Gaps between college and starting an MD-PhD program are adding years to physician-scientist training time

Abstract

The average age when physician-scientists begin their career has been rising. Here, we focused on one contributor to this change: the increasingly common decision by candidates to postpone applying to MD-PhD programs until after college. This creates a time gap between college and medical school. Data were obtained from 3544 trainees in 73 programs, 72 program directors, and AAMC databases. From 2013 to 2020, the prevalence of gaps rose from 53% to 75%, with the time usually spent doing research. Gap prevalence for MD students also increased but not to the same extent and for different reasons. Differences by gender, underrepresented status, and program size were minimal. Most candidates who took a gap did so because they believed it would improve their chances of admission, but gaps were as common among those not accepted to MD-PhD programs as among those who were. Many program directors preferred candidates with gaps, believing without evidence that gaps reflects greater commitment. Although candidates with gaps were more likely to have a publication at the time of admission, gaps were not associated with a shorter time to degree nor have they been shown to improve outcomes. Together, these observations raise concerns that, by promoting gaps after college, current admissions practices have had unintended consequences without commensurate advantages.

Keywords: Aging; Complex traits.

Figure 1. Gaps between college and MD-PhD program matriculation.

(A) Gap prevalence by matriculation year. Data on matriculating medical students (MD and MD-PhD) were obtained from the AAMC MSQ (n = 12,779–16,668). MD-PhD data from current survey respondents (matriculation year 2013–2020, n = 306, 355, 391, 419, 424, 472, 451, and 519, respectively). (B) Shorter versus longer gaps by matriculation year. (C) Comparison of gap prevalence for current MD-PhD matriculants derived from present survey data with gap prevalence derived from AAMC data on all MD-PhD program matriculants from 2013–2020 (n = 605–707/year, 5223 total). (D) Comparison of AAMC gap prevalence data for MD-PhD matriculants from 2013 to 2020 (n = 605–707/ year, 5223 total) with gap prevalence data for those who were not accepted (n = 957–1064/year, 8007 total). Comparison of gap duration for MD-PhD matriculants compared with those who were not admitted. (E) Average gap prevalence for trainees in NIGMS MSTP training grant-supported programs (n = 49 programs, n = 3068 respondents, 70% participation rate) versus trainees from programs without MSTP grants (n = 25 programs, n = 474 respondents, 76% participation rate) in 2021. Mean + SD. Boxes indicate the 25th to 75th percentiles; lines within the boxes indicate medians, and whiskers indicate the 10th and 90th percentiles. Points outside of whiskers are shown. Differences are not statistically significant by ANOVA with Tukey’s HSD post hoc test. (F) Programs were grouped by those with fewer than 60 trainees (group 1, n = 33 programs, n = 803 respondents, 75% participation rate), programs with 60–99 trainees (group 2, n = 26 programs, n = 1482 respondents, 72% participation rate), and programs with 100 or more trainees (group 3, n = 14 programs, n = 1257 respondents, 67% participation rate). Mean + SD. Differences are not statistically significant by 1-way ANOVA. (G) Gap prevalence by gender. NB, nonbinary; NA, declined to answer. (H) Gap prevalence by race/ethnicity. NA, declined to answer. (G and H) Parenthetical number represents total number of respondents in each group.

Figure 2. Gap length distribution by gender and by race and ethnicity status.

(A and C) Total number of survey respondents in each group who had a gap of the indicated duration following college graduation. (B and D) Percentage of each group who took a gap of the indicated duration. Total number of survey respondents in each group is indicated in Figure 1, G and H.

Figure 3. Extent of research experiences during college and likelihood of having publications at the time of MD-PhD application by gap duration.

Survey respondents were divided into those with no gap (n = 1196), a 1- to 2-year gap (n = 1719), and a gap of ≥3 years (n = 629). (A–D) Comparison of women (n = 1574) and men (n = 1923). (E–H) Comparison of those from groups considered to be underrepresented in medicine (UIM) (n = 545) to those who are not (non-UIM) (n = 2868). (A and E) Average number of semesters of research during college. (B and F) Percentage of those surveyed who reported a summer research experience between freshman and sophomore years. (C and G) Average number of summers of research. (D and H) Percentage of those surveyed who reported having a publication at the time of submission of their MD-PhD program application.

Figure 4. Distribution of undergraduate majors among MD-PhD students and gap length as a function of category of undergraduate major.

(A) Percentage of students reporting an indicated undergraduate major from a dropdown list of majors. Number of respondents and percentage of total (n = 3544). (B) Respondent’s majors were grouped into three categories: social sciences (includes humanities, social sciences, and psychology), biological sciences, and physical sciences (includes chemistry, computer sciences, engineering, mathematics, and physics). The percentage of students in a given category who had no gap (blue), a 1- to 2-year gap (orange), or a gap of 3 or more years (gray) is shown. The total number of students in each category is shown in parentheses. Note that the number of students with majors in the social sciences is much smaller than in the other two categories.

Figure 5. MD-PhD program trainees’ primary reasons for taking a gap, primary activity during the gap, and primary source of advice.

The students who had taken a gap were asked for (A) their primary reasons for doing a gap, (B) what they did during it, and (C) from where their advice came. Their choices and the percentage of respondents who selected that choice are shown on the pie charts. See also Tables 1, 3, and 6 for all of the reasons that were listed.

Figure 6. Trainee views on the necessity and advisability of taking a gap.

For the students who took a gap, responses to whether they felt a gap was necessary to maximize their candidacy and whether they would recommend taking a gap to future applicants. Number of respondents and the percentage of the total respondents for each response is shown. (A) The percentage who responded that a gap was necessary broken out by UIM status, gender, and gap duration. (B) The percentage of respondents who would recommend a gap by UIM status, gender, and gap duration.

Figure 7. Program directors’ survey.

(A) Responses to the following question in the program director’s survey: “Are gaps a factor when deciding whom to interview and admit?” Blue bars show the percentage of directors who responded yes or no. The box-and-whisker plots to the right of each blue bar show the gap prevalence in the programs whose directors answered yes or no. Boxes indicate the 25th to 75th percentiles; lines within the boxes indicate medians, and whiskers indicate the 10th and 90th percentiles. Points above and below the whiskers are shown. Numbers indicate average gap prevalence for the programs (n = 70). (B) Directors who responded that gaps were a factor were asked to indicate the impact on decision making from a dropdown list of responses. The possible choices are shown, with the percentage of respondents choosing a given response (n = 51).

Figure 8. Program directors’ survey.

Are gaps a prognostic factor, and what is the preferred activity during the gap? (A) Directors’ opinions regarding the extent to which an applicant choosing to take a gap indicates a commitment to complete the program (blue) and to pursue a career as a physician-scientist (red). Percentage of total responses is shown. (B) Directors’ choices from dropdown lists of preferred activity (red) and activity viewed most favorably (blue) (n = 71).

Figure 9. Program directors’ survey.

The importance of an applicant having publications or abstracts. The directors’ survey asked whether secondary applications asked about publications. If they responded yes (n = 43 of 71), they were asked to rate the importance of first author papers (blue), coauthored papers (orange), and abstracts (gray). The percentage of responses is shown.

Figure 10. Relationship between gap duration and time to degree.

Forty-one programs provided deidentified data on 2391 program graduates who entered training after 2006 and graduated by 2021, 1103 with no gap, 581 with a 1-year gap, 401 with a 2-year gap, and 306 with a gap of 3 or more years. Boxes indicate the 25th to 75th percentiles, and whiskers indicate the 10th and 90th percentiles. Points above and below the whiskers are shown. The “+” in each box is the mean; this value is shown above each box. The time to degree was the same for those who took either no gap after college or a gap lasting 1 or 2 years. The average time to degree for those with a gap of 3 or more years was approximately 0.6 years (7 months) shorter (P < 0.001 by 1-way ANOVA).