The Relationship Between Third-Year Clerkship Allocation Method and Specialty Choice

Original Research

Francis S. Nuthalapaty, MD; Tameika T.F. Lewis, MD; Anne Zinski, PhD; and James R. Jackson, PhD

From Greenville Health System/University of South Carolina School of Medicine Greenville, Green- ville, SC (F.S.N., T.T.F.L.), and Department of Medical Education, University of Alabama School of Medicine, Birmingham, Ala (A.Z., J.R.J.)

Abstract Background: A lack of research surrounds the impact of third-year clinical clerkship scheduling tactics on timing of specialty choice. The objective of this study was to compare the timing of clerkship experience and eventual specialty choice between 2 medical schools that use random and non-random third-year clerkship allocation methodologies. Methods: We conducted a retrospective cohort study using data for the graduating classes of 2003– 2006 at the University of Alabama School of Medicine (UASOM) and the Medical University of South Carolina (MUSC). The third year was divided into 6 sequential clerkship blocks at both institutions. At UASOM, clerkship allocation was non-random, with students making 3 “wishes” in which they specified the clerkship, the block, and an alternate. At MUSC, clerkship positions were allocated randomly. Clerkship block order was determined for all graduates who matched into 1 of 4 specialties (Internal Medicine, Pediatrics, Obstetrics and Gynecology, and General Surgery). For each specialty at each institution, clerkship block order was compared to a uniform distribution using a Chi-square goodness of fit test. Results: When the clerkship allocation method was non-random, the distribution of specialty choice across blocks differed significantlyP ( < .05) from a uniform distribution for each specialty, with the largest percentage matching in the fourth block for each specialty. When clerkships were assigned randomly, the distribution of specialty choice across blocks did not differ significantly from a uniform distribution for each specialty. Conclusions: Random allocation of clerkship order was associated with an even distribution of specialty interest throughout all clerkship blocks. Allowing students a choice in clerkship scheduling resulted in clustering of specialty interest into the middle 2 clerkship blocks, with the largest percentages matching in the fourth block.

he process by which medical students make graduation.4 This fact suggests a range of both a specialty choice is complex and multifac- commitment and exploration to a preferred spe- Ttorial. This decision is influenced by both cialty before beginning clerkship rotations. intrinsic (eg, socioeconomic background, age, motivation, personality) and extrinsic (eg, edu- Faculty and administrators who inform policies cational debt, length of residency, medical fac- regarding the scheduling of the third-year clerk- ulty, expected income, lifestyle) factors.1,2 Chief ships must consider an array of program and among the factors is students’ clerkship experi- student-level factors to maximize outcomes. For ences during the third year of medical school.3 instance, students with a firm specialty choice More than half of students change their specialty may prefer to time their specialty clerkship experi- choice from the time they complete their preclin- ence in a way that will maximize potential for res- ical coursework in the first 2 years to the time of idency match. Recent investigations have posited

38 GHS Proc. June 2017; 2 (1): 38-42 CLERKSHIP ALLOCATION METHOD AND SPECIALTY CHOICE that the sequence of specific clerkships are asso- tion within clerkship blocks between 2 medical ciated with exam scores in that discipline.5,6 It is schools that used random and non-random third- also established that third-year medical students’ year clinical clerkship allocation methods. scores on the National Board of Medical Exam- iners subject examinations tend to increase as Methods 7 they progress through the third-year curriculum. We designed a retrospective cohort study that Medical schools must also consider faculty grad- used data gathered from the Medical University ing tendencies and associated outcomes to neu- of South Carolina (MUSC) and the University 7 tralize potential effects of clerkship block order, of Alabama School of Medicine (UASOM), both as students may strategically delay clerkships in of which offer Liaison Committee for Medical their preferred specialty.3 In addition, as the letter Education-accredited 4-year medical degree of recommendation is cited as a factor in grant- programs. This investigation was reviewed and ing residency interviews by 86% of Residency granted exempt status by Greenville Health Sys- Program Directors,8 students who seek to garner tem’s Institutional Review Board and the Uni- a favorable recommendation letter through their versity of Alabama at Birmingham Institutional clerkship performance may choose to schedule Review Board. the associated clerkship in the second half of the year or when they are likely to be at their peak The study sample included all students graduat- 7 clinical performance. ing with an MD degree from the 2 institutions in However, for the balance of students who have yet 2003–2006, and who matched into 1 of the 4 pri- to solidify their specialty choice by the beginning mary specialties required in the third-year curric- of the third year, the priority may be to provide ulum (Internal Medicine, Pediatrics, Obstetrics adequate opportunities for breadth and quantity and Gynecology, and General Surgery). Students of clerkship experiences to encourage exploration who had a prolonged or interrupted third or of all career options and maximize performance, fourth year or who lacked a complete set of data regardless of timing. For example, Campos-Out- (eg, transfer students and dual-degree students) lat et al demonstrated that the implementation of were not included. a required third-year family practice clerkship Both institutions scheduled third-year clerkship in any sequence led to an immediate, significant increase in the proportion of students choosing rotations in 6 blocks through the academic year. family practice.9 Additionally, Paiva et al showed At UASOM, clerkship allocation was facilitated that, for previously undecided students, clerkships by the use of a locally developed computerized taken early in the clinical experience sequence scoring algorithm. Students were allowed to make were associated with eventual specialty choice, 3 “wishes” in which they specified the clerkship, and students who had identified other specialties the block, and an alternate block for each of their as a potential career path were likely to switch to 3 wishes. In addition, students were permitted to psychiatry following clerkship exposure.3 Fur- indicate which clerkship they did not want during ther, other investigations showed no association block 1. The scoring algorithm then tried to meet with clerkship timing, performance, and ultimate as many of a student’s wishes as possible, starting specialty choice, indicating that random clerkship with the top-ranked wish. sequencing may be equally appropriate.10-12 It was common practice for some medical student In the United States, 2 primary approaches to advisors at UASOM to recommend that students clerkship scheduling are commonly employed.13 who had identified their desired specialty pursue a The first allows students an opportunity to give clerkship in that specialty in the middle of the aca- input to their clerkship block schedule to deter- demic year. Conversely, at MUSC, students were mine their clerkship timing. The second approach not allowed to select the order of their clerkship eliminates student choice and assigns clerkship sequence; rather, the rotations were randomly sequence randomly. Some educators advocate for assigned using a locally developed computerized the former system as it may allow students the algorithm. Students were, however, able to switch opportunity to maximize their clerkship grade rotations among one another to a very limited and opportunities to obtain letters of recommen- extent for special circumstances. dation.14 Although both approaches are common, very little research has compared clerkship sched- The third-year clerkship schedules for all eligible uling tactics with regard to eventual specialty MUSC and UASOM graduates from 2003–2006, choice. Therefore, the objective of this study was along with their specialty matches, were obtained to compare the distribution of specialty selec- from institutional records.

GHS Proc. June 2017; 2 (1): 38-42 39 The percentage of students matching into the 4 significant, indicating that the percentages of stu- specialties was determined at each institution, and dents matching into the 4 specialties at each insti- the association between institution and specialty tution were similar (Table 1). Internal Medicine was examined with a Chi-square test of indepen- was the most common specialty choice, account- dence. Clerkship block order was determined for ing for at least one third of the 4 specialties in all graduates who matched into 1 of 4 specialties. each institutional cohort. Pediatrics accounted for For each specialty at each institution, clerkship more than one quarter of all included graduates. block order was compared to a uniform distribution using a Chi-square goodness of fit test. IBM The pattern of distribution of graduates across SPSS Statistics (version 22.0, (IBM, Armonk, NY) the 6 rotation blocks for each specialty was sim- was used for all analyses, and an alpha level of .05 ilar within each institutional cohort. At UASOM, was used as the criterion for statistical significance. where a non-random clerkship allocation method was used, the distribution of specialty choice Results across blocks differed significantlyP ( < .05) from a uniform distribution for each specialty (Table 2 During the study period, 277 students at UASOM and Fig. 1). A consistent clustering in the 2 mid- and 224 students at MUSC met the eligibility cri- dle blocks, with the highest percentages in block teria for inclusion. The association between insti- 4 (30.2%–41.9%) occurred and the second-highest tution and specialty choice was not statistically percentages in block 3 (20.0%–29.7%).

Table 1 By comparison, at MUSC, where a random clerk- Specialty match distribution for graduates of the University of Alabama ship allocation method was used, the distribution School of Medicine (UASOM) and the Medical University of South Carolina of specialty choice across blocks did not differ sig- (MUSC), 2003–2006. nificantly P( > .05) from a uniform distribution for each specialty (Table 2 and Fig. 2). UASOM MUSC (N = 277) (N = 224) P Value Discussion Specialty Match Distribution, no. (%) .220 These findings demonstrate that contrasting Internal Medicine 106 (38.2) 74 (33.0) approaches to clerkship allocation are associated with patterns of specialty interest within clerk- Pediatrics 74 (26.7) 65 (29.0) ship blocks. In the non-random allocation sys- General Surgery 50 (18.1) 33 (14.7) tem, higher percentages of students participated in clerkships in their eventual specialty match Obstetrics & Gynecology 47 (17.0) 52 (23.3) during the middle 2 blocks of the academic year,

Table 2 Number and percent of students who matched in a specialty by clerkship block for specialty and institution.

Clerkship Block for Specialty

1 2 3 4 5 6 Total P Value

Institution

UASOM, no. (%)

Internal Medicine 3 (2.8) 14 (13.2) 24 (22.6) 32 (30.2) 19 (17.9) 14 (13.2) 106 (100.0) <.001

Pediatrics 6 (8.1) 5 (6.8) 22 (29.7) 31 (41.9) 8 (10.8) 2 (2.7) 74 (100.0) <.001

Obstetrics & Gynecology 1 (2.1) 3 (6.4) 10 (21.3) 16 (34.0) 7 (14.9) 10 (21.3) 47 (100.0) .002

General Surgery 4 (8.0) 4 (8.0) 10 (20.0) 20 (40.0) 9 (18.0) 3 (6.0) 50 (100.0) <.001

MUSC, no. (%)

Internal Medicine 9 (12.2) 10 (13.5) 14 (18.9) 17 (23.0) 9 (12.2) 15 (20.3) 74 (100.0) .747

Pediatrics 8 (12.3) 10 (15.4) 15 (23.1) 15 (23.1) 7 (10.8) 10 (15.4) 65 (100.0) .366

Obstetrics & Gynecology 8 (15.4) 11 (21.2) 8 (15.4) 10 (19.2) 10 (19.2) 5 (9.6) 52 (100.0) .439

General Surgery 2 (6.1) 5 (15.2) 6 (18.2) 3 (9.1) 10 (30.3) 7 (21.2) 33 (100.0) .183

40 GHS Proc. June 2017; 2 (1): 38-42 CLERKSHIP ALLOCATION METHOD AND SPECIALTY CHOICE

Figure 1 Percent of students who matched in a specialty by clerkship block for specialty at UASOM.

45% Figure Legend: The Internal Medicine (n=106; p<.001*) distribution of specialty 40% Pediatrics (n=74; p<.001*) matches across 6 sequential Obstetrics & Gynecology (n=47; p<.002*) clerkship blocks, among 35% 277 student graduates of General Surgery (n=50; p<.001*) the UASOM Birmingham Campus, from 2003–2006. 30% This institution used a non- random clerkship allocation 25% method. A clustering appears in the middle 20% rotation blocks, regardless of specialty, with 60% of 15% graduates completing their clerkship in their eventual 10% chosen specialty during rotation blocks 3 and 4, as 5% compared to 14% in rotation blocks 1 and 2, and 26% in Percent of Students Who Matched in a Specialty rotation blocks 5 and 6. 0% 1 2 3 4 5 6 Clerkship Block for Specialty Figure 2 Percent of students who matched in a specialty by clerkship block for specialty at MUSC.

45% Figure Legend: The Internal Medicine (n=74; p=.439) distribution of specialty 40% Pediatrics (n=65; p=.366) matches across 6 Obstetrics & Gynecology (n=52; p=.747) sequential clerkship 35% blocks, among 224 student General Surgery (n=33; p=.183) graduates of the MUSC, 30% from 2003–2006. This institution used a random clerkship allocation 25% method. A uniform distribution appears across 20% the year, with 28% of students completing their 15% clerkship in their chosen specialty during rotation 10% blocks 1 and 2, compared with 39% in rotation 5% blocks 3 and 4, and 33% in

Percent of Students Who Matched in a Specialty rotation blocks 5 and 6. 0% 1 2 3 4 5 6 Clerkship Block for Specialty

whereas the cohort with random allocation dis- competition for patient care and procedural expe- played evenly distributed specialty interest across riences. Second, a lack of student enrollment in the academic year. specific clerkships early or later in the academic year could negatively impact faculty and resident While evidence suggests that non-random clerk- motivations, as well as yield very large or small ship order may benefit students who have a clear cohorts for which to provide clerkship oversight specialty choice, it may also lead to practical chal- and feedback. lenges. First, a concentrated group of students may select their desired clerkship specialty in the Importantly, although the non-random allocation middle of the academic year, building increased method appears to show lower percentages of spe-

GHS Proc. June 2017; 2 (1): 38-42 41 Abbreviations and cialty match interest among students in early and approach has the potential to negate faculty bias Acronyms late blocks, these students constitute a notable por- toward students, based on the timing of their UASOM = University tion of graduates. This finding seems to support clerkships, and could presumably increase the of Alabama School of the notion that the clerkship experience remains quality of interactions, and hence, the education Medicine; MUSC = an important determinant of specialty choice. of students overall. This approach might also Medical University of Therefore, it would be beneficial for faculty and eliminate the competitive pressure created when South Carolina residents to remain motivated in their teaching too many students in a clerkship block share the throughout the academic year, not only to enhance same specialty aspirations. Finally, for those stu- Correspondence learning, but also to stimulate student interest in dents who have yet to make a career choice, the Address to: their specialty. random allocation methods may decrease stress Francis S. Nuthalapaty, and anxiety regarding clerkship selection in their MD, Greenville Health The primary limitation of this study is in the poten- second year, before they have had a chance to per- System, Department tial generalizability of the findings to diverse cur- form any rotations at all. Further investigation of of OB/GYN, 890 W ricular environments. This study is representative these potential benefits is recommended. Faris Rd, Suite 470, of 2 accredited medical programs in the southern Greenville, SC 29605 United States with random and non-random Conclusion ([email protected]) clerkship scheduling during the third year. These This investigation showed that the patterns of do not speak to the variety of current approaches specialty choice across clerkship blocks varied to third-year clinical curricula, including longi- between institutions with random and non-ran- tudinal clerkships, community-based preceptor dom scheduling allocation. This finding is clerkships, and subspecialty electives, which may important in a little-studied area, as the asso- impact student specialty interest. ciation of clerkship timing and preference with eventual specialty match has not been explored Faculty who design policies regarding the sched- extensively, but may affect grading, faculty burden uling of third-year clerkships may want to give and motivation, and availability of patient cases in consideration to allocation method. The benefits each clerkship. As clerkship timing has been asso- of even distribution of student specialty inter- ciated with test achievement, specialty choice, and est across the academic year seem to be in the clinical performance, this investigation posits that best interest of both students and faculty. This clerkship allocation model must also be explored. References 1. Kiker BF, Zeh M. Relative income expectations, ington, DC. http://www.nrmp.org/wp-content/ expected malpractice premium costs, and other deter- uploads/2014/09/PD-Survey-Report-2014.pdf. Pub- minants of physician specialty choice. J Health Soc lished June 2014. Accessed March 15, 2017. Behav. 1998;39:152-67. 9. Campos-Outlat D, Senf J. A longitudinal, national 2. Maiorova T, Stevens F, Scherpbier A, van der Zee J. study of the effect of implementing a required third- The impact of clerkships on students’ specialty prefer- year family practice clerkship or a department of fam- ences: what do undergraduates learn for their profes- ily medicine on the selection of family medicine by sion? Med Educ. 2008;42:554-62. medical students. Acad Med. 1999;74:1016-20. 3. Paiva RE, Vu NV, Verhulst SJ. The effect of clinical 10. Potts MJ, Brazeau NK. The effect of first clinical clerk- experiences in medical school on specialty choice ship on medical students’ specialty choices. Med Educ. decisions. J Med Educ. 1982;57:666-74. 1989;23:413-5. 4. McLaughlin MA, Daugherty SR, Rose WH, Good- 11. Allen SS, Sherman MB, Bland CJ, Fiola JA. Effect of man LJ. Specialty choice during the clinical years: a early exposure to family medicine on students’ atti- prospective study. Acad Med. 1993;68:10. tudes toward the specialty. J Med Educ. 1987;62:911-7. 5. Kies SM, Roth V, Rowland M. Association of third- 12. Park RS, Chibnall JT, Morrow A. Relationship of rota- year medical students’ first clerkship with overall tion timing to pattern of clerkship performance in clerkship performance and examination scores. psychiatry. Acad Psychiatry. 2005;29:267-73. JAMA. 2010;304:1220-6. 13. Clerkship Selection. Organization of Student Rep- 6. Ouyang W, Cuddy MM, Swanson DB. US medical stu- resentatives (OSR) Listserv Questionnaire Results. dent performance on the NBME subject examination Association of American Medical Colleges, Washing- in Internal Medicine: do clerkship sequence and clerk- ton, DC. https://www.aamc.org/download/268830/ ship length matter? J Gen Intern Med. 2015;30:1307-12. data/clerkship_selection.pdf. Published December 7. Cho JE, Belmont JM, Cho CT. Correcting the bias 2011. Accessed March 15, 2017. of clerkship timing on academic performance. Arch 14. Baciewicz FA Jr, Arent L, Weaver M, Yeastings R, Pediatr Adolesc Med. 1998;152:1015-8. Thomford NR. Influence of clerkship structure and 8. Results of the 2014 NRMP Program Director Sur- timing on individual student performance. Am J Surg. vey. National Resident Matching Program, Wash- 1990;159:265-8.

42 GHS Proc. June 2017; 2 (1): 38-42