This study employed a quasi-experimental pre-test–post-test design conducted among security personnel at CMH. The design involved a single group of participants who were assessed before and after a structured educational intervention, without a control group. This approach was suitable for evaluating the impact of training in a real-world hospital setting, where randomisation was not feasible.

The study was conducted at CMH, located in Mdantsane, Buffalo City Metropolitan Municipality, Eastern Cape province, South Africa. Cecilia Makiwane Hospital is a regional academic hospital affiliated with Walter Sisulu University and Lilitha College of Nursing, providing a wide range of clinical and allied health services, including psychiatry, emergency medicine, surgery, internal medicine, obstetrics, paediatrics, and trauma care.

The hospital houses the only public-sector mental health unit serving the East London region, receiving frequent referrals of aggressive and violent patients from community clinics, correctional facilities, and law enforcement agencies. While private mental health facilities exist within the broader region, this study focused exclusively on the public healthcare sector, where resource constraints and patient volumes present unique challenges for the management of aggression and violence. Security personnel at CMH are responsible for maintaining safety and preventing violence, and they play an important role in containing incidents of aggression, thereby making the hospital an appropriate site for this intervention.

The study targeted all security personnel employed at CMH (n = 170). A census-based convenience sampling approach was employed, whereby all eligible and available security personnel were invited to participate in the data collection. This approach was appropriate given the finite size of the workforce and the exploratory nature of the intervention.

Both newly employed and more experienced security personnel were eligible for inclusion. Years of work experience were captured as a demographic variable and categorised to assess baseline attitudes and examine whether experience influenced post-intervention changes.

A formal sample size calculation was not performed, as the intention was to recruit the entire accessible population of security personnel at the study site. The final sample size was therefore determined by staff availability, operational requirements, and willingness to participate. Security personnel at the hospital are responsible for protecting hospital property, ensuring the safety of staff and patients, and responding to incidents involving aggression or violence.

Of the total workforce, 80 security personnel did not participate: 3 were off sick during the data collection period, 12 were unavailable because of shift patterns, 5 were unavailable because of operational demands, and 60 declined to participate. Consequently, 90 participants were enrolled and completed both the pre- and post-intervention assessments.

The educational programme was informed by established international literature and clinical guidelines on the non-pharmacological management of aggression in healthcare settings. Core components of the intervention were grounded in the biopsychosocial model of aggression and evidence-based de-escalation frameworks, which emphasise early identification of agitation, verbal de-escalation, empathetic communication, and minimisation of coercive practices. These principles align with the National Institute for Health and Care Excellence (NICE) guideline on violence and aggression,²² the consensus statement on verbal de-escalation by Richmond et al.,²⁵ and thematic syntheses of effective de-escalation techniques described by Price and Baker²³ and Price et al.²⁷ The overall structure and focus on non-coercive management strategies are further supported by systematic reviews evaluating de-escalation training in healthcare settings.^24,28,33

The programme was delivered over approximately 4 h and conducted on two separate days to accommodate alternating work shifts without disrupting hospital operations. Training was facilitated by the principal investigator, a psychiatry registrar with a Diploma in Mental Health, under the supervision of a consultant psychiatrist.

The intervention comprised two structured phases, guided by a standardised training pamphlet. Phase 1 focused on understanding the nature, causes, and progression of aggressive behaviour. Participants were introduced to different forms of violence, the anger escalation cycle (incident, escalation, crisis point, and recovery), and early warning signs of aggression, including verbal threats, intimidating posture, and behavioural cues. Emphasis was placed on differentiating behaviour from the individual, recognising situational triggers, and interpreting the social and environmental context in which aggression occurs.

Phase 2 focused on the practical application of de-escalation strategies. Participants were trained in calm verbal communication, active listening, validating emotions, empathy, and respectful engagement, as well as awareness of nonverbal cues, including tone, body posture, and physical proximity. Guidance was also provided on recognising situations in which de-escalation may be unsafe or ineffective, such as severe agitation or an imminent risk of harm. The overarching objectives were to prevent escalation, stabilise potentially volatile situations, and reduce reliance on coercive or physical interventions.

Teaching methods included brief didactic presentations, facilitated group discussions, scenario-based exercises, and role-play demonstrations to promote experiential learning and the integration of skills. These methods were selected in accordance with adult learning principles and prior evaluations of de-escalation training in healthcare environments.^25,27

Data were collected using the MAVAS, a validated instrument developed to assess staff attitudes towards the causes and management of aggression in healthcare settings. The MAVAS has demonstrated acceptable construct validity and internal consistency, with reported Cronbach’s alpha coefficients ranging from approximately 0.60 to 0.80 across its domains.¹² The instrument has been widely applied in international studies involving healthcare professionals and support staff, supporting its reliability and applicability across diverse clinical settings. The MAVAS consists of 30 items grouped into four subscales: patient-related causes (8 items), environmental and interpersonal causes (7 items), ward practices and systems (8 items), and management of aggression (7 items). Each item is rated on a four-point Likert scale ranging from 1 (strongly agree) to 4 (strongly disagree).

The MAVAS was administered in both English and isiXhosa. The isiXhosa version was developed through forward translation and review by bilingual mental health professionals to ensure conceptual equivalence and clarity. Formal psychometric validation of the isiXhosa version has not yet been conducted and is acknowledged as a limitation.

Data were collected during scheduled training sessions at CMH. Eligible security personnel were approached during duty hours and provided with verbal and written information about the study. Written informed consent was obtained from all participants prior to data collection.

Baseline data were collected immediately before the educational intervention using a self-administered questionnaire. The questionnaire consisted of two sections: a demographic data form and the MAVAS. Participants completed the questionnaires individually in a designated training venue to ensure privacy and minimise distractions. The principal investigator was present to provide clarification if required, but did not influence responses.

Following completion of the baseline assessment, participants underwent the structured de-escalation training intervention. The MAVAS was re-administered to the same participants immediately after the intervention using identical procedures. Pre- and post-intervention questionnaires were linked using unique study codes to preserve anonymity while enabling paired analysis.

All completed questionnaires were reviewed for completeness at the time of collection, securely stored, and later entered into an electronic database for statistical analysis.

Prior to the main study, a brief pilot assessment was conducted among a small group of security personnel who were not included in the final sample. The pilot aimed to evaluate the clarity, feasibility, and comprehension of the questionnaire, including the isiXhosa version. No modifications were required following the pilot assessment, and pilot data were excluded from the final analysis.

Subscale scores were calculated as the mean of items within each domain, with lower scores indicating stronger agreement with statements reflecting patient-centred, contextual, and non-coercive approaches to aggression management. Changes in mean subscale scores between pre- and post-intervention assessments were used to evaluate the impact of the educational programme.

Data were captured, checked for completeness, and cleaned before analysis. Statistical analysis was performed using IBM SPSS Statistics for Windows, v27.0 (IBM Corp., Armonk, New York, US). Descriptive statistics were used to summarise participant demographics and baseline characteristics.

Changes in attitudes across MAVAS domains were assessed using paired-samples t-tests with a two-tailed alpha of 0.05 and 95% confidence intervals (CIs). For non-normally distributed data, the Wilcoxon signed-rank test was applied as a sensitivity analysis.

Multiple linear regression analysis was conducted to identify predictors of change in MAVAS domain scores. All a priori selected demographic variables (age, sex, job rank, and years of work experience) were entered simultaneously into the regression models, regardless of their statistical significance in bivariate analyses, to control for potential confounding and ensure theoretical relevance.

Ethical clearance to conduct this study was obtained from the Walter Sisulu University Health Research Ethics Committee (Ref. No. 029/2025). The research protocol was submitted to the National Health Research Database and received approval. Institutional permission was granted by Cecilia Makiwane Hospital management and the contracted security company. Participation was voluntary, and written informed consent was obtained before data collection. Confidentiality and anonymity were maintained using unique participant codes, and electronic data were password-protected. No financial incentives were offered, although refreshments and certificates of participation were provided.

Participation was voluntary, and written informed consent was obtained from all participants before data collection. Confidentiality and anonymity were maintained by using unique participant codes rather than personal identifiers. Data were stored securely, and all electronic files were password-protected. No financial incentives were offered, although participants received refreshments and a certificate of participation.

A total of 90 participants were included in the study (Table 1). The mean age of respondents was 39.7 years (SD = 7.8). More than half were female (55.6%), and the majority were junior staff (86.7%). Most participants (54.4%) had more than 5 years of work experience. Participants were drawn from several hospital departments, including psychiatry wards, emergency (casualty) units, outpatient departments, and other high-risk areas for patient aggression.

Baseline (pre-intervention) MAVAS domain scores indicated moderate agreement with statements relating to patient-related, environmental, and systemic contributors to aggression, as well as approaches to managing aggressive behaviour. These findings reflect participants’ initial attitudes towards aggression management prior to the educational intervention.

To determine the effect of the educational intervention, MAVAS domain scores were compared before and after training. The intervention produced statistically significant reductions across all MAVAS domains (p ≤ 0.01), indicating improved attitudes towards recognising and managing aggression.

The paired-samples t-test showed significant reductions in mean scores across all MAVAS domains after the intervention (Table 2). As MAVAS uses a four-point Likert scale (1 = strongly agree to 4 = strongly disagree), a lower post-test score indicates stronger agreement with the items in each domain. In the patient-related causes of aggression domain, the mean score decreased from 2.29 (SD = 0.44) before the intervention to 2.13 (SD = 0.38) after the intervention (p = 0.001), indicating increased agreement that aggression can be linked to patient-related factors. The environmental and interpersonal causes domain declined from 2.06 (SD = 0.35) to 1.89 (SD = 0.36) (p < 0.001), indicating increased recognition of contextual factors contributing to aggression. In the ward practices and systems domain, mean scores dropped from 2.24 (SD = 0.31) to 2.08 (SD = 0.36) (p < 0.001), reflecting stronger agreement that institutional and procedural influences impact aggression. Lastly, the management of aggression domain decreased from 1.89 (SD = 0.38) to 1.79 (SD = 0.32) (p = 0.010), suggesting more positive attitudes towards strategies for managing aggression.

Figure 1 presents the mean post–pre differences in MAVAS domain scores, with 95% CIs. All mean differences were negative, and none of the CIs crossed zero, showing consistent improvement across domains. The largest change occurred in environmental and interpersonal causes (−0.18; 95% CI: −0.26 to −0.09), followed by patient-related causes (−0.16; 95% CI: −0.25 to −0.07), ward practices and systems (−0.16; 95% CI: −0.25 to −0.08), and management of aggression (−0.10; 95% CI: −0.18 to −0.02). Because lower MAVAS scores represent stronger agreement with positive statements, these findings suggest improved attitudes towards the recognition and management of aggression post-intervention.

Overall, the educational intervention significantly improved participants’ attitudes towards recognising and managing aggressive behaviour in hospital settings. Improvements were consistent across all MAVAS domains, with greater effects among female and more experienced staff. Age and professional ranking did not significantly influence outcomes. These results support the effectiveness of structured training programmes in enhancing staff attitudes towards de-escalation and aggression management.