The process of attrition in pre-medical studies: A large-scale analysis across 102 schools (2024)

Charlene Zhang, Formal analysis, Investigation, Methodology, Project administration, Validation, Visualization, Writing – original draft,^* Nathan R. Kuncel, Conceptualization, Funding acquisition, Investigation, Methodology, Supervision, Writing – review & editing, and Paul R. Sackett, Conceptualization, Funding acquisition, Methodology, Supervision, Writing – review & editing

Sample

Data from students who began their post-secondary education in academic years between 2006 and 2009 were provided by The College Board. Of the 917,459 individuals whose intended major choice information was available, 170,866 individuals had four years of complete coursework data available and attended a school using a standard semester system. Of those, 153,512 students did not indicate any intention in studying pre-medicine at the time of SAT, 1,912 indicated pre-medical studies as their secondary or tertiary major choice, or indicated pre-medical studies as their first choice major but were not certain of their choice, and 15,442 indicated pre-medicine as their first choice major and were very or fairly certain of their choice. This resulted in our primary sample of 15,442 students spanning 102 institutions.

Measures

Procedure

Students’ eligibility to apply to medical schools (persistence in studying medicine) was operationalized as whether they fulfill the standard coursework required by most medical programs. To determine medical school prerequisites, we tallied definitions used by prior research [8,9,25], admission requirements specified by the Association of American Medical Colleges (AAMC), and course requirements of 105 medical schools across the United States in 2013.

AAMC suggested general prerequisites to be one year of Biology, one year of Physics, and two years of Chemistry that includes Organic Chemistry courses [26]. Some medical programs also allowed one biochemistry course to substitute for the second organic chemistry course. Therefore, we operationalized fulfillment of medical school prerequisites to include one year (two semesters) of biology, physics, general chemistry, and either one year of organic chemistry or one semester of organic chemistry with one semester of biochemistry.

The number of courses in each subject was counted to indicate continued eligibility for medical school at a number of milestones during coursework progression. For a course to be included, a numerical grade of 0.7 or higher on a scale of 0 to 4 needed to be achieved, as it is the lowest passing grade.

Further, many students receive college credit for achieving satisfactory grades on AP exams. AP exam scores of 3 or above were accepted by most schools as course credits [27], and were counted toward prerequisite fulfillment.

Analyses

Descriptive statistics were calculated to examine differences between fulfillers of medical school course pre-requisites and non-fulfillers.

Subsequently, logistic regression analyses at several important coursework milestones were performed to examine the effects of demographics, academic preparedness, and college course performance on course fulfillment at each stage.

Rates of prerequisite fulfillment

Overall, among the 15,442 students who indicated initial pre-med interests, 2,555 (16.5%) completed the full set of medical school prerequisites, while a small proportion (7.7%) never completed any chemistry, biology, or physics courses. As shown in Fig 1, the process of dropping out is examined at four major milestones, namely:

Open in a separate window

Fig 1

Pattern of continuing fulfillment of pre-med coursework at various milestones.

Pre-Med Intention = students who indicated intentions of majoring in pre-medicine at the time of taking SAT, Milestone 1 = whether students took a first semester of general chemistry, biology, and physics, Milestone 2 = whether students who completed Milestone1 took a second semester of general chemistry, biology, and physics, Milestone 3 = whether students who completed Milestone2 took a first semester of organic chemistry, Milestone 4 (full fulfillment) = whether students who completed Milestone3 took a second semester of organic chemistry or a semester of biochemistry.

1. whether students passed one semester of general chemistry, biology, and physics (39% did so);

2. among students who fulfilled milestone 1, whether they passed a second semester of general chemistry, biology, and physics (64% did so);

3. among students who fulfilled milestone 2, whether they passed one semester of organic chemistry (77% did so);

4. among students who fulfilled milestone 3, whether they passed a second semester of organic chemistry or a semester of biochemistry, thereby satisfying the entire fulfillment requirement (85% did so).

Proportions of fulfillers at each milestone were also calculated for each of the 102 institutions individually. Across all institutions, an average of 34% of students fulfilled milestone 1 (SD = 19%), 23% fulfilled milestone 2 (SD = 16%), 19% fulfilled milestone 3 (SD = 16%), and 17% fulfilled milestone 4 (SD = 16%).

To allow comparisons, the number of students who fulfilled prerequisites for medical schools was also tallied for the group of 1,912 students who indicated some intention in studying pre-medicine but were not certain, and the group of 153,512 students who had no intention of studying pre-medicine. 267 (14.0%) fulfilled the full set of medical school prerequisite coursework in the former group and 2,633 (3.9%) fulfilled it in the latter group.

Comparison between fulfillers and non-fulfillers of medical school prerequisites

Demographic information of the 2,555 fulfillers and 12,887 non-fulfillers of medical school prerequisites can be found in Table 1. A larger proportion of males (21%) who reported pre-med intentions fulfilled the prerequisites than females (14%).

With regards to race/ethnicity, rates of fulfilling prerequisites among those who had intention of pursuing pre-medical studies were the highest for Asians (23%), followed by the Other Minority group (20%), then Whites (16%) and Hispanics (13%), and the lowest for Blacks (9%).

There were meaningful differences between fulfillers and non-fulfillers on a variety of variables, both prior to college and throughout college (see Table 2). Fulfillers scored higher than non-fulfillers on the SAT-Combined (d = .44), with the largest difference in the Math section (d = .51). Fulfillers also reported higher GPA, both in high school (d = .33), and in all four years of college (average d = .37). Higher socioeconomic status (d = .18) was reported by fulfillers than non-fulfillers. In cases where AP grades were available, fulfillers obtained higher AP scores than non-fulfillers, including AP Biology (d = .52), AP Chemistry (d = .38), and AP Physics B and C (average d = .50).

When college performance is broken down specifically by course, fulfillers were found to have taken a greater number of medical school prerequisite courses (average d = 1.31) as coursework definitionally distinguishes fulfillers from non-fulfillers. They were also found to have performed better in these courses (average d = .42).

Logistic regression models for predicting fulfillment progression

Logistic regression analyses were performed for the four fulfillment milestones. Predictors included gender, ethnicity, the composite SAT score, high school GPA, SES, and mean college pre-med course grades when applicable. Continuous variables—SAT, GPA, SES, and course grades—were standardized to aid the comparison between regression coefficients. Variable intercorrelations with each subset of the sample can be found in S1, S2, S3 and S4 Tables in the supplement. The progression of logistic regression models can be found in Table 3. The same models with individual prior course grades as predictors were also tested and yielded similar results.

Table 3 reports the odds ratios (OR) and their 95% confidence intervals (CI) for various milestones. To aid the interpretation of these results, we also computed predicted likelihoods based on the regression weights. To compute fulfillment likelihoods of various demographic groups, we plugged in average values of all other predictors. To compute the continuous variables’ effects on fulfillment likelihoods, we plugged in the reference groups (i.e., females for gender, White for race), average values of all other predictors, and the average and 1 standard deviation (SD) above average values of the predictor of interest.

Model 1 examined whether students took first-semester general chemistry, biology, and physics courses with demographic and pre-college predictors (milestone 1). The predicted likelihood of males fulfilling milestone 1 was 8.50% higher than that of females (odds ratio (OR) = 1.44), and the likelihood of Asian students completing the milestone was 11.97% higher than that of White students (OR = 1.65), net of all other predictors. Further, controlling for all other predictors, the predicted likelihood of fulfilling milestone 1 increased by 4.82% with every 1 SD increase in SAT score (OR = 1.23), 5.63% with every 1SD increase in high school GPA (OR = 1.28), and .98% with every 1SD increase in SES (OR = 1.04).

Model 2 examined whether students who completed a first semester of prerequisites proceeded to complete a second semester of general chemistry, biology, and physics (milestone 2), using not only demographic and pre-college predictors, but also the mean of first-semester general chemistry, biology, and physics grades. Among students who had fulfilled milestone 1, the predicted likelihood of males fulfilling milestone 2 was 4.48% higher than that of females (OR = 1.21), and the likelihood of Hispanic students was 8.34% higher than that of White females (OR = 1.44) when holding all other predictors constant. Further, 1SD increase in average first-semester course grades improved the likelihood of students fulfilling milestone 2 by 10.04% net of all other predictors (OR = 1.56).

Subsequently, model 3 used the same set of predictors with course grades computed as the average of grades across both semesters of general chemistry, biology, and physics to determine whether students who have fulfilled milestone 2 took any organic chemistry courses (milestone 3). Among students who had completed milestone 2, the likelihood of Hispanic students taking and passing an organic chemistry course was 8.58% lower than that of White females when holding all other predictors constant at average (OR = .64). The likelihood of fulfilling milestone 3 also increased by 4.51% with every 1SD increase in SAT score (OR = 1.33) and by 2.54% with every 1SD increase in high school GPA (OR = 1.17), net of all other variables.

Finally, Model 4 used the same previous set of predictors along with the first organic chemistry grade to predict whether students took a second organic chemistry course or a biochemistry course, thereby fulfilling all science coursework prerequisites for medical school, conditional on having completed all required courses thus far (milestone 4). There were no statistically significant difference in fulfillment likelihood between gender and racial majority and minority groups. Controlling for all other predictors, 1SD increase SAT scores decreased predicted likelihood of prerequisite completion by 2.24% (OR = .88), and 1SD increase in average coursework grades increased likelihood by 3.09% (OR = 1.21).

To explore the potential demographic variables’ interactions in predicting coursework fulfillment likelihoods, the same four models were also tested including the interaction terms between gender and race dummy variables. The only statistically significant and meaningful interaction was between being male and being Asian for fulfilling coursework milestone 2 (OR = .71, p < .001) and milestone 3 (OR = .69, p < .001). In other words, holding all other predictors constant at their average values, while the predicted likelihood of White males fulfilling milestone 2 exceeds that of White females by 8.67%, the difference in likelihood between Asian males and females is only 2.14%. Similarly, the difference between White males’ and females’ predicted likelihood for completing milestone 3 is 7.81%, while that between Asian males and females is 1.53%.

Additional logistic regression models with various interactions terms between demographic dummy variable (i.e., gender and race) and the continuous predictors (i.e., SAT, high school GPA, SES, and grades) were also tested individually. For predicting milestones 1, 2, and 3, there was a significant interaction between gender and SAT score such that the difference in fulfillment likelihood between males and females (in favor of males) was reduced as SAT score increased (OR = .88, .84, and .88, respectively). A similar effect was found in the White–Asian comparison for milestones 1 and 2, such that the higher fulfillment likelihood of Asian students than White students was reduced with higher SAT scores (OR = .83 and .84, respectively). In addition, for predicting fulfillment of milestone 3, the higher likelihood of Asian than White students was reduced as average college course grade increased (OR = .88). Lastly, the higher likelihood of Asian students fulfilling milestone 4 than White students was enhanced with higher high school GPA (OR = 1.31).

Predictors of pre-med persistence

Similar to previous findings on the association between gender and attrition [3,10] and consistent with the normative alternatives approach to explaining the persistence gap [9], being male was linked with a significantly higher likelihood of persisting in a pre-medical education at nearly all stages throughout college. In other words, women’s lower likelihood to persist in medicine may be construed as a higher likelihood to accept alternative career choices.

With regards to ethnic and racial identities, while previous investigations suggest that ethnic and racial minority students are less likely to persist [3,8], results of the current study were less consistent. For Asian students, the odds of fulfilling the first semester of coursework were more likely than those of White students, with persistence likelihood decreasing slightly throughout the later milestones but never significantly lower than those of White students. The odds of Hispanic students fulfilling the first year of coursework were higher than those of White students, but the odds of their completing organic chemistry were lower. Lastly, African American students did not differ from White students in their likelihood of persistence at any point after controlling for socio-economic status, SAT, and grades in high school and college.

Further, students who fulfilled all required coursework reported higher SES. Although SES predicted completion of the first semester of required coursework, it did not predict persistence at any of the later milestones. Thus, the advantage of coming from a family with higher SES found by [3] seems to wear off early on during college. However, its link with persistence in a medical education is likely to strengthen when students decide whether or not to attend medical school, as a $200,000 to $300,000 cost for medical school is no small expense [30].

Consistent with the reputation of a degree in pre-medical studies for being cognitively intensive and challenging, coursework fulfillers entered college with higher scores on all components of the SAT as well as higher high school GPA. Continuing with this advantage, college GPA’s of students who eventually fulfilled all required coursework were higher than the GPA’s of those who did not for all four years. However, the differences declined over time, with the GPA difference in the fourth year of college being half as large as the GPA difference in the first year of college. This may be an indirect reflection of the high levels of difficulty in pre-medical courses compared with other courses. In terms of college performance, coursework fulfillers both by definition completed a larger number of courses in relevant science subjects and obtained higher grades in them than non-fulfillers.

When examining the predictive validities of academic preparedness measured by variables prior to college (SAT and high school GPA) and college performance (course grades), an interesting pattern is observed. For predicting completion of the first semester of coursework in the absence of any college grades, higher academic preparedness was associated with a greater likelihood of completion. However, the predictive validities of such more distal, pre-college factors were overtaken by that of the more proximal college grades. The exception is whether students complete the first organic chemistry, for which academic preparedness rather than college grades was predictive. This may be due to the notoriously difficult organic chemistry being overwhelmingly identified as culprit in the leaky pipeline [3,11]. Students might be extra cautious in deciding whether they would be able to succeed in the course and resort to information about their academic effectiveness over a longer period of their lives from the past rather than grades from the recent one or two years to make the decision. In a recent study examining persistence in undergraduate STEM courses, it was found that grades in the first general chemistry course was related with subsequent persistence in STEM more strongly among underrepresented individuals than among well-represented students [20]. Our examination of such interactions between demographic group and other predictors (e.g., academic preparedness before college, grades in college) yielded mixed results.

Limitations and future directions

There are several limitations in the current study. First, because we lacked information about whether students actively pursued pre-medicine throughout college, we focused on completion of coursework required for medical programs. This operationalization is indirect and imperfect. It is possible that students of other science majors completed a similar set of coursework, cases that were considered noise in the current study. On the other hand, it is also possible that some students do not or partially complete prerequisite coursework during their undergraduate years, but later complete all required courses in a postbaccalaureate premedical program. A large number of such programs already existed during the time that data used in the current study was collected [31]. Thus, by focusing on students’ medical school eligibility in terms of their coursework at their undergraduate institution, the experiences of those that only become eligible later on were not captured.

Second, as we were not provided information about whether students who obtained satisfactory scores on AP exams actually used their AP credits toward their college degree, we assumed that any AP exam score of 3 or higher counted as a fulfillment course. There is the possibility that some students may forfeit their AP credits or take the equivalent college course.

Third, since the collection of data used in the present study, a new version of the MCAT that includes a larger social and behavioral science component has been implemented [32]. It is reasonable to assume that requirements by medical schools were also revised to contain psychology and sociology coursework. It will be valuable to examine the effects of these changes. Furthermore, major shifts have taken place with regards to the philosophy underlying the evaluation of medical school applicants. In the past decade, AAMC has endorsed the value of “holistic review,” which considers “applicants’ experiences, attributes, and academic metrics as well as the value an applicant would contribute to learning, practice, and teaching” [33]. As such, academic record and performance is considered alongside many other factors for admission decisions, such as “distance traveled” or cumulative life experiences, and other contextual information for the applicants’ accomplishment [34–36]. In response to the call for holistic evaluation, a number of medical schools have revised their admissions statements and requirements. For example, the Perelman School of Medicine at the University of Pennsylvania define competencies “not based on specific courses, but rather on the cumulative achievement of knowledge and skills needed to become a physician” [37]. The Boston University School of Medicine emphasize “experiential and personal qualities” in addition to academic rigor in their selection process [38]. There are even institutions like Stanford University School of Medicine that explicitly removed any specific prerequisite requirements, and only provide course recommendations instead [39]. Since the data used in the current study was collected prior to these changes, it will be important for future research to differently operationalize and examine pre-med intention and persistence.

Finally, as our study relied on archival data, we were limited in the extent to which we could explore underlying mechanisms that explain attrition in pre-medical studies. While the regression analyses revealed certain demographic and academic preparedness factors as predictors, no information about the specific reasons behind students’ dropping out was available. However, our study provides a valuable quantitative complement to the prior qualitative body of research that described such reasons.

S1 Table

Variable intercorrelations using sub-sample included in the logistic regression model for Milestone 1.

(DOCX)

S2 Table

Variable intercorrelations using sub-sample included in the logistic regression model for Milestone 2.

(DOCX)

S3 Table

Variable intercorrelations using sub-sample included in the logistic regression model for Milestone 3.

(DOCX)

S4 Table

Variable intercorrelations using sub-sample included in the logistic regression model for Milestone 4.

(DOCX)

1. Roese NJ, Summerville A. What We Regret Most… and Why. Personality and Social Psychology Bulletin. 2005;31(9):1273–85. 10.1177/0146167205274693 [PMC free article] [PubMed] [CrossRef] [Google Scholar]

3. Barr DA, Gonzalez ME, Wanat SF. The Leaky Pipeline: Factors Associated With Early Decline in Interest in Premedical Studies Among Underrepresented Minority Undergraduate Students. Academic Medicine. 2008. May;83(5):503 10.1097/ACM.0b013e31816bda16 [PubMed] [CrossRef] [Google Scholar]

4. Dyrbye LN, Thomas MR, Power DV, Durning S, Moutier C, Massie FS, et al. Burnout and Serious Thoughts of Dropping Out of Medical School: A Multi-Institutional Study: Academic Medicine. 2010. January;85(1):94–102. 10.1097/ACM.0b013e3181c46aad [PubMed] [CrossRef] [Google Scholar]

5. AAMC. Matriculant Student Questionnaire: 2019 All Schools Summary Report. Association of American Medical Colleges; 2019 p. 26.

7. Williams D. Craking the Code on the “Hidden Curriculum” in the Medical Education Pipeline and its Contribution to Attrition. Journal of Unschooling and Alternative Learning. 2016;10(19):18. [Google Scholar]

8. Alexander C, Chen E, Grumbach K. How Leaky Is the Health Career Pipeline? Minority Student Achievement in College Gateway Courses. Academic Medicine. 2009. June;84(6):797 10.1097/ACM.0b013e3181a3d948 [PubMed] [CrossRef] [Google Scholar]

9. Fiorentine R. Men, Women, and the Premed Persistence Gap: A Normative Alternatives Approach. American Journal of Sociology. 1987. March1;92(5):1118–39. [Google Scholar]

10. Fiorentine R, Cole S. Why fewer women become physicians: Explaining the premed persistence gap. Sociol Forum. 1992. September1;7(3):469–96. [Google Scholar]

11. Lovecchio K, Dundes L. Premed Survival: Understanding the Culling Process in Premedical Undergraduate Education. Academic Medicine. 2002. July;77(7):719 10.1097/00001888-200207000-00016 [PubMed] [CrossRef] [Google Scholar]

12. Grace MK. Parting ways: Sex-based differences in premedical attrition. Social Science & Medicine. 2019. June1;230:222–33. 10.1016/j.socscimed.2019.04.030 [PubMed] [CrossRef] [Google Scholar]

13. Chang MJ, Sharkness J, Newman CB, Hurtado S. What Matters in College for Retaining Aspiring Scientists and Engineers. In Denver, CO; 2010. p. 43. [Google Scholar]

14. Russell ML, Atwater MM. Traveling the road to success: A discourse on persistence throughout the science pipeline with African American students at a predominantly white institution. Journal of Research in Science Teaching. 2005;42(6):691–715. [Google Scholar]

15. Jacob JA. AAMC Report Examines How to Increase the Pipeline of Black Men Entering Medical School. JAMA. 2015. December1;314(21):2222 10.1001/jama.2015.12005 [PubMed] [CrossRef] [Google Scholar]

16. Cole D, Espinoza A. Examining the Academic Success of Latino Students in Science Technology Engineering and Mathematics (STEM) Majors. Journal of College Student Development. 2008;49(4):285–300. [Google Scholar]

18. Barr DA, Matsui J, Wanat SF, Gonzalez ME. Chemistry courses as the turning point for premedical students. Adv in Health Sci Educ. 2010. March1;15(1):45–54. 10.1007/s10459-009-9165-3 [PMC free article] [PubMed] [CrossRef] [Google Scholar]

19. Thurmond VB, Cregler LL. Why students drop out of the pipeline to health professions careers: a follow-up of gifted minority high school students. Academic Medicine: Journal of the Association of American Medical Colleges. 1999;74(4):448–451. 10.1097/00001888-199904000-00044 [PubMed] [CrossRef] [Google Scholar]

20. Harris RB, Mack MR, Bryant J, Theobald EJ, Freeman S. Reducing achievement gaps in undergraduate general chemistry could lift underrepresented students into a “hyperpersistent zone.”Science Advances. 2020. June;6(24):eaaz5687 10.1126/sciadv.aaz5687 [PMC free article] [PubMed] [CrossRef] [Google Scholar]

21. Gonzalez M, Barr DA, Wanat SF. Attrition From Premedical Studies Among Latinas: Case Studies. Hispanic Journal of Behavioral Sciences. 2010. November1;32(4):571–85. [Google Scholar]

22. Andriole DA, Jeffe DB, Hageman HL, Klingensmith ME, McAlister RP, Whelan AJ. Attrition During Graduate Medical Education: Medical School Perspective. Arch Surg. 2008. December15;143(12):1172 10.1001/archsurg.143.12.1172 [PubMed] [CrossRef] [Google Scholar]

24. Sackett PR, Kuncel NR, Arneson JJ, Cooper SR, Waters SD. Does socioeconomic status explain the relationship between admissions tests and post-secondary academic performance?Psychological bulletin. 2009;135(1):1 10.1037/a0013978 [PubMed] [CrossRef] [Google Scholar]

25. Liboff AR, Chopp M. Should the premed requirements in physics be changed?American Journal of Physics. 1979. April1;47(4):331–6. [Google Scholar]

28. Malgwi CA, Howe MA, Burnaby PA. Influences on Students’ Choice of College Major. Journal of Education for Business. 2005. May;80(5):275–82. [Google Scholar]

29. Arulampalam W, Naylor RA, Smith JP. Dropping out of medical school in the UK: explaining the changes over ten years. Medical Education. 2007;41(4):385–94. 10.1111/j.1365-2929.2007.02710.x [PubMed] [CrossRef] [Google Scholar]

32. Frazer NB, Twohig MP. The New Social and Behavioral MCAT Requirements: Inspired Innovation, Missed Opportunity, or Both?Peer Review. 2012;14(4):14–17. [Google Scholar]

34. Witzburg RA, Sondheimer HM. Holistic Review—Shaping the Medical Profession One Applicant at a Time. New England Journal of Medicine. 2013. April25;368(17):1565–7. 10.1056/NEJMp1300411 [PubMed] [CrossRef] [Google Scholar]

35. Bastedo MN, Bowman NA, Glasener KM, Kelly JL. What are We Talking About When We Talk About Holistic Review? Selective College Admissions and its Effects on Low-SES Students. The Journal of Higher Education. 2018. September3;89(5):782–805. [Google Scholar]

36. AAMC. Roadmap to excellence: Key concepts for evaluating the impact of medical school holistic admissions. Washington, DC: Association of American Medical Colleges; 2013. [Google Scholar]

30 Sep 2020

PONE-D-20-15864

The process of attrition in pre-medical studies: A large-scale analysis across 102 schools

PLOS ONE

Dear Dr. Zhang,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Your paper received comments from four reviewers. They all found the theme of the paper important. However, they shared several concerns with the paper. I am particularly concerned the missing issues in the paper as these issues if addressed could improved the quality of the paper's message.

==============================

Please submit your revised manuscript by Nov 14 2020 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at gro.solp@enosolp. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

• A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

• A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

• An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see:http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

We look forward to receiving your revised manuscript.

Kind regards,

Luisa N. Borrell, DDS, PhD

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. We note that you have indicated that data from this study are available upon request. PLOS only allows data to be available upon request if there are legal or ethical restrictions on sharing data publicly. For information on unacceptable data access restrictions, please see http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions.

In your revised cover letter, please address the following prompts:

a) If there are ethical or legal restrictions on sharing a de-identified data set, please explain them in detail (e.g., data contain potentially identifying or sensitive patient information) and who has imposed them (e.g., an ethics committee). Please also provide contact information for a data access committee, ethics committee, or other institutional body to which data requests may be sent.

b) If there are no restrictions, please upload the minimal anonymized data set necessary to replicate your study findings as either Supporting Information files or to a stable, public repository and provide us with the relevant URLs, DOIs, or accession numbers. Please see http://www.bmj.com/content/340/bmj.c181.long for guidelines on how to de-identify and prepare clinical data for publication. For a list of acceptable repositories, please see http://journals.plos.org/plosone/s/data-availability#loc-recommended-repositories.

We will update your Data Availability statement on your behalf to reflect the information you provide.

3. Please include captions for your Supporting Information files at the end of your manuscript, and update any in-text citations to match accordingly. Please see our Supporting Information guidelines for more information: http://journals.plos.org/plosone/s/supporting-information

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1:Yes

Reviewer #2:Yes

Reviewer #3:Yes

Reviewer #4:Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1:Yes

Reviewer #2:Yes

Reviewer #3:Yes

Reviewer #4:Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1:Yes

Reviewer #2:No

Reviewer #3:Yes

Reviewer #4:No

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1:Yes

Reviewer #2:Yes

Reviewer #3:Yes

Reviewer #4:Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1:This manuscript provides a valuable analysis of the academic trajectories of students who enter college with an interest in pursuing premedical studies, and subsequently to attend medical school. Using a large data set provided by the College Board, the authors were able to identify students who had expressed the interest in premedical studies based on students’ response to a question administered at the time students took the SAT exam. As the authors indicate, more than 15,000 students indicated the intent to pursue premedical studies once they entered college.

The metric the authors utilize to gauge students’ continued level of interest in being premed is the extent to which students complete the science courses that have traditionally been required by medical schools for entry. They analyze the successful completion of these courses in a stepwise manner, as described in the Methods section. Of the students included in the data set, 16.5% completed the full sequence of courses identified by the authors by the time they had completed their undergraduate education. The authors then did an analysis of the academic and demographic factors associated with full completion of the course sequence. They identified gender, race/ethnicity, and prior academic record as among the factors associated with completion of the full course sequence.

This manuscript provides valuable information, potentially useful to those who advise undergraduates in career options. The data set the authors use is unique in its size and the variables it contains. As such, the manuscript adds value to our knowledge of the factors associated with persistence in interest in premedical studies, which is an important factor in the eventual diversity of the physician workforce.

The manuscript reports data on students who entered college between 2006 and 2009. Most of these students would have completed their undergraduate education by 2013-2014. As such, the analysis has missed some fundamental changes that have taken place in the premedical experience. As a result, some of their assumptions and findings may need re-examination by the authors. As one example, the authors state in the Abstract, “Attrition rates are highest initially but drop as students take more relevant courses.” The issue is not the relevance of the courses. The issue is how advanced the courses are. The AAMC has done research indicating that inorganic chemistry courses have more relevance to the practice of medicine than do organic chemistry courses. Organic chemistry courses are more advanced than inorganic chemistry courses, but they are not more “relevant”.

A second thing the authors miss in their discussion is the fundamental shift that has been taking place in the medical school admission process, with the shift to holistic review. In April 2013, Witzburg and Sondheimer published an article in the New England Journal of Medicine titled, “Holistic Review — Shaping the Medical Profession One Applicant at a Time.” Under holistic review, as described by the AAMC, a student’s grades in the premedical science courses, while still relevant, are only one of a series of factors used in selecting candidates for medical school. Consistent with the holistic review model, a growing number of medical schools no longer have a specific list of prerequisite courses required for admission. For example, the Perelman School of Medicine at the University of Pennsylvania reports on its Admissions website, “Admissions competencies are not based on specific courses, but rather on the cumulative achievement of knowledge and skills needed to become a physician.” (https://www.med.upenn.edu/admissions/admissions.html)

Similarly, the Admissions webpage of the Stanford University School of Medicine states explicitly, “Stanford Medicine does not have specific course requirements, but recommends appropriate preparation for the study of medicine.”

(http://med.stanford.edu/md-admissions/how-to-apply/academic-requirements.html)

Those medical schools that have dropped their explicit course requirements are instead relying on the newly formatted MCAT to provide a metric of scientific and behavioral competencies. Accordingly, in order for this manuscript to sustain its relevance to the current premedical experience, the authors should include a full discussion of the changes that have taken place since their study cohort graduated from college.

There is another important factor the authors have left out of their analysis and discussion. While most students who enter college with an intent to pursue premedical studies will complete the sequence of science courses at their undergraduate institution, a substantial number will elect to take either no science courses or some science courses at their undergraduate institution, while completing the full list of traditional science courses as part of a postbaccalaureate premedical program. The AAMC currently identifies 267 programs nationally (https://apps.aamc.org/postbac/#/index) that offer students the opportunity to complete their premedical course sequence after they have graduated from their undergraduate institution. Most of these programs have been active for more than a decade, and thus were available to the students included in the authors’ data set. The authors are unable to identify which of the students elected to take only some courses as an undergraduate, while completing the course sequence in a past-bac program.

I should point out that, due to poor quality graphics of Figure 1 in the manuscript, I was unable to decipher the figure. The text and images were so faint that I could not read them. I would ask the authors to revise the figure with improved graphics.

As one final issue, I suggest the authors include in their discussion the results of a paper that was published this month that reported on the association between grades in undergraduate chemistry classes and persistence in STEM majors:

R. B. Harris et al. Reducing achievement gaps in undergraduate general chemistry could lift underrepresented students into a “hyperpersistent zone”. Science Advances. 10 Jun 2020:

Vol. 6, no. 24, eaaz5687 DOI: 10.1126/sciadv.aaz5687

Reviewer #2:0. This paper is important for empirically investigating the 'pre-med pipeline,' where many students enter into pre-med (an unspecified 'major' at many schools, including mine) -- and many drop out. Who are these students, and how does dropout unfold over time? That is the focus of the work, which should be of great interest to a wide range of stakeholders: e.g., universities, students and parents, STEM researchers, and medical educators and institutions. Authors have large-scale transcript data across 102 institutions; they have students' major intentions prior to college enrollment -- an impressive data set to answer these questions.

1. A passing grade (.7 GPA or higher) was used as a screen for eligibility to continue taking pre-med courses, but authors might clarify how this was used (being eligible is necessary but not sufficient for taking the next course). Related to this (but entirely optional), authors might indicate % failure (and didn't retake) of initial courses by demographics as one barrier to entry; and/or investigating course grades more generally might predict staying in pre-med.

2. p. 13 - One can easily appreciate how SAT, HSGPA, and CGPA would be higher for course-fulfillers, to the extent non-fulfilling is driven by obtaining low grades (or failures) in pre-med courses; this would suggest examining or at least suggesting course grades as an explanatory mechanism (see previous point). Regarding SES differences, could they be higher if college dropouts were included (those who couldn't afford to continue at all, let alone in pre-med)? Dropout might be a significant problem for pre-med and STEM majors, and for URM students, but the current sample does not include this group (maybe include this point somewhere in the discussion section, perhaps with some data on % dropout in the College Board student data).

3. Table 3 and related text - (a) interpretations of odds ratios need to be refined (they are with respect to the reference group of white females, not overall; they also need to take other predictors into account); (b) odds ratios can be usefully compared against base rate %s from Table 2 and the raw data (e.g., plug in values into the equation to get predicted probabilities, to compare against base rate%s); (c) the two previous points get at whether the pseudo-R2 values are practically useful; but in addition to this, I'd recommend a formal statistical comparison between adjacent models (e.g., compare AIC values).

4. A bit surprisingly, no institutional differences were provided -- even just adding something such as the distribution of %fulfill by institution would be interesting to readers (and this could be relatively anonymous, given 102 institutions).

Side notes:

5. Table 1 - proportion row/column information was useful, but the labeling was a little confusing

6. Table 2 - (a) the # of courses help define the fulfill/non-fulfill categories so the d-values may not be surprising (if you include them, also check out the distributions of N, as maybe there's positive skew and you consider a more robust d-value involving medians); (c) format the rows a little more descriptively (headers? better labels?)

Reviewer #3:This study examines a key outcome in the medical education track, and brings to light quantitative research findings that should provide a valuable contribution to the literature for policy-makers and admissions officers alike. The submission is clearly structured, and the analyses are appropriate to the research question, which complements existing work with a more quantitative and large scale data-based approach. I only have a handful of comments.

1) The limitations mentioned in the study are appropriate to mention, but I would have liked to also see further discussion on the relevance of data that is between 11 and 14 years old to the world of medical education today. How stable are the trends that are noted in comparison to the trends of today? Beyond noting the structural changes to MCAT, are there other factors that would lead to caution in drawing strong conclusions from this data, such as the final sentence of the conclusion?

2) When comparing fulfillers and non-fulfillers of medical school prerequisites, the paper highlights and contextualizes the larger proportion of male fulfillers, but is there a clarifying sentence that can also be added regarding meaningful differences with regards to race/ethnicity?

3) Examining subgroup differences in attrition is a key valuable component of the study, and to that end the paper would be improved by some examination or at the very least mention of interactive effects between race/ethnicity and gender and attrition. For example, the report Altering the Course: Black Males in Medicine highlights the percentage of black male medical school applicants as the lowest among any subgroup. Would conclusions such as ““Lastly, African American students did not differ from White students in their likelihood of persistence at any point.” (p 19) differ when differentiating between black men and black women?

Reviewer #4:This study investigates the completion rates of students completing the sequence of undergraduate natural science coursework prerequisite for entry to medical school in the U.S. using a large-scale, multi-institution dataset. It examines the associations of multiple sociodemographic, pre-college achievement, career aspiration, and college-level achievement measures in relation to students’ completion of four milestone events defined by natural science coursework from the first semester through advanced organic or biochemistry courses.

The study showed that, of those who reported intending to pursue a medical degree when they took the SAT in high school, only 39% completed the first in a sequence of courses that together provide the foundation needed to be ready for medical school. Completion rates were progressively higher for the (likely) increasingly select group of students achieving each of the three subsequent milestones of coursework. In the end, the overall completion rate of all four milestones was 16.5%. Observed proportions completing the full sequence were higher for males and Asian students, and lower for females and students from races/ethnicities underrepresented in medicine (i.e., American Indian, Black, and Hispanic).

High school GPA, SAT scores, SES, Gender (Male), and Race (Asian) were associated with greater likelihoods of completing the first set of milestone courses. Gender (Male), Race (Hispanic), and college grades in the first milestone courses were associated with greater likelihoods of completing the second milestone courses, which included the second semester courses of the subjects comprising the first milestone. SAT scores and high school GPA were associated with greater likelihoods of completing the third milestone courses, and race (Hispanic) with a lower likelihood. Finally, gender (Male) and intention to pursue medical school were associated with greater likelihoods of completing the fourth milestone courses, while SAT scores were associated with a lower likelihood of completing the fourth milestone courses.

Although this study has the potential to enhance our understanding of the points at which those interested in the medical profession drop out, the paper fails to make a strong case for it.

1. The authors should consider rewriting the introduction and discussion with a clearer focus and relevant citations, considering the following as suggestions.

• Integrating studies about the factors that contribute to a lack of diversity in STEM or medicine that occur during or before college with the present study, including those referenced in the current about “the leaky pipeline.” References 4, and 11-19 seem germane to the study’s focus.

• Addressing, alongside things like gender norms that might affect persistence, comparable treatment of the lower rates of completion for those from racial/ethnic backgrounds underrepresented in medicine who, more often have lower-quality middle and high school education. Addressing the role that high school preparation might play in students first-semester performance in natural science coursework and beyond would strengthen the paper.

• Removing or clarifying the relevance of literature on burnout/risk of attrition in medical school (references 3, and 5-9), and references 3, 10, and 20, which about attrition in the medical school in the U.K., which differs substantially from medical school in the U.S. The authors also should consider eliminating references about attrition in U.S. medical school (given that 95% of medical students graduate within five years) or make a stronger case for their relevance.

2. With a clearer focus, the authors might present their results differently to highlight important findings. For example, Table 1 shows that Black and Hispanic students completed all four milestones at lower rates than White or Asian students. It might be important, given the research questions, to know at which milestone(s) they did not progress, in addition to showing their prevalence at the two end points. Similarly, it was confusing on page 19 to read that African American students did not differ from White students in their likelihood of persistence at any point, even though only 9% of Black students completed all four milestones, compared to 16% of White students. The authors should address how the observed and predicted rates of completion lead to different interpretations.

3. Similar suggestions would improve the discussion of the paper’s findings.

4. Finally, the conclusions are unfounded and should be rewritten.

5. Other minor suggestions include:

• addressing how the increasingly select sample of students might affect the results of the logistic regression analyses. For example, SES, SAT scores, and high school GPA may be restricted as the sample reduces from more than 13,000 to about 2,500 students.

• Replacing outdated references with more recent research (e.g., 18) and confirming the appropriateness of journals cited (e.g., 9, which is missing the Journal Name “Journal of Unschooling and Alternative Learning”).

• Clarifying the relevance or eliminating the text describing challenges identifying premed majors. The study documented the process for identifying courses prerequisite for medical school (and that are well described on each medical school’s website).

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1:No

Reviewer #2:No

Reviewer #3:No

Reviewer #4:No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool,https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS atgro.solp@serugif. Please note that Supporting Information files do not need this step.

23 Oct 2020

Your paper received comments from four reviewers. They all found the theme of the paper important. However, they shared several concerns with the paper. I am particularly concerned the missing issues in the paper as these issues if addressed could improved the quality of the paper's message.

Dear Dr. Borrell,

Thank you for initiating reviews on our paper “The process of attrition in pre-medical studies: A large-scale analysis across 102 school” and giving us the opportunity to revise and resubmit. We appreciated the reviews and constructive comments provided by the reviewers and yourself.

It was encouraging to see that reviewers shared our view in the importance of this work. We had a unique opportunity to quantitatively examined the process through which undergraduate students fulfill prerequisite coursework for medical schools using a large multi-cohort, multi-institution sample. Given the popularity of the informal pre-med major and little prior work focusing on attrition in the early pre-medical study stage, we felt that this work is valuable in adding to the current higher education literature.

In our response to reviewer comments, we have made effort to address every reviewer comment point-by-point and to incorporate the feedback into our revised manuscript. Notably, we substantially restructured our Table 1 to describe the key fulfillment rates in a way that is clearer and also include rates at the intersection of gender and race and at various fulfillment milestones. We also tested additional models that included interaction terms between the various predictors. The results were not tabled given the large number of complex models, but we described key findings. Lastly, we added discussions of the recent shift to holistic evaluation of applicants in medical education and its implications for the ongoing relevance of our results.

A minor detail to note is that while “degree goal” was a predictor included in our original submission because it was a variable that existed in the archival data given to us, and has often been found to be predictive of academic performance in past research. However, upon reexamination, we deemed it irrelevant to the current work. We define our total sample as individuals who have intention of becoming physicians, therefore their “degree goal,” or the highest degree they hope to pursue, had very little variance. We excluded it in the revised manuscript, but none of the reported findings changed.

We believe that these edits, along with other minor revisions, have significantly improved our manuscript. Thank you again for your consideration.

Reviewer #1: This manuscript provides a valuable analysis of the academic trajectories of students who enter college with an interest in pursuing premedical studies, and subsequently to attend medical school. Using a large data set provided by the College Board, the authors were able to identify students who had expressed the interest in premedical studies based on students’ response to a question administered at the time students took the SAT exam. As the authors indicate, more than 15,000 students indicated the intent to pursue premedical studies once they entered college.

The metric the authors utilize to gauge students’ continued level of interest in being premed is the extent to which students complete the science courses that have traditionally been required by medical schools for entry. They analyze the successful completion of these courses in a stepwise manner, as described in the Methods section. Of the students included in the data set, 16.5% completed the full sequence of courses identified by the authors by the time they had completed their undergraduate education. The authors then did an analysis of the academic and demographic factors associated with full completion of the course sequence. They identified gender, race/ethnicity, and prior academic record as among the factors associated with completion of the full course sequence.

This manuscript provides valuable information, potentially useful to those who advise undergraduates in career options. The data set the authors use is unique in its size and the variables it contains. As such, the manuscript adds value to our knowledge of the factors associated with persistence in interest in premedical studies, which is an important factor in the eventual diversity of the physician workforce.

We appreciate the positive feedback on the value of this work. We agree with the reviewer that this dataset provides a unique opportunity to examine the academic trajectories of pre-medical students.

The manuscript reports data on students who entered college between 2006 and 2009. Most of these students would have completed their undergraduate education by 2013-2014. As such, the analysis has missed some fundamental changes that have taken place in the premedical experience. As a result, some of their assumptions and findings may need re-examination by the authors. As one example, the authors state in the Abstract, “Attrition rates are highest initially but drop as students take more relevant courses.” The issue is not the relevance of the courses. The issue is how advanced the courses are. The AAMC has done research indicating that inorganic chemistry courses have more relevance to the practice of medicine than do organic chemistry courses. Organic chemistry courses are more advanced than inorganic chemistry courses, but they are not more “relevant”.

We have changed “relevant” to “advanced” in the abstract as we agree that later, more advanced courses are not necessarily more relevant that earlier courses. We use the description “relevant” at several other places throughout the manuscript to refer to all courses used to operationalize medical school eligibility that are common prerequisites.

A second thing the authors miss in their discussion is the fundamental shift that has been taking place in the medical school admission process, with the shift to holistic review. In April 2013, Witzburg and Sondheimer published an article in the New England Journal of Medicine titled, “Holistic Review — Shaping the Medical Profession One Applicant at a Time.” Under holistic review, as described by the AAMC, a student’s grades in the premedical science courses, while still relevant, are only one of a series of factors used in selecting candidates for medical school. Consistent with the holistic review model, a growing number of medical schools no longer have a specific list of prerequisite courses required for admission. For example, the Perelman School of Medicine at the University of Pennsylvania reports on its Admissions website, “Admissions competencies are not based on specific courses, but rather on the cumulative achievement of knowledge and skills needed to become a physician.” (https://www.med.upenn.edu/admissions/admissions.html)

Similarly, the Admissions webpage of the Stanford University School of Medicine states explicitly, “Stanford Medicine does not have specific course requirements, but recommends appropriate preparation for the study of medicine.”

(http://med.stanford.edu/md-admissions/how-to-apply/academic-requirements.html)

Those medical schools that have dropped their explicit course requirements are instead relying on the newly formatted MCAT to provide a metric of scientific and behavioral competencies. Accordingly, in order for this manuscript to sustain its relevance to the current premedical experience, the authors should include a full discussion of the changes that have taken place since their study cohort graduated from college.

We thank the reviewer for pointing out the substantial shift in the evaluation process of medical school applicants. Given when our data was collected and how relatively uniform the prerequisite requirements were across medical schools during that time period, we were able to operationalize medical school eligibility using coursework. As the emphasis on prerequisites relaxes, this approach will become no longer reliable. Per the reviewer’s suggestion, we add some language in the limitations and future directions section to highlight this issue.

There is another important factor the authors have left out of their analysis and discussion. While most students who enter college with an intent to pursue premedical studies will complete the sequence of science courses at their undergraduate institution, a substantial number will elect to take either no science courses or some science courses at their undergraduate institution, while completing the full list of traditional science courses as part of a postbaccalaureate premedical program. The AAMC currently identifies 267 programs nationally (https://apps.aamc.org/postbac/#/index) that offer students the opportunity to complete their premedical course sequence after they have graduated from their undergraduate institution. Most of these programs have been active for more than a decade, and thus were available to the students included in the authors’ data set. The authors are unable to identify which of the students elected to take only some courses as an undergraduate, while completing the course sequence in a past-bac program.

It is certainly true that our approach of identifying students with intention in studying medicine and operationalizing their continued persistence is not perfect. As the first of the limitations we pointed out, by focusing on coursework, we inevitably include those that have the required coursework for most medical schools but nevertheless lost their intention to study medicine along the way. We thank the reviewer for pointing out another group considered as noise, namely those that do not complete the required coursework during their undergraduate years, but eventually do in a postbaccalaureate program. We added language to reflect this limitation.

I should point out that, due to poor quality graphics of Figure 1 in the manuscript, I was unable to decipher the figure. The text and images were so faint that I could not read them. I would ask the authors to revise the figure with improved graphics.

The figure has been regenerated to be higher-quality.

As one final issue, I suggest the authors include in their discussion the results of a paper that was published this month that reported on the association between grades in undergraduate chemistry classes and persistence in STEM majors:

R. B. Harris et al. Reducing achievement gaps in undergraduate general chemistry could lift underrepresented students into a “hyperpersistent zone”. Science Advances. 10 Jun 2020:

Vol. 6, no. 24, eaaz5687 DOI: 10.1126/sciadv.aaz5687

We have added reference of the study. We found the key results of interaction between race and grades for predicting subsequent persistence intriguing. Therefore, we also tested models additional to the ones reported in Table 3 that included various interactions between demographic dummies and continuous predictors. Because this yielded 32 additional models that contained largely redundant and statistically insignificant results, they are not tabled and presented. However, we added a paragraph at the end of the Results section describing the significant interactions.

Reviewer #2: 0. This paper is important for empirically investigating the 'pre-med pipeline,' where many students enter into pre-med (an unspecified 'major' at many schools, including mine) -- and many drop out. Who are these students, and how does dropout unfold over time? That is the focus of the work, which should be of great interest to a wide range of stakeholders: e.g., universities, students and parents, STEM researchers, and medical educators and institutions. Authors have large-scale transcript data across 102 institutions; they have students' major intentions prior to college enrollment -- an impressive data set to answer these questions.

1. A passing grade (.7 GPA or higher) was used as a screen for eligibility to continue taking pre-med courses, but authors might clarify how this was used (being eligible is necessary but not sufficient for taking the next course). Related to this (but entirely optional), authors might indicate % failure (and didn't retake) of initial courses by demographics as one barrier to entry; and/or investigating course grades more generally might predict staying in pre-med.

As we operationalized medical school eligibility as the fulfillment of various relevant courses, the passing grade was merely used as a minimum standard for which a course can be counted toward achievement. For example, the first milestone requires the completion of one semester of general chemistry, biology, and physics with a passing grade. We revised the language of the four milestones to make this clearer.

2. p. 13 - One can easily appreciate how SAT, HSGPA, and CGPA would be higher for course-fulfillers, to the extent non-fulfilling is driven by obtaining low grades (or failures) in pre-med courses; this would suggest examining or at least suggesting course grades as an explanatory mechanism (see previous point). Regarding SES differences, could they be higher if college dropouts were included (those who couldn't afford to continue at all, let alone in pre-med)? Dropout might be a significant problem for pre-med and STEM majors, and for URM students, but the current sample does not include this group (maybe include this point somewhere in the discussion section, perhaps with some data on % dropout in the College Board student data).

We agree with the reviewer about the merit of including course grades as a predictor. Our current logistic regression models include average prior relevant course grade as a predictor when possible. We also tested but did not report results of models that included individual course grades separately, as the models currently reported are more parsimonious and yielded similar results.

Unfortunately, the data we were given do not let us identify students that drop out. By examining coursework data, we are able to identify individuals that have coursework up to some point during their undergraduate years, but do not complete all coursework. However, we have no means of distinguishing between those that truly dropped out and those that transferred to a different institution.

3. Table 3 and related text - (a) interpretations of odds ratios need to be refined (they are with respect to the reference group of white females, not overall; they also need to take other predictors into account); (b) odds ratios can be usefully compared against base rate %s from Table 2 and the raw data (e.g., plug in values into the equation to get predicted probabilities, to compare against base rate%s); (c) the two previous points get at whether the pseudo-R2 values are practically useful; but in addition to this, I'd recommend a formal statistical comparison between adjacent models (e.g., compare AIC values).

We thank the reviewer for urging us to report regression findings with more accurate language. We have revised the results section to specify the reference group of the dummy variables as well as the interpretation of coefficients when controlling for all other predictors.

We also agree that odds ratios can be unintuitive and difficult to interpret, and describing the effects in terms of predicted probabilities can facilitate interpretation. Therefore, we have revised the results section by translating some of the key coefficients and odds ratios into differences in likelihood.

We have also added AIC values of the regression models that we tested in Table 3.

4. A bit surprisingly, no institutional differences were provided -- even just adding something such as the distribution of %fulfill by institution would be interesting to readers (and this could be relatively anonymous, given 102 institutions).

We agree with the reviewer that some attention should be paid to institutional differences. The sample sizes per institution varied substantially in the current data, ranging from 2 to 1,475. Therefore, proportions of students that fulfilled coursework milestones in institutions for which we only had sample sizes in the single digits could lead to misleading conclusions. However, we provide an added paragraph near the beginning of the results section with the mean and standard deviation of fulfillment proportion across institutions. We hope this sheds some light on the fulfillment distributions.

Side notes:

5. Table 1 - proportion row/column information was useful, but the labeling was a little confusing

We agree that the original Table 1 could benefit from clearer labels. In addition, we received suggestions from other reviewers to expand this table to also look at the fulfillment breakdown at the intersection of gender and race, as well as to look at fulfillment proportions at the intermediate milestones. In an effort to take into consideration all of these suggestions, we completely restructured our Table 1 to provide the sample size and percentage of individuals that fulfilled each of the four coursework milestones for the entire sample, by gender, by race, and by gender and race. By doing so, we focus the purpose of the table on displaying the proportion who remain med-school-eligible at each stage, and how the proportions compare across demographic groups. We think that this significantly improves the interpretability and readability of the table.

6. Table 2 - (a) the # of courses help define the fulfill/non-fulfill categories so the d-values may not be surprising (if you include them, also check out the distributions of N, as maybe there's positive skew and you consider a more robust d-value involving medians); (c) format the rows a little more descriptively (headers? better labels?)

The reviewer is correct in pointing out that the number of courses taken are quite different between fulfillers and non-fulfillers by definition and they are skewed. We have revised the table to show Cliff’s delta rather than Cohen’s d for those variables. Row names have been revised to be more descriptive and sub-headings have been added better describe the variables.

Reviewer #3: This study examines a key outcome in the medical education track, and brings to light quantitative research findings that should provide a valuable contribution to the literature for policy-makers and admissions officers alike. The submission is clearly structured, and the analyses are appropriate to the research question, which complements existing work with a more quantitative and large scale data-based approach. I only have a handful of comments.

1) The limitations mentioned in the study are appropriate to mention, but I would have liked to also see further discussion on the relevance of data that is between 11 and 14 years old to the world of medical education today. How stable are the trends that are noted in comparison to the trends of today? Beyond noting the structural changes to MCAT, are there other factors that would lead to caution in drawing strong conclusions from this data, such as the final sentence of the conclusion?

We thank the reviewer for the thoughtful comment. There certainly are additional factors that might influence the generalizability and relevance of the findings. A major one is the endorsem*nt of “holistic review” by medical applicant evaluation and the shift from primary emphasis on academic performance to a consideration of all aspects of applicants’ experiences and potential. We added some language in the limitations section explaining this shift in the past decade and its implications.

2) When comparing fulfillers and non-fulfillers of medical school prerequisites, the paper highlights and contextualizes the larger proportion of male fulfillers, but is there a clarifying sentence that can also be added regarding meaningful differences with regards to race/ethnicity?

We have revised the description of those results to reflect the differences.

3) Examining subgroup differences in attrition is a key valuable component of the study, and to that end the paper would be improved by some examination or at the very least mention of interactive effects between race/ethnicity and gender and attrition. For example, the report Altering the Course: Black Males in Medicine highlights the percentage of black male medical school applicants as the lowest among any subgroup. Would conclusions such as ““Lastly, African American students did not differ from White students in their likelihood of persistence at any point.” (p 19) differ when differentiating between black men and black women?

We think that the reviewer brings up an important issue regarding the interaction between gender and race for likelihood of persistence. We tried to address this issue in two different ways. First, we expanded and restructured Table 1 to describe not only fulfillment rates by gender and by race separately, but also at the intersection of race and gender. This allows us to compare both the number of individuals who declare some intention to pursue medicine as well as persistence patterns over time across demographic groups (e.g., between Black men and Black women). In addition to these descriptives, we also tested the same four models whose results are displayed in Table 3 when including the interaction terms between gender and race dummy variables. Because the inclusion of interaction terms would add an additional table that is similar to but much larger than Table 3 and more cumbersome to interpret, and most of the coefficients are statistically not significant, we did not table those results but describe the significant effects at the end of the Results section.

Reviewer #4: This study investigates the completion rates of students completing the sequence of undergraduate natural science coursework prerequisite for entry to medical school in the U.S. using a large-scale, multi-institution dataset. It examines the associations of multiple sociodemographic, pre-college achievement, career aspiration, and college-level achievement measures in relation to students’ completion of four milestone events defined by natural science coursework from the first semester through advanced organic or biochemistry courses.

The study showed that, of those who reported intending to pursue a medical degree when they took the SAT in high school, only 39% completed the first in a sequence of courses that together provide the foundation needed to be ready for medical school. Completion rates were progressively higher for the (likely) increasingly select group of students achieving each of the three subsequent milestones of coursework. In the end, the overall completion rate of all four milestones was 16.5%. Observed proportions completing the full sequence were higher for males and Asian students, and lower for females and students from races/ethnicities underrepresented in medicine (i.e., American Indian, Black, and Hispanic).

High school GPA, SAT scores, SES, Gender (Male), and Race (Asian) were associated with greater likelihoods of completing the first set of milestone courses. Gender (Male), Race (Hispanic), and college grades in the first milestone courses were associated with greater likelihoods of completing the second milestone courses, which included the second semester courses of the subjects comprising the first milestone. SAT scores and high school GPA were associated with greater likelihoods of completing the third milestone courses, and race (Hispanic) with a lower likelihood. Finally, gender (Male) and intention to pursue medical school were associated with greater likelihoods of completing the fourth milestone courses, while SAT scores were associated with a lower likelihood of completing the fourth milestone courses.

Although this study has the potential to enhance our understanding of the points at which those interested in the medical profession drop out, the paper fails to make a strong case for it.

1. The authors should consider rewriting the introduction and discussion with a clearer focus and relevant citations, considering the following as suggestions.

• Integrating studies about the factors that contribute to a lack of diversity in STEM or medicine that occur during or before college with the present study, including those referenced in the current about “the leaky pipeline.” References 4, and 11-19 seem germane to the study’s focus.

• Addressing, alongside things like gender norms that might affect persistence, comparable treatment of the lower rates of completion for those from racial/ethnic backgrounds underrepresented in medicine who, more often have lower-quality middle and high school education. Addressing the role that high school preparation might play in students first-semester performance in natural science coursework and beyond would strengthen the paper.

• Removing or clarifying the relevance of literature on burnout/risk of attrition in medical school (references 3, and 5-9), and references 3, 10, and 20, which about attrition in the medical school in the U.K., which differs substantially from medical school in the U.S. The authors also should consider eliminating references about attrition in U.S. medical school (given that 95% of medical students graduate within five years) or make a stronger case for their relevance.

We have revised the introduction and discussion following some of the reviewers’ suggestions, including removing references of studies of medical school in UK, removing review of the literature on attrition in medical school, and adding some language on systemic underrepresentation of ethnic minority groups in medicine.

2. With a clearer focus, the authors might present their results differently to highlight important findings. For example, Table 1 shows that Black and Hispanic students completed all four milestones at lower rates than White or Asian students. It might be important, given the research questions, to know at which milestone(s) they did not progress, in addition to showing their prevalence at the two end points. Similarly, it was confusing on page 19 to read that African American students did not differ from White students in their likelihood of persistence at any point, even though only 9% of Black students completed all four milestones, compared to 16% of White students. The authors should address how the observed and predicted rates of completion lead to different interpretations.

3. Similar suggestions would improve the discussion of the paper’s findings.

We have expanded and restructured Table 1 and present fulfillment rates by racial subgroups at various milestones, following the reviewer’s suggestions. We agree that by doing so, we gain a more complete understanding of the process of attrition.

We thank the reviewer for pointing out the confusion in presenting proportion of fulfillment by group and including those subgroup dummy variables in the regression models. For example, the African American group had the overall lowest proportion of overall eligibility based on their coursework. However, this difference seems to go away in the predictive models once the other predictors (e.g., SES, SAT scores, and grades) are controlled. We have revised the language in the Results and Discussion sections to clarify this.

4. Finally, the conclusions are unfounded and should be rewritten.

5. Other minor suggestions include:

• addressing how the increasingly select sample of students might affect the results of the logistic regression analyses. For example, SES, SAT scores, and high school GPA may be restricted as the sample reduces from more than 13,000 to about 2,500 students.

• Replacing outdated references with more recent research (e.g., 18) and confirming the appropriateness of journals cited (e.g., 9, which is missing the Journal Name “Journal of Unschooling and Alternative Learning”).

• Clarifying the relevance or eliminating the text describing challenges identifying premed majors. The study documented the process for identifying courses prerequisite for medical school (and that are well described on each medical school’s website).

References have been edited and the Conclusion section was rewritten.

24 Nov 2020

The process of attrition in pre-medical studies: A large-scale analysis across 102 schools

PONE-D-20-15864R1

Dear Dr. Zhang,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at gro.solp@gnillibrohtua.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact gro.solp@sserpeno.

Kind regards,

Luisa N. Borrell, DDS, PhD

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

You have addressed the reviewers' comments satisfactorily. However, I have some comments for thee tables. Please see attached.

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #2:All comments have been addressed

Reviewer #3:All comments have been addressed

Reviewer #4:All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #2:Yes

Reviewer #3:Yes

Reviewer #4:Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #2:Yes

Reviewer #3:Yes

Reviewer #4:Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #2:No

Reviewer #3:(No Response)

Reviewer #4:No

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #2:Yes

Reviewer #3:Yes

Reviewer #4:Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #2:my comments were very well addressed - thank you

if I have a quibble, it's to ensure that the mean of %fulfill across institutions is sample-size weighted, to take small-N into account

Reviewer #3:(No Response)

Reviewer #4:The revision was thorough and thoughtful. Three minor changes might improve the text:

1. On page 16, admissions criteria are described as "harsh". Stringent might be a more neutral term.

2. On page 17, last line of first partial paragraph, states "demographic characteristics such as having family members who were doctors and higher family income have been found to aid with persistence. The term "aid" implies an active role in persistence. Do you mean "are associated with"?

3. Reference 3 is a study about persistence in the U.K., and reference 5 is about burnout and thoughts of dropping out among U.S. medical students. Neither of these studies is about actual attrition in the U.S. On page 18, References 3 and 5 are used to support a statement about attrition rates in U.S. medical schools.

(Given the low acceptance rates into medical schools (7) and the attrition rates in medical schools (3,5), this early change in education track may actually prevent additional personal resources from being wasted in the process of applying to medical programs or institutional resources from being wasted when students drop out of medical programs.)

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #2:No

Reviewer #3:No

Reviewer #4:No

10 Dec 2020

PONE-D-20-15864R1

The process of attrition in pre-medical studies: A large-scale analysis across 102 schools

Dear Dr. Zhang:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact gro.solp@sserpeno.

If we can help with anything else, please email us at gro.solp@enosolp.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Luisa N. Borrell

Academic Editor

PLOS ONE