG. The radiology manpower equation: a
new look. AJR 1990; 154:11111115
3. Garrison LP, Wills J, Perrin EB, Peterson ML. Physician requirements: 1990for five hospital-based
specialties. Final report under contract no. 232-810027. Seattle, WA: Battelle Human Affairs Research Centers, 1982
4. Graduate Medical Education Advisory Committee.
GMENACs summary report, vol. 1. Rockville,
MD: United States Department of Health and Human Services, 1980: publication no. (HRA) 81-651
5. Janower ML. Too many radiologists? Radiology
1973; 108:219221
6. Janower ML, Sunshine JH. Too many radiologists?
Update. Radiology 1996; 200:545549
7. National Advisory Committee on Radiation. Protecting and improving health through the radiological sciences: a report to the Surgeon General.
Washington, DC: Government Printing Office,
1966; Public Health Service publication
8. Smith WL. Factors affecting the future numbers of
diagnostic radiologists. AJR 1994; 163:777779
9. Sunshine JH, Evens RG, Chan WC. How accurate
was GMENAC? A retrospective review of supply
projections for diagnostic radiologists. Radiology
1992; 182:365368
10. Sunshine JH, Maynard CD, Paros J, Forman HP.
Update on the diagnostic radiologist shortage. AJR
2004; 182:301305
11. Meghea CI, Sunshine JH. Whos overworked and
whos underworked among radiologists? An update on the radiologist shortage. Radiology 2005;
236:932938

12. Saket DD, Nwanze CC, Maynard CD, Sunshine

JH, Forman HP. Update on the diagnostic radiologist employment market: findings through
2004. AJR 2005; 185:14081415
13. Bhargavan M, Sunshine JH. Workload of radiologists in the United States in 19981999 and trends
since 19951996. AJR 2002; 179:11231128
14. Bhargavan M, Sunshine JH. Utilization of radiology services in the United States: levels and trends
in modalities, regions, and populations. Radiology
2005; 234:824832
15. Bhargavan M, Sunshine JH. Workload of radiologists in the United States in 20022003 and trends
since 19911992. Radiology 2005; 236:920931
16. Sunshine JH, Cypel YS, Schepps B. Diagnostic radiologists in 2000: basic characteristics, practices,
and issues related to the radiologist shortage. AJR
2002; 178:291301
17. Sunshine JH, Lewis RS, Bhargavan M. A portrait of
interventional radiologists in the United States. AJR
2005; 183;11031112
18. Deitch CH, Chan WC, Sunshine JH, Shaffer KA.
Profile of U.S. radiologists at mid decade: overview
of findings from the 1995 survey of radiologists. Radiology 1997; 202:6977
19. Pamuk ER. Cautiously adjusting to the new millennium: changing to the 2000 population standard.
Am J Public Health 2001; 91:11741176
20. Krieger N, Williams DR. Changing to the 2000
standard million: are declining racial/ethnic and
socioeconomic inequalities in health real
progress or statistical illusion? Am J Public
Health 2001; 91:12091213
21. U.S. Department of Labor: Bureau of Labor Statistics Web site. Civilian labor force participation rates
by sex, age, race, and Hispanic origin. Available at:
www.bls.gov/emp/emplab2002-03.htm. Accessed
December 2005
22. Gendell M. Retirement age declines again in 1990s.
Monthly Labor Review Online 2001; 24:1221
23. Burtless G, Quinn JF. Is working longer the answer
for an aging workforce? Issues in brief. Boston,
MA: Center for Retirement Research at Boston College, 2002

1411