A longitudinal study of the possible escalation of aggressive behaviors – from bullying and conflicts to workplace violence. Is emotional exhaustion a mediator?

Abstract

BACKGROUND:

Aggression theories and cross-sectional studies imply an escalatory pattern of aggressive behaviors; however, this has not been investigated in a follow-up study.

OBJECTIVE:

To investigate whether bullying or conflicts are antecedents of threats and physical violence, and whether threats mediate the relationship between bullying or conflicts and violence. Lastly, it was explored whether associations could be explained through the effect of emotional exhaustion.

METHODS:

Survey data was collected from a follow-up sample of 3,584 employees from four human service sectors, namely psychiatry, special schools, eldercare, and the prison and probation services. The main analysis uses hierarchical logistic regression.

RESULTS:

The analyses showed that frequent /intense conflicts, not bullying, at baseline were significantly related to higher exposure rates of threats (OR = 4.98, CI = [3.19–7.76]) and violence (OR = 3.01, CI = [1.96–4.76]) at follow-up. Emotional exhaustion was not confirmed as a substantial mediator. However, the proportion mediated by threats was significant (70%) for the relationship between frequent /intense conflicts and violence.

CONCLUSION:

This study finds that aggressive workplace behaviors may indeed escalate, particularly within a similar victim-perpetrator relationship, such as between employees and clients. The study highlights the need for de-escalation techniques that transcend specific encounters, recognizing that aggressive behavior may escalate over time.

Keywords

Violence threats aggression bullying conflicts escalation emotional exhaustion

1 Introduction

Workplace violence is widespread, costly and impacts employees and employers within a variety of work sectors. Reviews have shown that exposure to workplace violence and threats is particularly high in service and human service sectors, such as healthcare, education, public safety, retail, and justice industries [1 –4]. Exposure rates as high as 66.9% for nonphysical violence and 36.4% for physical violence have been found for nurses [4], among special education teachers nearly 40% report verbal abuse [5] and 50% report being kicked, while nearly one third of staff in eldercare are pinched or scratched [6]. The perpetrators were patients, pupils and elder clients respectively.

Workplace violence may be defined as “Incidents where staff are abused, threatened or assaulted in circumstances related to work, including commuting to and from work, involving explicit or implicit challenges to their safety, well-being or health” [7]. Researchers have suggested differentiating aggression and violence, such that workplace violence may be defined as physical aggression (physical assaults or the threat of assault), while aggression is the more general term also encompassing a variety of interpersonal behaviors that may cause psychological harm [8, 9]. Accordingly, all violent behaviors are, by definition, aggressive whereas not all aggressive behaviors are violent. Conflicts may be disagreements about the right treatment or the right way to perform the work whereas bullying acts may be humiliating or ridiculing someone in connection with his/her work.

The current study applies the above definitions of workplace violence in that it must take place in relation to work, and refers to behaviors that explicitly or implicitly imply physical harm, i.e. threats of violence and physical violence, while the general term of aggression refers to a variety of interpersonal behaviors that may cause psychological harm to someone who is motivated to avoid such treatment.

Distinguishing between types of aggressive workplace behaviors, such as bullying, harassment, or violence, is supported in light of studies that find distinct associations with prevalence estimates, antecedents and outcomes (e.g., [4 , 10–13]). Another important distinction regards the type of relationship in question, i.e. insider- or outsider-initiated aggression. Within the literature on workplace violence most studies focus on violence toward employees by service-recipients, such as clients, inmates, customers, or patients [2 , 14]. These perpetrators account for more than 50% of non-physical and physical violence within Anglo and European regions [4]. However, within some domains of aggressive behaviors, such as bullying, studies typically focus on co-worker or supervisor-subordinate relationships (e.g., [15]).

The varieties of domains within aggressive behaviors may reflect an escalation of aggressive workplace behaviors. Escalation in this context does not refer to increased frequency of behavior, but reflects that behaviors will occur in an orderly fashion progressing from less to more intense aggressive behaviors [16 –19]. This could for example reflect a range of aggressive behaviors from being scolded or bullied to threats of assaults and further to physical assault. Limited research exists on the relationship between less and more intense aggressive behaviors at work. However, Glomb (2002) studied employees at manufacturing companies and found evidence of an escalatory pattern to aggression, in which more severe behaviors were preceded by less severe behaviors within a particular incident [17]. Dupre & Barling (2006) also studied potential escalation of aggression and found that psychological aggression toward supervisors was positively associated with physical acts of aggression directed toward supervisors [18]. In addition, they found that psychological aggression partially mediated the relationship between interpersonal injustice and physical aggression, also supporting the notion of a progression of aggression from less serious to more extreme acts [18]. Lanza and colleagues (2006) found that health care workers who had experienced non-physical aggression were 7.17 times more likely to also have experienced physical violence, which may imply escalation [20]. However, these studies focus on specific incidents and/or use cross-sectional designs, which does not allow for conclusions regarding directionality. Moreover, they cannot account for possible escalation due to the accumulation of aggressive incidents over a period of time [16]. This accumulation may be understood in terms of a “triggering mechanism” or “tipping point”, in that one incident after a series of aggravating encounters may cause an employee or client to lose motivation to maintain control over his or her actions. Thus, the tipping point prompts a more intense behavioral response, such as evolving from non-physical aggression to threats of assaults and further to physical assault [16, 20]. In sum, there is need for a follow-up study testing the possible progression of aggressive behaviors and whether less intense aggression mediates more extreme aggressive (violent) behaviors.

Different theoretical frameworks may explain the emergence of aggressive workplace behaviors. According to the Frustration Aggression Theory [21, 22], tense and frustrated employees may show more aggressiveness and by venting negative emotions on co-workers, becoming a perpetrator or instigator of aggressive interactions. On the other hand, frustrations may encourage workplace violence and threats through their indirect effect on the target, as suggested by the social-interactionist framework [23, 24]. Feelings of frustration and being distressed may make it difficult to feign positive emotions and cause less competent work performance and violation of social norms, which others may consider provocative and thus react aggressively toward the distressed person. Punishing the perceived norm violation constitutes the first attack, which may be retaliated in order to deter future attacks, achieve justice and/or save face [23, 25]. Thus, the perception of and reaction to the primary punishment, and whether retaliatory reactions are initiated, may result in an escalatory cycle of aggressive behavior [25]. This interplay is particularly relevant for employees in the human service sectors, where clients or patients may have severe impulse problems due to a variety of diagnoses and/or history of violence. These clients may perceive minor negligence as norm violations and thus react with escalated aggressive behavior, which an employee may sanction in order to regain control and deter unwanted behaviour; however, this may instigate further counter-attacks.

These theoretical frameworks emphasize the mediating role of frustration and distress, making strain responses relevant for understanding aggression. Emotional exhaustion, a dimension of burnout, may be seen as a strain response that resembles job stress, reflecting a depletion of emotional and physical resources [26]. This state of emotional exhaustion may lead to increased depersonalization, which often manifests itself as a withdrawal from work, not only emotionally distancing, but also physically distancing takes place [26, 27]. Although depersonalization conceptually should affect the quality of interpersonal relationships, meta-analysis has shown that exhaustion significantly reduces in-role behavior, organizational citizenship behavior, and customer satisfaction, while results on depersonalization were inconclusive [28]. However, Halbesleben and Rathert (2008) found a positive association between emotional exhaustion and depersonalization with the latter being associated with lower patient satisfaction and longer postdischarge recovery time [29]. These results imply that emotional exhaustion may both have direct and indirect relationships with behavioral outcomes. As such, detecting emotional exhaustion may reflect an early stage in a possible escalating pattern of aggressive behaviors. Moreover, the relationship between emotional exhaustion and reduced customer satisfaction is evidence of possible link between employee strain and client frustration, thus implying conditions for aggression from organizational outsiders (e.g. clients).

Studies show a positive correlation between the time the perpetrator and target spend together and workplace aggression [17, 18], suggesting that long-term relationships with frequent contact, such as working relationships, are more prone to interpersonal conflicts. However, within some human service industries the relationship between employees and clients, patients, students, or inmates are long-term, stretching from weeks to months and even years of daily contact. Results from Lanza and colleagues (2006), showing associations between non-physical aggression and physical violence from patients directed at staff, suggests that client relationships may also involve target-specific escalation [20]. In addition, these researchers found that if the perpetrator of the most recent non-physical aggression was a staff member, the perpetrator of the most recent physical violence was almost as likely to be a patient as a staff member [20]. This result suggests a non-specific increased vulnerability to aggression, as aggression seemingly is displaced among unassociated targets [21, 30]. Again, these results do not allow for conclusions regarding directionality, but they do suggest an association between being exposed to non-physical aggression by employees and exposure to physical aggression by clients.

1.1 Conflicts and bullying as antecedents of workplace threats and violence

Both bullying and conflicts may be defined as an interaction between two individuals, an individual and a group or two groups in which at least one of the parties feels obstructed or irritated by the other [15 , 32]. In terms of aggression, both bullying and conflicts represent interpersonal behaviors that may cause psychological harm to anyone who is motivated to avoid such treatment. Furthermore, both bullying and conflicts are defined in terms of a subjective experience, which does not necessarily have an objective basis [15]. Although, bullying and conflicts share similar characteristics, there are also important distinctions. Workplace bullying, is by definition long-standing escalated behavior and refers to the outcome of a subsequent number of episodes in which negative acts become more frequent over time. Research has found that most often the perpetrator is a co-worker, the supervisor or subordinates [31, 32].

In contrast, conflicts may be more or less escalated and thus the magnitude of frustration will vary [33 –35]. Conflicts may arise between employees, between an employee and a client, between an employee and a supervisor. Moreover, conflicts may be short as well as rather long-standing, including a single episode (e.g., a quarrel regarding unclear procedures that are quickly clarified) or a series of episodes (e.g., a long-lasting debate between a patient and employees regarding no smoking rules) [31 , 34]. Another important distinction is that workplace bullying involves stigmatization of one particular employee, the target, into an inferior position, which hampers the target’s possibilities to counteract the bullying [36]. Conversely, unequal power is not a defining element of conflicts.

According to Buss (1961) aggressive behavior in general may be conceptualized along three dimensions: physical-verbal, active-passive, and direct-indirect [38]. The aggressive content may vary across bullying cases and social conflicts, but the behaviors would more often be defined by verbal behavior, indirect and passive dimensions [15]. In sum, the current study regards both conflicts and bullying as non-physical aggressive behaviors, which may constitute antecedents of even more extreme behaviors, such as threats of violence and physical violence.

1.2 Aims and hypotheses

This study takes on the challenge raised by Barling et al. (2009) [8], who in their review encourage future research to examine whether verbal or psychological aggression can escalate into physical aggression. Thus, the current study aims to explore the hypothesized escalatory pattern of aggressive behaviors by investigating whether bullying or conflicts are antecedents of threats of violence and physical violence, and also, by testing whether threats act as a meditator of the association between less physical and more physical aggression. In other words, we want to examine the relation between conflicts, threats of violence and violence against employees perpetrated by the same kind of perpetrator. In the study population the perpetrators can be pupils with special needs, elder people, psychiatric patients and inmates. Furthermore, this study aims to investigate the possible mediating role of emotional exhaustion, which may constitute a non-specific vulnerability to victimization. As such, we hypothesize that:

Hypothesis 1: Employees exposed to bullying or conflicts will report higher exposure rates of threats and violence at one-year follow-up

Hypothesis 2: The relationship between bullying or conflicts at baseline and exposure rates of violence at follow-up will be mediated by threats.

Hypothesis 2.1: The relationship between bullying or conflicts at baseline and exposure rates of threats and violence at follow-up will be mediated by emotional exhaustion

The findings will also be discussed with respect to non-specific and target-specific escalation of aggressive behaviors.

2 Method

2.1 Design and participants

This study is based on a prospective cohort of employees within four areas of human service sectors: psychiatry, special schools, eldercare, and the prison and probation services. These work sectors were chosen due to a relatively high risk of workplace threats and physical violence and that their service users, i.e. patients, elderly, inmates and special education students, all are likely to have a propensity to low impulse control due to various diagnoses and/or history of violence.

The baseline study, described in [6], consisted of 5497 employees from the above-mentioned work sectors. Eligible for the follow-up study were those employees still at the same worksite, still in jobs with client contact and no more than three weeks absence at the time of survey distribution. Altogether, 3584 participated in the follow-up study with an overall follow-up response rate of 65% (In the elder sector the response rate was 63%, psychiatry 75 %, special schools 71 %, prison and probation services 61%; for details, See flowchart of the cohort in Gadegaard et al. (2015) [39]). The four work sectors were not equally represented in the total sample, with the prison and probation services (n = 1741) representing close to 50% of the full sample, while the other three sectors were more similar in size, i.e. psychiatry (n = 698), special schools (n = 535), eldercare (n = 610).

2.2 Data collection

The baseline data collection took place in the period between May and October 2010 and the follow-up was conducted in the same period in 2011. At both rounds of data collection participants in the prison and probation services received a web-based questionnaire, while other participants received and filled out paper-and-pencil questionnaires during a planned meeting at the worksite. It was stated in the cover letter of the questionnaire that participation in the study was voluntary and that the data would be treated confidentially. In Denmark questionnaire surveys do not require approval by ethic committees. However, the study was approved by the Danish Data Protection Agency, and followed the regulations for data storage and protection. Respondents were identified by questionnaire numbers, which only members of the research group could link to civil registration numbers. This procedure was to ensure accurate matching of questionnaires from the two rounds of data collection.

2.3 Measures

2.3.1 Bullying and conflicts at work

Bullying and conflicts were measured using items from the second version of the Copenhagen Psychosocial Questionnaire (COPSOQ-II) [40]. Respondents were given the following introduction to bullying: “Bullying means that a person repeatedly is exposed to unpleasant or degrading treatment and that the person finds it difficult to defend him or herself against it”. Subsequently, respondents were asked “Have you been exposed to bullying at your workplace during the last 12 months”? One item was used for conflicts and quarrels: “Have you been involved in quarrels or conflicts at your workplace during the last 12 months”? This formulation has a negative connotation such that it excludes those conflicts that may reflect a positive, although intense, social exchange. Response options were similar for both items: Yes, daily; Yes, weekly; Yes, monthly; Yes, a few times; No. Furthermore, respondents were asked:” If yes, with whom? (You may tick off more than one); Colleagues; Manager/supervisor; Subordinates; Clients/patients”. For the statistical analyses, exposure categories were collapsed into the following categories: (1) Frequently exposed (daily-weekly); (2) Occasionally exposed (monthly - a few times); and (3) Never exposed.

2.3.2 Emotional exhaustion

COPSOQ-II [40] also has a dimension referred to as burnout. We used three of the four items: “How often have you been physically exhausted”?; “How often have you been emotionally exhausted”?; “How often have you felt tired”? These items reflect a state of emotional exhaustion as opposed to the depersonalization component of burnout, and therefore we label this dimension as emotional exhaustion. Response options were: All the time; A large part of the time; Part of the time; A small part of the time; Not at all. The scale was scored from 0–100. For all analyses, the scale was measured at baseline. Internal consistence was tested (α= 0.84).

2.3.3 Threats and physical violence

Respondents were asked whether they had experienced the following acts of threatening behaviors at their current worksite within the past 12 months: threats of beatings, written threats, threats over the phone, threats involving objects, and indirect threats (toward family). Similarly, respondents were asked whether they had experienced the following acts of physically violent behaviors: spitting, hitting, hitting with an object, scratching/pinching, shoving, being held, punching with a fist, kicking, biting, having a hard object thrown at you, and use of a weapon or weapon like object [13]. Response categories were: (1) Yes, daily; (2) Yes, weekly; (3) Yes, monthly; (4) Yes, now and then; and (5) No, never. If yes, with whom? (You may tick off more than one); Colleagues; Manager/supervisor; Subordinates; Using this as a continuous measure we saw problems with the statistical assumptions of normality and homoscedasticity, therefore the scale was dichotomized. For the statistical analyses, we chose to dichotomize at the 75th percentile recognizing that in these high-risk work sectors, some exposure to violence or threats is widespread.

2.3.4 Control variables

Reviews on workplace violence have shown that males are relatively more at risk, although this difference is larger for physical than non-physical forms of aggression [1, 2]. Results on whether age is a risk factor are inconsistent [1], however, for the current analyses, we controlled for both gender and age, measured at baseline.

2.4 Data analyses

One-way ANOVAs were performed in order to explore the relationship between levels in emotional exhaustion according to the frequency of bullying and conflicts. The Games-Howell post hoc procedure was used due to unequal group sizes and different population variances.

The main analysis uses hierarchical logistic regression to compute odds ratios (OR) and 95% confidence intervals (95% CI). Statistical significance was judged by whether or not the 95% confidence interval included the value of 1.00. All analyses were adjusted for age and gender. The analysis consisted of three main models. In model 1, conflicts and bullying were tested as antecedents of threats or violence. In case of statistically significant associations, in model 2, we added our potential mediator, emotional exhaustion. In the analyses concerning conflicts and bullying as antecedents of violence, we also tested possible mediation effects of threats (model 2.1). Similar to the work by Clausen et al. (2012) [10], we calculated the proportion of the significant associations that was mediated by our mediating variables by using the following formula: (OR2 – OR1) / (OR1 – 1) where OR1 designates the odds ratio of a given predictor variable in model 1, and OR2 designates the odds ratio of the predictor variable in model 2. Finally, in model 3, we controlled for the dependent variable at T1, thus ruling out the influence of the dependent variable at T1 on increased levels of the dependent variable at T2. The presumed causal chain was further explored by use of longitudinal regression analyses of the association between the antecedents at baseline and mediators at follow-up [10 , 41]. All statistical analyses were conducted using SPSS 21.

3 Results

Of the 3584 respondents, 64.6% were female with an average age of 45; at baseline (T1) (See Table 1). The group with high levels of threats at T1 (n = 826) was slightly lower than at T2 (n = 856). The groups with high levels of physical violence was also slightly lower at T1 (n = 744) than at T2 (n = 772). Approximately 85% at both T1 and T2 had never experienced being bullied, while the reverse was true for conflicts, such that at both T1 and T2 approximately 80% had experienced conflicts. These aggressive behaviours also differed concerning who were involved; about 70% of experiences of bullying were from other employees, while about 70% of experiences of conflicts were from clients (See Table 2). Both threats and physical violence were almost exclusively from clients. Thus, clients are involved in the majority of conflicts and incidents of workplace violence, indicating a similar victim-perpetrator relationship between these types of aggressive incidents. Bullying however, does not involve the same victim-perpetrator relationship as the other types of aggressive incidents, in that bullying most often involves other employees.

Table 1
Descriptive statistics for main study variables (follow-up study sample, N = 3584)

Baseline 2010 (T1) Follow-up 2011 (T2)

Female [n (%)] 2313 (64.6) 1

Age [Mean (SD)] 45 (10) 1

Bullying [n (%)]

Never 2949 (82,3) 2

Occasional 475 (13.3) 2

Frequent 52 (1.5) 2

Conflicts [n (%)] 2

Never 710 (19.8) 2

Occasional 2127 (59.3) 2

Frequent 655 (18.3) 2

Emotional exhaustion [Mean (SD)] 32 (21) 2

Threats, high exposure levels [n (%)] 826 (23.0) 856 (23.2)

Violence, high exposure levels [n (%)] 744 (20.7) 772 (20.1)

	Baseline 2010 (T1)	Follow-up 2011 (T2)
Female [n (%)]	2313 (64.6)	1
Age [Mean (SD)]	45 (10)	1
Bullying [n (%)]
Never	2949 (82,3)	2
Occasional	475 (13.3)	2
Frequent	52 (1.5)	2
Conflicts [n (%)]		2
Never	710 (19.8)	2
Occasional	2127 (59.3)	2
Frequent	655 (18.3)	2
Emotional exhaustion [Mean (SD)]	32 (21)	2
Threats, high exposure levels [n (%)]	826 (23.0)	856 (23.2)
Violence, high exposure levels [n (%)]	744 (20.7)	772 (20.1)

For analytical purposes the dependent variable was dichotomized and when controlling for baseline levels, we matched the categories of the variables measured both at T1 and T2. ¹ = we use a follow-up data. Same participants in both 2010 and in 2011. ² = we only analyzed these variable at baseline.

Table 2

Perpetrators of bullying, conflicts, threats and physical violence [n (%)]

	Clients	Employee(s)	Both employee(s) and clients
Bullying T1	61 (12.0)	354 (69.7)	93 (18.3)
Bullying T2	35 (6.6)	394 (74.1)	103 (19.4)
Conflicts T1	1852 (70.1)	229 (6.4)	562 (21.3)
Conflicts T2	1845 (68.8)	11 (0.4)	826 (30.8)
Threats T1	2197 (97.0)	16 (0.7)	51 (2.3)
Threats T2	2150 (96.5)	21 (0.9)	57 (2.6)
Physical violence T1	1355 (99.4)	0	8 (0.6)
Physical violence T2	1227 (99.4)	3 (0.2)	4 (0.3)

The above categories of clients and employees are mutually exclusive, such that if both employees and clients were ticked by the respondent then they would only be categorized as “both employee(s) and clients”. “Employees” refers to colleagues, manager/supervisor and subordinates.

Results of the ANOVA confirmed the expected relationship that higher frequencies of both bullying and conflicts were related to increased levels of emotional exhaustion (See Table 3).

Table 3

One-way ANOVAs of differences in levels of emotional exhaustion according to the frequency of conflicts and bullying

		Frequency	n	Mean	SD	F	df
Conflicts	0	Never^1,2	709	26.61	18.65	54.71^***	2
	1	Occasional^0,2	2027	32.41	20.28
	2	Frequent^0,1	678	37.90	21.79
Bullying	0	Never^1,2	2917	30.70	19.97	59.71^***	2
	1	Occasional^0,2	469	38.97	21.06
	2	Frequent^0,1	51	52.45	23.71

Numbers 0-1 indicate for the frequency: 0 = Never, 1 = Occasional, and 2 = Frequent. Superscript numbers indicate significant differences by the Games-Howell post hoc test. All variables are measured at baseline.

Table 4 shows the results of longitudinal logistic analyses of the association between bullying and conflicts at baseline and likelihood of threats at follow-up, adjusted for age and gender. Model 1, shows that occasional and frequent conflicts were associated with significantly increased odds ratios for threats, while no associations were significant for bullying. Model 2, shows that mediation effects by emotional exhaustion were minor (– 3.9% and – 5.5%); Model 3, shows that associations were still significant, except for emotional exhaustion, when adjusting for baseline levels of threats.

Table 4

Bullying and conflicts as antecedents of threats (N = 3175). Results of longitudinal logistic regression analyses

	Model 1		Model 2^a			Model 3^b
	OR	95% CI	OR	95% CI	MP	OR	95% CI
Bullying
Never	Ref	–	Ref	–		Ref	–
Occasionally	1.23	0.97–1.56	1.17	0.92–1.49		1.16	0.88–1.53
Frequent	1.23	0.65–2.33	1.04	0.55–1.99		0.76	0.37–1.59
Conflicts
Never	Ref	–	Ref	–		Ref	–
Occasionally	5.57	3.77–8.23	5.39	3.64–7.97	–3.9%	3.14	2.09–4.73
Frequent	18.96	12.61–28.52	17.97	11.93–27.06	–5.5%	4.98	3.19–7.76
Emotional exhaustion	1.01	1.00–1.01	1.00	1.00–1.01
Threats T1
Low						Ref	–
High						11.21	9.09–13.83
Step χ² (df)	469.37 (6)^***		10.43(1)^**			563.59(1)^***
R2 (Nagelkerke)	0.20		0.21			0.42

OR, Odds Ratio; CI, confidence interval; MP = Mediated proportion (OR² – OR¹) / (OR¹ – 1). ^*p < 0.05, ^**p < 0.01, ^***p < .001. All analyses are adjusted for age and gender. ^aTesting emotional exhaustion as a mediator of the relationship between bullying or conflicts and both threats and violence. ^bAdjusting for baseline levels of the dependent variable.

Table 5 shows the results of longitudinal logistic analyses of the association between bullying and conflicts at baseline and likelihood of violence at follow-up, adjusted for age and gender. Model 1 shows that occasional and frequent conflicts were associated with significantly increased odds ratios for violence, while no associations were significant for bullying. Model 2, shows that emotional exhaustion was significantly associated with violence, and that mediation effects were again minor (– 3.3% and – 4.2 %). Model 2.1, however, shows substantial partial mediation by threats (– 51.3% and – 69.9%), while emotional exhaustion no longer was significantly associated with violence. Further, model 2.1 revealed an association between frequent bullying and significantly reduced odds ratios for violence (OR = 0.34). Model 3, shows that associations were still significant when adjusting for baseline levels of violence.

Table 5

Bullying and conflicts as antecedents of physical violence (N = 3135). Results of longitudinal logistic regression analyses

	Model 1		Model 2^a			Model 2.1^b			Model 3^c
	OR	95% CI	OR	95% CI	MP	OR	95% CI	MP	OR	95% CI
Bullying
Never	Ref	–	Ref	–		Ref	–		Ref	–
Occasionally	0.88	0.68–1.14	0.85	0.66–1.12		0.78	0.60–1.06		1.03	0.74–1.42
Frequent	0.53	0.25–1.12	0.47	0.22–1.01		0.34	0.15–0.76	–27.7%	0.23	0.09–0.55
Conflicts
Never	Ref	–	Ref	–		Ref	–		Ref	–
Occasionally	3.72	2.62–5.26	3.63	2.56–5.14	–3.3%	2.28	1.58–3.28	–51.3%	1.77	1.19–2.63
Frequent	16.41	11.36–23.70	15.76	10.89–22.80	–4.2%	5.44	3.65–8.11	–69.9%	3.01	1.96–4.76
Emotional exhaustion			1.01	1.00–1.01		1.00	1.00–1.01		1.00	1.00–1.01
Threats T1
Low						Ref	–		Ref	–
High						7.90	6.39–9.76		2.70	2.08–3.50
Violence T1
Low									Ref	–
High									15.32	11.86–19.80
Step χ² (df)	425.74(6)^***		4.40(1)^*			389.72(1)^***			479.86(1)^***
R2 (Nagelkerke)	0.19		0.20			0.35			0.52

OR, Odds Ratio; CI, confidence interval; MP = Mediated proportion if OR > 1 then (OR² – OR¹) / (OR¹ – 1), but if OR < 1 then (OR² – OR¹) / (OR¹), All models are adjusted for age and gender. ^*p < 0.05, ^**p < 0.01, ^***p < 0.001. ^aTesting emotional exhaustion as a mediator of the relationship between bullying or conflicts and both threats and violence. ^bTesting baseline threats as a mediator of the relationship between bullying or conflicts and physical violence. ^cAdjusting for baseline levels of the dependent variable.

Exploring the relationships between antecedents and mediators, longitudinal regression analyses (not shown) showed that associations with emotional exhaustion at follow-up were similar to results in Tables 4 and 5, thus showing significant OR’s of 1.00 to 1.03 and 95% CI’s of 1.00–1.01 to 1.01–1.04. Moreover, Table 4 shows that the antecedents at baseline were significantly associated with threats at follow-up.

4 Discussion

The aim of this study was to explore whether non-physical aggression, defined as bullying or conflicts, were antecedents of threats of violence and physical violence, while also testing for the potential mediating effect of emotional exhaustion. The results showed that conflicts at baseline were significantly related to higher self-reported exposure rates of threats and violence at follow-up, while this was not the case for bullying. Thus, partly giving support for hypotheses 1. Moreover, threats did show substantial mediation and was thus confirmed as a partial mediator of the relationship between conflicts and violence (in support of hypothesis 2). The latter result is consistent with the cross-sectional finding by Dupre & Barling (2006) [18], who found that psychological aggression partially mediated the relationship between interpersonal injustice and physical aggression. The current findings suggest an escalation effect, in that behaviors may progress from less to mere extreme, i.e. from conflicts to threats and further to violence. The findings also show a dose-response relationship in that higher frequencies of conflicts are associated with higher odds ratios for threats and violence. This indicates a cumulative effect of aggressive behavior. These findings are in line with and expand existing evidence on escalating aggression [17, 18].

One possible explanation for the finding that bullying was not associated with higher levels of threats or violence may be that bullying primarily involved other employees. In contrast, conflicts in this study involved a similar victim-perpetrator relationship as violence and threats. The associations between conflicts and threats/violence may therefore reflect a more closed circuit of interpersonal escalating and reciprocal behaviors. This implies that frustration from one source (bullying from employees) was not displaced against unassociated targets (clients). As such, this follow-up study could not replicate the findings from the cross-sectional study by Lanza & colleagues (2006), which suggested an association between being exposed to non-physical aggression by employees and being exposed to physical aggression by clients [20]. This may be explained by differences in measures and study populations. However, the current results did confirm that non-physical violence is a risk factor for physical violence in the client-employee relationship.

The accumulation of aggressive incidents over time may involve specific challenges for employees in the human service sectors. These employees may perceive client’s aggression as a result of illness [42], which as mitigating information may lessen the degree of frustration, negative affect and aggressive tendency [21]. However, qualitative results from one study on nurses showed that after experiencing aggression many times it became more difficult to see aggression as an illness. And after a while they would use more stern replies to this behavior [42]. The time span of the present study suggests that this accumulation and possible escalation of behavior may persist over an extensive time frame.

In the human service sectors, verbal or physical limit setting for clients is an inherent task, which according to both the Frustration Aggression Theory [21, 22] and the social interactionist framework [23, 24] may constitute an aversive event for the client, which he/she may react to with escalated aggression. Risk of assault has been associated with situations such as enforcing smoking rules, requesting patients to go to certain areas or not eat or drink certain foods, and also assisting with activities of daily living (ADL) [43 –45]. This suggests that irrespective of employee distress, these working situations may involve risk of conflicts and escalation to violence.

Although escalation pertains to reciprocal behaviors, in the client relationship there may not be equal responsibility for the exchange of behaviors. In contrast to co-worker relationships, the de-escalation of a conflict may rest more on one party of the exchange, namely the employee. Thus, irrespective of the instigation or cause of the conflict, employees may be required to competently de-escalate the situation [46]. However, results from the current study should be interpreted with caution, so as to not further assign responsibility of victimization on the victim but rather focus on the role of the workplace in preventing future aggression from clients. As such, workplaces may offer training in verbal and physical management of clients [47]. Moreover, worksites should have violence prevention policies and practices, where formal reporting of incidents are encouraged and taken seriously [48 –50]. Also, victims of aggression should receive social support from supervisors and co-workers, thus alleviating strain responses [51, 52]. Thus, the current findings may be used by organizations to inform effective prevention by recognizing the accumulating and escalating nature of aggression in the client relationship.

As expected, this study found that experiences of bullying and conflicts were associated with increased levels of emotional exhaustion, although surprisingly, this did not convincingly explain the increased risk of threats or violence (Hypothesis 2.1). With regard to conflicts, the lack of significant mediation may be due to insufficient levels of emotional exhaustion in order to influence behavior. In fact, the mean level for the emotional exhaustion scale was 32, while the most intense level for conflicts was about 37.9. In contrast, bullying experiences showed by far more taxing levels of emotional exhaustion. These differences in strain may be due to different opportunities to cope with the negative encounters. In a client-related interpersonal conflict, an employee may refer to a professional understanding of the client’s problems (cognitive appraisal) [53], while also getting support from co-workers, thus buffering strain responses [51, 54]. In contrast, bullying experiences may obstruct sources of intra-organizational support and induce negative feelings of being ostracized, thus intensifying strain responses [15, 37]. However, even rather high levels of emotional exhaustion related to bullying at baseline did not convincingly mediate the risk of threats or violence at follow-up. This may reflect that targets of bullying, although experiencing high strain, may have qualitatively different client relationships in comparison to employees involved in ongoing conflicts with clients. Thus, for targets of bullying from co-workers it may be a relief to be in a position of somewhat power and control, as inherent in the client relationship, in contrast to being trapped in an inferior position in relation to colleagues and/or supervisors [37]. However, future studies should further explore the qualitative aspects of the caregiver-client relationship when an employee is bullied by co-workers.

Surprisingly, the current study found that frequent bullying was associated with significantly decreased risk of violence, after adjusting for threats and baseline violence. In terms of burnout, this may reflect an effect of depersonalization, i.e. distancing or withdrawal from work, both emotionally and physically [26, 55]. This could entail frequent or prolonged absenteeism from work [10, 56], declining certain tasks and/or declining to deal with certain aggressive clients. However, for the current study we did not include the measure of depersonalization since the translation of this dimension into Danish has proven problematic [57]; although, it is possible that emotional exhaustion at baseline evolved into depersonalization at follow-up [26, 55]. Furthermore, some evidence suggests that the combination of emotional exhaustion and depersonalization is related to less dominant behavior, which paradoxically may lead to less escalation in conflict situations [58]. Thus, this detached behavior may become an effective strategy by acting in a neutral, less dominant and less directive way, thereby preventing escalation and ending a negative interaction. Another explanation may be that a supervisor is trying to protect and take care of a vulnerable employee by not requiring that employee to do tasks with more inherent risk of violence. In terms of the social-interactionist framework [23, 24] this withdrawal or favorable treatment may be considered a nuisance for co-workers, who may react with intensifying the bullying behavior. Thus, while this may entail less exposure to violence from clients, it may escalate bullying behaviors from co-workers, thereby intertwining these domains of aggressive behavior. However, more research is needed to unravel issues regarding upward and downward spiraling across domains of aggressive behaviors and type of perpetrator.

The assumption of less and more severe (intense) aggressive behavior warrants a few comments. Studies have found that bullying as opposed to threats and (non-fatal) violence may have more severe consequences [10, 11], however, for the current purposes the focus was on the possible escalation from non-physical aggression (bullying and conflicts) to violence. Therefore, in the current context, escalating severity does not refer to potential consequences, but only to the progression of behaviors from non-physical aggression to physical aggression, in which behaviors become more physical, active and direct [38].

5 Strengths and limitations

A significant strength of the present study is the use of a follow-up design with acceptable response rates (Flowchart described in [39]). This reduces bias among measures and provides support for causal interpretations. Moreover, although an optimal mediation analysis requires three data points, the current study explored follow-up relationships between antecedents at baseline and mediation variables at both baseline and follow-up [41].

Another strength is one-year follow up. Research has shown that too long or short time span between baseline and follow up increases the risk of recall errors. In relation to work environment, previous research found that the best fit for analytic models was one-year time lag [59].

In all sectors the response rates were relative high. However, if only employees with interest in the topic under study responded this may have inflated the associations. An additional analysis showed no difference in gender, work experience and education between responders and non-responders at baseline in the largest sector (Prison and Probation Service) (data not shown).

However, the study also has limitations. The fact that 50% of the participants were from the prison and probations services questions whether the associations are more relevant for this sector than the other three sectors. However, as this study investigates behaviors that are more individual-specific than context-specific, we believe that the associations are relevant for the entire sample. An additional sector specific analysis confirms this (data not shown), but due to lower number in each analysis, the associations didn’t remain statistical significant. However, the directions of the associations remain the same. Moreover, although our main predictors (bullying and conflicts) did not correlate substantially (r_s= 0.15, p < 0.01), as we adjusted for baseline levels of the dependent variables, we found correlations between threats and conflicts (r_s= 0.49, p < 0.01). Although, this still leaves 50% of unique variance, it does entail some multicollinearity, in which some of the shared effect is subsumed in the stronger association. Furthermore, we examined the effect of stratifying as opposed to adjusting for baseline levels of the dependent measure. This did not change the overall results; however, in the case of occasional bullying we did find differences for high and low baseline exposure groups, but these results were still not significant. A further limitation is the use of dichotomizing measures, i.e. threats and violence, which means that we lose some information and therefore we are perhaps simplifying the relationships in question. However, dichotomizing also has some advantages in that measures, such as the odds ratio, often are realistic and meaningful measures of strength of relationship [60]. Moreover, the measures do not allow for conclusions about who is the primary instigator or who escalates the aggressive exchange; this entails subjective accounts of perceived retaliatory reactions [25] and observed behavior. Future studies may use interview and observational reports to account for the processes between feelings of frustrations and escalation of aggressive behavior on part of both the employee and client. Lastly, the optimal time span between baseline and follow-up in relation to workplace aggression is unknown. Studies on escalating negative behaviours have used cross-sectional aggregate measures [18] or analyses of critical incidents [17]. While the analyses of specific encounters is valuable in understanding situated escalating behaviour, the assumption of the accumulation of negative encounters and tipping points warrants a longer timeframe. We chose a one-year follow-up since this has been recommended in relation to psycho-social work environment and mental health [59].

6 Conclusion and Implications

As far as we know, this is the first study using a follow-up design showing that aggressive workplace behaviors may progress from non-physical aggression (conflicts) toward physical aggression (threats of violence and violence). Furthermore, this study suggests that there is some domain specificity with regard to aggressive behaviors, such that escalation of behaviors is more likely within the same circuit of interpersonal relationships. This is contrary to the notion that strain responses may mediate a general vulnerability to future victimization irrespective of the source of frustration.

These findings imply that when working in the human service industries, particularly with clients that have low impulse control, it is essential to train conflict management. This pertains to de-escalating specific encounters, while the current results further suggest that de-escalation also should cover a longer timeframe. The latter may entail identifying patterns of problematic client-employee relationships, which allows for pro-active prevention efforts, such as practicing non-aggressive communication, before the rise of a new conflict, in an effort to break a vicious pattern of escalating aggressive behavior. In practice, this requires that managers encourage reporting of minor incidents and take these reports seriously as a means to prevent workplace violence.

Conflict of interest

None to report.

Footnotes

Acknowledgments

The authors would like to thank the participating worksites for their involvement in the study.

This work was supported by the Danish Working Environment Research Fund under Grant number 20-2009-03.

References

Hogh

, Viitasara

. A systematic review of longitudinal studies of nonfatal workplace violence. European Journal of Work and Organizational Psychology. 2005;14(3):291–313.

Piquero

, Piquero

, Craig

, Clipper

. Assessing research on workplace violence, 2000–2012. Aggression and Violent Behavior. 2013;18(3):383–394.

Renker

, Scribner

, Huff

. Staff perspectives of violence in the emergency department: Appeals for consequences, collaboration, and consistency. Work. 2015;51(1):5–18.

Spector

, Zhou

, Che

. Nurse exposure to physical and nonphysical violence, bullying, and sexual harassment: A quantitative review. Int J Nurs Stud. 2014 1;51(1):72–84.

Tiesman

, Konda

, Hendricks

, Mercer

, Amandus

. Workplace violence among Pennsylvania education workers: Differences among occupations. J Saf Res. 2013;44:65–71.

Rasmussen

, Hogh

, Andersen

. Threats and physical violence in the workplace: A comparative study of four areas of human service work. J Interpers Violence. 2013;28(13):2749–69.

Wynne

, Clarkin

, Cox

, Griffeths

. Guidance on the prevention of violence at work. 1997.

Barling

, Dupré

, Kelloway

. Predicting workplace aggression and violence. Annu Rev Psychol. 2009;60:671–692.

Schat

, Kelloway

. Workplace aggression. Handbook of Work Stress. 2005:189–218.

10.

Clausen

, Hogh

, Borg

. Acts of offensive behaviour and risk of long-term sickness absence in the Danish elder-care services: A prospective analysis of register-based outcomes. Int Arch Occup Environ Health. 2012;85(4):381–7.

11.

Clausen

, Hogh

, Carneiro

, Borg

. Does psychological well-being mediate the association between experiences of acts of offensive behaviour and turnover among care workers? A longitudinal analysis. J Adv Nurs. 2013;69(6):1301–13.

12.

Matthiesen

, Einarsen

. Psychiatric distress and symptoms of PTSD among victims of bullying at work. British Journal of Guidance & Counselling. 2004;32(3):335–56.

13.

Menckel

, Viitasara

. Threats and violence in Swedish care and welfare. Magnitude of the problem and impact on municipal personnel. Scand J Caring Sci. 2002;16(4):376–85.

14.

Vidal-Martí

, Pérez-Testor

. Patient assaults on professionals. Type II workplace violence: A known phenomenon in Barcelona nursing homes. Work. 2015;51(4):879–86.

15.

Matthiesen

, Einarsen

. Bullying in the workplace: Definition, prevalence, antecedents and consequences. International Journal of Organization Theory & Behavior. 2010;2018/09;13(2):202–48.

16.

Andersson

, Pearson

. Tit for tat? The spiraling effect of incivility in the workplace. Academy of Management Review. 1999;24(3):452–71.

17.

Glomb

, Liao

. Interpersonal aggression in work groups: Social influence, reciprocal, and individual effects. Academy of Management Journal. 2003;46(4):486–96.

18.

Dupré

, Barling

. Predicting and preventing supervisory workplace aggression. J Occup Health Psychol. 2006;11(1):13.

19.

Schilpzand

, De Pater

, Erez

. Workplace incivility: A review of the literature and agenda for future research. J Organ Behav. 2016;37:S57–88.

20.

Lanza

, Zeiss

, Rierdan

. Non-physical violence: A risk factor for physical violence in health care settings. AAOHN J. 2006;54(9):397–402.

21.

Berkowitz

. Frustration-aggression hypothesis: Examination and reformulation. Psychol Bull. 1989;106(1):59.

22.

Fox

, Spector

. A model of work frustration-aggression. J Organ Behav. 1999;20(6):915–31.

23.

Felson

. “Kick’em when they’re down”: Explanations of the relationship between stress and interpersonal aggression and violence. Sociological Quarterly. 1992;33(1):1–16.

24.

Felson

, Tedeschi

. Aggression and violence: Social interactionist perspectives: American Psychological Association; 1993.

25.

Felson

, Steadman

. Situational factors in disputes leading to criminal violence. Criminology. 1983;21(1):59–74.

26.

Leiter

. Coping patterns as predictors of burnout: The function of control and escapist coping patterns. J Organ Behav. 1991;12(2):123–44.

27.

Maslach

, Schaufeli

. Historical and conceptual development of burnout. Professional burnout: Recent developments in theory and research Philadelphia, PA, US: Taylor & Francis, Philadelphia, PA; 1993, pp. 1–16.

28.

Taris

. Is there a relationship between burnout and objective performance? A critical review of 16 studies. Work & Stress. 2006;20(4):316–34.

29.

Halbesleben

, Rathert

. Linking physicvian burnout and patient outcome: Exploring the dyadic relationship between physicians and patients. Health Care Management Review. 2008;31:29–39.

30.

Dollard

, Miller

, Doob

, Mowrer

, Sears

. Frustration and aggression. New Haven, CT, US: Yale University Press, New Haven, CT; 1939.

31.

Baillien

, De Witte

. The relationship between the occurrence of conflicts in the work unit, the conflict management styles in the work unit and workplace bullying. Psychol Belg. 2009;49(4):207.

32.

Miller

. Theory and experiment relating psychoanalytic displacement to stimulus-response generalization. The Journal of Abnormal and Social Psychology. 1948;43(2):155.

33.

Garden

. How to Resolve Conflict in Organizations: The Power of People Models and Procedure: Routledge; 2018.

34.

Baillien

, Escartín

, Gross

, Zapf

. Towards a conceptual and empirical differentiation between workplace bullying and interpersonal conflict. European Journal of Work and Organizational Psychology. 2017;26(6):870–81.

35.

Liu

, Wang

, Chang

, Shi

, Zhou

, Shao

. Work–family conflict, emotional exhaustion, and displaced aggression toward others: The moderating roles of workplace interpersonal conflict and perceived managerial family support. J Appl Psychol. 2015;100(3):793.

36.

Einarsen

, Raknes

BrI

, Matthiesen

. Bullying and harassment at work and their relationships to work environment quality: An exploratory study. European Journal of Work and Organizational Psychology. 1994;4(4):381–401.

37.

Zapf

, Gross

. Conflict escalation and coping with workplace bullying: A replication and extension. European Journal of Work and Organizational Psychology. 2001;10(4):497–522.

38.

Buss

. The psychology of aggression. Hoboken, NJ, US. 1961.

39.

Gadegaard

, Andersen

, Hogh

. Effects of violence prevention behavior on exposure to workplace violence and threats: A follow-up study. Journal of Interpersonal Violence. 2015.

40.

Pejtersen

, Kristensen

, Borg

, Bjorner

. The second version of the Copenhagen Psychosocial Questionnaire. Scand J Public Health. 2010;38:8–24.

41.

Taris

, Kompier

. Games researchers play— Extreme-groups analysis and mediation analysis in longitudinal occupational health research. Scand J Work Environ Health. 2006:463–72.

42.

Åkerström

. Slaps, punches, pinches? But not violence: Boundary-Work in nursing homes for the elderly. Symbolic Interaction. 2002;25(4):515–36.

43.

Bensley

, Nelson

, Kaufman

, Silverstein

, Shields

. Patient and staff views of factors influencing assaults on psychiatric hospital employees. Issues Ment Health Nurs. 1995;16(5):433–46.

44.

Lanza

, Kayne

. Patient assault: A comparison of patient and staff perceptions. Issues Ment Health Nurs. 1995;16(2):129–41.

45.

Lanza

. Factors relevant to patient assault. Issues Ment Health Nurs. 1988;9(3):239–57.

46.

Fauteux

. De-escalating angry and violent clients. Am J Psychother. 2010;64(2):195–213.

47.

Wassell

. Workplace violence intervention effectiveness: A systematic literature review. Saf Sci. 2009;47(8):1049–55.

48.

Chang

, Eatough

, Spector

, Kessler

. Violence-prevention climate, exposure to violence and aggression, and prevention behavior: A mediation model. J Organ Behav. 2012;33(5):657–77.

49.

Kessler

, Spector

, Chang

, Parr

. Organizational violence and aggression: Development of the three-factor Violence Climate Survey. Work & Stress. 2008;22(2):108–24.

50.

Yang

, Spector

, Chang

, Gallant-Roman

, Powell

. Psychosocial precursors and physical consequences of workplace violence towards nurses: A longitudinal examination with naturally occurring groups in hospital settings. Int J Nurs Stud. 2012;49(9):1091–102.

51.

Schat

ACH

, Kelloway

. Reducing the adverse consequences of workplace aggression and violence: The buffering effects of organizational support. J Occup Health Psychol. 2003;8(2):110–22.

52.

Leather

, Lawrence

, Beale

, Cox

, Dickson

. Exposure to occupational violence and the buffering effects of intra-organizational support. Work & Stress. 1998;12(2):161–78.

53.

Chapman

, Styles

, Perry

, Combs

. Nurses’ experience of adjusting to workplace violence: A theory of adaptation. Int J Ment Health Nurs. 2010;19(3):186–94.

54.

Frese

. Social support as a moderator of the relationship between work stressors and psychological dysfunctioning: A longitudinal study with objective measures. J Occup Health Psychol. 1999;4(3):179.

55.

Maslach

, LEITER

. Burnout. Stress and Quality of Working Life: Current Perspectives in Occupational Health. 2006:37.

56.

Ortega

, Christensen

, Hogh

, Rugulies

, Borg

. One-year prospective study on the effect of workplace bullying on long-term sickness absence. J Nurs Manag. 2011;19(6):752–9.

57.

Kristensen

, Borritz

, Villadsen

, Christensen

. The Copenhagen Burnout Inventory: A new tool for the assessment of burnout. Work & Stress. 2005;19(3):192–207.

58.

Euwema

, Kop

, Bakker

. The behaviour of police officers in conflict situations: How burnout and reduced dominance contribute to better outcomes. Work Stress. 2004;18(1):23–38.

59.

De Lange

, Taris

, Kompier

MAJ

, Houtman

ILD

, Bongers

. The relationships between work characteristics and mental health: Examining normal, reversed and reciprocal relationships in a 4-wave study. Work & Stress. 2004;18(2):149–66.

60.

Farrington

, Loeber

. Some benefits of dichotomization in psychiatric and criminological research. Criminal Behaviour and Mental Health. 2000;10(2):100–22.