Impact of Generalist Physician Initiatives on Residency Selection
Michael H. Malloy, M.D., M.S*., Christine A. Stroup-Benham, Ph.D.†
*Department of Pediatric
†Office of Institutional Analysis
University of Texas Medical Branch
Galveston, TexasObjective:To compare the residency selection choices of students who experienced courses resulting from generalist physician initiatives to choices made by students prior to the implementation of those courses and to describe the characteristics of students selecting primary care residencies.
Background:In the fall of 1994 a first year Community Continuity
Experience course was initiated and in the summer of 1995 a third year Multidisciplinary Ambulatory Clerkship was begun at the University of Texas Medical Branch in Galveston. These courses were inserted into the curriculum to enhance and promote primary care education.
Design/Methods:We examined the residency selections of cohorts of graduating medical students before (1992-1996) and after (1997-1999) the implementation of the primary care courses. Survey information on career preferences at matriculation and in the fourth year of medical school were available for students graduating after the programs began. We compared the career preferences and characteristics of those students who selected a primary care residency to those who did not.
Results:Prior to the implementation of the programs, 45%(425/950) of students graduating selected primary care residencies compared to 45% (210/465) of students participating in the programs (p=0.88). At matriculation, 45% of students had listed a primary care discipline as their first career choice. Among the students who had indicated this degree of primary care interest 61% ended up matching in a primary care discipline. At year 4, 31% of students indicated a primary care discipline as their first career choice and 92% of these students matched to a primary care residency. By univariate analysis, minority students (53%) were more likely to select a primary care residency than non-minority students (40%); students in the two lowest grade point average quartiles (55% and 50%) selected primary care residencies compared to 37% and 38% of students in the top 2 quartiles; and students who stated that income potential had little or no impact on their choice were more likely to select a primary care residency (48%) than those who said income potential was important (37%).
Conclusions:We observed no significant trend towards higher proportions of graduating students selecting primary care discipline residencies as a result of implementing courses that emphasized primary care. Those students expressing an interest in a primary care discipline at their entrance into medical school were more likely to select a primary care residency. A more significant impact on graduating students interested in primary care may be made through the medical student selection process than by altering the curriculum.
Key Words: Generalist physician initiative, primary care, residency selection, career choice.
The Robert Wood Johnson Generalist Physician Initiatives were implemented to emphasize primary care education during medical school and to encourage selection of primary care disciplines as career choices. Previous research has documented the strong effect that demographic variables have on residency selection.1-4 Older students, married students, and students from rural areas appear more likely to select primary care residencies. Thus, there appear to be several avenues available to influence the number of students selecting primary care careers. The first avenue is through the medical student admission process. Selecting medical students for admission who have demographic characteristics associated with primary care residency selection and who express an interest in primary care appear to be the most viable means of influencing residency selection. Providing increased exposure to primary care experiences during the first and second years of medical school and emphasizing primary care ambulatory experiences during the clerkship years also seem to be viable options for influencing career selection.
At the University of Texas Medical Branch at Galveston, both avenues were used to influence primary care selection. In 1994 UTMB was awarded a Robert Wood Johnson Generalist Physician Initiative (GPI) grant to modify the curriculum in order to encourage generalist physician education and career choice. This analysis examines the impact of these modifications on primary care residency selection and some of the characteristics of entering medical students who stated an interest in primary care as a career choice.
Methods
Two courses were implemented at the University of Texas Medical Branch as a result of funding from the Generalist Physician Initiative to promote primary care education. Implemented in 1994, the first was a longitudinal two-year Community Continuity Experience Course in which first and second year medical students shadowed community physicians on a monthly basis. The second was a Multidisciplinary Ambulatory Clerkship which consisted of a 12 week rotation during the third year in which students spent equal amounts of time in the offices of community physicians in the disciplines of family medicine, internal medicine and pediatrics. The latter course was piloted among a subset of students (one-third of class) in 1995 academic year and implemented for all students in 1996. Thus, only part of the graduating class of 1997 may have experienced any effect of curricular change on their residency selection decisions. In addition, in 1995, a problem-based learning (PBL) track was implemented as a parallel curriculum to the traditional curriculum. Each year 24 students were admitted to the PBL track and approximately 176 students were admitted to the traditional lecture-based curriculum. The two tracks were run separately for the first two years, but students were integrated into a single curriculum during the 3rd and 4th years.
There were several objectives of this analysis. The first was to examine the residency selections of cohorts of graduating medical students before (1992-1996) and after (1997-1999) the implementation of the primary care courses. A second objective was to relate survey information collected from the graduating classes of 1997-1999 regarding career preferences at matriculation and during the 4th year of medical school to residency selection. Last, we were interested in comparing characteristics of students selecting a primary care residency or expressing a high interest in primary care to characteristics of those who did not.
To accomplish this analysis we obtained information on graduating student residency selection and student demographics from the UTMB Registrar’s Office. Career preference surveys were administered to students at the time of matriculation to medical school and during the 4th year for the graduating classes of 1997 - 1999. Response rates for the first year surveys were 79%, 71%, and 77% for the graduating classes of 1997, 1998 and 1999 respectively. Response rates for the fourth year surveys were 82%, 84%, and 79% for the same respective classes. The surveys were extensive questionnaires on student career/specialty preferences (Appendix B). Only those variables showing a statistically significant association with residency selection or variables that had been reported in the literature as associated with residency selection were used in the multivariate analysis. The racial category of Minority was further subclassified for analytic purposes into categories of African-American, Hispanic, and other (predominantly Asian). The Non-Minority category consisted of Caucasians. Grade point averages were obtained from the Registrar’s office for students graduating in the years 1997 – 1999 and categorized into quartiles, as were USMLE part II scores. For the years 1997-1999 only records with complete data were used. Approximately 15% of the records were missing complete data.
Comparison of discreet variables for measures of association or trends over time were done by chi-square tests of general association. Logistic regression modeling was used to measure the strength of association of between the dependent variable concerning residency selection and the various independent demographic, survey, and school performance variables.5 Analyses were conducted using SAS™ 6 A p-value of <0.05 was set as the criteria for statistical significance.
Results
A residency was defined as a primary care residency if it was a residency in family medicine, internal medicine, pediatrics or meds/peds. There was no significant trend in the proportion of students selecting a primary care residency from 1992 to 1999 (Table 1). Prior to the implementation of the GPI programs in 1992-1996, 45% (425/950) of students selected primary care residencies compared to 45% (210/465) in 1997-1999 (p=0.88).
During the first and fourth years of medical school for the graduating classes of 1997-1999, surveys were administered to determine the career interests of students and their attitudes towards various aspects of medicine. Among the students responding to the survey, there were 209 students or 45% of the 3 cohorts who selected a primary care discipline as their first choice during their first year of medical school. By the 4th year the number had decreased to 142 or 31% of the graduating classes. Only 90 of the students (19% of the graduating classes) indicated an interest in both the 1st and 4th years. Among those students selecting a primary care discipline as their top choice for a future career during their first year of medical school, 61% ended up matching in a primary care residency. Among students selecting a primary care discipline as their top choice for a career in the 4th year, 92% matched to a primary care residency.
From the survey results of the classes of 1997 -1999, 53% of Minority students selected a primary are residency compared to 40% of Non-Minority students (Table 2). This was true for all sub-categories of Minority students. There were no significant differences between men and women in their selection of primary care residencies, 42% of men selected primary care residencies compared to 49% of women, p=0.11. Although it did not attain statistical significance, only 32% of students < 22 years of age chose primary care residencies compared to 47% of students in the 22-23 year age category and 46% of students in the over 23 year of age category, p=0.15.
Grade point averages were computed on a 100 point scale and averages less than 70 were considered failing. The mean ± s.d grade point average (GPA) at graduation for students selecting a primary care residency was 86.9 ± 3.3 compared to 87.9 ± 3.1 for students who did not (p<0.01). When examined by quartiles, 55 % and 50% of students in the third and fourth GPA quartiles, respectively, selected primary care residencies compared to 38% and 37% of students in the first and second quartiles respectively (p<0.01). The mean ± s.d. USMLE Part II score for students selecting a primary care residency was 207.1 ± 21.3 compared to 209.8 ± 19.4 for students selecting other residencies (p=0.16). When examined by quartiles, 44%, 38%, 43%, and 55% selected primary care residencies in the first, second, third and fourth quartiles, respectively (p=0.07). These data are presented in Table 2 .
Multiple logistic modeling was used to examine the odds ratios for matching in a primary care residency for various student characteristics that were available in the first year of medical school. Independent of other characteristics, a student’s stated preference for seeking a primary care career in the first year of medical school was the most significant predictor for matching in a primary care residency (odds ratio=3.34, 95% confidence interval=2.22, 5.03), Table 2. Within the subcategories of the Minority racial groups, the "Other" category was the only racial category significantly associated with selecting a primary care residency independent of other characteristics. This "Other" category was composed primarily of Asian students. The youngest category of students (< 22 years of age) was least likely to select a primary care residency (odds ratio=0.45, 95% confidence interval=0.22, 0.93). Independent of other characteristics, students in the third GPA quartile had a greater odds of selecting a primary care residency than students in the first quartile, odds ratio=2.03, 95% confidence interval=1.15, 3.59). Students in the lowest GPA quartile also had a higher odds of matching in a primary care residency than the highest quartile students, however, the odds ratio was not statistically significant odds ratio=3.23, the 95% confidence interval=0.40, 26.23.
Conclusions
In conclusion, we observed no significant trend towards higher proportions of graduating students selecting primary care discipline residencies as a result of implementing courses that emphasized primary care. Those students expressing an interest in a primary care discipline during their first year of medical school were more likely to select a primary care residency. Being a minority student, particularly Asian; not being among the youngest students; and being in the lower GPA quartiles were also associated with selecting a primary care residency.
These observations confirm previous reports of the importance of demographic characteristics in the type of residency selected 1 2 3 4. Older students, married students, and students from rural communities have been previously reported more likely to select a primary care residency. Based on the results of a survey of 4th year medical students who had already been through the matching process, Targett et al reported that two groups of student were apparent 1. Those influenced by medical school faculty and the medical school curriculum had a greater tendency to select specialty oriented residencies, while another group influenced more by their social values and factors outside of the medical school curriculum were more likely to select a primary care residency. The implied question posed by these authors was whether or not the medical school curriculum could be altered to promote primary care residency selection. Our results would suggest that curriculum modification has not made a significant difference on residency choice.
Other factors affecting medical student residency selection include financial burden. Rosenthal et al. reported the higher the level of debt among students at Jefferson Medical College, the less likely they were to select a family practice residency.7 Kiker et al. confirmed that observation using the responses obtained from the Association of American Medical Colleges’ Medical School Questionnaire administered to 3,200 graduating medical students in 1995 4. That is, the odds ratios for selecting a non-primary care residency increased as the concern about student indebtedness increased . We were unable to confirm this observation since no differences were observed across the levels of indebtedness.
The accuracy of early declarations of interest in a primary care career in predicting residency selection appears quite reliable. Among the 209 students (45% of the graduating class) who indicated an interest in a primary care discipline during their first year of medical school, 127 (61%) selected a primary care residency. As one might expect when students were surveyed in the 4th year the percentage of students actually entering a primary care residency who said they were interested in a primary care career was very high (92%). The proportion of students expressing an interest in a primary care career, however, had dropped to 31% of the graduating seniors. Interestingly, 45% of the graduating students ended up doing primary care residencies. This additional 79 students who did not indicate an interest in a primary care career, but who accepted a primary care residency, may represent students who did not get their first choice of some other subspecialty residency or perhaps they had a change of heart prior to graduation. The question that presents itself is whether the medical school curricular experience had any affect on moving students away or toward an interest in a primary care career. For these students, we do not have data that would allow us to determine this.
Other studies surveying student interest in primary care careers during the first year of medical school have found wide ranges of interest by medical school. Grayson et al. reported 26.6% of first year medical students at New York Medical College expressed an interest in a primary care career compared with 52.2% of 1st year students at East Carolina University (ECU).8 Major differences in the population characteristics that might account for the difference in career preferences were a larger proportion of African-American students at ECU, a larger rural population at ECU, and a lower level of college indebtedness at ECU. A rural background and level of indebtedness did not attain significance as predictors of residency selection in our study.
Role model physicians have also been reported to be influential on medical students selection of residency. Wright et al. reported odds ratios for students interacting with role models in a given clinical field and choosing that same field for residency to be 12.8 for pediatrics, 5.1 for family medicine, 4.7 for internal medicine and 3.6 for surgery.9 The impact of role models appears to be related more to the individual physician’s role model characteristics than to the curriculum within which the role model functions. Basco and Reigart further evaluated the effect of career-influencing physician role models.10 They observed that many students had met a physician role model in the first and second years of schools prior to making their residency selection and were thus "influencable". The authors concluded that medical schools should track the degree to which students are exposed to all specialties, not just primary care, to determine if primary care role model exposure is adequate.
Academic performance has also been suggested as a determinant of residency selection patterns. The highest achieving students have been reported to be more likely to select internal medicine residencies than family medicine residencies.11-13 Fenderson et al. reported a greater percentage of the top ranked students at Jefferson Medical College selected internal medicine residencies than family medicine residencies. When internal medicine was included in the category of primary care residencies, the percentage of top ranked students selecting primary care residencies (58%) appears comparable to the percentage of lower ranked students selecting primary care residencies (57%) 13. Jarwecky et al, however, reported that over time, Alpha Omega Alpha students compared to the bottom 10% of students at the University of Kentucky College of Medicine have selected specialties other than primary care.14 Primary care in their study included internal medicine residencies. In our population of students we did observe a greater proportion of the students in the lower half of the classes selecting primary care residencies than in the upper half.
Two additional issues should be considered. First, how well do students know what they will do after four years of medical school? Based on this study’s results and others observations , there are student characteristics that appear to be related to a selection of a primary care residency that precede exposure to the medical school curriculum. In addition, the strongest predictor of selecting a primary care residency at entrance to medical school was a student’s expression of interest in a primary care career. However, the validity of this observation might be questioned. Ree, et al. ina theretical discussion of medical student specialty choice, suggest that the "human inability to observe and, perhaps , accurately report the cognitive process responsible for one’s own behavior", makes student observations of career choice suspect.15 The second issue is, do students attempt to deceive the admissions system in order to assure themselves a place in medical school? That is to say, if students perceive an advantage in portraying themselves as interested in primary care, would they do so even when they were not interested in primary care? We would suggest that even if this overstatement does occur, an observation supported by at least one subjective article in the literature16 the student’s stated interest still appears to be the best predictor.
In summary, this analysis suggests that as the percentage of minority students, older students, and students expressing an interest in primary care careers increase, the percentage of students selecting primary care residencies will increase. Thus, the medical student selection process appears to play a greater role in producing graduating students interested in primary care as a career than modifications of the curriculum.
References
1. Taggart MP, Wartman SA, and Wessen AF. An analysis of medical students’ residency and specialty choices. Soc Sci Med 1987;25:1063-1068.
2. Greer T and Carline JD. Specialty choice by medical students: recent graduate follow-up survey at the University of Washington. Fam Med 1989;21:127-131.
3. Campos-Outcalt D. Family practice specialty selection: a research agenda. Fam Med 1991;23:609-619.
4. Kiker BF and Zeh M. Relative income expectations, expected malpractice premium costs, and other determinants of physician specialty choice. J Health and Soc Behav 1998;39:152-167.
5. Kleinbaum DG, Kupper LL, and Morgenstern. Epidemiologic Research: Principles and Quantitative Methods. Lifetime Learning Publ, Belmont, CA. 1982: 447-457.
6. SAS Institute Inc. SAS/STAT User’s Guide, Release 6.12 Edition. Cary, NC: SAS Institute Inc., 1996.
7. Rosenthal MP, Marquette PA, and Diamond JJ. Trends along the debt-income axis: implications for medical student’s selections of family practice careers. Acad Med 1996;71:675-677.
8. Grayson MS, Newton DA, and Whitley TW. First-year medical students’ knowledge of and attitudes toward primary care careers. Fam Med 1996;28:337-342.
9. Wright S, Wong A, and Newill C. The impact of role models on medical students. J Gen Intern Med 1997;12:53-56.
10. Basco WT and Reigart JR. When do medical students identify career-influencing physician role models? Acad Med 2001;76:380-382.
11. Golden WE. Initial career choices of medical school honors graduates in the early 1970s and 1980s. Acad Med 1989;64:616-621.
12. Babbott D, Weaver SO, Baldwin DC. Person characteristics, career plans, and specialty choices of medical students electoed to alpha omega alpha. Arch Intern Med 1989;149:576-580.
13. Fenderson BA, Hojat M, Damjanov I, and Rubin E. Characteristics of medical students completing an honors program in pathology. Hum Pathol 1999;30:1296-1301.
14. Jarecky RK, Donnelly MB, Rubeck RF, and Schwartz RW. Changes in the patterns of specialties selected by high and low academic performers before and after 1980. Acad Med 1993;68:158-160.
15. Reed VA, Jernstedt GC, and Reber ES. Understanding and improving medical student specialty choice: a synthesis of the literature using decision theory as a referent. Teaching and Learning in Medicine 2001;13:117-129.
Acknowledgments: This paper was pre-sented in part at the Annual Meeting of the Robert Wood Johnson Generalist Physician Initiatives, January 8, 2000, and supported in part by a grant from the Robert Wood Johnson Generalist Physician Initiative
