Effects of mistreatment in medical schools: how to evaluate? A brief review

Introduction: The occurrence of abuse, harassment and mistreatment of medical students is a widespread phenomenon and not a problem limited to certain countries or particular schools. Such behavior during medical training creates hostile learning environments, induces stress, depressive symptoms, may impair performance and patient care. Objective: To analyze the methodology used in recent studies to describe the consequences of mistreatment on medical student’s life and academic performance. Method: A brief review of the literature indexed in 6 international databases was carried out (PubMed, Scopus, Web of Science, SciELO, PsycINFO and Cochrane Library). The descriptors were categorized into two groups, one containing different types of violence and the other contemplating the studied population. Results: A total of 20 articles were selected for this study and all of them based their research methodology on the use of questionnaires, scales and/or interviews. The strengths and weaknesses of these methodologies were discussed, and the use of simulation was suggested as a new methodological alternative. Conclusion: This review reinforces that mistreatment of medical students has remained frequent over time and it is closely related to the students’ mental health and performance impairment. The authors recommend a new methodological approach to collect data related to the effects arising from a hostile learning environment.


INTRODUCTION
The mistreatment of medical students is prevalent throughout the world and has been pointed out since the early 1980s 1 . In 2011, The Association of American Medical Colleges 2 more specifically described mistreatment through a list of behaviors with an effort to clarify the definition. Examples include sexual harassment, humiliation, psychological or physical punishment, and harassment based on race, religion, gender, or sexual orientation; behaviors which portray disrespect for the dignity of others and unreasonably interfere with the learning process.
A considerable number of studies [3][4][5][6] have found that most medical students around the world experience high incidence rates of mistreatment during training. A metaanalysis carried out in 2014, including 59 studies, reported that the combined prevalence of harassment and discrimination of medical students was about 59.6% 7 .
Although mistreatment situations are not exclusive to medical training environments, Rautio et al. 8 found a higher prevalence in medical courses when compared to courses in human science and technology areas. Some teachers have stated that these abuses are an unavoidable part of medical education 9,10 . However, abusive behaviors result in negative effects on the students' well-being and learning abilities. Reports indicate 9,11,12 that perceived abuse has been associated with long-term mental health consequences such as depression, alcoholism, and suicidal attempts. Students who have been victims of mistreatment, belittlement, harassment, or bullying report a loss in their relationship with teachers, are more dissatisfied with their professional choice, and more frequently consider dropping out of the course, feeling more stressed and depressed.
Therefore, it is noted that in addition to representing a common phenomenon in the medical training environment, mistreatment can result in several types of damage with different characteristics. Some result from sporadic situations, while others, from persistent ones. Some represent immediate damage, such as impairment in the relationship with the teacher and in the acquisition of knowledge. Others, however, are identified later and can have a long-term impact on the students' lives, including depression, increase consumption of alcoholic beverages and even dropping out of the course. Thus, the question arises: how to collect data involving such a sensitive subject and with such diversified outcomes, using an ideal approach? What has been practiced in the existing literature?
Thus, the main goal of this study was to analyze/evaluate the methodology used in recent research describing the consequences of a hostile learning environment on quality of life, mental health and/or academic performance and propose a new approach.

METHOD Study type: short review
A literature search was carried out from November 2019 to January 2020 to evaluate the recent literature related to the effects that any type of mistreatment can cause to undergraduate medical students and residents.

Keyword selection and database search
A comprehensive literature search was performed in 6 databases: PubMed, Scopus, Web of Science, SciELO, PsycINFO and Cochrane Library. The terms initially used in the search included descriptors and words from the text associated with key concepts related to the review topic. The terms were divided into 2 broad categories: 1) mistreatment types in the academic environment ("mistreatment", "harassment", "sexual harassment", "belittlement", "abuse", "intimidation", "bullying" and "violence"); 2) target population ("medical students", "medical school", "residents", "clinical internship" and "medical interns"). The search terms were refined during the bibliographic search and different combinations of descriptors and keywords were used according to the database and available controlled vocabularies. The Endnote® program was used to organize and manage references, as well as to eliminate duplicates.

Eligibility criteria
Inclusion/exclusion criteria were applied using a modified PICO -Population, Intervention, Comparison, Outcomeframework, which offers a useful, structured, and tailored approach to determine whether an article could be included or excluded from the research. However, the PICO formula was modified by substituting 'comparison' for 'environment' (Table 1). Research letters, articles without abstracts (because the abstracts were checked to review the article) and book chapters were not included in this review.
After this first step, a manual search of the references from the selected studies was also performed and the relevant data were summarized and organized.

RESULTS
One thousand, two hundred and seventy-eight (1,278) articles were found in the six databases (PubMed: 203; Scopus: 347; Web of Science: 355; SciELO: 53; Cochrane Library: 31; PsycINFO: 289). After excluding the duplicates, 706 articles remained. From these, based on the eligibility criteria, 688 articles were excluded. Through a manual search of the 18 selected articles, two more were included, resulting in a total of 20 articles that were analyzed in this research ( Figure 1).
This brief review showed that all abovementioned authors used questionnaires or scales as instruments to collect data. Some complemented their research with interviewing techniques. A description of all comparative parameters related to the methodologies used in these 20 studies was performed.
The results are organized and displayed in Table 2.

DISCUSSION
Training to become a physician and the practice of

selected articles
Careful reading of the abstracts (exclusion of some articles because they did not meet the inclusion criteria).

selected articles
Careful reading of the method and results sections (inclusion only of articles that described deleterious effects caused by mistreatment).    33 . First, the existence of strongly hierarchical relationship patterns and a medical culture that permeates teaching and learning relationships, perpetuating mistreatment situations as "rites of passage".

included articles
Olasoji's survey 10 pointed out that almost half of the studied sample did not report mistreatment incidents because they thought they were not important enough to be reported and considered them as inherent to medical training. The second is related to the low success rates of implemented strategies to reduce inappropriate attitudes in the academic environment.
This same survey showed that the majority (66.7%) of students who reported a "toxic" experience felt neutral to the outcome of reporting. Besides that, Wilkinson et al. 15  However, methodological weaknesses can also be raised.
Prodanov 35 states that questionnaires cannot suggest or induce responses. It was found that some questionnaires 11,18 addressed the impacts caused by mistreatment by directing answers to negative effects, giving the respondent no other choice of response. Only in the most recent studies, such as the one by Olasoji 10 , answers involving possible "positive" aspects were added, and what could have been considered by many authors as not possible so far, surprisingly pointed out that 37.5% felt the "toxic" practice was useful for learning and made them stronger.
Andrade 36 warned that in surveys using questionnaires only, where you cannot have an interviewer present, answers can be biased, leading to an apparent uniformity. That can happen because of the difficulty some respondents may have to understand some questions. The data obtained from most articles 13,22,27 that constituted this brief review result from questionnaire analyses only, without the association of a subsequent interview that would allow the participant to solve any type of doubt that arose during the filling out of the questionnaires, or to expose any type of feelings that they were unable to explain when completing the standard questionnaire. And here, it is worth mentioning that the impacts generated by any type of mistreatment represent a complex type of data to be collected, which can vary a lot, whether regarding the affected area or its intensity, as well as variations from individual to individual.
Although it does not represent the majority, some Some authors 11,19 have proposed a longitudinal survey, with questionnaires being applied at two or more moments, and theoretically, this could minimize the bias of a crosssectional analysis. However, invariably, these same articles faced the difficulty in maintaining good response rates, as they tend to decline throughout the study. Besides that, studies always benefit from the use of different methodologies, whatever the area of interest is.
The use of simulation could be a methodological alternative that would avoid many of the weaknesses pointed out so far, especially when determining the immediate damage from mistreatment situations in the learning environment (What actually happens to the student at the moment of the mistreatment?; Are learning and performance impaired?).
Riskin et al. 39,40 have already proven the usefulness of simulation for this type of research by conducting studies that aimed to assess the impact of rudeness on the performance of NICU (Neonatal Intensive Care Unit) medical teams. During training workshops, some teams were randomly assigned to an exposure to rudeness. In both surveys, they showed that rude attitudes had deleterious effect on the teams' performance.
Diagnosis and intervention, information sharing, workload, help, and communication were affected.
Setting up scenes of mistreatment in a simulation environment would provide prospective and recorded observation in a nonbiased fashion, typically using audio and video capture. Additionally, some other parameters could be monitored for further analysis, such as heart rate variability, facial and behavioral analysis, and assessment of the proposed task performance. They would be of great value to study with a little more depth the real effects of the stress caused by mistreatment in the academic environment.
It's also necessary to make a very pertinent caveat when discussing the general belief that individuals learn better from stressful events, especially when the source of stress is mistreatment. According to Christianson 41 the information of an event to be remembered will be retained very well if the event itself is the cause of the individual's stress response. If the stress is caused by something peripheral to the information to be remembered, the consolidation of that information will not be improved. So, it is possible to infer that a student who feels stressed during a class where they are learning how to intubate a patient, for example, due to the presence of the teacher's intimidating or abusive attitudes, they may develop a strong memory of the teacher's attitudes and actions, but the memory they will have of the class itself, and of the associated essential learning points is unlikely to be improved. Therefore, simulation would allow differentiating environments where the student must perform a certain action under intense pressure and stress because their patient's life is at risk from those in which frequent abuse and verbal aggression occur, showing that such attitudes are unnecessary and closely related to learning impairment. Perhaps, it will help to end years of a hierarchical culture that confounds these concepts and postulates these attitudes as necessary for medical training.

CONCLUSIONS
Based on 3 important considerations, it is clear that a new approach is needed to study the impact of a hostile learning environment on students: 1. The prevalence indicated in the majority of the studies continue to show worryingly high rates of mistreatment of medical students; 2. Sufficient documentation already points out that mistreatment to students can result in a wide range of negative outcomes, both in the short and long-term; 3. This review allowed the identification of some methodological weaknesses of studies carried out in area to date.
For future studies, the simulation of learning scenarios showing inadequate teachers' behaviors must be effectively put into practice in order to assess their effects on the students.