Specialty and Lifestyle Preference Changes during Medical School

Abstract

Purpose

Medical student specialty choices have significant downstream effects on the availability of physicians and, ultimately, the effectiveness of health systems. This study investigated how medical student specialty preferences change over time in relation to their demographics and lifestyle preferences.

Method

Students from ten medical schools were surveyed at matriculation (2012) and graduation (2016). The two surveys included questions about specialty and lifestyle preferences, demographics, educational background, and indebtedness. Student data from 2012 to 2016 were paired together and grouped into those whose specialty preferences remained constant or switched.

Results

Response rates in 2012 and 2016 were 65% (997/1530) and 50% (788/1575), respectively. Fourth-year students ranked “enjoying the type of work I am doing” as less important to a good physician lifestyle than did first-year students (from 59.6 to 39.7%). The lifestyle factors “having control of work schedule” and “having enough time off work” were ranked as more important to fourth-year students than first-year students (from 15.6 to 18.2% and 14.8 to 31.9%, respectively). The paired dataset included 19% of eligible students (237/1226). Demographic and lifestyle factors were not significantly associated with specialty preference switching. Additionally, no significant association existed between changing lifestyle preferences and switching specialty preference (p = 0.85).

Conclusions

During the course of medical school, lifestyle preferences became more focused on day-to-day factors and less on deeper motivational factors. Neither demographics nor lifestyle preferences appear to relate to a student’s decision to switch specialty preference during medical school. These findings represent an important step in uncovering causes of specialty preference trends.

Electronic supplementary material

The online version of this article (10.1007/s40670-019-00790-6) contains supplementary material, which is available to authorized users.

Introduction

Medical student specialty preferences have changed significantly over the past several decades [1, 2]. These changes have resulted in a redistribution of physicians, with students gravitating towards more specialized careers that have better perceived lifestyles [3–6]. Although the overall effects of these trends remain to be seen, emerging data have revealed several negative consequences [2, 7–9]. For example, reduced access to comprehensive primary care (PC) is associated with worsened patient health, higher costs, and reduced health equity [9]. Insufficient numbers of primary care physicians (PCPs) contribute to longer wait times, which leads some patients to forgo health services [10]. Diminished access to general surgeons is associated with measurably worsened patient outcomes such as increased death rates following traumatic events [11]. These trends may also create a negative feedback loop in which more subspecialization further narrows the scope of practice for generalists, which in turn makes PC less desirable and pushes even more graduates into subspecialization [8].

Although in theory, undergraduate medical education provides an opportunity to alter these trends, interest in high-need specialties such as PC remains constant or decreases during medical school [12–14]. The process of specialty selection is not only complex but varies between individual students [7, 15]. Thus, better understanding this process has been the topic of considerable research. A number of studies have investigated how specialty preferences change during medical school [12, 13, 16–21]. The Association of American Medical Colleges (AAMC)’s Annual Report on Residents also provides national data on specialty preference switching during medical school [22]. These sources detail which specialties students switch into and out of, but do not address the question of why students switch specialties. A 2014 study by our group began addressing this question by investigating the correlation between lifestyle factors and specialty preferences in a cross-sectional analysis of first- and fourth-year medical students at a representative sample of US medical schools [14]. Fourth-year medical students rated factors such as time off, schedule control, and financial compensation as more important than did first-year students. This study provided a snapshot of first- and fourth-year students but did not examine the association between specialty preference changes and lifestyle preferences.

The current study extends the 2014 study by following students from medical school matriculation to graduation, making it possible to track individual-level changes in both specialty and lifestyle preferences. We aimed to determine whether lifestyle and/or demographic factors relate to student decisions to change specialty preferences between the first and fourth year. We hypothesized that specific demographic factors, such as age and gender, would correlate with specialty switching. We furthermore hypothesized that lifestyle preferences would vary between students who switch specialty preference and those who do not. Finally, we hypothesized that students who change lifestyle preferences would be more likely to also change specialty preferences.

Methods

Participant Selection and Sampling Frame

This study is part of the larger InSPIRE (Investigating Specialty Preferences In tRainEes) project; results of the earlier phases and a full description of the survey methods have been published elsewhere [14, 23]. Eleven schools were originally included in the InSPIRE project to form a nationally representative medical student sample. Schools were selected based on research, social mission, and US News and World reports rankings as of 2012 [23]. The current study utilized data from 2012 along with new data collected in 2016. The 2012 survey was administered to students in their first year; results have been described elsewhere [23]. The follow-up survey was administered to fourth-year students at 10 of the original 11 schools in 2016, 4 years after the initial study. Students did not need to have completed the 2012 survey in order to be eligible to participate in the 2016 phase of the study. Approval or exemption for the study was obtained from each participating school’s Institutional Review Board.

Questionnaire Development and Content

Survey instruments were developed through an iterative process starting with a literature review and ending with pilot testing of each survey for clarity [24]. The final 36-question fourth-year survey contained four sections that gathered information about specialty preferences, perceptions of physician lifestyles, as well as demographic, educational, and debt burden information [14, 23]. The 2016 survey questions duplicated survey questions used in 2012, with some items modified to be relevant to fourth-year students [23]. For example, rather than asking, “If you had to choose today, what specialty would be your first choice?” fourth-year students were asked, “What residency was your first choice in the match?” Five additional questions were added to the fourth-year survey to determine how much research and volunteering students had done during medical school. This same survey was used in a previous study of fourth-year medical students [14]. The electronic survey was hosted on www.SurveyMonkey.com [24]. Both surveys are available as supplemental material.

Data Collection

The 2016 survey was administered beginning the week after the National Residency Matching Program’s 2016 Main Residency Match in March. One of the schools that had participated in the previous survey, the Uniformed Services University of the Health Sciences, was unable to participate in the 2016 survey. Site investigators at the ten remaining schools invited all medical students in their final year of medical school to participate (N = 1575). The survey remained open for 6 weeks, and two reminder emails were sent. To encourage participation, one gift card was raffled at each school.

Statistical Analyses

Data from the 2012 and 2016 questionnaires were paired using an anonymous personal identifier provided by the student. The identifier consisted of the first two letters of the participant’s mother’s first name, the first two letters of the father’s last name, and the two-digit month of the participant’s birth. In cases where identifiers did not match exactly, data were paired if all demographic and school data matched exactly, the two numbers matched, and at least one of the two letter combinations matched. In cases where the identifier and demographics matched exactly (i.e., student filled out the 2016 survey twice), one of the items was removed. Students who did not provide a valid response to the question about preferred specialty choice were not included in the paired sample.

Specialty preferences that were selected as “other” were recoded to their appropriate specialty when the specialty that was written matched one of the options provided. When the “other” category was a double or triple board certification, responses were coded as the first specialty listed. PC specialties were defined as family medicine, internal medicine, medicine-pediatrics, and pediatrics. Surgical subspecialties were defined as ENT, neurosurgery, orthopedics, and plastics. Specialties that were not PC, general surgery, or surgical subspecialties were grouped as “other.”

Participant data were grouped into respondents who completed both the first- and fourth-year surveys, and those who completed only the fourth-year survey. Descriptive data were analyzed to assess non-response bias. Data were grouped into students who switched their specialty choice between first and fourth year and students who did not. Data were also grouped by change in most important lifestyle factor preference between the first and fourth year. These groups were analyzed to determine the associations between specialty preference changes and demographics, lifestyle preferences, or lifestyle preference changes. Chi-square analyses were conducted as appropriate. Significance was determined to be p ≤ 0.05. For all analyses, the most up-to-date version of R was used (R Foundation for Statistical Computing, Vienna, Austria).

Results

Among the ten schools, 997 of 1530 eligible students (65.2%) completed the 2012 survey and 788 of 1575 eligible students (50.0%) completed the 2016 survey. Of the students surveyed in 2016, 1226 were also in the original 2012 cohort; of these, 237 (19.3%) were included in the paired sample. Since not all students answered every question, denominators varied across questions.

All Respondents (Paired and Unpaired)

Table 1 shows the demographic characteristics of all respondents for the 2012 and 2016 surveys. There were more non-white respondents in 2012 compared with 2016 (40.5 to 31.8%); differences in other demographic characteristics varied by 3% or less between the two surveys. Demographic characteristics for those in the paired sample compared with all participants in the 2016 survey are shown in Table 2. Significantly more females compared with males responded to both surveys (p = 0.02); differences between race, age, marital status, having a physician-parent, and pre-medical school debt were not significant.

Table 1.

Demographics of respondents to a national survey of medical students, 2012–2016

Demographic category	No. (%) of surveyed students, 2012a	No. (%) of surveyed students, 2016a
Female	476 (51.9%)	386 (54.8%)
Ethnic background non-White	371 (40.5%)	224 (31.8%)
No physician-parent	700 (76.3%)	525 (74.6%)
Age at matriculation < 27 years	733 (80.0%)	547 (78.4%)

^aNot all students answered all questions, so denominators vary across questions

Table 2.

Demographics of respondents to a national survey of medical students, paired data, 2012–2016

Demographic category	No. (%) of students who participated in 2016 surveya,b	No. (%) of students included in the paired dataseta	Significance
Female	240 (51.8%)	142 (61.5%)	p = 0.02
Ethnic background non-White	132 (29.2%)	68 (30.1%)	p = 0.88
Married/domestic partner	112 (24.3%)	67 (29.1%)	p = 0.36
No physician-parent	344 (74.0%)	176 (76.2%)	p = 0.76
Age at matriculation < 27 years	190 (41.3%)	115 (49.8%)	p = 0.09
No pre-medical school debt	282 (62.0%)	157 (69.2%)	p = 0.40

^aNot all students answered all questions, so denominators vary across questions

^bExcludes students who did not provide valid responses to the specialty choice question, so totals vary from those of Table 1

When comparing overall first- and fourth-year survey results, several lifestyle preferences changed. Fourth-year students ranked “enjoying the type of work I am doing” as less important to a good physician lifestyle than first-year students (from 59.6 to 39.7%); “having control of work schedule” was more important to fourth-year students than first-year students (from 15.6 to 18.2%). Students also ranked “having enough time off work” as more important in their fourth year (from 14.8 to 31.9%). Students ranked “enjoying the work environment” and “financial compensation” the same in their first and fourth years.

Paired Data

Table 3 displays the number of students switching into and out of specialties, grouped into PC, general surgery, surgical subspecialties, and other specialties. Roughly one-third of students (30.8%) reported having the same top specialty choice between their first and fourth year. The remaining 69.2% of students reported different top specialty choices. Of note, the number of students switching into PC and general surgery was very similar to the number switching out of those specialties (73 switching into PC and 76 switching out; 11 switching into general surgery and 12 switching out). Thus, although the total number of students interested in these specialties remained constant, the individual students who comprised these groups differed between 2012 and 2016.

Table 3.

Specialty preference changes reported in a national survey of medical students grouped by specialty type, 2012–2016

Specialty	No. (%) of first-year students who selected this field as their top specialty choice (2012)	No. (%) of fourth-year students who selected this field as their top specialty choice (2016)	No. of students who never switched	No. of students who switched into this field	No. of students who switched out of this field
Primary Care*a	114 (48.1%)	111 (46.8%)	38	73	76
General Surgery	13 (5.5%)	12 (5.1%)	1	11	12
Surgical subspecialties†b	44 (18.6%)	27 (11.4%)	11	16	33
Other specialties‡c	66 (27.8%)	87 (36.7%)	23	64	43
Total	237	237	73 (30.8%)	164 (69.2%)	164

^aFamily medicine, internal medicine, medicine-pediatrics, pediatrics

^bENT, neurosurgery, orthopedics, plastic surgery

^cAnesthesia, dermatology, EM, neurology, OB-GYN, ophthalmology, pathology, PM&R, psychiatry, radiology, radiation oncology, urology, other

Table 4 displays lifestyle preference and demographic differences between “specialty switchers” (students whose top specialty choice changed between their first and fourth years) and “specialty non-switchers.” There were no significant differences between specialty switchers and specialty non-switchers in race (p = 0.07), gender (p = 0.44), age (p = 0.12), marital status (p = 0.92), having a physician-parent (p = 0.59), and premedical school debt (p = 0.93). In addition, lifestyle preferences were compared between specialty switchers and specialty non-switchers. There were no significant differences between these two groups in which lifestyle factors they considered the most important (p = 0.21). Finally, students were grouped into those whose most important lifestyle factor changed from 2012 to 2016 and those whose most important lifestyle factor stayed the same. There was no significant difference between these two groups in the number of students who switched specialty preferences (p = 0.85).

Table 4.

Lifestyle preferences and demographic characteristics among medical students whose specialty preferences switched and did not switch for a national survey of medical students, 2012–2016

Lifestyle or demographic characteristic	Specialty switchera	Specialty non-switcher	Significance
Age under 27	55.0%	38.0%	p = 0.12
Gender, female	63.1%	57.8%	p = 0.44
Race White	65.8%	78.9%	p = 0.07
Marital status single	60.3%	60.6%	p = 0.92
No premedical school debt	70.5%	66.2%	p = 0.93
Parent who is a physician	23.8%	23.9%	p = 0.59
Most important lifestyle factor ranking			p = 0.21
Financial compensation	4 (2.5%)	0 (0%)	–
Having control of work schedule	36 (22.6%)	9 (12.9)	–
Having enough time off work	43 (27.0%)	26 (37.1%)	–
Enjoying the work environment	15 (9.4%)	7 (10.0%)	–
Enjoying the type of work I am doing	61 (38.4%)	28 (40.0%)	–
Most important lifestyle factor change from 2012–2016	–	–	p = 0.85

^aStudents whose first-choice specialty was different in 2012 and 2016

Discussion

This study is the first to investigate the associations between specialty preference changes during medical school and individual factors such as demographics and lifestyle preferences. It uncovered several new findings that further the discussion on specialty selection trends. When comparing overall survey results between the first and fourth year, we found that during the course of medical school, lifestyle preferences become more focused on day-to-day factors such as scheduling and vacation time and less on deeper motivational factors such as enjoyment in the type of work being done. This finding is consistent with our previous InSPIRE study [14]. Thus, as students gain exposure to the realities of day-to-day medical practice, it appears that concerns about daily lifestyles begin to replace the more idealistic vision of their career. A career is defined as being a calling when it provides both personal meaning and prosocial benefit [25]. Although the current study did not ask explicitly about calling, the question about “enjoyment in the type of work being done” aligns with the “personal meaning” portion of this construct. Thus, the current study’s findings could represent a possible shift away from a sense of calling during medical school. Previous studies have found that physicians who view medicine more as a calling are less likely to experience burnout and more likely to report high career satisfaction [26, 27]. Burnout has been shown to increase during medical school and is increasingly more prevalent among physicians than non-medical professionals; the current study’s results add to the discussion on why this could be occurring [28, 29].

Contrary to our hypothesis, we found that demographic factors, lifestyle preferences, and lifestyle preference changes did not relate to a student’s decision to switch specialty preference. Previous studies have shown that both demographics and lifestyle preferences influence a student’s specialty choice, but our findings indicate that they do not relate to a student’s decision to switch their specialty preference during medical school [5–7, 30, 31]. This suggests that other factors may be motivating students to change specialty preferences during medical school. A myriad of possibilities exists as to what these factors could be but may include mentorship, school-specific factors such as a PC-driven mission, grades, or board exam scores.

Although demographics and lifestyle factors may not influence specialty preference switching during medical school, it is possible that they exert their influence on specialty choices before medical school even begins. For example, pre-medical students are exposed to medical specialties through a variety of means, from shadowing experiences to television dramas. This exposure could create lifestyle preferences long before medical school begins and contribute to the observed trends towards specialties with better perceived lifestyles. It could also allow them to identify role models in specific specialties with demographics similar to their own. Thus, future efforts to increase interest in high-need specialties may be effective if geared towards pre-medical students whose preferences are still being formed. If interventions are targeted towards medical students, our study suggests that focusing on lifestyle and demographic factors may be less effective.

Another important finding this study revealed is that the overall rate of specialty preference switching of 69.2% is consistent with prior work, suggesting a relatively constant rate of switching across most studies over the past several decades [12, 13, 15, 18–23]. Although specialty selection trends have changed dramatically over time, the number of students with stable specialty preferences during medical school has not. Thus, whatever factors drive students to switch specialty preference appear to be stable over time. In addition, our results showed that interest in PC decreased slightly during medical school, which also aligns with previous work [12–14]. In spite of many schools’ efforts to increase the supply of PCPs, this does not appear to be happening at the schools in our sample.

This study highlights the need for future research into several areas. First, our findings revealed important new information about specialty selection trends but were not designed to investigate these trends on a specialty-specific level. Doing so would require a much larger dataset. The dataset used by the AAMC’s Annual Report on Residents provides a unique opportunity to build on our findings, as it includes over eleven thousand students with paired data from the beginning and end of medical school [22]. It could also allow for research into gender- and race-specific trends. Additionally, the finding that the number of students switching into PC and general surgery nearly equaled the number switching out highlights the importance of longitudinal data such as the one used by the AAMC, as cross-sectional data would have been unable to detect these significant shifts. Second, our findings point to the possibility of pre-medical school exposure influencing specialty and lifestyle preferences. More research is needed into pre-medical student perceptions of lifestyles and specialty preferences. Third, our study suggested that demographics and lifestyle factors do not relate to specialty preference switching, but more research is needed into what factors do relate to this decision. Future studies may benefit from qualitative methods to effectively address this question.

This study had several limitations. First, although our response rates for each survey were 65% and 50%, only 1 in 5 students was included in the paired dataset. However, between those in our paired sample and all those who responded to the 2016 survey, only gender significantly varied; all other demographics were similar. Second, our paired sample included only students who completed medical school in four consecutive years. Thus, our results are not generalizable to students who took time off or failed to advance for other reasons. Third, this study did not ascertain how confident students were in their specialty choice during their first year as we did not provide an “undecided” option. Thus, it is possible that the specialty switcher and non-switcher groups were heterogeneous, as some students may have been very confident in their specialty choice during their first year while others were not. Finally, we do not know how many students will ultimately go on to practice primary care among those we grouped into primary care specialties.

Despite these limitations, this study advances the body of literature around specialty selection. The process of specialty selection for medical students is indeed complex, and these results highlight the fact that student decisions to switch specialty preference cannot be explained by demographic or lifestyle factors alone. This information is useful in guiding medical educators interested in recruiting students into high-need fields like PC. In order to achieve better care for patients, it is important that the specialties medical schools produce match the needs of society. Our findings further the discussion on how medical educators can best achieve this aim.

Compliance with Ethical Standards

Conflict of Interest

The authors declare that they have no conflict of interest.

Ethical Approval

This study was approved or deemed exempt by the institutional review boards of all participating institutions.

Disclaimers

None.