Nuthalapaty, Francis S. MD; Jackson, James R. PhD; Owen, John MD
“Lifestyle” during residency has emerged as a focal issue in graduate medical education. The aspects of a favorable lifestyle that pertain to medical professionals have been broadly defined as control over the number of hours devoted to professional responsibilities to enable time for leisure, family, and avocational pursuits.1–3 There is a perception that today's generation of trainees places a greater emphasis on these aspects of lifestyle during residency and subsequent practice than generations past.4 This perception is likely the result of convincing data that lifestyle preferences have been an influential aspect of specialty selection since the 1980s1–3 best exemplified by the established trend of decreasing interest in general surgery. Bland et al.5 reported a drop in the percentage of U.S. senior students who selected this specialty as their first choice, from a peak of 12.1% in 1989, to 6.1% in 2001. In contrast, those specialties that allow more control over work hours and other lifestyle preferences in conjunction with higher compensation (e.g., radiology, anesthesiology, and emergency medicine) all displayed a trend of increasing interest since the mid-1990s.
Although lifestyle preferences appear to play a primary role in specialty selection, available research suggests that they have not played as significant a role in selecting a specialty-specific graduate medical education (GME) training environment. Between the early 1980s and mid-1990s, several investigators reported factors students used in selecting residency programs (see Table 1). 6–15 These studies confirmed that, of the numerous characteristics that defined their preferred GME training environments, applicants consistently emphasized three major issues: the geographic location of the training program, the perceived satisfaction of the residents with their training environment, and the quality of the academic experiences. In contrast, they assigned less importance to on-call frequency and to issues that could indirectly impact lifestyle (e.g., salary and opportunities for supplemental income like moonlighting).
The importance medical graduates give to lifestyle concerns during residency and to concerns during subsequent practice appear to differ. This may be due to either the limited scope of prior research (see Table 1) or a lack of recent evaluation. Understanding medical students’ perceptions of an optimal GME environment can help educators effectively structure their programs to provide the most effective balance between education, patient care, and residents’ well-being. The aim of our study was to identify which factors influence an applicant's residency program rank order. Given the established trend away from surgical specialties, a secondary aim of the study was to determine if there were significant differences in selected factors between students who applied to surgical versus nonsurgical specialties.
We reviewed the literature and identified previously published research in this area (see Table 1). Lifestyle factors used in these studies were cataloged, and we selected 20 that were broadly representative of three possible categories of influence (quality of life, the academic experience, and the workplace environment) to form the basis of our questionnaire. The questionnaire was converted into a dynamic, Internet-based tool that consisted of five sections. Section 1 questions sought demographic information such as age, gender, race, marital status, number of dependents, and geographic region of the medical school. Responses were optional for this section, but were required for Sections 2–5. The participant's responses to questions in Section 2 determined the series of questions posed subsequently.
Section 2 presented the participant with a series of yes-or-no questions to determine whether any of the 20 factors were important in their selection of the residency programs on their rank list. If a participant responded “yes” to any one of seven specific factors (amount of conference/didactic teaching, emphasis on medical student interaction, emphasis on research in the residency curriculum, level of patient management responsibility, amount of faculty supervision in patient care, frequency of in-house call, and size of the patient case load), he or she was presented with a five-point Likert scale in Section 3, which elicited the preferred magnitude of each (1 = least to 5 = most).
In Section 4, the participant was asked to rate the relative importance of each of the factors initially selected in Section 2 for determining the final rank order (1 = minimally important, 7 = extremely important). Finally, in Section 5, the participant was presented with a list of all the factors that received the highest rating in the previous section and asked to select the single most important factor used in determining the order of the rank list.
Survey Participants and Protocol
The questionnaire and study protocol were approved by the University of Alabama at Birmingham Institutional Review Board. The eligible population was all graduates of U.S. allopathic medical schools who registered with the Electronic Residency Application Service (ERAS) for the 2003 Match. ERAS staff queried their applicant database to identify eligible participants who provided an e-mail address (N = 16,363). An automated e-mail invitation to participate in the survey was then sent to each eligible participant following the deadline for submitting their rank-order list (February 27, 2003). One hundred eighty invitations were returned due to an invalid e-mail address, limiting the potential study population to 16,183. The e-mail invitation contained a brief explanation of the survey and provided a universal resource locator (URL) to our survey Web site. Each invitation contained a unique identification code that was embedded into the URL. Duplicate submissions were prevented through automated crosschecking of the unique identification code prior to allowing access to the questionnaire. All nonresponders were sent a second and, finally, a third e-mail invitation at one-week intervals (March 6 and March 13, 2003). The Web site was closed at 12 noon (CST) on March 20, 2003 (Match Day).
Only participants who completed the questionnaire in its entirety were included in the analysis. Eleven participants who indicated they had chosen a transitional year program were excluded due to the limited scope and variable nature of these types of programs. The selected prevalence of each of the 20 factors was calculated and the factors ranked in descending order. A mean response score and standard deviation were calculated for all Likert-scale responses. To assess whether differences in factor selection were associated with the type of specialty, specialties were categorically defined as either surgical or nonsurgical.16 Comparisons between categorical variables were performed using the chi-square test. The t test was used to compare continuous data. Statistical significance was set at p < .05. Analyses were performed using SPSS, version 11.5 (SPSS Inc., Chicago, IL).
Of the 16,183 eligible participants, 7,486 (46.3%) entered the Web site, and 7,183 (44.3%) qualified for inclusion in the analysis. We compared demographic characteristics and specialty distribution of the participants with the overall population of 2003 U.S. allopathic medical school graduates16 and found similar age, sex, ethnic, regional characteristics, and specialty (see Tables 2 and 3).
The mean number of factors selected by each participant in Section 2 was 14 ± 3. Fifteen of the 20 factors were selected by at least 50% of the participants (see Table 4). Four of the five factors selected by at least 90% of respondents were related to impressions of the work environment, and the fifth was geographic location of the residency program. The five factors selected by less than 50% of respondents included two academic factors (the amount of emphasis on research and the amount of emphasis on medical student interaction), and three factors related primarily to financial determinants of quality of life (employee benefits, salary, and supplemental income opportunities).
The relative importance of each of the 20 factors is listed in Table 5. The range of mean scores reported for each factor on the seven-point Likert-type scale was relatively small (4.9 ± 1.4 to 6.5 ± 0.9). In contrast, the selection prevalence of the single most important factor in determining rank-list order was more distinct. As shown in Table 6, how well the applicant thought he or she would fit into the residency program and the geographic location of the residency program each accounted for 35% and 30% of all responses, respectively. The mean scores for each of the seven characteristics of the GME environment showed that participants preferred relatively more resident management responsibility for patient care (4.3 ± 0.7), emphasis on medical student interaction (4.0 ± 0.8), conference/didactic teaching (3.9 ± 0.7), faculty supervision (3.7 ± 0.8), emphasis on research (3.5 ± 1.1), and a larger patient case load (3.5 ± 0.8), compared with time on-call (2.2 ± 1.0).
Following stratification of participants into surgical versus nonsurgical specialties, several distinctions were evident (see Table 4). Nonsurgical applicants were more likely than were surgical applicants to consider the geographic location of the residency program, the amount of faculty teaching on the wards, the frequency of on-call duty, the amount of clinical support services, the amount of emphasis on interacting with medical students, the types of employee benefits, salary, and supplemental income (moonlighting) opportunities. How well the applicant believed he or she would fit into the residency program was selected as the single most important factor in determining rank list order more frequently for surgical specialty applicants than for nonsurgical applicants versus geographic location of the residency program, which had lower selection prevalence for surgical applicants than nonsurgical applicants (see Table 6).
Small but statistically significant differences between surgical and nonsurgical applicants were also apparent in their preferences for various characteristics of the GME training environment. Surgical applicants as compared to nonsurgical applicants desired less conference/didactic teaching (3.8 ± 0.7 versus 3.9 ± 0.7, p < .001), a greater level of patient management responsibility (4.4 ± 0.6 versus 4.3 ± 0.7, p = .001), less faculty supervision in patient care (3.6 ± 0.7 versus 3.7 ± 0.8, p = .029), a greater frequency of on-call duty (2.3 ± 0.9 versus 2.1 ± 1.0, p < .001), and a larger patient case load (3.8 ± 0.8 versus 3.4 ± 0.8, p < .001).
To our knowledge, this study represents the first national survey of U.S. allopathic medical school graduates to determine factors that influenced their residency program rank-order list in the national residency matching program. We found that, although applicants considered a variety of factors in program selection, those most commonly cited indicated the importance of their impression of the residency work environment and the geographic location of the residency program. Issues dealing with the academic experience, such as the academic reputation of the program and curricular characteristics like conferences, case load, and supervision, tended to be in the middle tier. Applicants reported that they preferred a high level of resident management responsibility for patient care and a low on-call frequency. Financial issues such as salary, employee benefits, and supplemental income (moonlighting) opportunities were among the least commonly selected factors.
Comparisons of our results with those of prior studies (see Table 1) suggest both similarities and differences between today's medical graduates and their predecessors from the 1980s and early 1990s. First, there may be a trend toward greater emphasis on the quality of the work environment. The measure of a quality work environment appears to be in the level of sensitivity, collegiality, and satisfaction that residents and faculty have towards each other and their own professional appointments. Second, this increased emphasis on the work environment appears to parallel a decreased emphasis on the academic experience. The five factors related to the academic experience that had the highest selection prevalence ranked between seventh and 12th in the overall order of prevalence (see Table 4), which indicates their secondary importance in the process of residency selection. Third, factors that potentially impact lifestyle do not appear to have a significant influence in the process of residency selection. Only geographic location of the residency program is a top priority for applicants and has been since the earliest investigations on this topic.6,8,10,12–15 In contrast, no other factors that could potentially impact lifestyle, such as on-call frequency, cost of living, or employee benefits ranked higher than 13th of 20 in selection prevalence. The standardization of on-call frequency across residency programs through the recent implementation of duty hour limitations may have limited the importance of this issue as a discriminating factor. Similarly, salary and employee benefits also have limited variability between programs. To compensate for this issue, we also included cost of living and supplemental income opportunities, in an attempt to distinguish those applicants who felt that either earning more or spending less was a priority. The finding that all of these factors had uniformly low selection prevalences suggests that applicants continue to view residency as primarily a period of educational and not financial gain.
Overall, there was remarkable homogeneity in the selection prevalence of all factors between applicants to surgical versus nonsurgical specialties. Despite the several small but statistically significant differences identified between these two groups, few are likely to be of practical significance. For example, if we consider that a minimum difference of 5% or more in the selection prevalence of a particular factor is of practical significance, surgical and nonsurgical applicants differ only on four factors (none of which had a high overall selection prevalence): the amount of emphasis on research in the residency curriculum (54.4% versus 39.1%), the amount of moonlighting opportunities (12.3% versus 21.8%), the amount of clinical support services (46.1% versus 51.5%), and the amount of faculty teaching on the wards (68.3% versus 75.8%).
Undoubtedly, there are many potential reasons for the differences observed between applicants to surgical and nonsurgical programs. We speculate that applicants to surgical specialties may anticipate a greater quantity of required duty hours, and thus may not seek additional duty hours through moonlighting. Perhaps surgical applicants expect that performing clinical support services is a common part of their residency experience and have a greater tolerance for it. Additionally, surgical applicants may distinguish between intraoperative teaching from that which occurs pre- or postoperatively and value the former over the latter. Our general labeling of “teaching on the wards” does not accommodate such a discrimination. Finally, the 14.5%-higher selection prevalence for research emphasis among surgical applicants may indicate that more of them desire postresidency academic research careers.
Several issues may impact the validity of these findings. There are many other factors that could potentially influence residency selection and rank order. We chose to include factors we believed had limited covariance with other issues. For example, we felt that the needs of a spouse or significant other would likely be highly covariate with geographic location of the residency program and, therefore, excluded the former. There were 426 nonresponders to the demographic question on marital status. In their written comments, participants indicated that they were partnered and not married, which we failed to include as an option. Therefore, we suspect that many who did not answer this question fell into this category. There were also 434 nonresponders to the demographic question on race/ethnic background, although we had provided a diverse range of 20 possible responses. Therefore, we believe that this observation represents an inherent sensitivity on the part of some participants to this issue and not a lack of adequate options. We also did not assess whether the respondents were participating in a couple's match, which was at its highest rate ever in the 2003 Match.17 This factor certainly could have influenced residency selection in our cohort.
Our study has several strengths. The sample size is more than eight times that of any previous investigation (see Table 1). In addition, despite a 44% participation rate, the demographic characteristics of the study's population indicate that it is a highly representative sample of the overall U.S. medical school graduate population. Finally, unlike previous investigations, we minimized recall bias by timing data acquisition to the three weeks between rank list submission and the announcement of match results.
In summary, residency selection appears to be a complex process that involves consideration of many components. Our results suggest that residency applicants place significant value on their subjective impressions of the work environment rather than on more objective information regarding the academic experience. In addition, factors relating to lifestyle determinants appear to be among the least influential in this process. An effective GME environment that balances education, patient care, and residents’ well-being should measure the latter based on the satisfaction of the residents and faculty with the work environment, rather than issues relating to the academic experience or lifestyle. Research is needed to determine how this prioritization of the components of the GME environment may change as residents progress in their training.
We thank Alice R. Goepfert, MD, for her contributions to the survey design and Paul Jolly, PhD, and the ERAS staff for their assistance in the survey conduct. Dr. Owen's effort was supported by the National Institutes of Child Health and Human Development (K24 HD043314-01).
1.Schwartz RW, Jarecky RK, Strodel WE, Haley JV, Young B, Griffen WO Jr. Controllable lifestyle: a new factor in career choice by medical students. Acad Med. 1989;64:606–9.
2.Schwartz RW, Haley JV, Williams C, et al. The controllable lifestyle factor and students’ attitudes about specialty selection. Acad Med. 1990;65:207–10.
3.Dorsey ER, Jarjoura D, Rutecki GW. Influence of controllable lifestyle on recent trends in specialty choice by US medical students. JAMA. 2003;290:1173–8.
5.Bland KI, Isaacs G. Contemporary trends in student selection of medical specialties: the potential impact on general surgery. Arch Surg. 2002;137:259–67.
6.DiTomasso RA, DeLauro JP, Carter ST Jr. Factors influencing program selection among family practice residents. J Med Educ. 1983;58:527–33.
7.Weissman SH, Bashook PG. The 1982 first-year resident in psychiatry. Am J Psychiatry. 1984;141:1240–3.
8.Sledge WH, Leaf PJ, Sacks MH. Applicants’ choice of a residency training program. Am J Psychiatry. 1987;144:501–3.
9.Hitchcock MA, Kreis SR, Foster BM. Factors influencing student selection of family practice residency programs in Texas. Fam Med. 1989;21:122–6.
10.Simmonds AC, Robbins JM, Brinker MR, Rice JC, Kerstein MD. Factors important to students in selecting a residency program. Acad Med. 1990;65:640–3.
11.DeLisa JA, Jain SS, Campagnolo D, McCutcheon PH. Selecting a physical medicine and rehabilitation residency. Am J Phys Med Rehabil. 1992;71:72–6.
12.Flynn TC, Gerrity MS, Berkowitz LR. What do applicants look for when selecting internal medicine residency programs? A comparison of rating scale and open-ended responses. J Gen Intern Med. 1993;8:249–54.
13.Lebovits A, Cottrell JE, Capuano C. The selection of a residency program: prospective anesthesiologists compared to others. Anesth Analg. 1993;77:313–7.
14.Mayeaux EJ Jr, Arnold J. Why first-year family practice residents choose their residency programs. Fam Med. 1993;25:253–6.
15.Diebold SE, David CT, Ferraro CM. Factors influencing applicants’ match rank order. Acad Emerg Med. 1995;2:328–30.
16.Medical Student Graduation Questionnaire All Schools Report: 2003. Washington, DC: Association of American Medical Colleges, 2003.
17.National Residency Matching Program Data Tables: 2003 〈http://www.nrmp.org
〉. Accessed 15 May 2003. Washington, DC: Association of American Medical Colleges, 2003.