Workplace violence towards nurses in Hong Kong: prevalence and correlates

Student summary

Why This Research Matters

Workplace violence (WPV) against nurses is a recognized occupational hazard worldwide, but Hong Kong had few recent population-based studies on the topic when Teris Cheung and Paul S. F. Yip conducted this cross-sectional survey, published in BMC Public Health in 2017. The researchers wanted to know how common WPV was among Hong Kong nurses and what factors were linked to experiencing it.

The study used a web-based questionnaire sent to members of the Association of Hong Kong Nursing Staff in late 2013. Of the roughly 16,000 members with email accounts, 850 registered nurses aged 18-65 completed the survey, a response rate the authors themselves described as relatively low. The questionnaire combined items from an ILO/ICN/WHO/PSI workplace violence toolkit, translated and validated for a Chinese-speaking population, with the Depression Anxiety Stress Scale-21 (DASS-21) to measure psychological symptoms, plus questions on job satisfaction, shift patterns, clinical rank, and history of deliberate self-harm.

The headline finding was that 44.6% of nurses had experienced some form of workplace violence in the year before the survey. Verbal abuse or bullying was by far the most common form (39.2%), followed by physical assault (22.7%) and sexual harassment (1.1%). Patients were the most frequent source of violence (36.6%), followed by patients' relatives (17.5%), colleagues (7.7%), and supervisors (6.3%). Male nurses reported slightly higher rates of WPV (48.6%) than female nurses (44.0%); this modest difference may reflect differences in clinical role assignment or reporting behaviour rather than a true difference in exposure.

Using multivariate logistic regression, the researchers identified several factors independently associated with experiencing WPV: rotating shift work was the strongest predictor, followed by job dissatisfaction, a history of deliberate self-harm, recent conflict with colleagues, symptoms of anxiety, and clinical position (front-line nurses had lower odds of reporting WPV than charge nurses, possibly reflecting different patient contact patterns or reporting behaviour). Roughly three-quarters of nurses who experienced WPV described a moderate or severe personal impact.

For students, this study is a useful illustration of how survey-based occupational health research works, including its practical limits. The 5.3% response rate is genuinely low, and the authors themselves flag it as a source of possible selection bias, meaning nurses who felt strongly about the topic (either because they had been affected or because they cared about it) may have been more likely to respond. The sample also had far more male nurses proportionally than the actual Hong Kong nursing workforce, which required statistical weighting to correct. Because the study is cross-sectional, it can show that certain factors are associated with WPV, but it cannot prove that shift work or job dissatisfaction causes violence to happen, only that they occur together.

What makes the paper valuable for Canadian nursing students is the parallel with Canadian data: national surveys here have also found alarmingly high, and likely underreported, rates of both physical and emotional workplace violence against nurses, with patients as the leading source. The Hong Kong findings reinforce a now-familiar picture across health systems: WPV is common, underreported, often normalized as "part of the job," and linked to psychological distress among nurses. The authors call for zero-tolerance workplace policies, easier and safer reporting mechanisms, de-escalation training built into nursing curricula, and psychological support for affected staff. When appraising a study like this, students should always weigh the strength of the evidence (a cross-sectional, low-response-rate survey) against how consistent its findings are with the broader international and Canadian literature on the same problem, which in this case strengthens rather than undermines the overall message that WPV against nurses deserves serious institutional attention.

Faculty notes

Educational Relevance

Cheung and Yip's 2017 BMC Public Health paper offers a useful teaching case in occupational health epidemiology among nurses, combining descriptive prevalence estimation with correlate identification via multivariate logistic regression. The study surveyed 850 registered nurses in Hong Kong (745 female, 105 male) recruited via mass email invitation through the Association of Hong Kong Nursing Staff (approximately 16,082 potential recipients), yielding a response rate of only 5.3%. Data collection used a Chinese-translated, content-validated adaptation of the ILO/ICN/WHO/PSI workplace violence survey instrument (test-retest reliability 0.85) alongside the DASS-21 to capture depression, anxiety, and stress symptomatology, and items on job satisfaction, shift pattern, clinical rank, chronic illness, and deliberate self-harm history.

Headline prevalence: 44.6% of respondents reported experiencing WPV in the prior 12 months, with verbal abuse/bullying (39.2%) far exceeding physical assault (22.7%) and sexual harassment (1.1%). Patients (36.6%) and patients' relatives (17.5%) were the dominant perpetrator categories, exceeding intra-staff sources (colleagues 7.7%, supervisors 6.3%) — a pattern instructors can use to distinguish external, care-relationship-driven violence from internal, workplace-culture-driven violence, each warranting distinct mitigation strategies.

The adjusted logistic regression identified six independent correlates: rotating shift work (the strongest association, aOR approximately 2.67), job dissatisfaction (aOR approximately 1.72), history of deliberate self-harm (aOR approximately 1.66), recent workplace conflict with colleagues (aOR approximately 1.41), anxiety symptoms (aOR approximately 1.48), and clinical rank, with front-line nurses showing lower adjusted odds than charge nurses (aOR approximately 0.34). This is a good opportunity in class to discuss confounding and directionality: does anxiety predispose nurses to perceive or experience more conflict, or does experiencing WPV generate anxiety symptoms? The cross-sectional design cannot resolve this, and the authors themselves invoke negative-affectivity theory as one plausible interpretive lens without claiming causal proof.

Methodologically, this paper is valuable for teaching critical appraisal of survey research. The 5.3% response rate is low even by the standards of nursing workforce surveys and creates a real risk of self-selection bias in either direction (over-representation of nurses strongly affected by violence, or conversely of those with strong professional engagement generally). The achieved sample also over-represented male nurses relative to the Hong Kong nursing workforce, which the authors addressed through weighting — a good springboard for discussing sampling weight methodology and its limits in correcting for non-response bias, which weighting cannot fully address. Recruitment through a single professional association's email list also limits generalizability to non-members and nurses without institutional email access.

For discussion, faculty may want to connect this study to Canadian occupational health data (e.g., CFNU national surveys reporting very high rates of nurse-experienced violence, and provincial studies such as those on British Columbia nurses), since both contexts show patient- and family-driven violence as the leading source and consistent underreporting as a systemic barrier to accurate prevalence estimation. The paper's practice recommendations — cognitive-behavioural reframing of violence as unacceptable rather than an occupational norm, accessible and protected reporting channels, zero-tolerance institutional policy, and curriculum-embedded de-escalation training — map directly onto competencies increasingly expected in Canadian nursing education and regulatory frameworks around psychological health and safety in the workplace.