Title of Assignment: Research Proposal (What factors affect the experiences of assessment and achievement at medical school from the perspective of UK-originating BME undergraduate medical students?) and Critical Commentary (Risk factors at medical school for subsequent professional misconduct: multicentre retrospective case-control study)

Critical Commentary

Risk factors at medical school for subsequent professional misconduct: multicentre retrospective case-control study¹

The question to be addressed is clearly outlined in the first line of the abstract, allowing the reader to establish quickly what the issue is: are there risk factors in a doctor’s time at medical school that are associated with subsequent professional misconduct?

At this point it is worth identifying the target audience, the kind of reader the text is addressing. It is clearly aimed at the scientific community, and with a leaning towards those engaged in quantitative empirical research. It is not an easy paper to understand and it required at least three readings in order to be able to grasp fully what the authors were driving at. This presents a difficulty when it comes to making an assessment of the believability of the findings, particularly when having to judge the transparency of the research process and the sufficiency of the information provided.

Background and rationale for the study

From the abstract the paper develops the objective and its importance: public interest issues and, separately, public concern regarding doctors who fail to maintain professional standards. There is a slight danger of being accused of sensationalism in the initial justification for the study when it is noted that the authors highlight the public awareness of and concern for high profile cases such as Shipman and the Bristol paediatric cardiac surgery case. None of the doctors involved in either example have histories which bear any relevance to the area under study and indeed Shipman himself (had he been anonymously selected) would have been a “control” rather than a “case”.

The study authors note that prior to 2010 there had been no mechanism to track and evaluate doctors’ performance following graduation whereas there was at least speculation that some of those doctors who experienced professional difficulties may have had problems at medical school which might have acted as pointers as to the course of future events. There had, at the time, been no published research in the UK which had attempted to link students’ course records with subsequent proceedings for professional misconduct. As such this was a motivational driver for the research and also meant that this could be seen as entirely original work.

The study

The fundamental design was a retrospective matched case-control study looking at a sample of doctors who had graduated from UK medical schools 1958 – 1997 and had a proven finding of serious professional misconduct in GMC proceedings 1999 – 2004.

These were the “cases”, and both they and the “controls” were selected by the GMC’s research and development advisory board and all were fully anonymised before being sent to the study authors for data entry and analysis, thus ensuring total confidentiality and data protection compliance.

The cases were matched using systematic sampling from matching graduation cohorts in the ratio 1 case to 4 matches. The controls were doctors who had graduated from the same university in the same year and were fully registered, currently practising, and had never been under investigation. There was some difficulty in providing exact control matches because of variations in time spent at medical school due to academic delay, intercalated years and transfers-in from other medical schools. 90 of the controls showed a mismatch of up to two years with their “case”.

The number of cases examined was fairly small at 59, with 236 controls, but the criteria for inclusion as part of the research sample were specific and consequently any argument that potential participants had been overlooked ought normally to be approached with caution.

One might perhaps question the fact that only 8 medical schools were included (out of a potential 33).  The authors make clear that these eight were chosen because they “retained student records indefinitely and expressed an interest in participating in the research”. It does imply however that other medical schools could have been included.

For all cases “all available data” were extracted and entered onto a customised database. However this “all” is qualified to mean “sociodemographic factors” and course progress and, as far as can be seen or inferred from the paper “sociodemographic factors” is further qualified only to mean estimating social class from the parental occupation of the father using three independent reviewers and the registrar general’s scheme³, and later condensing the five classes therein into two – higher or lower. This may not be the correct interpretation but it is certainly how it reads.

The authors concede being unable to investigate the possible effects of ethnicity as “these data were not available”. This is somewhat surprising given that race and ethnicity have been at the forefront of the British national debate certainly over the period of time addressed in this study, and especially as the authors note that “ethnic minority status is known to influence performance at medical school and future attainment”^4,5,6, and merits further study.

This is a glaring gap and inconsistency in the rationale. It is not enough to say that findings linked to sociodemographic factors are “sensitive” whilst at the same time overlooking arguably the single most relevant and contentious sociodemographic factor relating to professional and career progression and, even prior to that, entry itself into a UK medical school.

Having identified the sociodemographic characteristics to be used (Sex (M/F), domicile (UK/ overseas), age (< or > 21) estimated social class (higher/ lower), negative comments on UCCA/ UCAS reference (relevant comments/ no relevant comments), the paper appears to be weak in describing the methodology.

This statement “all available data were entered onto a customised database” has already been highlighted but it sits in contrast to the description of data analysis the authors state “was done firstly as a univariate analysis, then a multivariate analysis initially including all explanatory variables then consecutively dropping the least significant one until all included variables were significant at p<0.05”. Despite this the paper does not identify what other explanatory variables were initially considered (in other words whether there were more identified in the “all available data” statement).

Results

The broad findings were that cases were more likely to be young men from lower social class groups. The presence or absence of negative comments on the medical school reference was largely irrelevant. Cases were more likely to have struggled through the course, particularly in the early (preclinical) years. Cases were less likely to have achieved consultant status or be on the general practice register.

Interpretation

There are a number of ways one could look at the results, most of which require the filling in of gaps in knowledge left unanswered by the study.

It is silent on whether allowance has been made for student intake proportions from a time when male students vastly outnumbered female.

The authors are right to point out that the findings in relation to social class grouping may be a reflection of social influences operating at the time, up to nearly 50 years ago. It has been identified as a “sensitive” finding, although it is the least significant statistically of the explanatory variables in the study. Although the authors offer no explanation for the finding, it is not entirely untoward to link it to the social class make-up of a medical school intake at the time when most of the cases were medical students.

The authors concede two important points which go to the limitation of their findings: the first is that social class is difficult to define and subject to frequent re-evaluation, and the second is that in the context of this study, estimated social class may have been acting as a proxy for some other influence which has not been explicitly examined, for example race or ethnicity.

The authors, in this regard, allude (but only go that far) to the phenomenon of stereotype threat, and also mention negative role-modelling, workplace bullying and harassment particularly to doctors from ethnic minorities.

Gender-based differences in personality and consulting styles have previously been suggested as a contributory reason for an increase in complaints against male doctors^{7 8}, as might social biases and the assumption that men were more likely to be aggressive or commit sexual misconduct. The authors have rightly negated this consideration by focussing only on proven misconduct, relying in doing so on the expected robustness of the GMC’s enquiry and investigative operation.

The authors conclude that further work is needed with larger cohorts and a longer timespan. They also raise the possibility of investigating harassment or bullying, but it ought to be said that this is an area which the paper did not set out to look into and only raised as a potential explanation for substandard performance in the workplace, after graduation, and therefore not something which could be called a medical school risk factor.

Impression

In terms of the research process itself, a degree of transparency had to be sacrificed in order to preserve anonymity and data protection compliance.

As to believability, there is a difficulty in insufficient information being sought for analysis and presented in results. It is difficult to come away with the impression that anything of significance has been added to the general discourse of what was already known in this area. On that particular point, the authors remind the reader that a tentative link has already been made in American studies of unprofessional behaviour at medical school and subsequent professional misconduct (in and of itself, unsurprising) but no comparable evidence exists from the UK. It would have been a valid exercise to look at this phenomenon from a UK perspective but it was overlooked, possibly in the quest for originality.

However, what the authors attempted to write up as their headline finding, that of the link between social class and professional misconduct, turned out to be the least sturdy of their findings, for which they could offer no explanation beyond suggesting that it may not mean anything in and of itself and may in fact signify something else which was not looked at.

To its credit the paper did not make any claim at the outset that such risk factors as it sought to find actually existed. It did not set out to convince any sceptics, though it did not need to as the results were already largely known or could easily be deduced from other information sources.

In conclusion the major drawback in reading this paper was the feeling that it did not go far enough. It lacked ambition. What it highlighted as areas for further research should have been the leading questions upon which the research was founded.

1.  Yates J, James D. Risk factors at medical school for subsequent professional misconduct: multicentre retrospective case-control study. BMJ. 2010;340:c2040. doi:10.1136/bmj.c2040

2.  Garfield E. The history and meaning of the Journal Impact Factor. J Am Med Assoc. 2006;295(1):2006-2009.

3.  Rose D. Official social classifications in the UK. In: University of Surrey; 1995.

4.  Yates J, James D. Predicting the “strugglers”: a case-control study of students at Nottingham University Medical School. Bmj. 2006;332(7548):1009-1013. doi:10.1136/bmj.38730.678310.63

5.  Yates J, James D. Risk factors for poor performance on the undergraduate medical course: Cohort study at Nottingham University. Med Educ. 2007;41(1):65-73. doi:10.1111/j.1365-2929.2006.02648.x

6.  Liddell MJ, Koritsas S. Effect of medical students’ ethnicity on their attitudes towards consultation skills and final year examination performance. Med Educ. 2004;38(2):187-198. doi:10.1111/j.1365-2923.2004.01753.x

7.  Firth-Cozens J. Effects of gender on performance in medicine. Bmj. 2008;336(7647):731-732.

8.  Kinnersley, P; Edwards A. Complaints Against Doctors. Bmj. 2008;336:841-842. doi:10.1136/bmj.3.5876.407-b

MA CLINICAL EDUCATION – UNDERSTANDING EDUCATION RESEARCH MODULE

RESEARCH PROPOSAL

Word Count: 3051

What factors affect the experiences of assessment and achievement at medical school from the perspective of UK-originating BME undergraduate medical students?

Introduction

There have been many studies over the years of undergraduate student performance, not confined to the medical student population. The clear majority focus on the positive predictors of success (A level grades, personality, learning style); very few have looked at factors that may predict impaired performance or identify students who may struggle both academically and personally.

The Nottingham University Medical School study (1) found that in UK medical students whose ethnicity was known, being non-white was a highly significant predictor of being a “struggler”, along with other significant (but not “highly significant”) predictors such as a lower mean examination grade at A level, being male and the late offer of a place at medical school.

This work and other studies around this subject area, (see for example (2) (3)) have highlighted the need for further research into the experiences of BME medical students particularly with regard to race and socio-economic grouping and how that relates to the phenomenon of stereotype threat, underachievement and inclusion or exclusion. The aim of this study is to be able to say something in detail about the perceptions and understanding of this group of students without making more general claims.

What this study seeks to explore: the main question is whether, from the participants’ perspectives, ethnicity is a predictor of educational underachievement at medical school and, if so, why? Whether BME/ SEG status at medical school is associated with an increased risk of being a struggler and, if so, why? Whether stereotype threat is an issue in the UK equivalents of what are defined in the US as Predominantly White Institutions (PWIs) and Historically Black Colleges and Universities (HBCUs)? (For this equivalence analysis consider BME medical students at UK medical schools as against UK-originating BME medical students at a Caribbean medical school).

Theoretical Framework

Overview of the literature

Although eventually the literature search was purposefully limited in which papers and writings were used to guide the research proposal, searches were originally made of the following databases:

Proquest

Social Sciences Citation Index (SSCI)

Education Resource Information Centre (ERIC)

British Educational Index

PsychInfo

Medline

Embase

When members of a stigmatised group find themselves in a situation where negative stereotypes provide a possible framework for interpreting their behaviour or their achievement, the risk of being judged in light of those stereotypes can elicit a disruptive state that undermines performance and aspirations in that domain (4). This is the essence of the phenomenon of stereotype threat. The theory has latterly been extended to examine how stereotype threat is related to identity and well-being and how it is associated with feelings of belonging in different environments.

A UK-based qualitative study (3) used the framework of stereotype threat and the constant comparison method to attempt to generate hypotheses to explain underperformance in BME medical students, however no clear consensus emerged. The study did acknowledge the subject area to be important and under-explored and felt that the strong theoretical underpinnings of its data usefully lay the foundations for future work and testing on the way that stereotyping, teacher-student interactions and performance are related.

Similarly Lempp and Seale (2) highlighted findings in their qualitative study conducted in a UK medical school, where non-white students reported perceived differences and disadvantages. In relation to black African students these focussed on themes of having to justify intelligence, not “fitting the image” and needing to conform to the expected template of a “real” medical student, perceptions which were not brought up in the semi-structured interviews by any white or Asian students. The study authors acknowledge that these perceptions and experiences “require further research with a targeted sample of this minority population in medical schools”.

Another qualitative study, this time conducted at a Canadian medical school (5), examined the issue of everyday racism and its attritional effect on student well-being, performance and sense of belonging. The study found that medical students from what it called “racialised minority groups” were subjected to “mundane daily practices which intentionally or unintentionally convey disregard, disrespect or marginality” and which are difficult to deal with. Individually they may seem trivial, minor and not worth making a fuss over but cumulatively they uphold social relations of power, privilege, marginality and oppression. The accumulation of such incidents can significantly increase the level of stress experienced by the recipient (6), but very often the response has been to allow the incident to pass in order to protect one’s time, energy, sanity or bodily integrity (7).

A student labouring under the burden of stereotype threat is likely to find such stresses wearing on their psychological well-being, with academic underperformance being the result. One way of insulating themselves from the stigma and effect of underachievement is to withdraw from the domain. An unfortunate aspect of stereotype threat is that the very people who tend to be the highest achieving and care the most are also those most affected by negative stereotypes. These students experience greater falls in their performance under stereotype threat because their performance in the domain is self-relevant. Osborne and Walker (8) found that BME students who were most academically strong were more likely to withdraw from their course of study. This offers some explanation as to why the mechanism of disidentifying from a domain, ceasing to connect success in a domain to one’s sense of self, is seen as being able to eliminate the negative performance effects of stereotype threat (9).

Thus, stereotype threat can undermine a person’s sense of belonging, affect motivation and make them more likely to withdraw from their academic setting, as well as making them uncertain of the quality of their social bonds.  These are features of marginalisation, which Essed identified as a key feature in her theory of everyday racism, a category of behaviours she calls “containment”, humiliation or belittling aimed at keeping people in their place (10).

There is research which posits that reducing concerns about being in potentially threatening environments can reduce vulnerability to stereotype threat and thus allow targeted individuals to perform to their full potential (11). Walton and Spencer  come back to this point when they suggest that, according to their latent ability theory, a black student undertaking a course in an “identity-safe” environment in which he did not risk being judged against a negative stereotype of his group would earn better grades than an equivalent white student (12).

This leads us to the understanding that if targeted, negatively stereotyped groups are not made to feel welcome in their academic milieu, chronic exposure to stereotype threat can lead to those students disidentifying from their domains, leaving their programs and eventually abandoning their course of study entirely.

It would be interesting therefore to investigate the perspective and perceptions of UK-originating BME students in what might be imagined to be an identity-safe environment.

Methodology

It is intended that this should be a study using the Interpretative Phenomenological Analysis (IPA) approach whilst drawing upon and adapting elements of Grounded Theory Method (GTM).

The conceptual framework of IPA is hermeneutic phenomenology (13). Hermeneutics is a theory of interpretation which puts meaning at the centre of the pursuit to understand human experience; phenomenology uses an open-ended approach to describing phenomena in the world as they are experienced by individuals: people cannot be understood in isolation and we must therefore always consider the context in which they live. Whereas mainstream psychology and IPA converge in their joint interest in how people think about what is happening to them, they diverge in the application of how this thinking can best be studied, with mainstream psychology still firmly committed to quantitative and experimental methodology and IPA employing an in-depth qualitative analysis.

IPA research focusses on the person within their world in order to understand the phenomenon of interest from their perspective. It is concerned with an individual’s personal perception of an event, and how that individual makes sense of their personal and social world. At the same time the researcher could be said to be trying to make sense of the participant trying to make sense of their world – a two-stage interpretation process or double hermeneutic.

IPA is also concerned with trying to take the side of the participant, to try to understand what things are like from their point of view, and again at the same time it may involve asking detailed critical questions in the analysis of the texts, in other words combining an empathic hermeneutics with a questioning hermeneutics.

The research question in this study is purposely framed broadly and openly. There will be no attempt to test any predetermined theory or hypothesis, rather to explore, flexibly and in detail, an area of concern. The mode of enquiry will be idiographic as opposed to the nomothetic approach which tends to predominate in psychology.

A distinctive feature of IPA is its commitment to a detailed interpretative account of the cases included. Many researchers have recognised that this can only realistically be done on a very small sample, in other words breadth is sacrificed for depth. The current thinking is that for researchers undertaking IPA for the first time a sample size of three participants is probably best. This allows for sufficient in-depth engagement with each individual case whilst also allowing a detailed examination of cross-case similarity and difference, convergence and divergence.

To that end, it is helpful that the target demographic of the study would automatically create an homogenous sample. If one is intending to interview a small number of participants, it is unhelpful to be thinking in terms of random or representative sampling. This study will therefore go in the opposite direction and use purposive sampling to find a more closely defined group for whom the research question will be significant. This is a logic not dissimilar to ethnographic research conducted by social anthropologists in any one community. The end result of that is the ability to report in detail about that particular culture or community without claiming the ability to say something about all cultures.

Grounded Theory Method (GTM) was defined by its originators Glaser and Strauss in 1967 as “the discovery of theory from data – systematically obtained and analysed in social research” (14). The emphasis on theory in the original book is in sharp contrast to the use of GTM these days, where it is known primarily as a method of qualitative data analysis. There is nothing inherently wrong with this – one of the benefits of GTM is that it is infinitely adaptable, but any adaptation should be explained says Cathy Urquhart (15). In her book is a reminder of the key features of GTM which should serve as a starting point before any adaptations are made:

• The aim is to generate or discover a theory.
• The researcher must set aside theoretical ideas to let the substantive theory emerge.
• Theory focusses on how individuals interact with the phenomenon under study.
• Theory asserts a plausible relationship between concepts and sets of concepts.
• Theory is derived from data acquired from fieldwork interviews, observation and documents.
• Data analysis is systematic and begins as soon as data is available.
• Data analysis proceeds through identifying categories and connecting them.
• Further data collection (or sampling) is based on emerging concepts.
• These concepts are developed through constant comparison with additional data.
• Data collection can stop when no new conceptualisations emerge.
• Data analysis proceeds from open coding (identifying categories, properties and dimensions) through selective coding (clustering around categories) to theoretical coding.
• The resulting theory can be reported in a narrative framework or a set of propositions.

In practical terms in relation to this study the most significant obstacle is the normal institutional requirement to have undertaken a literature review prior to embarking upon the research whereas GTM requires that the researcher does not impose theoretical concepts on the coding of the data. The institutional barrier may also manifest as a perception that GTM is not intellectually robust, that it ignores the literature. That is not so, but one way to deal with this and what this study proposes to do is rely on a non-committal, or limited, literature review whose relevance at write-up is determined by the emergent theory.

Strategy and Design

Sampling

The participants will be purposively recruited from the recommendation or suggestion of the respective medical school faculties. Initial letters of approach will be to Deans of University College London Medical School (UCLMS), Peninsula College of Medicine and Dentistry (PCMD) and University of the West Indies Faculty of Medical Sciences (UWI). One Year 3 medical student will be recruited from each location, making a total of three.

Data Collection and Analysis

IPA/ GTM studies require a flexible data collection instrument and the best method, certainly the most widely-used, is the semi-structured interview. It is proposed that this method will be used to conduct research in this study. This form of interviewing allows the researcher and participant to engage in a dialogue whereby initial questions can be modified in light of the participants responses and the researcher is able to probe interesting and important areas which arise. The participant is very much the experiential expert. By giving maximum time and opportunity to tell their story,

• rapport and empathy is facilitated
• the order of questions becomes less important
• there is greater flexibility of coverage and the ability to go into new areas,
• the interview can follow the participant’s interests or concerns
• there is a tendency to produce richer data.

The interviews will be professionally transcribed and prepared for coding, which I will undertake personally. It is intended that the coding will follow the GTM template of open coding, selective coding and finally theoretical coding. The aim will be to understand the context, context and complexity of the meanings behind the words rather than measure their frequency. Thus, an element of IPA is present in that I as researcher will need to engage in an interpretative relationship with the transcripts.

Ethical Issues

This study will be fully compliant with the British Educational Research Association (2011) Ethical Guidelines

Sampling, methods, recruitment: There is little to no potential for bias from the study author in the selection of the sample. The participants will be purposively recruited from the recommendation of the respective medical school faculties. I will not know any of the participants personally.

As to the participants, I have to consider whether any might choose to participate in order to gain an outlet to air strong negative feelings about their institutions and whether this might bias the findings one way or the other. I think the use of purposive sampling militates against this. The sample is not being recruited by an open process but by a process more akin to headhunting as seen in commercial recruitment. Only three participants will be interviewed, in other words sampling will not be continuing unto theoretical saturation. It will lean more towards Interpretative Phenomenological Analysis (IPA) than pure Grounded Theory Method (GTM).

Informed Consent: All participants will be fully briefed as to the study’s aims and objectives and their part in it. They will have the opportunity to ask questions or raise concerns which I will endeavour to address prior to the start of the interview process. They will understand that the signing of a consent form is an acknowledgment that the study has been explained to them but that it does not bind them in any way to continue to participate even after starting.

Confidentiality/ Anonymity, Data Storage and Security: I have a responsibility both morally and legislatively to ensure that the confidentiality of the participants and their personal data is protected. I intend to keep such data as I accumulate secured in password-protected files on a password-protected and encrypted USB device as well as on the password-protected UCL One Drive server. Data will be stored as audio files and as transcribed interviews in Microsoft Word format and possibly also as Adobe PDF files.

The participants will be assigned a coded identifier at the start of the process which cannot be linked to any one named individual. No names will be used when the interviews are being recorded. In terms of proportionality, this would be a satisfactory method of achieving anonymity.

The participants will be aware from the information sheet and also from me personally that the information being accumulated could potentially, as well as being used in a Masters dissertation, be published and disseminated. The consent form will point to this possibility and assure that under these circumstances as well as for the primary purpose of the research, data protection processes will be adhered to.

Dissemination, disclosures and limits to confidentiality: I intend to send a synopsis of my findings to the participants after write-up. They will of course have been anonymous throughout the process but having received the document they will be free to identify themselves and their participation to their peer group. From a practical and pragmatic point of view that is not a difficulty. No participant is able to identify any other participant. Therefore confidentiality of information persists beyond the formal end of the study.

Sensitive topics. This is a broad heading which potentially covers a wide range issues. At this stage what can usefully be said is that sensitive topics will be handled sensitively and in keeping with what I have learned through other modules on the MA course. Anything that cannot be handled competently by me will be referred on to better qualified parties.

This study will not, in the normal course of events, have to deal with issues of safeguarding or child protection. Neither will it be recruiting potentially vulnerable participants. I say “in the normal course of events”. There is a possibility that matters relating to these areas could surface in the course of an interview, or that a participant on interview turns out to be or appears to be emotionally compromised in a way that could not have been anticipated at the start of the process. There would be an unwritten contingency that in those events the process would be stopped immediately and the participant concerned referred for appropriate help and support. It is not expected that I would be able to provide that specialist help were it to be needed.

Appendices

Information Sheet

Consent Form

References

1. Yates J, James D. Predicting the “strugglers”: a case-control study of students at Nottingham University Medical School.BMJ2006;332:1009-101
2. Lempp H, Seale C. Medical students’ perceptions in relation to ethnicity and gender: a qualitative study. BMC Medical Education 2006;6:17-23
3. Woolf K, Cave J, Greenhalgh T, Dacre J. Ethnic stereotypes and the underachievement of UK medical students from ethnic minorities: qualitative study
4. Spencer SJ, Logel C, Davies PG. Stereotype threat. Ann Rev Psychol 2016;67:415-37
5. Beagan BL. “Is this worth getting into a big fuss over?” Everyday racism in medical school. Med Educ 2003;37:852-60
6. Harrell SP. A multidimensional conceptualisation of racism-related stress: implications for the well-being of people of colour. Am J Orthopsychiatry 2000;70:42-57
7. Sinclair S. Making doctors: an institutional apprenticeship. Oxford: Berg, 1997
8. Osborne JW, Walker C. Stereotype threat, identification with academics, and withdrawal from school: why the most successful students of colour might be most likely to withdraw. Educ. Psychol.2006;26(4):563–77
9. Steele CM. A threat in the air: how stereotypes shape intellectual identity and performance. Am. Psychol. 1997;52(6):613–29
10. Essed P. Understanding Everyday Racism: An Interdisciplinary Theory. New York: Sage 1991
11. Walton GM, Cohen GL. A question of belonging: race, social fit, and achievement. J. Personal. Soc. Psychol.2007;92:82–96
12. Walton GM, Spencer SJ. Latent ability: Grades and test scores systematically underestimate the intellectual ability of negatively stereotyped students. Psychol. Sci.2009;20:1132–39
13. Langdridge D. Phenomenological Psychology: Theory, Method and Research: Pearson Education 2007
14. Glaser B, Strauss A. The discovery of grounded theory. Strategies for qualitative research. London and New Brunswick: Transaction Publishers, 1967
15. Urquhart C. Grounded Theory for Qualitative Research: A Practical Guide. London: Sage 2013
16. Patel RS, Tarrant C, Bonas S, Shaw RL. Medical students’ personal experience of high-stakes failure: case studies using interpretative phenomenological analysis. BMC Medical Education 2015;15:86-93
17. Lempp H, Seale C. The hidden curriculum in undergraduate medical education: qualitative study of medical students’ perceptions of teaching. BMJ 2004;329:770-773