1.1 Impact of COVID-19 pandemic on Nurses' working conditions

The outbreak of coronavirus disease 2019 (COVID-19) has so far challenged all healthcare systems and raised unprecedented demands for healthcare providers (HCPs) worldwide (Nogee & Tomassoni, 2020), leading to highly complex and unpredictable conditions in the nursing profession (Shrestha, 2020). As the backbone of healthcare systems, HCPs are now expected to work in an environment transformed by the pandemic (Kalateh Sadati et al., 2021), where they must fulfil their duties at the forefront of treating and caring for patients under the conditions threatening to their own health or even life (Rodríguez-Bolaños et al., 2020). Nurses are thus the largest group of frontline HCPs, often in direct contact with patients and people in general; therefore, they play key roles in treating and preventing the progression of diseases (Abbaszadeh & Ehsani, 2015; Bella Magnaye et al., 2011). Besides, there is strong evidence suggesting a rising trend in the prevalence rate of aggressive behaviours among HCPs, particularly nurses, (Yenealem et al., 2019) due to the critical conditions caused by the COVID-19 pandemic, its prolongation, high infection rate, equipment shortages (especially at the onset of the pandemic), increased number of patients, elevated workload, higher mortality rates, mutation of the virus, re-infection in patients, fear of contracting the disease (8, 11), role overload, unrealistic expectations and role ambiguity (Hauge et al., 2009). Such aggressive behaviours demonstrated by these individuals towards each other are thus called horizontal or lateral violence (American Nurses Association, 2012).

1.2 Horizontal violence and turnover intention

Horizontal violence can be emotional, verbal or physical, or in the form of discrimination, prejudice and inadequate support (Bambi et al., 2013; Taylor & Taylor, 2017). The victims of this behaviour are often subjected to repeated harassment, abuse and insult, along with the violation of their physical or mental dignity or integrity (Hauge et al., 2009). However, nurses, mostly the experienced ones, do not see themselves as violent individuals or bullies, but tend to consider these negative behaviours a necessity to check their less experienced coworkers. Moreover, some nurses believe that constant criticism is a constructive attitude, or gossiping is a way to avoid and cope with work stress (Taylor, 2016; Taylor & Taylor, 2017). However, it has been shown that such attitudes can foster distrust among coworkers, erode professionalism and create hostile work environments (Morrison et al., 2017). Horizontal violence can further affect behavioural responses among victims (Hutchinson et al., 2010; Laschinger et al., 2012). As well, the responses provided by nurses to horizontal violence tend to be in the form of increased absenteeism and turnover intention, and ultimately leaving the workplace or even the profession (Berthelsen et al., 2011; Cheung & Yip, 2017; Kim & Park, 2016), which may be devastating for understaffed healthcare centres. In this line, a survey in Iran had explored that many nurses were less willing to attend shifts, regretted being in this profession, expressed their intention to leave the profession and were suffering from psychological turmoil, such as fear, anxiety, aggression, confusion, restlessness and apprehension, because of the challenges caused by this pandemic (Moradi et al., 2021). Other studies had also indicated that turnover intention had significantly amplified among nurses following COVID-19 (Falatah, 2021; Haji & Mohammadimehr, 2021).

Considering that limited studies have been so far completed in Iran in this field, it seems that overlooking this phenomenon poses the risk of the integration of the nursing culture and development into a generational phenomenon, transforming it into a silent epidemic in Iranian healthcare institutions.

1.3 Conceptual framework

Hostility between nurses at the same level is known as horizontal or lateral violence, which has been defined by some scholars as bullying in this profession. Previous studies have further confirmed the idea that nurses who have experienced horizontal violence can have higher turnover intention (Bashaw, 2019; Kabir et al., 2022), as explained by Kanter's theory of structural empowerment (Kanter, 2008). This theory takes account of three levels of empowerment, that is, the workplace that has the necessary structures to promote empowerment, the psychological belief in one's ability to be empowered, and the acknowledgement of power in the relationships and care that nurses provide (Amor et al., 2021; Favaro et al., 2021). Structural empowerment (at the first level), which generally describes access to four environmental factors, namely opportunity, information, support and resources in the organization, can thus initiate psychological empowerment in nurses (Favaro et al., 2021).

2.1 Participants and procedures

This study was conducted from September 6 to December 21, 2021, at the public referral hospitals in the city of Urmia, selected to process the COVID-19 confirmed and suspected patients. For this purpose, the eligible nurses were randomly recruited from five of these hospitals, considering the criteria of at least 1 year of work experience, working in the mentioned hospitals all through the COVID-19 pandemic and showing consent to participate in the research. The nurses who did not completely fill out the questionnaires or had simultaneously worked in private clinics or hospitals were also excluded. Using the information from previous studies and considering the 10% drop-out rate, the sample size was also calculated to be 335. Out of the questionnaires distributed among the participants, 31 cases were removed because of incompleteness and nine cases were excluded owing to no response, so ultimately 295 questionnaires were analysed (i.e. the response rate of 88%). Since the study was performed during the COVID-19 pandemic, there were more attempts to avoid face-to-face interactions as much as possible and took statistically significant precautions all through unavoidable in-person communications. Therefore, part of the data was collected through online workgroups on WhatsApp. To this end, the group managers were asked to share the digital version of the questionnaires. As well, the nurses who were not part of the groups were given paper questionnaires and asked to return them after 24 h. The research objectives were additionally attached to the questionnaires, and the participants were assured that their answers would remain anonymous and confidential.

2.2 Research instruments

The data collection was performed by a three-part questionnaire. The first part was related to the demographic and occupational characteristics information, including gender, age, marital status, education, work experience, type of employment contract and monthly overtime hours.

The second part was the Turnover Intention Questionnaire (TIQ), developed by Kim et al. (2007), consisting of 15 items measuring intention to leave the profession by a person, on a five-point Likert-type scale (viz., strongly agree = 5, agree = 4, neither agree nor disagree = 3, disagree = 2, strongly disagree = 1). The total score of this questionnaire ranged from 15 to 75. In this study, the mean score of all items was considered as the turnover intention value. Of note, this questionnaire had been already translated into Persian in Iran, and its validity and reliability had been confirmed (Haji & Mohammadimehr, 2021). In this study, the content validity of the Persian version of the questionnaire was established by consulting 12 faculty members at the School of Nursing and Midwifery, affiliated to Urmia University of Medial Sciences, Urmia, Iran, and experts on the subject. The reliability of the questionnaire was further determined to be 0.93 as measured by Cronbach's alpha coefficient.

The third part was the 22-item Negative Acts Questionnaire-Revised (NAQ-R; Einarsen, Hoel, & Notelaers, 2009), measuring a person's exposure to negative acts over the last 6 months, using items designed on a five-point Likert-type scale (namely, never = 1, occasionally = 2, monthly = 3, weekly = 4, and daily = 5). The NAQ-R total score also ranged from 22 to 120, with higher scores indicating higher exposure to horizontal violence. This questionnaire had three dimensions, viz., work-related bullying, person-oriented bullying, and physically-intimidating bullying. As well, Salimi et al. (2014) had reported Cronbach's alpha coefficients of 0.83, 0.91 and 0.85 for the questionnaire's internal consistency in these three dimensions, and Cronbach's alpha coefficient of 0.94 for the questionnaire as a whole. They had similarly verified the questionnaire's construct validity, convergent validity and correlation between its subscales (Salimi et al., 2019).

2.3 Ethical considerations

The study participants were assured of the anonymity and confidentiality of their information and were then asked to provide informed written consent. The study was further approved by the Research Committee of Urmia University of Medical Sciences, Urmia, Iran, with the ethics code of IR.UMSU.REC.1399.199.

2.4 Data analysis

The study data were analysed using the SPSS software package (ver. 19). The qualitative and quantitative data were thus organized in the frequency/percentage and mean/standard deviation formats, respectively. The relationship between the two main variables in this study, namely horizontal violence and turnover intention, was also investigated via the Pearson correlation test, regression analysis and stepwise multiple linear regression. The relationship between the demographic variables and both study variables was also examined through different tests, chosen according to the type of variables. Since the scores obtained for the three dimensions of horizontal violence had various ranges, they were normalized to 0–100 to become comparable. Notably, the statistical significance threshold was considered to be p = 0.05.

5.1 Limitations

The present study was based on self-report data and the respondents' subjective perceptions towards horizontal violence, so the results might not be entirely accurate. In this respect, further research, using different collection methods, such as in-depth individual interviews, can clearly explain the issue. The second limitation was that all participants were working in public hospitals in the city of Urmia, West Azerbaijan Province, Iran, which meant the findings might not have been generalizable to the nurses working in the private sector and those involved in other areas. To address this issue, similar studies need to be conducted in different areas and within all hospitals. The third limitation was that the NAQ-R asked the respondents to recall the negative behaviours they had experienced over the past 6 months, which might lead to a recall bias. Finally, it was impossible to comment on the causality of the observed relationships because of the cross-sectional nature of the data. Therefore, further research should be conducted on the way turnover intention is exactly related to horizontal violence.