Abstract
Workplace aggression can escalate from verbal to physical, adding risk of physical injury to negative psychological impact. Understanding escalation is necessary to develop effective violence intervention/prevention programs, but its occurrence is not well documented. An online survey determined if and how aggression escalates in occupational therapy (OT) workplaces. Because local culture can influence workplace events we compared surveys from northeastern states with the lowest rates of homicide to southeastern states with the highest. Ninety four OTs working in southeastern or northeastern states reported details of the most recent verbal and/or physical aggressive event initiated by clients (NIOSH Type 2) or staff (NIOSH Type 3) they witnessed or experienced in the workplace within the last year. Aggressor status, motivation, actions; therapist reactions, injuries and psychological sequelae were queried in logical order. Behavior inter-correlation followed by cluster analysis determined if levels of escalation could be detected. Analyses grouped events into five escalation levels in which progressively more hostile or damaging behaviors were added at each successive level. Higher levels involved known risk factors: Younger, reactively angry clients with prior aggression histories escalated attacks to involve restraint, calls for help and physical injury. Southeastern aggressor/target interactions appeared more combative, seemingly regardless of personal risk. Remarkably, northeastern OTs reported significantly more positive and fewer negative feelings across levels of escalation. OT workplace aggression occurs frequently, can escalate to injurious levels and requires prevention/intervention. How therapists maintain positive attitudes in the face of violence deserves further study.
Keywords
Introduction
Aggression in healthcare workplaces is all too common. A remarkable 93% of all assaults against employees in the US, and 75% of all aggravated workplace assaults, occur against health personnel (Phillips, 2016). Staff in an Australian brain injury ward were nearly four times more likely to need time off from aggression-related injuries than from other types of injury (Kerr et al., 2017). Aggression also occurs at elevated rates in mental health (Fry et al., 2002) and dementia-care facilities (Pulsford & Duxbury, 2006). The greater the frequency and severity of aggression, the greater the psychological trauma (Rippon, 2000) and career consequences. A dozen studies have found that 13%–60% of victims of violence in healthcare workplaces thought about quitting their jobs; 1%–21.2% actually did (Lanctôt & Guay, 2014). Witnessing violence also has a negative psychological impact (Zhou et al., 2017).
While numerous studies document aggression risks for other health professionals, less is known about experiences of occupational therapists (OTs). Reports from samples of 7–600 OTs in clinics and hospitals in Australia, Canada, the UK and US do indicate that 40–55.2% had been verbally and/or physically attacked in the last 12 months and up to 98% had been assaulted at least once during their careers (Beaulieu, 2007; Brown & Rivard, 2016; Demir et al., 2014; Dyrkacz et al., 2012; Fry et al., 2002; Kerr et al., 2017; Mullan & Badger, 2007; Schneider et al., 1999; Swain et al., 2014; Whittington et al., 1996). Bullying may be even more prevalent. Scanlon and Piersol (2021) documented American OTs experiences with disrespect and intimidation from nurses, physicians and other health professionals they worked with. They noted that such unfortunately common events are under-recognized by OTs and should be highlighted as a first step in addressing them. Clearly, aggression in OT workplaces requires more effective intervention, especially to identify and forestall events that can escalate to the level of physical injury. Unfortunately, a recent Cochrane review found evidence on intervention to be low quality (Spelten et al., 2020). De-escalation programs must be based on understanding what actually happens. The aim of this survey was to begin to provide such information.
The survey was organized to address several important issues, starting with the fact that different forms of aggression are perpetrated by individuals of different status with different motivations. Striking out by an anxious or agitated patient with dementia may be a form of defense while attacks on OTs by cognitively intact patients/family members (Rippon, 2000) may be angry reactions to treatment procedures, outcomes or costs. These forms of “reactive” aggression contrast with “proactive” aggressive acts, like bullying by co-workers or supervisors (Brown & Rivard, 2016; Demir et al., 2014). (For review of distinctions and concomitants of current aggression typologies, see Bobadilla et al., 2012). Because intimidation and interpersonal aggression arise out of “organizational cultures” in which OTs are disrespected (Scanlon & Piersol, 2021) and assaulted (Andersen et al., 2018), we assessed this factor by asking participants to rate their work environment on a scale that tapped the dimensions of such cultures. However, organizational culture may also incorporate attitudes of the larger surrounding milieu. The American southeast has consistently higher homicide rates than other parts of the country (Thomas et al., 2018). These rates are associated with generally higher levels of violence that reflect “hypermasculine” vigilance for any intended disrespect and aggression toward such threat when it is perceived in what has been called the Southern “culture of honor” (Miner & Smittick, 2016). OTs everywhere strive to maintain their workplaces as therapeutic havens in which both clients and therapists can feel safe. However, these efforts must contend with the influence of regional culture. Homicide rates are systematically reported by reliable sources, readily available and are unambiguous markers of overall levels of violence (Harries, 1997). To determine if aggressive behavior within OT clinics might be influenced by regional factors, we compared experiences of OTs in five contiguous northeastern states with the lowest US homicide rates to those in five contiguous southeastern states with the highest rates.
We also recognized that aggressive events begin with a verbal interchange that sometimes escalates to some level of physical violence as each act provokes a counter-response (Felson, 1982). The resulting higher prevalence of verbal than physical aggression is true for OT workplaces, too. Across the above-mentioned studies that covered the last year’s experience, a median 43% of respondents reported experiencing verbal aggression versus 21% who reported physical aggression. In the career-long retrospective studies, a median 73% of respondents had experienced verbal aggression versus 57% who had experienced physical aggression.
These data imply that aggression does escalate in OT workplaces. Significant correlations between verbal and physical aggression have been reported (Kerr et al., 2017). However, correlations alone do not capture the potentially hierarchic nature of escalation in which verbal elements like curses, threats and screaming may appear alone or in escalated combination with physical acts like throwing things, hitting and kicking. Rather than impose pre-determined groupings and classifications on the reported behaviors we used a sequence of correlation and hierarchical cluster analysis, a method of “unsupervised classification” (Jain et al., 1999), to determine if levels of escalation would emerge naturally from the behavioral data. When they did, as described below, we ascertained whether escalation was related to reactive versus proactive aggressive motivation and/or preceding emotional state. We also assessed the role in escalation of aggressor sex (because males are more physically aggressive, Archer, 2019), age range (because physical aggression prevalence peaks in late adolesence, Fahlgren et al., 2020), relative size (because larger men tend to be more aggressive, Hartin et al., 2020), and history of aggression (because previous aggression predicts future aggression, Tardiff, 2008). These findings in the literature were the basis of specific hypotheses that: (1) Perpetrators of escalated aggression would be large, late adolescent-young adult males with prior histories of aggression, (2) that escalated events would involve greater physical harm and negative emotional reactions and that (3) higher levels of escalation might occur in southeastern states.
Method
Survey Development
The survey was developed in three stages, each involving input from a different set of OTs in clinical practice. In Stage 1, four OTs from Minnesota were interviewed in depth about their experiences of aggression on the job. Pilot survey questions were then developed that framed these experiences in terms of the typologies and processes of aggressive behavior as noted above as well as information derived from the published surveys of OTs. For example, our list of aggressive behaviors was drawn in part from the Conflict Tactics Scale (Straus et al., 1996), which itemizes acts among adults in families. The list of responses to aggression was drawn in part from reactions to acts by dementia patients, e.g., effects on caregiver distress and confidence (Hansen et al., 2020; Pulsford & Duxbury, 2006). In Stage 2 seven OTs who had specific experience with aggression in adult/geriatric OT workplaces completed a pilot survey based on Stage 1 information. These OTs provided feedback on the format of questions included in the survey, how the questions related to their own experiences, and suggestions about items that were unclear or had been omitted. In Stage 3 the revised pilot survey was completed by 13 OTs who had been in practice from 3 months to 36 years. Their question completions and comments were used to determine final survey content and format.
Survey Format, Content and Access
Anonymously responding OTs first provided demographic information about their gender, SE or NE workplace location and number of years working as an OT. They then indicated the overall recency and frequency of aggressive events in their workplace using roughly geometric, easy to understand intervals, e.g., …1–3/month, 1–2/week, 3–6/week, 1–2/day… They rated their organizational culture by moving a slider on a continuous scale from 0 (“supportive/collaborative/respectful”) to 100 (“unsupportive/noncollaborative/disrespectful”). Responding to questions formatted as either single forced choice, “check all that apply” or fill-in, respondents then provided details of the most recent aggressive face-to-face and electronically mediated events they had witnessed or been involved in, thus tapping their freshest memories. Per Gideon (2012), questions followed a natural sequence of recall characteristics of those involved including aggressor’s sex, age, status (client, family, co-worker, supervisor) and, if a client, known diagnosis. No questions were asked about race because of the likelihood that focusing on race would bias the recall of events (Levinson, 2007; Silverstein & Flamenbaum, 1989). Respondents reported the specific verbal and physical aggressive behaviors that occurred, duration of the event and physical and emotional consequences. Face-to-face aggressive events were assessed in the first 36 questions, electronically-mediated aggression (phone calls, texts, emails, video-conferences) in the last 18. The latter section included a question about event frequency using the same geometric intervals as above. The survey was hosted online via the University of Minnesota supported Research Electronic Data Capture (REDCap) software (Harris et al., 2009). The survey is available as a PDF from the first author.
Participants and Regions
The five contiguous SE states with the highest homicide rates in 2018 were Alabama, Arkansas, Louisiana, Mississippi, and Missouri (9.1–13.4 homicides/100,000, National Center for Health Statistics, 2017). The contiguous NE states with the lowest rates were Maine, Massachusetts, New Hampshire, Rhode Island and Vermont (0–2.3 homicides/100,000). Between June–August 2019 1000 postcards inviting recipients in each set of states to complete the on-line survey were mailed to OTs at addresses obtained from rental lists of the American Occupational Therapy Association selected from Special Interest Groups of Community Health, Mental Health, Productive Aging, Rehabilitation and Disabilities and a wide range of work settings and practice areas. To increase return rates, we offered a $5 Starbucks gift card and short summary of results when they became available. Participants linked to the survey through a URL listed on the postcard. The University of Minnesota IRB determined that this survey was exempted from their review (3/19/2019). The Assurance of Compliance number is FWA00000312.
Aggression triggers, antecedent states and the Reactive-Proactive Aggression Index
From the literature and the 24 experienced OTs consulted during survey design we identified 12 likely triggers or motivations for aggression in the OT workplace, ranging from reactions to overstimulating environments, invasions of personal space, being teased or harassed or objecting to participating in OT activities through proactive demands for food, objects or activities to attempts to control or dominate the situation or to bully someone. Because participants were to choose “all that applied” and motivations for aggression can be mixed and complex, we simplified the assortments of triggers into the relative proportions of nine reactive motivations and four proactive motivations with a Reactive/Proactive Aggression Index (R/PI) as follows: R/P Index = (R - P)/(R + P) where R = reported number of reactive aggression triggers and P = number of proactive triggers.
Then, for only proactive triggers, R/PI = −1.0. For equal numbers of proactive and reactive triggers, R/PI = 0. For only reactive triggers R/PI = 1.0
Respondents also reported the aggressor’s physiological state preceding the event including hunger and/or pain (Bushman et al., 2014). His/her emotional state was scored as: Anxiety: Absent = 0, present = 1 Agitation: Absent = 0, confused/disoriented/agitated = 1, spacey/delusional/hallucinating = 2 Anger: Absent = 0, uncooperative/defiant = 1, irritable/angry = 2
Target/OT Reactions and Injuries
Respondent Feelings Associated With the Aggressive Event.
Correlation and Cluster Analyses
To reduce the 22 aggressive behaviors listed to the ≤6 variables recommended for a cluster analysis of <128 cases, per the rule of m variables for 2m cases (Formann, 1984), the behaviors were first grouped by an intercorrelation matrix tempered by judgments of likely co-occurrence so that the groupings maintained psychological face validity. For example, verbal threat was combined with “staring/glaring” and “standing too close” into a “Threat” grouping although this group is a mix of verbal and physical behaviors. Thus, each event was represented as a set of scores on six grouping variables with the score for each variable being the number of those behaviors reported as having occurred, e.g., 0–3 for Threat. Cluster analyses were then carried out using SPSS V25.
We then tested the purely behavioral escalation levels that were identified for differences in sex, age, relative size and status of aggressors; their known diagnoses (if a client), motivations (per the R/PI Index) and emotional state preceding the attack. We also tested for inter-level differences in target behavioral and emotional reactions. Frequency tests, parametric and non-parametric ANOVAs were used as appropriate. Note that the many variables generated by this survey invites analysis of multiple inter-relationships, with a concomitant increase in the risk of Type 1 statistical error. For this reason data analysis was generally restricted to assessing differences among levels.
Inconsistency Indices
Expecting that mental effort and recall accuracy would vary across respondents, we created 14 cross-checks on internal consistency. Thus, a report that workplace aggression occurred 1–2/day on average, but the most recent aggressive event occurred a month ago would be inconsistent. Similarly, reporting that both verbal threats and throwing occurred (Q27), but that the event consisted of verbal aggression only (Q28) would be inconsistent. Because estimating frequency and recency requires some mental calculation this consistency check was treated separately; the other consistency checks were combined into a 0–13 Inconsistency Index that was used to assess overall levels of inaccuracy and to identify the least consistent respondents.
Results
Return Rate and Demographics
Respondent Demographics.
Workplace columns show percent of respondents employed in that type of facility. Memory care facilities and group homes were workplaces for <4% of respondents in both regions. No respondents were employed in Veterans Affairs facilities or hospices.
Reliability, Inconsistency Indices, Recall Certainty and Severity Ratings
A frequency-recency scatterplot showed six respondents to have reported high aggression frequencies but discrepantly long times since the most recent event. Inconsistency scores among the 94 surveys were low, ranging from 0 – four out of a possible 13. The plurality of surveys (43%) had only one inconsistency; six surveys had three or four inconsistencies.
Face to face and electronically-mediated aggression correlated significantly both in frequency (r = 0.49, p < .001) and recency (r = 0.42, p < .001). Fifty nine percent of respondents reported being very certain about their recall, 34% were somewhat certain; their degree of certainty was independent of subsequently determined escalation level. Severity ratings were normally distributed around the mid-scale “about average.”
Correlation and Cluster Analysis of Events
Two groups of verbally aggressive behaviors (V1, V2), a group of threat (T) behaviors and three physically aggressive behavior groups (P1 – P3) were formed from the correlation matrix with 2–5 behaviors/group (see Figure 1 legend for details). Within-group correlation r’s ranged from 0.15–0.5. Confidence in the validity of these groupings was bolstered by finding that behaviors within each group were similarly prevalent: verbal behaviors were most common, threat behaviors less common and physical acts least common. Percent of events with one or more behaviors of each of the six behavior groups for. Each Escalation Level. Read each Level row left to right. V1: Sarcastic, hostile, harsh voice; talk over, cut off V2: Yell/scream; verbal abuse T: Verbal threat, stare/glare, stand too close P1: Damage property, throw things P2: Pinch/claw/scratch, harm self, push/poke/pull hair, spit, kick P3: Slap/hit/punch, choke, head-butt White-gray-black column fill indicates increasing harmfulness of behavior group. Hidden values for Cluster Level 4: V1 = 33.3%, V2 = 55.6%.
There were 30 events with verbal behavior only and 59 events with verbal behaviors and at least one physically threatening or assaultive behavior (just five events of physical aggression without verbal aggression were reported). A hierarchical clustering using Ward’s method with squared Euclidean distances yielded the most interpretable results with clusters being extracted at first, second or third nodes of the dendrogram such that there were at least nine surveys per cluster. Crucially, a hierarchy of escalation levels emerged from this analysis in which each level added additional behaviors to those of the level below. As Figure 1 shows, Level 1 events mostly contain only the mildest verbal behaviors (V1), represented by the leftmost white column. Progressively darker columns, indicating increasingly more hostile, threatening, or damaging behaviors, are added on the right of the behavioral profile at each successive level. Thus, 100% of Level 3 events also include threats (T, gray column) while 40% of Level 4–5 events also include the most violent physical behaviors (P3, black columns). This is how aggression escalates.
Status, Sex, Age, and Diagnosis
Seventy six percent of all conflicts between staff occurred at Levels 1–2, meaning they were mostly verbal exchanges. The R/P Index showed that proactive aggressive motivation, including bullying, predominated only in inter-staff conflicts and only at Levels 1–3 (−0.21 > R/P-I > - 0.34). Reactive aggression predominated in the three Level 4–5 events between staff and at all levels in client/family precipitated attacks on staff (0.5 ≥ R/P–I ≥ 0).
Escalation Levels 3 and 5 had more male aggressors than did Levels 1–2 but sex composition did not differ significantly across levels. In contrast, there were significant age effects; 8 of the 14 Level 5 events were committed by the youngest (≤29 yrs) aggressors [1-way ANOVA F(4,89) = 4.16, p < .005]. Clients and family were aggressors in 55% of events; staff in 39%. However, clients and family were responsible for a disproportionately significant 83% of Level 4–5 events (Fisher Exact test, p < .01). Among clients the four most frequently reported conditions were anxiety/depression, cognitive impairment, dementia and schizophrenia, together accounting for 50% of reported diagnoses. Diagnoses were not associated with escalation level.
Aggressor History, Event Trigger, Preceding State and Event Duration
Between 58%–75% of aggressors in Level 1–3 were reported to have histories of occasional or frequent aggression. In contrast, all Level 4–5 perpetrators had such previous histories. This difference among levels was significant [Friedman nonparametric ANOVA χ 2(2) = 9.1, p < .01].
Across all levels, just four aggressors were perceived as hungry and three in pain prior to the event. Prior emotional states of agitation, anxiety and anger were more prevalent and influential. Slightly more than half of aggressors showed at least one of these states. The combined emotional state score increased significantly across levels (1-way ANOVA F(4, 93) = 3.75, p < .01). Among the three states, anger was the most frequent and systematic driver of escalated aggression (data not shown).
Median event duration ranges at different levels were 1–2 or 3–5 min. Event duration and recency were unrelated to level, so escalation analysis was not biased by reporting of more recent or longer events
Target Responses and Injuries
Targets’ most common response was trying to de-escalate the conflict verbally; around a third of respondents at Levels 1–2 and a quarter of those at Levels 4–5 reported such attempts. Level-related shifts in other reactions were more pronounced. Ignoring or arguing with aggressors became less common across escalation levels (Spearman correlations of proportions of these two responses with Level were −.8 and −.82, respectively) while restraining the aggressor and calling for help became more common (Spearman correlations with Level were .72 and .8, respectively.) Correspondingly, the fraction of targets injured increased across levels, from 1/19 in Level 1 to 9/14 in Level 5. Level 1–2 injuries included scratches/scrapes/cuts, bruises/black eyes, and being knocked off balance. Levels 4–5 included all of those as well as torn muscles or ligaments/broken bones and concussions.
Regional Comparisons
There were no differences between the northeast (NE) and southeast (SE) in aggression frequency, recency or duration. Nonetheless, some notable regional differences in behavior and feelings occurred. On the instigator side, 83% of NE aggressors were large and/or strong for their sex and age, as might be expected from the usual size advantage in conflicts (Hartin et al., 2020). In contrast, 50% of SE aggressors were average in size/strength with the remainder evenly split between large/strong and small/weak, as if aggressors’ own size and strength had been disregarded in deciding to attack. This regional difference was highly significant (Fisher Exact Test p < .001). On the recipient side, six of the 52 NE targets argued with the aggressor; one of these was also reported to fight back. Thirteen of 42 SE targets argued and/or fought back, indicating significantly greater SE combativeness (Fisher Exact Test p < .02). Correspondingly, 36% of SE targets versus 19% of NE targets were hurt, resulting in a trend toward differences in injury rates [χ 2 (1) = 3.2, p = .07].
Respondents experienced mixes of negative and positive self and workplace-related feelings during and after the event. On all three questions NE OTs reported more positive feelings and fewer negative feelings, on average, than SE OTs based on the percentage of total possible feelings of each sort. When combined the three questions yielded a significant interaction between geographic region and overall proportions of positive and negative feelings (2 factor ANOVA with repeated measures on the Feelings factor [F(1,92) = 8.0, p < .01]). Both SE and NE OTs recalled proportionately more negative than positive feelings during Level 1 events. Contrary to our prior hypotheses, across higher levels NE OTs recalled progressively fewer negative self-feelings and more positive self-feelings during the event as shown by the respective descending and ascending slopes of those functions in Figure 2. To examine relationships among positive and negative feelings, escalation level, and region, we fit separate proportional odds logistic regression models for positive and negative feelings, with number of feelings as ordered categorical responses and escalation level, region, and level-by-region interaction terms as predictors. The upward slope of NE positive feelings and the downward slope of NE negative feelings were each significantly different from 0 (p < .02, p < .01, respectively). Neither SE slope was different from 0. The NE positive feeling slope was significantly steeper than the SE positive feeling slope (p < .05). Percentage of all possible positive and negative feelings about the aggressive event in the northeast (NE – thick lines, triangle markers) and southeast groups (SE-thin dotted lines, circle markers) across the five escalation levels.
Organizational Culture Ratings
The mean score of 34.4 (±25.5) indicates that respondents generally experienced their workplaces as supportive. Importantly, there were no differences in these ratings across escalation levels [1 way ANOVA F(4,76) = 1.08, NS]. So, OTs’ experience of organizational culture is unrelated to aggression escalation. Mean regional ratings were 35.7 ± 27.2 for NE, 33.2 ± 23.9 for SE. OTs’ organizational culture experience did not differ by region [t(78) = 0.44, NS].
Discussion
Among the methodological strengths of this survey are findings that, as expected, respondents’ estimates of the frequency and recency of face-to-face aggression were inversely correlated and frequencies of face-to-face and electronically mediated aggression were positively correlated. That is, both respondents and descriptions of their work environments appear consistent. Correspondingly, rates of internal inconsistency were low overall; eliminating the few respondents with problematic inconsistencies did not affect any of the results. These observations, together with high ratings of recall certainty, increase confidence in overall data reliability and credibility of the results. Similarly, the normal distribution of severity ratings around the mid-scale “about average” means that the escalation levels model is largely built on the most commonly occurring events in different work environments.
A major finding was that progressive levels of escalation did emerge from the “unsupervised classification” of aggressive behaviors by hierarchical cluster analysis. This was not a foregone result and suggests that aggression escalation is robust. Importantly, the identification of escalation levels was supported by systematic and interpretable relationships between levels and several highly aggression-relevant variables. Overall, younger, reactively angry clients with a prior history of aggression perpetrated the most escalated events in which targets had to restrain the aggressor and/or call for help and in which they were most likely to be injured. There are no surprises here. Youth, prior history and anger are all well known risk factors for aggression. These familiar outcomes indicate that the respondents were largely reliable reporters and cluster analysis was a valid approach. The news is that all this happens in OT workplaces.
The lack of differences in event duration across levels suggests that physical acts are not just added onto the end of verbal interchanges. Perhaps resorting to physical aggression actually truncates conflicts. More generally, the escalation pattern described here admits of at least two possibilities. On one hand, the behavioral organization of levels suggests that most conflicts develop along a continuum of escalation. In this context, we note that three-quarters of inter-staff conflicts (NIOSH Type 3) remained at verbal-only escalation levels 1–2, suggesting that most clinical personnel exercise the self-control necessary to prevent escalation (Denissen et al., 2018). However, a quarter of staff conflicts did escalate to include threat and some of these to physical attack while 42% of client/family-initiated conflicts (NIOSH Type 2) remained at levels 1–2. This substantial overlap in escalation level distributions suggests considerable commonality in the progression of inter-staff and client-staff conflicts. On the other hand, the shift from proactive to reactive aggression in inter-staff conflicts across levels implies that there may be different dynamics of aggressive interaction from onset to final level, i.e., conflicts that eventually escalate to high levels may begin differently from those that do not escalate so far. These alternatives have different implications for aggression prevention, so should be investigated further.
Ratings of organizational culture were unrelated to escalation and did not differ between northeast and southeast, but regional differences in aggressive interaction were observed. The apparent willingness of SE aggressors to attack regardless of their relative size, together with the significantly greater fraction of SE targets who argued, fought back, and were injured, might well indicate a generally more combative attitude. Such combativeness might entail suspicions about others’ hostile motives, an aspect of Southern honor culture which might also account for the lower SE positive attitudes and greater negative attitudes (Vandello et al., 2008). It might also reflect the prior degree of caring experienced by the respondents, with those caring more prior to the event being better able to preserve their feeling for clients after the event (Forté et al., 2017). However, the apparent regional differences could also relate to the somewhat higher percentage of males among the SE respondents, their shorter time on the job and the differences in workplace type. The biggest surprise was NE OTs increasingly positive attitude in the face of greater challenges. These laudable responses may protect against intrapersonal stress and impaired job performance.
Limitations and Future Studies
One reason for the relatively low response rate is that it was not possible to issue more than one invitation to each OT. However, this return rate is consistent with OSHA’s (2016) estimate that only 10% of workplace aggression events are reported. Among the unknowns contributing to non-response, OTs who have been subjected to workplace aggression may have felt that the events they experienced were either too trivial to report or too traumatic to recall (Lanctôt & Guay, 2014). If so, then non-response biases might only affect the extremes of the event distribution without altering conclusions about more common type of aggressive encounters.
The percentage of males among our respondents was somewhat higher than their 9.1% representation in the OT workforce (American Occupational Therapy Association, 2020). This may mean males are more likely to consider aggressive events important enough to take the time to complete a survey and/or that they are more likely to be involved in such events. Males were not over-represented among the inconsistent reporters nor among those reporting higher escalation events, so there is no significant evidence of bias in their reporting. However, true bias assessment would require comparing reports of male and female OTs from the same clinics.
There was only one report of an event with a sexual component; perhaps OTs are reluctant to report such events. Sexualized events might have a different escalation trajectory than that documented here, so would require a different sort of survey. The temporal dynamics of anger and aggression are complex (Potegal, 2010) and the questions raised about common versus multipily different escalation trajectories might be best answered in a prospective study in which participants were pre-trained to note the timing, or at least the order of occurrence, of behaviors.
While our sample of 94 is modest for a survey, it is still above the median number of OTs queried in previous studies of workplace aggression. To the best of our knowledge, this is the first survey to address aggression escalation in OT workplaces so our findings do need replication. The existence and nature of regional differences especially requires re-examination. If our results are replicated and SE OTs are indeed at higher risk it will be important to determine whether it is a matter of less restrained, more aggression–prone clients, combative therapist attitudes or a combination of these that present barriers to de-escalation. This is a potentially sensitive issue and we caution against any premature conclusions pending further investigation and extensions to other areas of the country, which would be necessary for geographic generalization. In any event, if future studies replicate NE OTs' increasingly positive attitude in the face of greater challenges, it should be determined if these attitudes protect against intrapersonal stress and impaired job performance.
Conclusions and Implications
In focusing on escalated/injurious aggression, this study differs from previous work on verbal bullying within organizational cultures. We intended to provide information that would be useful in the development of aggression prevention/intervention programs. What has been learned? Reaction to perceived provocation, not proactive bullying, is most clearly associated with higher level escalation. Anger is a more consistent predictor of incipient aggression than anxiety or agitation. Angry young male clients with histories of aggression present the highest risk of injurious attack. Aggression in OT workplaces escalates rapidly from verbal to physical, making it important to identify developing threats early and initiate de-escalation procedures promptly. Regional differences appear salient. In the northeast an angry person’s larger size is a clue to whether he will escalate but in the southeast his size may not help in estimating risk. Arguing and fighting back may increase the risk of injury. The responses of the NE OTs suggest that positive attitudes can be maintained in the face of aggression challenge. Learning how they manage to do this could be an important component in developing de-escalation programs for the OT workplace.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The Occupational Therapy Program of the University of Minnesota provided some support for participant recruitment and statistical consultation through its Strategic Initiative Fund.
