This cross-sectional mixed-methods study provided self-reported information on medical students’ experiences of harassment during their clinical training. It is clear from the findings in this study that many students in this sample had reported encountering some form of harassment in the areas of disrespect, isolation or exclusion, threats or bribes, and verbal aggression during their clinical placement, which ranged from 47 to 89%. Only physical aggression indicated a low level of experience at 6%. However, the sub-group analysis suggested there were no discernible differences in terms of sex, ethnicity, or year of study. Nevertheless, older students tended to report experiencing higher levels of harassment compared to younger students. There was no discernible reason for this given that some studies have reported that older students tend to report harassment less often [23]. One hypothesis is that the discrepancy relates to the willingness to report rather than the experience of more harassment and this could be linked to the idea of power, which will be subsequently discussed.

We will now consider each of the conceptual factors described by Rospenda and Richman [19] in more detail.

Rospenda and Richman [19] propose that this factor underscores associations with humiliation, belittlement, feeling inferior, unfair judgment and treatment, controlling behaviours, embarrassment, inequitable task allocation, and unwarranted blaming. The concept of disrespect overlaps with incivility, or being devalued or treated as less capable [24]. The findings in this study indicated that some students experienced persistent disrespectful behaviours, with three questionnaire items highlighting students’ feeling that they were talked down to (7%), treated as being less good (4%), or asked to do work which they perceived was not part of their job (3%).

One student (Table 2) saw disrespect as linked to a lack of involvement when tasks were being allocated, and thus feeling a lack of autonomy. Leape and colleagues [25] proposed that disrespect can be attributed to several sources and occur at many levels. For example, they cite the problem of learning through humiliation, when senior physicians model disdain when responding to patient questions. Disrespect can be linked to disruptive behaviour resulting in poorer health outcomes for patients [8]. For example, they cite a culture of education through humiliation, and senior physicians modelling of disdain when answering patients’ questions. Disrespect can occur at subtle levels by displaying passive-aggressive behaviours, such as not following through on previously agreed proposals. Disrespectful behaviours can be dysfunctional and entrenched within workplace cultural norms. This form of entrenchment may create an unprofessional hierarchy whereby healthcare workers are perceived as being more valuable than students and patients. Lastly, disrespectful behaviours can stem from the individual (e.g. personality) or environmental (e.g. cultural norms) factors. Leape highlights that students are particularly at risk of experiencing forms of disrespect, including belittlement which was of particular concern in our findings.

Rospenda and Richman [19] contend that this factor is linked to the concepts of being excluded from important conversations or meetings, being used by those in higher positions, having contributions plagiarised, and creating discord in the workspace. The findings indicated that a critical mass of students in this study experienced aspects of being isolated or excluded, for example, 64% of students had specifically experienced being ignored in reference to their work contributions. The quote in Table 2 clearly describes one student’s feelings regarding this type of occurrence (e.g. “am I invisible today?”). The feeling of being excluded or isolated can adversely impact a student’s sense of being a team member culminating in a perception that they are not making a meaningful contribution.

Romanski and colleagues [26] put forward the term the “invisible student” that aptly describes this felt phenomenon. They report students describing this type of treatment as often being covert, whilst stating that it obstructs learning and reinforces their low position in the team hierarchy. The nature of exclusion can also be perceived as being passive, although the impact may have an adverse effect on students’ development, morale, and ultimately their career choices. The converse also appears to be true, that is, that feeling part of the team during clinical placements is likely to increase the chance that students will choose to practice in that area once fully qualified [27].

Some recent literature has focussed on finding ways to meaningfully involve students during clinical training. Ooi and colleagues [28] reported ways in which students could be included when discussing patient issues in handover, such as making time for opportunistic teaching moments. In this way, students could feel valued and part of the team they belong to. In addition, the University of Auckland have put in place an anonymous survey whereby medical students can report experienced or witnessed behaviour associated with bullying, harassment, discrimination, and exclusion [29]. Even though this method may not capture unique instances of harassment, it can isolate clinical settings where students feel unsafe.

Physical aggression is a particularly worrying aspect of harassment with potential legal ramifications if reported. Rospenda and Richman [19] have defined physical aggression in terms of being pushed, grabbed, and physically assaulted. In this survey, we noted that ten students (5%) had reported being pushed or grabbed, three (1%) had reported something being thrown at them, and two (1%) had reported being hit physically — while low rates in absolute terms, these are unacceptably events. It is worrying that there are self-reported instances of this type as this represents one of the most extreme forms of harassment. The instance described by the student in Table 2 is very concerning given that this student was junior and felt unable to report the incident. The feeling of being unsafe and unsupported is a concern for any university or healthcare setting, suggesting that it is crucial to have mechanisms that optimise access to support and safe reporting systems.

Physical aggression is clearly a very explicit form of mistreatment and does occur in healthcare workplaces. Warshawski [30] suggested that many healthcare workers at some time in their career do experience physical assault, although they do not specify the perpetrators. In a further study, the authors stated that 18% of their sample of Nigerian medical students had experienced being slapped, pushed, kicked, or hit [3]. Nonetheless, as with this study, physical assault is typically reported as occurring less often than other forms of harassment. One of the important issues to consider is ensuring that students feel safe and empowered to report instances of physical assault. Students need to be provided with information at the beginning of, or prior to, their clinical rotations regarding this [30]. Reporting systems need to be shown to be effective; otherwise, students may perceive it to be harmful to their future careers or personal safety.

Rospenda and Richman [19] underscore this aspect of harassment in terms of feeling pressured to change one’s belief or opinion, being subjected to bribes, or being threatened. Students in this study revealed encountering certain harassment behaviours, such as feeling pressured to change one’s beliefs or ideas (46%), reporting of feeling bribed (3%), or feeling explicitly threatened (2%). Although none of these items met the mean score > 2 threshold for highlighting concerning items, 46% of respondents scored 2 or greater for feeling pressured to change one’s beliefs. The notion of feeling pressured to change one’s belief or idea is alarming, particularly given the example provided by the student in Table 2. However, the context of the experience needs to be fully investigated, given that in medicine ideas and beliefs about practice are often challenged, which is part of the ongoing process of being a competent and safe health professional.

In one study, Owaoje and colleagues [3] stated that 26% of their sample of medical students experienced threats of harm, which is demonstrably higher than the reports in this study. The source of these threats appeared to originate from those in power (e.g. consultants, lecturers, and registrars). Issues of power and hierarchy have been highlighted in the literature and often used to explain why some groups may be at risk of bullying [31, 32]. The issue of power is conditional, in that negative use of power can lead to evident cases of mistreatment while positive power dynamics can create constructive learning outcomes [32]. Moreover, the influence of power can occur in both the defined and hidden curricula [32].

Lastly, verbal aggression is an area often cited as being problematic in the healthcare workplace [33]. Rospenda and Richman [19] have defined verbal aggression in terms of certain behavioural instances, such as being yelled at, sworn at, insulted, and spreading unpleasant rumours. Three items with this scale were highlighted as being problematic, namely experiencing negative comments about performance and intelligence (70%), preventing the expression of one’s views (62%), and being told insulting jokes (59%). The exemplar student quote (Table 2) regarding their experience of verbal harassment is concerning. These experiences and their relatively high level of incidence may be linked to the notion of accepted cultural practices within teams, suggesting staff development in healthcare learning settings is crucial to minimising the occurrence of these modes of communication [8]. Nonetheless, negative comments are likely dependent on the context in which they are given and the use of language underlying the negative comments (e.g. use of personalised demeaning phrases).

Verbal harassment is clearly linked to workplace culture, especially when considering the responses in this survey and those examples found in the wider literature [2, 34]. The cultural norms surrounding ways of communicating can occur at all levels and are often found within the hidden curriculum. As with previous assertions, it is likely to be associated with notions of power and hierarchy. Undesirable consequences can result, such as becoming adversely socialised or using toxic behaviours when communicating with patients and future colleagues [34]. Therefore, the strategies put forward by Kassebaum and Cutler [34] are likely to be still valid today, such as promoting respectful communication at all levels of the learning spectrum, and ensuring methods of communication are audited at both formal and informal levels.

Our response rate of 25% may be perceived as a limitation. However, this yielded a margin of error for our survey of only 6% at the 95% level of confidence, well within the 8% margin of error considered acceptable for generalisable commercial surveys [14]. In addition, response rates are typically low in these types of survey studies aiming to obtain data in this topic area [35]. Fosnacht and colleagues [36] acknowledge that variation exists with defining nonresponse bias, but state that “low response rates may or may not lead to nonresponse bias because answers to survey items may not differ substantially between responders and nonresponders”. In their study, using simulated data, they found that simulated datasets that yielded response rates of 5% generated meaningful correlation coefficients (0.64 to 0.89) with full sample estimates. Therefore, they question the assumption that low response rates are symptomatic of low data quality and further state that response rates of 20 to 25% are likely to yield meaningful survey data.

Moreover, Fan and Yan [37] state that using online survey has several advantages when compared with traditional paper-based survey, such as being more accessible to those participants who have internet access, created quicker response time, lower costs in delivery, and easier data entry options. Nonetheless, Fan and Yan also note that online surveys can be problematic in reaching participants who do not have internet access and often generate low response rates. In reference to the present cohort, lack of internet access is unlikely to be an issue, although the sensitive nature of the questionnaire and students’ high study workload may have impeded students’ willingness to respond to the survey. Further factors that may have obstructed a willingness to respond relate to the sponsor of the survey and the time it would take participants to fill in the questionnaire. The sponsor of this current questionnaire was the university that manages the clinical placement, which may have affected the response rate. Nonetheless, employing a student researcher to orchestrate the communication protocol and data collection of the survey may have mitigated some of the risks associated with this factor. In addition, a monetary incentive was provided to students. The GHWQ has 29 items which may have deterred some students due to their busy schedules, and given that the university medical curriculum requires not only examination commitments but also evaluations of courses and placements beyond the current study. This may have hampered students’ enthusiasm to fill in a survey occurring outside their curriculum duties, even though this survey was explicitly sanctioned by this university.

A related issue that needs to be acknowledged is the potential presence of reporting bias, which may result in the under-reporting or overestimation of incidence [38]. To further validate the data presented in this study would require ongoing studies to audit the potential presence of harassment in clinical settings as experienced by medical students. Nonetheless, occurrences of harassment are evident within this sample establishing it is an issue worthy of further investigation. In addition, as aforementioned, Fosnacht and colleagues [36] have provided a convincing argument to suggest this may not be a critical problem in these types of studies.

Given this is a convenience sample, we made an assumption that the sample participating in this study had representation from key demographic groups (age, sex, ethnicity, and year of study). However, we acknowledge that the sample may not be statistically equivalent in proportionality to the population. We argue that the cross tabulations between age, sex, ethnicity, and year of study would be interesting, but would unlikely add value to the central thesis of the study, which is a mixed-methods approach aimed at exploring and identifying issues of harassment within the clinical context. However, cross tabulations could be scope for further research, although gaining statistical representation at this fine-grained level would likely be very difficult to achieve, especially when measuring this sensitive issue which can only be practically conducted using convenience sampling. Lastly, an explicit measure of sexual harassment was not included even though it is an important component of harassment.