Abstract
The recent study by Arnetz et al. (2015) concluded that hospital employees underreported incidents of workplace violence mainly because non-victims (witnesses) tended to report incidents far less often than victims. However, this conclusion is not well supported by the data in their study because of unclear definitions of violence, insufficient separation of the participants into the two groups of victims of violence and witnesses of violence, and plausible alternative explanations for the findings, as explained below.
Of the 446 hospital workers surveyed, a flowchart in Figure 1 in the above study indicates that 62% of them (n = 275) answered “Yes, I was a target of violence within the last year.” Of these 275 respondents, the same flowchart shows that 212 of them answered that “I did not report the violence” in the survey. However, when asked about reasons for not reporting incidents of violence (Table 3), 30% of the 212 respondents answered that they were neither a target nor witnesses to violence. The contradicting answers indicate a low reliability for the participants’ determination of whether they were a target of workplace violence. This problem can be addressed by giving the participants of the survey an option to indicate the nature of the violence experienced and their role (e.g., target, witness).
The low reliability regarding workplace violence may have been due to an unclear definition of violence. Although the flowchart in Figure 1 documents whether a participant had been a target of violence, the authors state in the “Methods” section that the question was whether the participant had been a target of either violence or aggression with no definition provided for aggression. The two forms of this question would be expected to produce different responses. Furthermore, the “Results” section reported that the question was whether the participant had experienced a violent incident, presumably including the witnessing of an incident without being a target. The confusing definitions of violence could explain the low reliability of the responses, but this is not the only possible explanation for the low reliability of the responses to the same question at two different places in the survey. Another possible explanation is faulty recall due to the lengthy recall period of 1 year. Recall periods for injuries or similar events, such as hospital stays, have been shown to be unreliable if more than 3 months prior (Moshiro, Heuch, Astrøm, Setel, & Kvåle, 2005; Overpeck et al., 1994).
The study reported that 88% of hospital workers who experienced workplace violence failed to report the incident using the official electronic system, and thus concluded that this finding indicated a need to take action. The estimate of 88% is likely exaggerated for the following reasons. First, one incident could have been witnessed by multiple employees but reported only once by one person in the electronic reporting system. If an employee knows that someone else reported it, they may decide that this is sufficient. This explanation is supported by the study conclusion that employees tended not to report incidents if they were not the victims of violence themselves. By the methods used in the study, this explanation would inflate the numbers of employees who failed to report an incident. Second, the study hospital system only encouraged workers to report via the electronic system, but does not require it. If it is only encouraged, workers cannot be expected to report anything or everything. It is probably unreasonable to conclude that a failure-to-report problem exits when reporting is not mandated. This explanation also would be expected to inflate the failure-to-report percentage.
The study by Arnetz et al. has many strengths including an adequate sample size, use of multiple sites within a hospital system, appropriate statistical analysis, clear discussion of workplace violence literature, and a focus on a critical occupational health and safety problem. Despite these strengths, due to the low response reliability of the key questions and evidence of alternative explanations for the findings, the study conclusion, that hospital employees drastically underreport workplace violence, is not well supported. Instead, an alternative conclusion, that workplace violence reporting may be acceptable, but the value of the electronic reporting system is low among hospital employees, particularly nurses, is proposed.