Full Terms & Conditions of access and use can be found athttps://www.tandfonline.com/action/journalInformation?journalCode=htlm20

Teaching and Learning in MedicineAn International Journal

ISSN: 1040-1334 (Print) 1532-8015 (Online) Journal homepage: https://www.tandfonline.com/loi/htlm20

Questioning Style and Pimping in ClinicalEducation: A Quantitative Score Derived from aSurvey of Internal Medicine Teaching Faculty

John W. McEvoy, John H. Shatzer, Sanjay V. Desai & Scott M. Wright

To cite this article: John W. McEvoy, John H. Shatzer, Sanjay V. Desai & Scott M. Wright (2019)Questioning Style and Pimping in Clinical Education: A Quantitative Score Derived from a Surveyof Internal Medicine Teaching Faculty, Teaching and Learning in Medicine, 31:1, 53-64, DOI:10.1080/10401334.2018.1481752

To link to this article: https://doi.org/10.1080/10401334.2018.1481752

Published with license by Taylor & Francis©John W. McEvoy, John H. Shatzer, Sanjay V.Desai and Scott M. Wright

Published online: 01 Oct 2018.

Submit your article to this journal Article views: 2176

View related articles View Crossmark data

Citing articles: 4 View citing articles

VALIDATION

Questioning Style and Pimping in Clinical Education: A Quantitative ScoreDerived from a Survey of Internal Medicine Teaching Faculty

John W. McEvoya , John H. Shatzerb, Sanjay V. Desaic and Scott M. Wrightc

aDepartment of Medicine, Division of Cardiology, Johns Hopkins University School of Medicine, Baltimore, Maryland, USA; bSchool ofEducation, Johns Hopkins University, Baltimore, Maryland, USA; cDepartment of Medicine, Johns Hopkins University School ofMedicine, Baltimore, Maryland, USA

ABSTRACTConstruct: Pimping is a controversial pedagogical technique in medicine, and there is a ten-sion between pimping being considered as “value adding” in some circumstances versusalways unacceptable. Consequently, faculty differ in their attitudes toward pimping, andsuch differences may be measurable and used to inform future research regarding theimpact of pimping on learner outcomes. Background: Despite renewed attention in medicaleducation on creating a supportive learning environment, there is a dearth of prior researchon pimping. We sought to characterize faculty who are more aggressive in their questioningstyle (i.e., those with a “pimper” phenotype) from those who are less threatening. Approach:This study was conducted between December 2015 and September 2016 at Johns HopkinsUniversity. We created a 13-item questionnaire assessing faculty perceptions on pimping asa pedagogical technique. We surveyed all medicine faculty (n¼ 150) who had attended oninpatient teaching services at two university-affiliated hospitals over the prior 2 years. Then,using responses to the faculty survey, we developed a numeric “pimping score” designed tocharacterize faculty into “pimper” (those with scores in the upper quartile of the range) and“nonpimper” phenotypes. Results: The response rate was 84%. Although almost half of thefaculty reported that being pimped helped them in their own learning (45%), fewerreported that pimping was effective in their own teaching practice (20%). The pimpingscore was normally distributed across a range of 13–42, with a mean of 24 and a 75th per-centile cutoff of 28 or greater. Younger faculty, male participants, specialists, and thosereporting lower quality of life had higher pimping score values, all p< .05. Faculty whoopenly endorsed favorable views about the educational value of pimping had sevenfoldhigher odds of being characterized as “pimpers” using our numeric pimping score(p� .001). Conclusions: The establishment of a quantitative pimping score may have rele-vance for training programs concerned about the learning environment in clinical settingsand may inform future research on the impact of pimping on learning outcomes.

KEYWORDSbedside teaching; pimping;faculty traits

Introduction

Pimping is a well-known term in the medical lexicon,described by Brancati as occurring when an “attendingposes a series of very difficult questions to an intern orstudent [emphasis added].”1(p89) It is also a contro-versial pedagogical technique in bedside medical edu-cation.2,3 On one hand, pimping can teach, motivate,and involve the learner in clinical rounds;4–7 on theother hand, it may be interpreted as learner mistreat-ment.8,9 With reference to the latter, Kost and Chensuggested that the definition of pimping should berestricted to “questioning with the intent to shame or

humiliate the learner to maintain the power hierarchyin medical education [emphasis added].”3(p21)

Heightened concerns for medical student and traineemistreatment have recently placed renewed scrutinyon the practice of pimping.10,11 Nonetheless, thereremains a spectrum of opinion toward pimping inclinical practice and in the literature, with some pro-moting the potential virtues of this practice and otherswho feel it is a shameful relic of the past that shouldbe banished from modern medical education. Thisspectrum is enabled by the lack of a universallyagreed-upon definition for what pimping actually is;for example, the term can mean different things to

CONTACT John W. McEvoy jmcevoy1@jhmi.edu Division of Cardiology, Johns Hopkins University School of Medicine, Blalock 524C, 600 N. WolfeStreet, Baltimore, MD 21287, USA. Twitter handle: @johnwmcevoy; Institution Twitter handle: @HopkinsMedicinePublished with license by Taylor & Francis � 2018 John W. McEvoy, John H. Shatzer, Sanjay V. Desai and Scott M. WrightThis is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License (http://creativecommons.org/licenses/by-nc-nd/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited, and is not altered, transformed,or built upon in any way.

TEACHING AND LEARNING IN MEDICINE2019, VOL. 31, NO. 1, 53–64https://doi.org/10.1080/10401334.2018.1481752

different people ranging from Socratic banter to amore malignant belittlement. Thus, there is a tensionbetween pimping being considered as “value adding”in some circumstances versus pimping as being alwaysunacceptable.

Contributing to the controversy and uncertaintyregarding the value of this technique is the paucity ofresearch addressing the role of pimping in medical edu-cation.2 Multiple gaps in our understanding exist as tothe prevalence of pimping in modern academic medicalcenters, faculty perceptions about pimping, demographicand attitudinal correlates with pimping, and the impactof pimping on student satisfaction and learning out-comes. The objective of the current study was to exam-ine faculty perceptions about pimping and to develop aninstrument designed to characterize clinical teaching fac-ulty as “pimpers” or “nonpimpers.” Such an instrumentcould in theory facilitate faculty development efforts,quality improvement directed toward the local learningculture, and further research on the impact of pimpingon learner outcomes. Second, we set out to correlatepimping behaviors with other professional factors, suchas faculty demographics and their well-being. Whileattempting to be unbiased as to the pros and cons ofthis pedagogical technique, we hypothesized that facultyreporting diminished professional well-being (e.g., burnout or poor quality of life) would be more likely to becharacterized as pimpers. This hypothesis was groundedin prior literature suggesting that pimping may occur orbe perceived more frequently in negative learningenvironments.5

Method

Study design, subjects, and setting

For this cross-sectional study, we surveyedDepartment of Medicine faculty at one of two loca-tions: Johns Hopkins Hospital (a large, tertiary-referralacademic medical center) and Johns Hopkins BayviewMedical Center (a smaller academic hospital). Bothlocations provide comprehensive clinical education tomedical students and graduate medical educationtrainees (interns, residents, and fellows). Each has ageographically distinct internal medicine residencyprogram. The survey was sent electronically to all fac-ulty who had attended for 2 or more weeks duringthe past 2 years on any of the following three teachingservices at either hospital: general medicine wards,medical intensive care units, or cardiac intensivecare units.

Survey development

The survey instrument was designed with three objec-tives in mind: (a) to probe faculty perspectives onpimping as a pedagogical strategy, (b) to characterizefaculty respondents as pimpers or nonpimpers basedon their self-reported bedside teaching questioningstyle (using a pimping scale), and (c) to test relation-ships of these characterizations with other variablesof interest.

The survey instrument was developed using aniterative process over several months. First, theresearch questions, hypotheses, and candidate surveyquestions were presented to a group of Johns Hopkinsmedical education experts at a research-in-progressacademic conference in December 2015. The choice ofcandidate survey questions was informed by examin-ation of the literature. The literature review and feed-back obtained from content experts provided contentvalidity evidence to the pimping scale. This processresulted in a pilot survey of 40 questions. Of these, 25were core questions addressing the pimping construct,and 15 were supplementary demographic questionsand questions designed to establish relation to othervariables validity evidence.

Next, we conducted two rounds of survey piloting,analyzing responses, and soliciting feedback from 18purposively preselected faculty members. These facultywere selected to pilot the survey by the authorshipteam on the basis of their reputations and predilec-tions toward or against pimping behaviors. Questionswere modified and iteratively revised based on feed-back obtained during these two rounds of testing,whereas others were eliminated because of eitherskewed responses by all or an apparent inability todiscriminate between respondents. Ultimately, 11 corequestions tied to pimping questioning style remainedin the final pimping scale. To minimize any socialdesirability bias, these 11 core questions did notinclude any reference to the term pimping. Two fur-ther attitudinal questions were included at the end ofthe survey that directly asked faculty about their per-spectives on pimping (here we defined the termloosely as “repeatedly asking challenging questions toreveal medical knowledge deficiencies that may resultin embarrassment,” so as to strike a balance betweenthe benign and malignant interpretations of the prac-tice). As such, the survey included 13 questionsaddressing the pimping construct. Response optionsfor the survey questions used Likert scales. Thosequestions tied to frequency of behaviors had sixresponse options (never, rarely, sometimes, often, mostof the time, and always) and those tied to level of

54 J. W. McEVOY ET AL.

agreement had four response options (strongly dis-agree, disagree, agree, strongly agree; see AppendixTable A1, which provides details on the full survey).

Data collection

The survey was disseminated in July 2016, and responseswere collected and recorded through September 2016.The survey was hosted on the Qualtrics platform (http://www.qualtrics.com; Qualtrics, Provo, Utah) using asecure institutional account that allowed anonymouscompletion by faculty respondents. To encourage partici-pation, there were weekly drawings of Amazon gift cer-tificates for respondents. We targeted a response rate ofgreater than 80%. The Johns Hopkins University Schoolof Medicine Institutional Review Board approved thestudy protocol (date April 20, 2016; protocol numberIRB00098488).

Data analysis

Demographics and other baseline variables were com-pared among the faculty respondents using analysis ofvariance, Kruskal-Wallis, or chi-square testing, as appro-priate. Using similar statistical approaches, we also

compared survey question responses of general internalmedicine faculty versus specialty medicine faculty.

To characterize faculty respondents as pimpers ornonpimpers, we calculated a summative pimping scorefor each faculty respondent by adding up theirnumeric responses to the first 11 core questions in thepimping scale. Specifically, for the frequency ques-tions, a response of never was assigned a value of 1through always, which had a value of 6. Similarly, forthe level of agreement questions, the four responseswere assigned values of 1 through 4 (Appendix TableA1). Higher values of this summative score suggestmore of a predilection toward pimping behaviors; thetheoretical range of possible values was 11 to 50.Exploratory analysis prior to dissemination of the finalsurvey revealed that preselected faculty in our pilotwith known predilections toward pimping had highermean values on this summative score than thoseknown to have more supportive and less intimidatingeducational approaches (two-sided t test, p¼ .03).

Among faculty who completed the final survey, wetested the distribution and normality of this pimpingscore using Kernel Density plots and the Shapiro-Wilk test. Associations between faculty exposure vari-ables (e.g., demographics, teaching setting, and self-

Table 1. Baseline faculty demographics according to self-reported attitude toward pimpingNegative Attitude Toward Pimpinga Positive Attitude Toward Pimpingb p

Age, Years 46.3 ± 10.3 46.0 ± 8.7 .87Male, % 39 (56%) 37 (66%) .36Race, %White 49 (71%) 39 (70%) 1.00Black 3 (4%) 3 (5%)Asian 14 (21%) 11 (20%)Other 3 (4%) 3 (5%)

Academic Hospital TypeTertiary, Larger 24 (35%) 39 (70%) < .001�Smaller 45 (65%) 17 (30%)

Clinician TypeInvestigator 23 (41%) 23 (41%) 1.00Educator 34 (49%) 27 (48%)Other 7 (10%) 6 (11%)

Weeks on Service 8.6 ± 8.3 8.0 ± 6.8 .66ServiceGeneral Internal Med 34 (49%) 17 (30%) .03�Specialty Med 35 (51%) 36 (70%)

SettingGeneral Ward 57 (83%) 41 (73%) .27ICU 12 (17%) 15 (27%)

Poor Quality of Lifec 1 (1%) 4 (7%) .17Poor Work–Life Balancec 14 (20%) 9 (16%) .65Feelings of Callousnessc 11 (16%) 16 (29%) .09Feeling Burnoutc 26 (38%) 17 (30%) .45Depression Screen Positivec 6 (9%) 2 (4%) .29

Note: Items are mean (standard deviation), number (%), or median (interquartile range). Faculty who agreed orstrongly agreed with the statement “Being pimped by my teachers helped me learn when I was a medicaltrainee” were allocated to the positive attitude category. �p < .05. ICU¼ Intensive Care Unit.

aN¼ 69.bN¼ 56.cPoor quality of life was defined as a score of either 1 or 2 on Question 22 of the survey (see Appendix TableA1). Poor work–life balance was defined as a score of on Question 1 or of 2 on Question 23. Callousness wasdefined as a score of 3 or 4 on Item 28. Feeling burnout was described as a score of 4–7 on Item 27.

TEACHING AND LEARNING IN MEDICINE 55

reported metrics of well-being) and the pimping scorevalue were determined using unadjusted and adjustedmultivariable linear and logistic regressions (depend-ing on whether the pimping score was analyzed as acontinuous or a categorical outcome variable). In thecategorical analyses, we defined the binary (yes/no)outcome of pimper phenotype based on whether afaculty member was in the fourth quartile of thepimping score distribution (i.e., a pimping score �28)and then used logistic regression to identify predictorsof being designated a pimper.

Relation to other variables validity evidence wasassessed by correlating the binary pimper phenotypewith responses to the two direct attitudinal pimpingquestions included at the end of the survey. Toaddress response process validity evidence, facultyparticipating in the pilot testing commented on theirinterpretation of what the questions were asking andrated clarity and lack of ambiguity for each. Further,for analytic purposes, we included data only from fac-ulty who answered all survey questions. In addition,we reviewed the distribution of responses for eachsurvey item to ensure that faculty were not selectingthe same option for each item to complete the surveyquickly. All analyses were performed using STATA-13(Stata, College Station, TX) and two-sided p valuesless than .05 were considered statistically significant.

Results

Of the 150 faculty who received the survey, one facultymember disclosed that they were ineligible because theyhad not served as an inpatient teaching attending withinthe last 2 years. Of the 149 eligible faculty, 125 responseswere obtained (84% response rate). The mean age of thefaculty sample was 49 years; 61% were male, and 70%were White. The median number of weeks per year thatfaculty reported serving as inpatient teaching attendingwas 6, with an interquartile range from 3 to 12weeks.Approximately 40% of respondents reported they weregeneral internal medicine physicians, and 49% self-iden-tified as clinician educators. With respect to well-beingin the sample overall, 4% of faculty reported poor qual-ity of life, up to 18% reported being dissatisfied withwork–life balance, 34% reported feeling burned outmore than once a month, and 22% reported feelingmore callous toward people over their time practicingmedicine (see Appendix Table A1, which tabulatesdescriptive data for aggregate responses to the sur-vey questions).

Comparison of faculty based on positive ornegative attitude toward pimping

Table 1 compares demographic and well-being dataamong faculty who agreed with the statement “Beingpimped by my teachers helped me learn when I was amedical trainee” (n¼ 56) versus those who disagreed(n¼ 69). The faculty with positive attitudes toward hav-ing been pimped were more likely to work in the largertertiary-referral academic medical center (p< .001) andwere more likely to be specialists (p¼ .03). Faculty whoappreciated being pimped also appeared more likely toendorse a poor quality of life (7% vs. 1%) and feelingsof callousness (29% vs. 16%), although these differencesdid not reach statistical significance.

Responses to survey questions addressingthe core construct of pimping

Table 2 summarizes the responses to each of the coresurvey questions pertaining to pimping, comparingthe general medicine to the specialty faculty. Specialtyfaculty were more likely to provide affirmativeresponses in 11 of the 13 questions, with four of thesereaching statistical significance. In addition, specialistswere also statistically more likely than generalists toassert that pimping of learners is an effective teachingmethod during rounds (p¼ .01).

Distribution of the summative pimping score

The range of values for the summative pimping scorederived from faculty responses to the first 11 pimpingsurvey questions was 13 to 42 (potential range of11–50), with a mean and median of 24 and a normaldistribution (Figure 1; Shapiro-Wilk p value¼ .4). Thequartile cutoffs for this score were as follows: first quar-tile �20 (n¼ 32), second quartile 21–23 (n¼ 30), thirdquartile 24–27 (n¼ 33), and fourth quartile �28(n¼ 30). The mean summative pimping score value wassignificantly higher among specialists (25) than that seenin the generalist faculty members (22, p¼ .05; Table 2).

Pimping score values and independent predictorsof the ‘pimper’ phenotype

Faculty who were younger, were male, were workingin the larger tertiary-referral academic medical center,and practice as specialists were statistically more likelyto have higher mean values of the pimping score (seeAppendix Table A2, which demonstrates results fromunadjusted linear regression models and modelsadjusted for faculty age, gender, and race). Similar to

56 J. W. McEVOY ET AL.

the findings for differences in mean pimping scorevalue, crude and adjusted logistic models demonstratedthat pimpers (i.e., faculty with pimping scores in fourth

quartile) were more likely to be younger (approximately5% lower odds of being a pimper per year increase inage) and working at the larger tertiary-referral academicmedical center (Table 3). Specialists were also 3 to 4times more likely to be pimpers than nonspecialists (e.g.,odds ratio [OR]¼ 3.7, p¼ .01), and there was a signifi-cantly lower odds of being a pimper among facultyreporting higher quality of life. As expected, facultymembers who answered often, most of the time, oralways (N¼ 12) to Question 3 of our survey were 9times more likely to be identified as pimpers using ourscore, after adjustment for age, gender, and race(p¼ .003).

Correlation between pimper phenotype andreported pimping attitude (construct validity)

Faculty designated as pimpers based on their pimpingscore value were more likely to agree that “Pimping of

Table 2. Responses to questioning style survey questions according to general versus specialist med-ical practice

General Internal Medicinea Specialty Medicineb p

“When serving as the teaching attending on clinical rounds with the team, I …”; most of the time or always, n (%)Ask general medical knowledge ques-

tions specifically directed to indi-vidual trainees

26 (51%) 51 (69%) .04�

Ask the most junior trainee questionsfirst, even if the question is abovetheir expected level of knowledge

15 (29%) 24 (32%) .84

Ask the same trainee another similaror more challenging medical know-ledge question if they don’t knowthe answer to my first question

6 (12%) 6 (8%) .55

“In the context of my role as teaching attending, …”; agree or strongly agree, n (%)My style of questioning is stressful for

trainees on clinical rounds1 (2%) 8 (11%) .05�

I am “hard” on my clinical trainees 3 (6%) 8 (11%) .52Clinical questions on rounds should

reveal deficiencies in a train-ee’s knowledge

11 (22%) 26 (35%) .11

Stress can improve the learning ofmedical students and house-staff

11 (22%) 29 (39%) .04�

Teaching on clinical rounds shouldhave an unspoken hierarchy

13 (26%) 29 (39%) .12

It is OK to embarrass trainees if neces-sary to improve learning

3 (6%) 14 (19%) .04�

If I avoid embarrassing my trainees atall costs, they will learn lessfrom me

11 (22%) 25 (34%) .16

Trainees should never “speak theirminds” during clinical rounds

5 (10%) 5 (7%) .74

Pimping of students or residents is aneffective teaching strategyon roundsc

4 (8%) 19 (26%) .01�

Being pimped by my teachers helpedme learn when I was a med-ical traineec

17 (34%) 39 (53%) .03�

Pimping score,d mean value ± stan-dard error

22 ± 0.6 25 ± 0.5 .05

Note: Other items as defined in Table 2. �p< .05. CI¼ confidence interval.aN¼ 51.bN¼ 74.cThe last two attitudinal questions were not included in the summative pimping score but, rather, were used as variablesto test the construct validity of this score.

dThe pimping score was derived by adding the numeric responses to each to the first 11 questions on the survey.

Figure 1. Distribution and range of pimping score derivedfrom faculty survey responses.

TEACHING AND LEARNING IN MEDICINE 57

students or residents is an effective teaching strategyon clinical rounds” (OR¼ 4.9, p¼ .002), 95% CI [1.8,13.4], adjusted for age, gender, and race. Similarly,pimper faculty were more convinced that “Beingpimped by my teachers helped me learn when I was amedical trainee” (OR¼ 6.6, p< .001), 95% CI [2.4,17.7], adjusted for age, gender, and race. Our pimpingscore also accurately discriminated faculty who agreedwith these latter two direct questions, with results pro-vided in the appendix.

Discussion

This study characterizes the questioning practices ofattending physicians who lead teams of learners ininpatient settings at two academic teaching hospitals.Almost half of the faculty surveyed reported somepositive attitudes about the value of pimping. The sur-vey responses were used to derive a novel quantitativepimping score that is supported by validity evidenceand allows for discrimination of faculty who are pim-pers from those who are not. This score may enablefurther medical education research, including the abil-ity to link faculty pimping behavior to higher leveleducational outcomes. Further work from our group

is under way in this regard. In the current study, wewere able to describe demographic features and othercorrelates that were associated with the pimpingphenotype in our sample.

To our knowledge, this is one of the first publishedworks to describe clinical teaching faculty’s viewsabout pimping.12 Accordingly, these results informour understanding of faculty perceptions about pimp-ing of medical trainees. For example, although almosthalf the faculty appeared to recall being the subject ofpimping in positive terms, they were much less likelyto report that they themselves thought pimping waseffective in their own teaching practice. This phenom-enon has been seen in a similar survey of pharma-cists.12 One may hypothesize that many faculty havepurposively chosen not to be the source of pimpingthemselves, perhaps because of their preference to beviewed as a source of support rather than an instiga-tor of stress. Nonetheless, faculty with stronger self-reported predilections toward pimping behavior intheir teaching practice (based on the numericresponses to our pimping scale) were more likely torecall and believe that being pimped had helped themto learn when they were training. As such, it appearsthat, for pimping, prior learning experiences mayinform future teaching behaviors.

Table 3. Unadjusted and adjusted associations of faculty demographics with odds of beingcharacterized a “pimper”

Likelihood of Being a PimperOdds Ratio [95% CI]

Unadjusted Model pModel Adjusted for Age, Gender,

and Race p

Age, Years 0.95 [0.90, 0.99] .04� 0.95 [0.90, 1.00] .05�Male 1.39 [0.59, 3.30] .45 1.85 [0.85, 4.64] .09RaceWhite 1 [reference] — 1 [reference] —Black 1.81 [0.31, 10.67] .51 1.50 [0.24, 9.35] .66Asian 2.04 [0.78, 5.34] .14 1.92 [0.68, 5.39] .22

Academic Hospital TypeSmaller Hospital 1 [reference] — 1 [reference] —Larger Tertiary-Referral 2.0 [0.91, 4.76] .09 2.32 [1.01, 5.88] .04�

Clinician TypeInvestigator 1 [reference] — 1 [reference] —Educator 0.71 [0.30, 1.67] .44 0.71 [0.29, 1.76] .46Other 0.20 [0.02, 1.68] .14 0.24 [0.03, 2.16] .20

Weeks on Servicea 1.42 [0.87, 2.34] .16 1.47 [0.88, 2.48] .14ServiceGIM 1 [reference] — 1 [reference] —Specialty 2.83 [1.11, 7.24] .03� 3.74 [1.33, 10.49] .01�

SettingGeneral Ward 1 [reference] — 1 [reference] —-ICU 0.66 [0.23, 1.94] .45 0.69 [0.22, 2.14] .52

Quality of Lifeb 0.61 [0.37, 1.00] .05 0.55 [0.30, 0.98] .04�Work–Life Balanceb 0.89 [0.63, 1.27] .54 0.94 [0.64, 1.37] .74Callousnessb 1.24 [0.75, 2.06] .39 1.16 [0.67, 1.99] .59Burnoutb 1.91 [0.64, 5.70] .25 2.23 [0.68, 7.32] .18Depression Screen Positiveb 1.06 [0.20, 5.55] .94 1.26 [0.22, 7.22] .80

Note: Faculty with a pimping score value at 28 or higher were characterized as pimpers. �p< .05.CI¼ confidence interval; GIM¼General Internal Medicine; ICU¼ Intensive Care Unit.

aWeeks on service is log transformed.bQuality of life point estimate is per unit increase in Item 22 of the survey, work–life balance is per unitincrease in Item 23 of the survey. Other items as defined in Table 2.

58 J. W. McEVOY ET AL.

The association between younger faculty, male fac-ulty, and specialists with higher pimping score valuesis of interest and may reflect, at least in part, a desireby faculty with these characteristics to be respected andfeared by their juniors.7,13 Such a position is hard toprove definitively, and there may also be other morebenign reasons behind this finding. Furthermore, theassociation between faculty reporting lower quality oflife with predictions toward pimping behavior is inkeeping with the more negative perspectives on thispedagogical practice.3,14 In the quarter century sinceBrancati’s publication on the topic,1 the pimping oftrainees has been increasingly called into question byvirtue of its presumed link with student mistreat-ment.3,8,11 However, whether pimping truly representslearner mistreatment is unclear, and there remains acase in favor of the practice depending on how onedefines it. We believe that pimping is an interactionalphenomenon and that its virtues or drawbacks mayultimately come down to one’s perceived definition of“pimping,” how it is internalized by the learner,15 thelearning environment, and the intention of the attend-ing physician.2

The few small studies published on pimping todate have reported inconsistent findings and evaluatedlower level educational outcomes only, such asstudents’ reactions to pimping.14–17 Our pimpingscore may be useful in future medical educationresearch on this topic or for educational programleaders wanting to learn the style and approaches ofthose selected to teach learners (and how theseapproaches may or may not be grounded in best prac-tices for learning). First, the survey is short and canbe easily disseminated via e-mail. Second, the 11 corequestions included in the scale do not make referenceto the term pimping and as such minimize any biasin responses. If the two attitudinal questions makingdirect reference to pimping are included in futurescholarly efforts using our pimping scale, we recom-mend that these questions be included at the end ofthe survey so that they do not influence responses tothe 11 core questions. Third, the pimping scorederived from these data appeared to help identifycohorts of faculty (e.g., younger, males, specialists) inwhom pimping may be more likely to occur andcould identify target groups where interventions toreduce pimping behaviors may be more necessary. Inaddition, it is known that learners themselves havevaried opinions about pimping2,14,15; thus, our scorecould be used to match students with faculty whoexhibit either pimping or nonpimping phenotypes,based on students’ preferences regarding the practice.

Finally, we are conducting follow-up studies testingwhether our pimping score can be used to assess therelationship between faculty pimping behaviors andhigher level educational outcomes (such as knowledgeacquisition) and clinical outcomes (like accuratediagnoses versus medical errors and more thoughtfultesting with efficient high value care).18 Futurestudies are also necessary to determine the signifi-cance of differences in our score on meaningful edu-cational outcomes.

Several limitations of this study should be consid-ered. First, self-report of teaching style is subject tosocial desirability and other biases. The pimping scoreis a surrogate for teaching behavior, and it is notbased on direct observations. Therefore, we cannotalways be sure that faculty identified as pimpers usingour score are pimpers in reality. However, because thescore is based largely on frequency of performing spe-cific acts, this may be more accurate than attitudinalassessments.19 Second, faculty who responded to thesurvey may have been more interested or favorablypredisposed toward pimping as compared with non-respondents. However, our 84% response rate amongall eligible faculty is reassuring that the resultsshould be internally generalizable. With regards toexternal generalizability, one cannot assume that theperspectives in this cohort would match those of fac-ulty affiliated with other departments or schools, soadditional validation studies in other cohorts isnecessary. Third, there are limitations to this type ofsurvey research when dealing with a topic like pimp-ing, which often incites emotional responses andwhich tends to polarize opinion. Finally, we under-stand that program-level attitudes, cultures, andbehaviors20 may have an even greater impact on thelearning environment or trainee well-being than doindividual, faculty-level behaviors.

In conclusion, pimping behaviors continue atteaching hospitals in contemporary medical education.This study describes a pimping scale as a tool thatallows for the identification and phenotyping of clin-ical teachers who use this questioning style in theireducational approach. Although there may be manyapplications of the tool and possibilities for futureresearch, it may be most useful for faculty to be awareof and reflect upon how they relate to learners.

Ethical approval

This study was approved by the Johns HopkinsSchool of Medicine Institutional Review Board onApril 20, 2016 (reference number 00098488).

TEACHING AND LEARNING IN MEDICINE 59

Previous presentations

Oral abstract presentation in Baltimore, Maryland onJuly 2017 at the JHU Masters of Education in theHealth Professions (MEHP) Summer Conference:Advancing Careers through Interprofessional HealthProfessions Education-Teach, Research, Lead.

Acknowledgments

This paper represents the thesis work Dr. McEvoy under-took as part of his MEHP degree course. Dr. Wright is theAnne Gaines and G. Thomas Miller Professor of Medicineand is supported through the Johns Hopkins Center forInnovative Medicine.

Funding

Dr. McEvoy is supported by the following grants: NIH/NIDDK-R01DK089174, NIH/NIDDK-R01DK108784, andAHA-17MCPRP33400031.

ORCID

John W. McEvoy http://orcid.org/0000-0001-6530-5479

References

1. Brancati FL. The art of pimping. JAMA.1989;262(1):89–90.

2. McCarthy CP, McEvoy JW. Pimping in medical edu-cation: lacking evidence and under threat. JAMA.2015;314(22):2347–2348.

3. Kost A, Chen FM. Socrates was not a pimp: changingthe paradigm of questioning in medical education.Acad Med. 2015;90(1):20–24.

4. Antonoff MB, D’Cunha J. Retrieval practice as ameans of primary learning: Socrates had the rightidea. Semin Thorac Cardiovasc Surg. 2011;23(2):89–90.

5. McConnell MM, Eva KW. The role of emotion in thelearning and transfer of clinical skills and knowledge.Acad Med. 2012;87(10):1316–1322.

6. Hautz WE, Schroder T, Dannenberg KA, et al. Shamein medical education: a randomized study of theacquisition of intimate examination skills and its

effect on subsequent performance. Teach Learn Med.2017;29(2):196–206.

7. Healy JM, Yoo PS. In defense of “pimping.” J SurgEduc. 2015;72(1):176–177.

8. Reifler DR. The pedagogy of pimping: educational rigoror mistreatment? JAMA. 2015;314(22):2355–2356.

9. Lucey C, Levinson W, Ginsburg S. Medical studentmistreatment. JAMA. 2016;316(21):2263–2264.

10. Mavis B, Sousa A, Lipscomb W, Rappley MD.Learning about medical student mistreatment fromresponses to the medical school graduation question-naire. Acad Med. 2014;89(5):705–711.

11. Fnais N, Soobiah C, Chen MH, et al. Harassment anddiscrimination in medical training: a systematic reviewand meta-analysis. Acad Med. 2014;89(5):817–827.

12. Williams EA, Miesner AR, Beckett EA, Grady SE.“Pimping” in pharmacy education: a survey and com-parison of student and faculty views. J Pharm Pract.2018;31(3):353–360.

13. Angoff NR, Duncan L, Roxas N, Hansen H. Powerday: addressing the use and abuse of power in med-ical training. J Bioeth Inq. 2016;13(2):203–213.

14. Scott KM, Caldwell PH, Barnes EH, Barrett J.Teaching by humiliation and mistreatment of medicalstudents in clinical rotations: a pilot study. Med JAust. 2015;203(4):185–186.

15. Lo L, Regehr G. Medical students’ understanding ofdirected questioning by their clinical preceptors.Teach Learn Med. 2017;29(1):5–12.

16. Wear D, Kokinova M, Keck-McNulty C, Aultman J.Pimping: perspectives of 4th year medical students.Teach Learn Med. 2005;17(2):184–191.

17. Zou L, King A, Soman S, et al. Medical students’ pref-erences in radiology education a comparison betweenthe Socratic and didactic methods utilizing power-point features in radiology education. Acad Radiol.2011;18(2):253–256.

18. Parker K, Burrows G, Nash H, Rosenblum ND. Goingbeyond Kirkpatrick in evaluating a clinician scientistprogram: it’s not “if it works” but “how it works.”Acad Med. 2011;86(11):1389–1396.

19. Ajzen I, Fishbein M. The prediction of behavior fromattitudinal and normative variables. J Exp SocialPsychol. 1970;6(4):466–487.

20. Slavin SJ, Schindler DL, Chibnall JT. Medical studentmental health 3.0: improving student wellness throughcurricular changes. Acad Med. 2014;89(4):573–577.

60 J. W. McEVOY ET AL.

TableA1.

Aggregaterespon

sesto

thefullsurvey

Cor

e Pi

mpi

ng Q

uest

ions

(N=1

3*)

Supp

lem

enta

ry Q

uest

ions

(N=1

5)"W

hen

serv

ing

as th

e te

achi

ng a

tten

ding

on

clin

ical

rou

nds a

nd w

ith m

ost o

r al

l oft

he te

am p

rese

nt, I

……

"14

. Age

(mea

n an

d SD

)46

(9) y

ears

1. A

sk g

ener

al m

edic

al k

now

ledg

e qu

estio

ns

spec

ifica

lly d

irect

ed to

indi

vidu

al tr

aine

es

(ver

sus t

he e

ntire

gro

up)

Nev

er2

(1.6

%)

Rar

ely

14 (1

1.2%

)So

met

imes

32 (2

5.6%

)O

ften

29 (2

3.2%

)M

ost o

f the

tim

e 31

(24.

8%)

Alw

ays

17 (1

3.6%

)

15. G

ende

rM

ale

76 (6

1%)

Fem

ale

4

9 (3

9%)

2. A

sk th

e m

ost j

unio

r tra

inee

(e.g

., m

edic

al

stud

ent o

r int

ern)

que

stio

ns fi

rst,

even

if th

e ge

nera

l med

ical

kno

wle

dge

ques

tion

is a

bove

th

eir e

xpec

ted

leve

l of k

now

ledg

e

Nev

er11

(8.8

%)

Rar

ely

29 (2

3.2%

)So

met

imes

46 (3

6.8%

)O

ften

24 (1

9.2%

)M

ost o

f the

tim

e 10

(8%

)A

lway

s5

(4%

)

16. R

ace

Whi

te

88

(70%

)B

lack

6

(5

%)

Asi

an25

(2

0%)

Oth

er6

(5

%)

3. A

sk th

e sa

me

train

ee a

noth

er si

mila

r or m

ore

chal

leng

ing

med

ical

kno

wle

dge

ques

tion

if th

ey

does

n’t k

now

the

answ

er to

my

1st q

uest

ion

Nev

er47

(37.

6%)

Rar

ely

51 (4

0.8%

)So

met

imes

15 (1

2%)

Ofte

n8

(6.4

%)

Mos

t of t

he ti

me

3 (2

.4%

)A

lway

s1

(0.8

%)

17. W

hat i

s you

r D

ivis

iona

l aff

iliat

ion

in th

e D

epar

tmen

t of M

edic

ine?

Gen

eral

Inte

rnal

Med

icin

e

51

(

41%

)C

ardi

olog

y22

(

17%

)Pu

lmon

ary

23

(1

8%)

Ger

iatri

cs13

(11%

)O

ther

16

(1

3%)

“In

the

cont

ext o

f my

role

as t

each

ing

atte

ndin

g, I

belie

ve th

at…

.”18

. Wha

t is t

he a

vera

gew

eeks

per

yea

r th

at

you

are

teac

hing

att

endi

ng?

(Med

ian,

IQR

)6

(3-1

2) w

eeks

4. M

y st

yle

of q

uest

ioni

ng is

stre

ssfu

l for

tra

inee

s on

clin

ical

roun

ds

Stro

ngly

Dis

agre

e39

(31.

2%)

Dis

agre

e77

(61.

6%)

Agr

ee8

(6.4

%)

Stro

ngly

Agr

ee1

(0.8

%)

19. W

hich

hos

pita

l do

you

pred

omin

antly

se

rve

as te

achi

ng a

tten

ding

? JH

H†

6

3

(50

%)

JHB

MC

†

62

(

50%

)

5. I

am ‘h

ard’

on

my

clin

ical

trai

nees

Stro

ngly

Dis

agre

e47

(37.

6%)

Dis

agre

e67

(53.

6%)

Agr

ee10

(8%

)St

rong

ly A

gree

1 (0

.8%

)

20. I

n w

hat s

ettin

g do

you

pre

dom

inan

tly

serv

e as

the

teac

hing

att

endi

ng?

G

ener

al W

ard

98

(7

8%)

Inte

nsiv

e C

are

Uni

t

27

(2

2%)

6. C

linic

al q

uest

ions

on

roun

ds sh

ould

reve

al

defic

ienc

ies i

n a

train

ee’s

kno

wle

dge

Stro

ngly

Dis

agre

e26

(20.

8%)

Dis

agre

e62

(49.

6%)

Agr

ee28

(22.

4%)

Stro

ngly

Agr

ee9

(7.2

%)

21. I

see

mys

elf a

s pri

mar

ily a

: C

linic

ian

Educ

ator

61

(49

%)

Clin

icia

n In

vest

igat

or49

(39%

)O

ther

15

(1

2%)

(Continued)

TEACHING AND LEARNING IN MEDICINE 61

TableA1.

Continued.

7. S

tress

can

impr

ove

the

lear

ning

of m

edic

al

stud

ents

and

hou

se-s

taff

Stro

ngly

Dis

agre

e38

(30.

4%)

Dis

agre

e47

(37.

6%)

Agr

ee35

(28%

)St

rong

ly A

gree

5 (4

%)

22. W

hich

of t

he fo

llow

ing

best

des

crib

es y

our

over

all q

ualit

y of

life

?

As b

ad a

s it c

an b

e2

(1.6

%)

Som

ewha

t bad

3 (2

.4%

)So

mew

hat g

ood

52 (4

1.6%

)A

s goo

d as

it c

an b

e68

(54.

4%)

8. T

each

ing

on c

linic

al ro

unds

shou

ld h

ave

an

unsp

oken

hie

rarc

hy (e

.g.,

from

stud

ent,

to in

tern

, to

resi

dent

, to

fello

w, t

o at

tend

ing)

Stro

ngly

Dis

agre

e26

(20.

8%)

Dis

agre

e57

(45.

6%)

Agr

ee38

(30.

4%)

Stro

ngly

Agr

ee4

(3.2

%)

23. H

ow sa

tisfie

d ar

e yo

u w

ith th

e ba

lanc

e be

twee

n yo

ur p

erso

nal a

nd p

rofe

ssio

nal l

ife?

Ver

y di

ssat

isfie

d6

(4.8

%)

Som

ewha

t dis

satis

fied

17 (1

3.6%

)So

mew

hat s

atis

fied

61 (4

8.8%

)V

ery

satis

fied

41 (3

2.8%

)

9. It

is O

K to

em

barr

ass t

rain

ees i

f nec

essa

ry to

im

prov

e le

arni

ng (e

.g. c

orre

ctin

g w

rong

ans

wer

s to

impo

rtant

clin

ical

que

stio

ns)

Stro

ngly

Dis

agre

e82

(65.

2%)

Dis

agre

e26

(20.

8%)

Agr

ee16

(12.

8%)

Stro

ngly

Agr

ee1

(0.8

%)

24. I

’m b

ette

r th

an m

ost p

eopl

e at

mos

t th

ings

.

Stro

ngly

Dis

agre

e3

(2.4

%)

Dis

agre

e67

(53.

6%)

Agr

ee50

(40%

)St

rong

ly A

gree

5 (4

%)

10. I

f I a

void

em

barr

assi

ng m

y tra

inee

s at a

ll co

sts,

they

will

lear

n le

ss fr

om m

e

Stro

ngly

Dis

agre

e50

(40%

)D

isag

ree

39 (3

1.2%

)A

gree

30 (2

4%)

Stro

ngly

Agr

ee6

(4.8

%)

25. I

feel

une

asy

whe

n I a

m th

e fo

cus o

f at

tent

ion.

Stro

ngly

Dis

agre

e10

(8%

)D

isag

ree

61 (4

8.8%

)A

gree

48 (3

8.4%

)St

rong

ly A

gree

6 (4

.8%

)

11. T

rain

ees s

houl

d al

way

s ‘sp

eak

thei

r min

ds’

durin

g cl

inic

al ro

unds

(e.g

., sp

eaki

ng u

p if

they

di

sagr

ee)

Stro

ngly

Agr

ee‡

4

3 (3

4.4%

)A

gree

72 (5

7.6%

)D

isag

ree

6 (6

.4%

)St

rong

ly D

isag

ree

2 (1

.6%

)

26. H

ave

you

felt

depr

esse

d or

sad

muc

h of

th

e tim

e in

the

past

yea

r?Y

es8

(6.4

%)

No

117

(93.

6%)

*12.

Pim

ping

(e.g

. rep

eate

dly

aski

ng

chal

leng

ing

ques

tions

to re

veal

med

ical

kn

owle

dge

defic

ienc

ies t

hat m

ay re

sult

in

emba

rras

smen

t) of

stud

ents

or re

side

nts i

s an

effe

ctiv

e te

achi

ng st

rate

gy o

n cl

inic

al ro

unds

Stro

ngly

Dis

agre

e61

(48.

8%)

Dis

agre

e41

(32.

8%)

Agr

ee22

(17.

6%)

Stro

ngly

Agr

ee1

(0.8

%)

27. H

ow o

ften

do

you

feel

bur

ned

out f

rom

w

ork?

Nev

er12

(9.6

%)

A fe

w ti

mes

a y

ear o

r les

s54

(43.

2%)

Onc

e a

mon

th16

(12.

8%)

A fe

w ti

mes

a m

onth

26 (2

0.8%

)O

nce

a w

eek

11 (8

.8%

)A

few

tim

es a

wee

k4

(3.2

%)

Ever

y da

y2

(1.6

%)

*13.

Bei

ng p

impe

d by

my

teac

hers

hel

ped

me

lear

n wh

en I

was a

med

ical

trai

nee

Stro

ngly

Dis

agre

e29

(23.

2%)

Dis

agre

e40

(32%

)A

gree

44 (3

5.2%

)St

rong

ly A

gree

12 (9

.6%

)

28. I

hav

e be

com

e m

ore

callo

used

tow

ards

pe

ople

ove

r m

y tim

e w

orki

ng in

med

icin

e.

Stro

ngly

Dis

agre

e51

(40.

8%)

Dis

agre

e47

(37.

6%)

Agr

ee25

(20%

)St

rong

ly A

gree

2 (1

.6%

)

� The

last

2qu

estio

nswereno

tinclud

edin

thesummativepimping

scorebu

t,rather,w

ereused

asvariables

totest

theconstructvalidity

ofthisscore

†JHH¼TheJohn

sHop

kins

Hospital,JHBM

C¼John

sHop

kins

Bayview

MedicalCenter

‡Notethat

theorderof

preference

isreversed

forthisspecificqu

estio

n,so

asto

facilitatederivationof

thesummativepimping

score.

62 J. W. McEVOY ET AL.

Table A2. Unadjusted and adjusted associations of faculty demographics with mean differences in self-reported pimpingscore value

Difference in Pimping Score Value* Mean (95% CI)

Unadjusted Model p-value Model adjusted for age, gender, and race p-value

Age, years† −0.07 (-0.17, 0.03) 0.16 -0.11 (-0.21, -0.14) 0.04Male† 2.40 (0.53, 4.27) 0.01 2.78 (0.82, 4.73) 0.005Race-White 1 (reference) - 1 (reference) --Black −0.09 (-4.54, 4.35) 0.97 −0.36 (-4.74, 4.02) 0.97-Asian −0.62 (-3.01, 1.77) 0.61 −0.55 (-3.00, 1.91) 0.66-Other −1.26 (-5.71, 3.18) 0.58 −0.05 (-4.81, 4.71) 0.98

Academic Hospital Type-Smaller hospital 1 (reference) - 1 (reference) --Larger tertiary-referral 2.57 (0.76, 4.39) 0.006 2.02 (0.12, 3.92) 0.03

Clinician Type-Investigator 1 (reference) - 1 (reference) --Educator −0.56 (-2.55, 1.42) 0.58 −0.74, (-2.76, 1.28) 0.47-Other −1.76 (-5.01, 1.49) 0.29 −1.66 (-4.98, 1.65) 0.33

Weeks on service‡ 0.54 (-0.58, 1.67) 0.34 0.49 (-0.64, 1.61) 0.40Service-GIM 1 (reference) - 1 (reference) --Specialty 2.84 (0.00, 3.72) 0.05 1.99 (0.09, 3.88) 0.04

Setting-General Ward 1 (reference) - 1 (reference) --ICU 1.47 (-0.79, 3.73) 0.20 1.28 (-1.05, 3.62) 0.28

*Values are mean differences in pimping score between groups derived from linear regression beta-coefficients.†Difference in mean score for every additional year of age and between men and women faculty.‡Weeks on service is log transformed.

TEACHING AND LEARNING IN MEDICINE 63

Figure A1. (a) The ability to predict faculty who agree or strongly agree with the first attitudinal question about pimping was sig-nificantly improved by adding the pimping score to a base model that included age, gender, medical practice (GIM versus spe-cialty), and teaching location (c-statistic for base model of 0.69 vs. c-statistic of 0.83 for base model plus pimp score p=0.006)(eFigure 1a). (b) The ability to predict faculty who agree or strongly agree with the second attitudinal question about pimpingwas also significantly improved by adding the pimp score to the same base model (c-statistic for base model of 0.66 vs. c-statisticof 0.81 for base model plus pimp score p<0.001) (eFigure 1b).

64 J. W. McEVOY ET AL.