- Email: [email protected]
Career choices of 135 cardiology trainees at Duke University Medical Center from 1970 to 1984
CareerChoicesof 135 CardiologyTraineesat Duke UniversityMedicalCenterfrom 1970 to 1984 EDWARD L.C. PRITCHEll, MD, GALEN S. WAGNER, MD, ANDREW G. WALLACE, MD, and JOSEPH C. GREENFIELD, Jr., MD
Alarm has been expressed at recently presented evidence showing that diminIshin numbers of physicians are enterin9 academk careers. Ihe experience of the cardkkgy trainkrg program at a university medkal center between 1970 and 1984 was reviewed to determins ths career paths chosen by its trainees. During the study period, 135 physkians received train@ Between 1970 and 1978 the percentage of trainees makin academk medkine their initlal career choke ffU&ed conskkrably. Ebeginning in 1978, the percenta9e enterin9 academic medicine steadily increased; in ths most re-
cent class, 8 of 9 trainees accepted academic facultypositkns.Amon972formertraineeswhojoinsd an academic faculty after finishin train& approximately 7% per year left academic medkfne for cllnkal practice. The median lergth of an academic career was 10 years. Individual MltuWns may be able to reverse the national trend of trainees makIn clinkal practice their initial career choke. However, physicians who leave academic medicine for clhlcal practke may continue to deplete faculty ranks. (Am J Cardkl
T
hat academic physicians (in general) and clinical investigators (in particular) are an “endangered species” has become a popular topic for editorial comment.1-3 Data have been presented that showed fewer physicians are entering academic careers and more are entering clinical practice. Few data are available on a related problem: the attrition rate at which academic physicians leave their careers. We therefore reviewed the initial career choices and the subsequent career changes of trainees in the cardiology division at Duke University.
1988;57:313-315)
the national sessions of major cardiology societies. The position held by trainees for each year was available in correspondence kept in divisional personnel files or in professional directories.4*5 Career choices of trainees were classified into the following categories: (1) Academic faculty at Duke University Medical Center. (2) Academic faculty at another university medical center in the United States, a “dean’s committee” Veteran’s Administration hospital, the National Institutes of Health (NIH), or the Uniformed Services School of Health Sciences. (3) Clinical practice of cardiology. (4) Federal hospitals including those maintained by the Veteran’s Administration (when not part of a university medical center), Public Health Service and the uniformed services. (5) Private industry (e.g., pharmaceuticals or technology). (6) Medical practice in a foreign country. Data analysis: First, we calculated the percentage of trainees entering each of the 6 categories every year to determine if the initial career choice changed during the study period. We then examined data from the trainees who chose academic careers (combined category 1 and category 2) to determine the rate at which those trainees left academia for clinical practice (category 3). We used the Kaplan-Meier life table method to calculate the cumulative proportion who remained in an academic position for each subsequent year.6 For trainees who had graduated from foreign medical schools, we only calculated raw numbers because the number was relatively small.
Methods We studied 135 physicians who trained in the division of cardiology at this medical center between July 1,197O and July 1,1984. We used the following criteria to identify “trainees”: (1) The trainee must have held the M.D. degree at the time he or she worked in the division, (2) the trainee must have worked in the division for at least 12 months, and (3) the trainee eventually must have received training in clinical cardiology equivalent to that required to take the American Board of Internal Medicine subspecialty of Cardiovascular Medicine examination, The division maintains contact with former trainees through a society that meets in conjunction with From the Division of Cardiology, Department of Medicine, Duke University Medical Center, Durham, North Carolina. Manuscript received May 10,1985; revised manuscript received June 13,1985, accepted June 18.1985. Address for reprints: Edward L.C. Pritchett, MD, Box 3477, Duke University Medical Center, Durham, North Carolina 27710.
Results The initial career choice of trainees varied during the study period (Fig. 1). The proportion that chose 313
314
CARDIOLOGY
CAREER
CHOICES
loo8060-
n
INDUSTRY
0
FOREIGN
q
FEDERAL
fl
$;&$‘;
FIGURE 1. lnltlal career choice of tralnees by year that they completed the program (1970 through 1994). Beglnnlng In 1979 the percentage of tralnees Jolnlng academic faculties rteadlly Increased.
OTHER UNIVERSITY i?d FACULTY
q
40-
DUKE FACULTY
20 O-
academic careers(categories1 and 2) fluctuated substantially. However, beginning in 1978 there was a steadyincreasein the proportion that choseacademic careers; and this trend continued to the end of the study period (July 1, 1984).During the first 7 years of the study period 35% of traineesmade academicmedicine their initial career choice;during the last 7 years 58% made this choice. In the most recent class,8 of 9 traineesacceptedacademicpositions.During the most recentyearsabouthalf of the traineeselectedto accept academic appointments at this medical center (category 1) and half acceptedappointments at other academic medical centers(category2). The programsize tended to increase during this period, although there were annual fluctuations (Fig. 2). Between 1970and 1977the averageclasssize was 7.7 f 2.9persons.Between 1978and 1984,the averageclass size had increasedto 10.6 f 2.3 persons(p’
study period. They left these academic positions for clinical practice(category3) at a rate of approximately 7% per year beginning in the second year (Fig. 3). The median length of an academic.career was 10 years. Nineteen of the 135trainees(14%) were graduates of foreign medical schools,locatedin Canada(8).Australia (5), United Kingdom (31,New Zealand (I), the Philippine Islands (1) or Venezuela (11.Fourteen of these 19 returned to their country of origin, 3 were in academicpositionsin the United States,and 2 were in clinical practicein the United States.One of theselast 2 had held academicpositionsin the United Statesfor 6 years before entering clinical practice.
0 Z
15r
2v
lb
L
1
II
11
TIME :INCE dFlNdlNG
0
70
I
1
72
84
I
74
I
76
I1
11
78
1 80
1
82
*
1 84
YEAR FIGURE 2. Total number of trainees (1970 through 1994). The number gradual Increase has occurred.
flnlsftlng thelr tralnlng each year of trainees has fluctuated, but a
1
‘1
0
8TRAIbtNG
‘1
‘J
12 ( YEARSi
FIGURE 3. Cumulatlve proportlon of tralnees remalnkg In academic positions durlng follow-up (1970 through 1994). Seventy-two tralnees accepted an academk posltkn at some time after flnlshlng their tralnlftg program. Each step In the curve represents a year when 1 or more former tralnees left academic medklne for cllnlcal practice. The number of tralnees who have actually been followed for 10 or more years Is relatlvely small because the classes were smaller In the early 1970s. The behavlor of the curve Is heavlly Influenced In Its later years by the career changes of 1 or 2 persons who entered the program near the beglnnlng of the study . period.
Discussion During our study period (1970 through 1984), an increasing proportion of trainees in this cardiology division made academic medicine their initial career choice (Fig. 1). This experience appears to contrast sharply with national trends described by Wyngaarden7 who reported that the number of physicians being supported by NIH research training grants declined from 86% in 1968, to 77% in 1971, and to 42% in 1978. Additional data on this point came from surveys showing that the number of physicians reporting research as their primary activity declined from 15,400 in 1968 to 7,900 in 1975.* We do not interpret our data to mean that the conclusions about national trends were wrong. Rather, we believe that some institutions, including many others besides our own, likely have experiences that are opposite from the national trend. During the study period a number of changes took place in academic medicine and in our division that may have caused our findings. In 1974 grants for research training (National Research Service Awards) were authorized by Congress with the National Research Act. This program replaced earlier NIH programs with more liberal provisions for supporting clinical training. This new type of training grant was implemented nationally and funded in this division on July 1,1975. Trainees whom it supported began graduating from the program in the late l%‘Os, which was about the same time when the number of trainees entering academic medicine began to rise sharply. Coincident with external incentives to emphasize research training, a number of changes were made within this division. The selection process progressively emphasized identifying applicants with a long and serious academic commitment. Applicants’ records were examined for evidence of academic associations (such as research experiences) dating back to medical school.g Trainees were matched, if possible, with a specific research mentor at the time of selection. The program of clinical cardiology training for persons wishing to enter clinical practice was discontinued; the last such applicant who was accepted into the program was the lone trainee who entered clinical practice July 1, 1984. Each trainee in our program is now assured of 18 to 24 months of research training. [The same commitment is made to each trainee regardless of whether he or she is supported by the National Research Service Award training grant or by institutional funds.) If a research mentor was identified at the time of selection,
it is considered preferable for a trainee to have his or her research experieni:e at: the very beginning of the fellowship. The research time is taken as an uninterrupted block, while clinical training may be broken up into 6-month segments. Clinical training emphasizes fundamental diagnostic and therapeutic skills required for a consultant cardiologist. Between 1970 and 1976, 15 trainees entered military service. Only 3 of these 15 later entered academic medicine. The draft law expired in July, 1973; after that time few trainees chose to enter military service. When we examined the careers of the 72 former trainees who entered academic medicine, we found that the proportion who remained in academic positions (Fig. 3) declined at a rate of about 7% per year. The median length of an academic career was IO years, which is slightly longer than the median length of time (8.5 years) that principal investigators continued to hold NIH research grant support in another study.* The number of trainees who have actually been followed for 10 or more years is relatively small because the fellowship classes were smaller in the early 1970s. The behavior of the curve in Figure 3 is therefore heavily influenced in its later years by career changes of 1 or 2 persons who entered the program near the beginning of the study period. It is too soon to determine whether changes in the selection of trainees and structure of the training program will affect this attrition rate. Most graduates of foreign medical schools who were trained in our program have returned to their home country, where they are working in teaching hospitals. Most foreign nationals were selected by scientific councils in their home countries and sponsored by these councils during their training.
References 1. Symposium on the Academic Physician: An Endangered Species. Bull NY Acod Med 1981;57:411-502. 2. Forbes M Sr. Back to research where the product is the byproduct. Forbes 1981;March 2:ZO. 3. Petersdorf RG. Academic medicine: No longer threadbare or genteel. N Engl r Med 1981;304:841-843. 4. Directory of Medical Specialists, Zlst ed., 1983-1984. Chicago: Marquis Who’s who. 1983. 5. Who’s Who in America, 42nd ed., 1982-1983. Chicago: Marquis Who’s Who, 1982. 6. Kaplan EL, Meier P. Non-parametric estimation from incomplete observations. r Am Stat Assoc 1958;53:457. 7. Wyngaarden JB. The clinical investigator as an endangered species. Bull NY Acad Med 1981;57:415-426. 8. Burns TW. Manpower needs for academic medicine. Ann Intern Med 1982;97:611-612. 9. Davis WK, Kelley WN. Factors influencing decisions to enter careers in clinical investigation. r Med Educ 1982;57:275-281.
Alarm has been expressed at recently presented evidence showing that diminIshin numbers of physicians are enterin9 academk careers. Ihe experience of the cardkkgy trainkrg program at a university medkal center between 1970 and 1984 was reviewed to determins ths career paths chosen by its trainees. During the study period, 135 physkians received train@ Between 1970 and 1978 the percentage of trainees makin academk medkine their initlal career choke ffU&ed conskkrably. Ebeginning in 1978, the percenta9e enterin9 academic medicine steadily increased; in ths most re-
cent class, 8 of 9 trainees accepted academic facultypositkns.Amon972formertraineeswhojoinsd an academic faculty after finishin train& approximately 7% per year left academic medkfne for cllnkal practice. The median lergth of an academic career was 10 years. Individual MltuWns may be able to reverse the national trend of trainees makIn clinkal practice their initial career choke. However, physicians who leave academic medicine for clhlcal practke may continue to deplete faculty ranks. (Am J Cardkl
T
hat academic physicians (in general) and clinical investigators (in particular) are an “endangered species” has become a popular topic for editorial comment.1-3 Data have been presented that showed fewer physicians are entering academic careers and more are entering clinical practice. Few data are available on a related problem: the attrition rate at which academic physicians leave their careers. We therefore reviewed the initial career choices and the subsequent career changes of trainees in the cardiology division at Duke University.
1988;57:313-315)
the national sessions of major cardiology societies. The position held by trainees for each year was available in correspondence kept in divisional personnel files or in professional directories.4*5 Career choices of trainees were classified into the following categories: (1) Academic faculty at Duke University Medical Center. (2) Academic faculty at another university medical center in the United States, a “dean’s committee” Veteran’s Administration hospital, the National Institutes of Health (NIH), or the Uniformed Services School of Health Sciences. (3) Clinical practice of cardiology. (4) Federal hospitals including those maintained by the Veteran’s Administration (when not part of a university medical center), Public Health Service and the uniformed services. (5) Private industry (e.g., pharmaceuticals or technology). (6) Medical practice in a foreign country. Data analysis: First, we calculated the percentage of trainees entering each of the 6 categories every year to determine if the initial career choice changed during the study period. We then examined data from the trainees who chose academic careers (combined category 1 and category 2) to determine the rate at which those trainees left academia for clinical practice (category 3). We used the Kaplan-Meier life table method to calculate the cumulative proportion who remained in an academic position for each subsequent year.6 For trainees who had graduated from foreign medical schools, we only calculated raw numbers because the number was relatively small.
Methods We studied 135 physicians who trained in the division of cardiology at this medical center between July 1,197O and July 1,1984. We used the following criteria to identify “trainees”: (1) The trainee must have held the M.D. degree at the time he or she worked in the division, (2) the trainee must have worked in the division for at least 12 months, and (3) the trainee eventually must have received training in clinical cardiology equivalent to that required to take the American Board of Internal Medicine subspecialty of Cardiovascular Medicine examination, The division maintains contact with former trainees through a society that meets in conjunction with From the Division of Cardiology, Department of Medicine, Duke University Medical Center, Durham, North Carolina. Manuscript received May 10,1985; revised manuscript received June 13,1985, accepted June 18.1985. Address for reprints: Edward L.C. Pritchett, MD, Box 3477, Duke University Medical Center, Durham, North Carolina 27710.
Results The initial career choice of trainees varied during the study period (Fig. 1). The proportion that chose 313
314
CARDIOLOGY
CAREER
CHOICES
loo8060-
n
INDUSTRY
0
FOREIGN
q
FEDERAL
fl
$;&$‘;
FIGURE 1. lnltlal career choice of tralnees by year that they completed the program (1970 through 1994). Beglnnlng In 1979 the percentage of tralnees Jolnlng academic faculties rteadlly Increased.
OTHER UNIVERSITY i?d FACULTY
q
40-
DUKE FACULTY
20 O-
academic careers(categories1 and 2) fluctuated substantially. However, beginning in 1978 there was a steadyincreasein the proportion that choseacademic careers; and this trend continued to the end of the study period (July 1, 1984).During the first 7 years of the study period 35% of traineesmade academicmedicine their initial career choice;during the last 7 years 58% made this choice. In the most recent class,8 of 9 traineesacceptedacademicpositions.During the most recentyearsabouthalf of the traineeselectedto accept academic appointments at this medical center (category 1) and half acceptedappointments at other academic medical centers(category2). The programsize tended to increase during this period, although there were annual fluctuations (Fig. 2). Between 1970and 1977the averageclasssize was 7.7 f 2.9persons.Between 1978and 1984,the averageclass size had increasedto 10.6 f 2.3 persons(p’
study period. They left these academic positions for clinical practice(category3) at a rate of approximately 7% per year beginning in the second year (Fig. 3). The median length of an academic.career was 10 years. Nineteen of the 135trainees(14%) were graduates of foreign medical schools,locatedin Canada(8).Australia (5), United Kingdom (31,New Zealand (I), the Philippine Islands (1) or Venezuela (11.Fourteen of these 19 returned to their country of origin, 3 were in academicpositionsin the United States,and 2 were in clinical practicein the United States.One of theselast 2 had held academicpositionsin the United Statesfor 6 years before entering clinical practice.
0 Z
15r
2v
lb
L
1
II
11
TIME :INCE dFlNdlNG
0
70
I
1
72
84
I
74
I
76
I1
11
78
1 80
1
82
*
1 84
YEAR FIGURE 2. Total number of trainees (1970 through 1994). The number gradual Increase has occurred.
flnlsftlng thelr tralnlng each year of trainees has fluctuated, but a
1
‘1
0
8TRAIbtNG
‘1
‘J
12 ( YEARSi
FIGURE 3. Cumulatlve proportlon of tralnees remalnkg In academic positions durlng follow-up (1970 through 1994). Seventy-two tralnees accepted an academk posltkn at some time after flnlshlng their tralnlftg program. Each step In the curve represents a year when 1 or more former tralnees left academic medklne for cllnlcal practice. The number of tralnees who have actually been followed for 10 or more years Is relatlvely small because the classes were smaller In the early 1970s. The behavlor of the curve Is heavlly Influenced In Its later years by the career changes of 1 or 2 persons who entered the program near the beglnnlng of the study . period.
Discussion During our study period (1970 through 1984), an increasing proportion of trainees in this cardiology division made academic medicine their initial career choice (Fig. 1). This experience appears to contrast sharply with national trends described by Wyngaarden7 who reported that the number of physicians being supported by NIH research training grants declined from 86% in 1968, to 77% in 1971, and to 42% in 1978. Additional data on this point came from surveys showing that the number of physicians reporting research as their primary activity declined from 15,400 in 1968 to 7,900 in 1975.* We do not interpret our data to mean that the conclusions about national trends were wrong. Rather, we believe that some institutions, including many others besides our own, likely have experiences that are opposite from the national trend. During the study period a number of changes took place in academic medicine and in our division that may have caused our findings. In 1974 grants for research training (National Research Service Awards) were authorized by Congress with the National Research Act. This program replaced earlier NIH programs with more liberal provisions for supporting clinical training. This new type of training grant was implemented nationally and funded in this division on July 1,1975. Trainees whom it supported began graduating from the program in the late l%‘Os, which was about the same time when the number of trainees entering academic medicine began to rise sharply. Coincident with external incentives to emphasize research training, a number of changes were made within this division. The selection process progressively emphasized identifying applicants with a long and serious academic commitment. Applicants’ records were examined for evidence of academic associations (such as research experiences) dating back to medical school.g Trainees were matched, if possible, with a specific research mentor at the time of selection. The program of clinical cardiology training for persons wishing to enter clinical practice was discontinued; the last such applicant who was accepted into the program was the lone trainee who entered clinical practice July 1, 1984. Each trainee in our program is now assured of 18 to 24 months of research training. [The same commitment is made to each trainee regardless of whether he or she is supported by the National Research Service Award training grant or by institutional funds.) If a research mentor was identified at the time of selection,
it is considered preferable for a trainee to have his or her research experieni:e at: the very beginning of the fellowship. The research time is taken as an uninterrupted block, while clinical training may be broken up into 6-month segments. Clinical training emphasizes fundamental diagnostic and therapeutic skills required for a consultant cardiologist. Between 1970 and 1976, 15 trainees entered military service. Only 3 of these 15 later entered academic medicine. The draft law expired in July, 1973; after that time few trainees chose to enter military service. When we examined the careers of the 72 former trainees who entered academic medicine, we found that the proportion who remained in academic positions (Fig. 3) declined at a rate of about 7% per year. The median length of an academic career was IO years, which is slightly longer than the median length of time (8.5 years) that principal investigators continued to hold NIH research grant support in another study.* The number of trainees who have actually been followed for 10 or more years is relatively small because the fellowship classes were smaller in the early 1970s. The behavior of the curve in Figure 3 is therefore heavily influenced in its later years by career changes of 1 or 2 persons who entered the program near the beginning of the study period. It is too soon to determine whether changes in the selection of trainees and structure of the training program will affect this attrition rate. Most graduates of foreign medical schools who were trained in our program have returned to their home country, where they are working in teaching hospitals. Most foreign nationals were selected by scientific councils in their home countries and sponsored by these councils during their training.
References 1. Symposium on the Academic Physician: An Endangered Species. Bull NY Acod Med 1981;57:411-502. 2. Forbes M Sr. Back to research where the product is the byproduct. Forbes 1981;March 2:ZO. 3. Petersdorf RG. Academic medicine: No longer threadbare or genteel. N Engl r Med 1981;304:841-843. 4. Directory of Medical Specialists, Zlst ed., 1983-1984. Chicago: Marquis Who’s who. 1983. 5. Who’s Who in America, 42nd ed., 1982-1983. Chicago: Marquis Who’s Who, 1982. 6. Kaplan EL, Meier P. Non-parametric estimation from incomplete observations. r Am Stat Assoc 1958;53:457. 7. Wyngaarden JB. The clinical investigator as an endangered species. Bull NY Acad Med 1981;57:415-426. 8. Burns TW. Manpower needs for academic medicine. Ann Intern Med 1982;97:611-612. 9. Davis WK, Kelley WN. Factors influencing decisions to enter careers in clinical investigation. r Med Educ 1982;57:275-281.