Prevalence and health correlates of workplace violence and discrimination – a cross-sectional study among hospital employees in Switzerland

Violence and discrimination are common events at work, especially in the health care setting. Nevertheless, comprehensive data on their prevalence and health correlates among the entirety 15 of hospital staff is missing. This study aimed to estimate the prevalence of different self-reported 16 forms of workplace violence and discrimination among hospital employees in Switzerland and to 17 investigate the association between such experiences at work and the general and mental 18 health status. This cross-sectional study was based on secondary data from a company survey among five 22 public hospitals and rehabilitation clinics in German-speaking Switzerland conducted in 2015/16 23 (N = 1,567). Relative frequencies of different forms of violence and discrimination at work were 24 calculated for the entire study population and for the occupational subgroups. These prevalence 25 data were compared with a representative sample of the general Swiss working population as a 26 reference group. Multiple logistic regression analyses were further computed to investigate the 27 association between the number of different experienced forms of violence and/or discrimination 28 at work and several poor general and mental health outcomes (poor self-rated health, sleep 29 disorders, psychological stress, mental problem, burnout). ratios for strong sleep disorders, strong stress feelings and increased burnout symptoms were 39 between three and four times higher among the most exposed compared to those hospital 40 employees who did not make any of these experiences at work at all. 41 Conclusions 43 Study findings underline the importance of an active combat against violent and discriminatory 44 behaviors in health care. Prevention strategies should particularly focus on nurses and 45 midwives, which turned out to be the most affected and exposed group of all health professions.

affected from workplace violence with a prevalence in an international comparison up to 17% for 67 physical and up to 67% for psychological violence within a year (5, 6). Sexual harassment 68 during the occupational career was reported by every second academic medical faculty women 69 in the United States (7). Previous evidence suggests a prevalence of discrimination in hospital 70 workers around 14% during the past year (8). 71 However, the true extent of workplace violence in the health care sector is difficult to assess, as 72 a high number of unreported cases must be assumed (9). Reasons for systematic 73 underreporting might be that employees do not anticipate a change by the reporting of violence 74 experiences (10) or they underestimate its negative consequences (11). Underreporting is a 75 particular problem in health workers with direct patient contact, as they tend to excuse the 76 behavior of their attackers, for example due to the mental status, pain or emergency situation of 77 their patients (12). 78 There is evidence that workplace violence and discrimination among health professionals have 79 a huge (public) health impact. A systematic review revealed a broad range of negative 80 consequences of workplace violence in the health care setting, which can be divided into seven 81 categories: physical consequences (direct injuries of the body), psychological consequences 82 (as depressive symptoms), emotional affecting (as anxiety), adverse effect on the workplace 83 functioning (as decrease in productivity), negative impact on health care quality (as patient 84 safety), social consequences (on private and family life) and financial consequences (as the 85 loss of income due to absences from work) (13). Similarly, a meta-analysis on the effect of 86 workplace discrimination due to race showed a negative impact on the attitude to work and on 87 loss of productivity in consequence of experiencing violence at work is a serious challenge in 94 health care. 95 In the Swiss Health Survey from 2017 the experience of at least one form of violence or 96 discrimination at work in the previous year was reported by 21.1% of women and 17.5% of men, 97 with an increase in both sexes compared to the previous survey in 2012 (15). 98 Although workplace violence and discrimination have been identified and recognized as a 99 common phenomenon in the health care sector, the prevalence and health correlates of this 100 specific psychosocial risk factor among health care workers are largely unexplored and 101 underresearched, particularly in Switzerland. For this reason, the present study has been 102 conducted, based on survey data collected among hospital employees and particularly health 103 professionals from German-speaking Switzerland. 104 Against the background of the lack of evidence in this regard, this study aimed to investigate the 105 following research questions: 106 1. What is the prevalence of different aspects and accumulated experiences of workplace 107 violence and discrimination among hospital employees and particularly among health 108 professionals in Switzerland, and in comparison with the entire working population? 109 2. Is there consistently a pronounced and negative association and a halfway linear dose-110 response relationship between accumulated experiences of workplace violence and/or 111 discrimination on the one hand and different health outcomes on the other hand among 112 hospital employees? 113 Figure 1 illustrates the theoretical path model of the postulated association between 114 experienced workplace violence and/or discrimination and health status, which is assumed to 115 be potentially confounded by chronic disease. 116

Figure 1 118
Theoretical path model showing the assumed and studied associations between exposure, 119 outcome and confounding variables 120 researchers by a prepaid reply envelope within four weeks, and got a reminder one week before 139 submission deadline. For the statistical analysis, 1567 completed questionnaires were available, 140 corresponding to an overall response rate of 41%. The response rates ranged from 36% to 49% 141 between the participating hospitals. 142 143

Exposure variable(s) 145
Violence and discrimination at work. Experiences of workplace violence and discrimination were 146 measured by the question "Have you experienced the following in the past 12 months at work?" 147 with a note that multiple answers were possible. Ten answer categories were given: 148 "discrimination due to age", "discrimination due to gender", "discrimination due to nationality, 149 ethnicity or skin color", "discrimination due to disability", "verbal violence", "threats and 150 humiliation", "physical violence", "intimidation or mobbing", "sexual harassment", and "none of 151 them". The question and response options were adopted from the Swiss Health Survey, 152 allowing to make a direct comparison with secondary survey data representing the working 153 population of (German-speaking) Switzerland (15). For the association analyses, a sum scale 154 which simply adds up the number of experienced and surveyed different aspects of workplace 155 violence and/or discrimination was constructed. The sum scale starting with 0 and a possible 156 maximum score of 10 was classified into three categories of "none" (0), "a single" (1) and "two 157 or more" (2+). 158 • Self-rated health. Self-reported general health status was measured by asking "How is your 166 health status in general?" with response categories from 1 ("very good") to 5 ("very bad"). 167 Self-rated health (SRH) is an established and well-validated measure of general health, 168 showing a strong association with both mortality (16) and morbidity (17). Due to its strongly 169 skewed distribution, SRH was dichotomized (and binary coded) into two categories, 170 combining answers from "very good" and "good" (value 0) and from "moderate" to "very bad" 171 (1), labeled as "poor SRH", as suggested by the research literature (18). 172 • Sleep disorders. Sleeping problems were assessed by asking the respondents about having 173 had complaints in the past four weeks such as difficulties in falling asleep or sleeping 174 through, with three answer categories dichotomized into "none at all/only a little" (0) and 175 "strong" (1), in order to calculate logistic regression analyses. 176 • Psychological stress. General stress in a psychological (and not physiological sense) was 177 measured by a given definition ("Stress means a condition in which a person feels tense, 178 restless, nervous or anxious or is unable to sleep at night because his/her mind is troubled 179 all the time."), followed by the question: "Did you feel stressed in the past 12 months?" For 180 the analysis, response categories, initially on a five-point Likert Scale from 1 ("not at all") to 181 5 ("very strong"), were then dichotomized and binary coded, distinguishing between 0 "less 182 stressed" ("not at all", "a little", "moderate") and 1 "strongly stressed" ("strong", "very 183 strong"). This single-item measure of general stress is a widely used and well-validated 184 indicator of mental strain (19). 185 • Mental problem. In order to measure a psychological problem, survey participants were 186 directly asked if they had been treated due to a mental problem in the past 12 months, with 187 the answer options "no" (0) and "yes" (1). 0 ("never") to 4 ("always"). The sum score out of these answers of the CBI was calculated 194 and ranged between 0 and 24, with values above 16 being considered as an increased risk 195 of burnout. 196

Confounding variable 198
Chronic disease as a potential confounder was measured by the question "Do you have a 199 chronic disease or health problem?" (yes/no), followed by the explanation that this is a condition 200 which is already lasting or still ongoing for at least 6 months. 201 202

Control variables 203
Sex, age and education were used as control variables. Age was measured by asking about the 204 age category the respondent belongs to (< 25 years, 25-34 years, 35-44 years, 45-54 years and 205 ≥ 55 years). Education was measured by asking participants about their highest degree of 206 education. The 12 given educational qualifications were categorized into four levels of 207 education: 1 "low" (no vocational education), 2 "medium" (basic vocational 208 education/apprenticeship), 3 "high" (higher vocational education or high-school diploma), and 4 209 "very high" (university degree). 210 For stratified analyses or rather differentiated descriptive statistics, study or survey participants 211 were further categorized into four occupational groups (nurses and midwives, physicians and 212 other academics, medical-therapeutic and medical-technical staff, administrative and other 213 service staff). 214

215
Analyses 216 To answer the first research question regarding the prevalence rates of workplace violence and 217 discrimination among health care workers in German-speaking Switzerland, relative frequencies

Descriptive statistics 233
The prevalence of experiences of different forms or aspects of violence and discrimination at 234 work among hospital employees (in the past twelve months) is shown in Table 1. With a look at 235 the entirety of hospital employees, the most frequently reported form of discrimination was due 236 to age (5%), followed by discrimination due to gender (4%), nationality, ethnicity or skin color 237 (3%) and disability (less than 1%). In physicians and other academic staff, discrimination due to 238 gender was the most prevalent form (8%). 239 With regard to experiences of violence at work among the studied hospital employees, 240 intimidation or mobbing was the most commonly reported form (10%), followed by verbal 241 violence (7%), threats and humiliation (5%), sexual harassment (1%), and physical violence 242 (1%). In the working population of German-speaking Switzerland, intimidation or mobbing was 243 also the most frequently reported form of violence (7%). In comparison with other occupational 244 groups, nurses and midwives were by far the most affected from all forms of violence. 245 Overall, almost a quarter (23%) of the surveyed hospital employees reported at least one form 246 of discrimination or violence in the past year, whereby nurses and midwives were most 247 frequently affected (24%), followed by physicians and other academic staff (23%), medical-248 therapeutic and medical-technical staff (21%) and administrative and other service staff (19%).
Hospital employees and particularly health professionals were found to be more frequently 250 affected by experiences of violence and discrimination at work than employed persons and 251 working people in general, which make such experiences at work on average in "only" 18% of 252 the cases. 253 Table 1 255 One-year prevalence of workplace discrimination and violence among occupational groups of 256  Based on weighted and extrapolated data from the Swiss Health Survey 2017

Forms of violence
258 Table 2 illustrates the associations between the experience of workplace violence and/or 261 discrimination and different dimensions of health among hospital employees: After adjusting for 262 sex, age, education (control variables) and chronic disease (potential confounding variable), 263 experiencing one single form of discrimination or violence at work (compared to having not 264 experienced any violence or discrimination) was significantly associated with strong sleep 265 disorders (19% vs. 11%, aOR 2.0), strong psychological stress (19% vs. 11%, aOR 1.7) and 266 increased burnout symptoms (14% vs. 6%, aOR 2.6). These associations were clearly more 267 pronounced when having reported accumulated experiences of workplace violence and/or 268 discrimination, i.e. more than one form. These most affected or exposed hospital employees 269 show almost consistently -although not always significantly -the highest prevalence rate and and adjusted odds ratios as proxies for the relative risk were not significantly increased for the 278 most affected and exposed from the very beginning and in both models. 279 In other words, a strong association and clear and stable dose-response relationship was 280 observed between the number of experiences of workplace violence and/or discrimination and 281 three of the five studied health outcomes. And this relationship was not substantially 282 confounded by chronic disease (extended model) which in turn was found to be a strong and 283 significant risk factor of poor general and mental health outcomes itself.  Odds ratios adjusted for sex, age and education (control variables)

292
One purpose of this study was to assess the frequency of workplace violence and discrimination 293 among hospital employees and particular health professions in a hospital setting in German-294 speaking Switzerland. Almost every fourth (23%) of the study population reported at least one 295 form of discrimination or violence at work in the past 12 months before the survey, compared to 296 only every sixth (18%) in the entire working population of German-speaking Switzerland. With a 297 view to the occupational groups, nurses and midwives were most often affected by violence at 298 work, whereas physicians and other academics were most often affected by discrimination due 299 to nationality, ethnicity or skin color and particularly due to gender. The most frequent form of 300 violence among hospital employees was intimidation or mobbing (10%), whereas ageism was 301 the most commonly reported type of discrimination (5%). 302 Hence, the finding of a comparably high prevalence of workplace violence and discrimination in 303 hospital employees compared with the general working population is in accordance with 304 previous research, which has shown that health care workers are at special risk for workplace 305 violence, as they work with people who are in distress (6). Working in direct patient contact 306 means to be faced with people whose behavior can be affected by acute illness and pain, 307 psychiatric and neurological disorders, intoxications and substance abuse (6, 21). Nurses are at 308 particular risk, as this is usually the professional group temporally most exposed to patients 309 (22). There is also evidence that certain hospital units are more confronted with violence from 310 patients or visitors, such as emergency departments and psychiatric wards (23, 24). Intimidation 311 or mobbing in this study was found to be the most commonly reported form of workplace 312 violence and twice as often than in another study conducted in nursing homes in Switzerland, 313 which found a prevalence of mobbing in the past 6 months of nearly 5% among care workers 314 (25). Regarding ageism, earlier studies showed that discrimination on the grounds of being "too 315 young" is at least as common as on the grounds of being "too old" (26). Although employees of 316 these two age groups are confronted with different prejudices and potential occupational 317 disadvantages, there is evidence that ageism is associated with a lower level of affective commitment in both of them (26). Another finding of this study is that gender discrimination is 319 most commonly reported among physicians and other academic staff. A possible explanation 320 might be that this form of discrimination is becoming increasingly important in employees with 321 higher education. Equal rights for women and men and gender equality is an important concern 322 in politics and policies since many years in Switzerland (27). However, with a look at the Gender 323 Monitoring Report from Swiss University faculties of medicine in 2014, there is still a 324 considerable gender gap in higher positions: While over 50% of medical graduates with a 325 master's degree are women, the proportion drops to 10% on full professor level (28). 326 With a look at the one-year prevalence of violence expected by health personal in an 327 international comparison, a very broad range can be observed, ranging from 3% (Portugal) to 328 17% (South Africa) for physical attacks, from 17% (Portugal) to 67% (Austria) for verbal 329 violence and from 11% (Australia) to 31% (Bulgaria) for mobbing (6). One reason for these 330 large differences between countries might be a limited comparability of the underlying studies, 331 for example in relation to the methodology (study design, definitions used), setting (differences 332 in health care systems, hospital versus outpatient sector, medical specialties), sample (personal 333 characteristics of the study population) and cultural peculiarities (including differences in 334 awareness and reporting systems). 335 336 Besides increased prevalence rates of specific forms of violence and/or discrimination at work 337 and/or among particular occupational groups, accumulated experiences of workplace violence 338 and/or discrimination among hospital employees and particularly health professionals were 339 found to be strongly associated with poor mental health outcomes such as strong sleep 340 disorders, strong stress feelings or increased burnout symptoms. The prevalence (or relative 341 frequency) and the odds or likelihood (or relative risk) of these poor health outcomes were 342 shown to gradually, substantially and significantly increase with the self-reported number of Partly in contrast to this Chinese study, we did not find a significant association of experienced 359 violence or discrimination at work with self-reported health, at least not after adjusting for 360 chronic disease. In other words: Although prevalence rates of poor self-rated health were 361 significantly increased among hospital employees who experience and report at least one form 362 of violence and/or discrimination at work, this was mainly due to their higher proportion of 363 chronically diseased who in turn showed an almost tenfold higher risk of being in poor general 364 and self-reported health than those without a chronic disease. Regarding the other studied poor 365 health outcomes, chronic disease not turned out to be an important confounder, as having a 366 chronic disease only about doubled (and not tenfold increased) the risk of having strong sleep 367 disorders, feeling strong psychological stress, being treated for a mental problem or showing 368 increased burnout symptoms. For these health measures, chronic disease only slightly 369 explained and therefore reduced the strong association found between workplace violence or 370 discrimination and poor health outcomes. But in case of poor self-rated health, a gradually and 371 significantly increased risk with the increasing number of experiences of violence and 372 discrimination at work turned out to be not statistically significant anymore when the association In sum -it is not really surprising but at the same time has not been shown before at least for 375 health care workers in Switzerland -we found that accumulated experiences of violence and 376 discrimination at work are a strong stressor and risk factor for sleep disorders, psychological 377 stress and burnout, even though it does not seem to cause severe mental problems. This study has some limitations that have to be considered with regard to the study results: 395 First, the study is based on cross-sectional data, which do not allow to test for causality. 396 Moreover, reverse causality cannot be excluded either. For instance, a high level of stress can 397 trigger unsocial behavior towards colleagues or impatient behavior towards patients, which 398 themselves could increase the risk of discriminating statements from colleagues or aggressive 399 behavior in patients. This raises the question whether psychological stress is a consequence of 400 discrimination or violence experiences or if a high stress level could also be the starting point of Secondly, the question measuring experiences of workplace violence or discrimination did not 403 distinguish between internal and external violence and did not assess the true extent of the 404 experiences, which would have been helpful for a more accurate estimation of the exposure or 405 strain and for the interpretation of the results. Additionally, there is a risk for potential recall bias, 406 as the question on experienced violence and discrimination refers to a period of 12 months 407 before completion of the questionnaire. 408 In view of the described overall response rate of around 41%, there is a risk for potential 409 selection bias (non-response bias). Also, as participants were allowed to complete the 410 questionnaire during working time, a potential response bias should be taken into account, as 411 people could have answered in a way they considered to be desirable by their company. Having in mind the negative consequences of workplace violence and discrimination on mental 416 health, the present study underlines the importance of an active combat against these 417 undesired but still common behaviors in the health care setting. Managing violence and 418 discrimination at workplace is a challenging task not only for clinical practice, but also for health 419 policy, requiring a holistic approach according to the complexity of these phenomenon. 420 Therefore, prevention and dealing strategies in hospitals should not only focus on organizational 421 factors, but also on the level of the individual employees and their interactions. With regard to 422 the latter, previous studies suggest to provide training programs for health care workers, for 423 conclusions on their identity. For all these reasons, informed and explicit consent from 459 respondents was not needed or obtained but implicitly given by participating voluntarily in the 460

survey. 461
The study and survey were carried out in accordance with the Declaration of Helsinki. The study 462 was exempted from requiring ethical approval because the used data do not fall within the 463 scope of the Swiss Federal Act on Research involving Human Beings whose purpose is to 464 protect the dignity, privacy and health of human beings involved in research. This so-called 465 Human Research Act (HRA) explicitly declares in Art. 2 that it does NOT apply to research 466 which involves anonymously collected or subsequently anonymized health-related data. Since 467 the data used were no register data and did not involve medical records or human tissues from 468 patients but instead were self-reports from employees and collected completely anonymously, 469 so that they cannot be traced to a specific person, no formal approval or authorization of the 470 study is required or will be issued, neither by the cantonal ethics committees nor by the cantonal 471 commissioners for data protection. This is not even recommended by the medical-ethical 472 guidelines for scientific integrity of the Central Ethics Committee and the Swiss Academies of 473

Consent for publication 476
This manuscript does not include details, images, or videos relating to an individual person, 477 therefore no written informed consent for the publication of these details must be obtained from 478 the study participants.

University of Zurich (Epidemiology, Biostatistics and Prevention Institute) and the collaborating 487
Careum Research, a division of the Careum Foundation. As contracted, the use of the data is 488 basically limited to the two research institutions and disclosure and delivery of the data therefore 489 is not permitted. In order to get an exceptional permission and possible conditional access to 490 the survey data for scientific purposes the corresponding author as the principal investigator 491 and the responsible for the data collection needs to be contacted. 492