Application of the Plan-Do-Check-Act Management Model to Violence Risk Management in a Psychiatric Ward: Impact on Patient Safety and Nurse–Patient Relationships Authors: Junhua Li, Huiming Niu, Xiaojuan Liu, Zhijie Lei, Hongfang Li, Ying Cheng, Hongjuan Du, Yalan Wang Published: 2026 ABSTRACT AbstractBackgroundViolence in psychiatric wards poses significant risks to patients, staff and the therapeutic environment. Patients with schizophrenia frequently exhibit violent behaviours that compromise safety and disrupt therapeutic relationships. Although reactive crisis management remains the norm, systematic, proactive strategies for violence prevention are scarce. The Plan-Do-Check-Act (PDCA) cycle provides a structured quality improvement framework, yet its effectiveness in psychiatric violence risk management has not been well established.ObjectiveThis study aimed to evaluate the effectiveness of the PDCA cycle management model in reducing violent incidents and improving patient satisfaction with nursing care in inpatients with schizophrenia.MethodsA quasi-experimental study was conducted between April 2021 and April 2022 at Tianshui Third People’s Hospital, Gansu Province, China. Four hundred patients with schizophrenia were enrolled and allocated to either a research group (n = 200), receiving PDCA cycle management in addition to standard care, or a control group (n = 200), receiving standard care alone. The Modified Overt Aggression Scale, violence behaviour recording forms and patient satisfaction surveys were used to assess outcomes over an 8-week period.ResultsOf the 400 enrolled patients, 331 completed the 8-week study (research group: n = 162; control group: n = 169). After implementing PDCA cycle management, the research group demonstrated significantly lower aggression scale scores (7.54 ± 2.07 vs 10.53 ± 3.99, p < 0.001) and reduced frequencies of verbal aggression, property damage, self-harm and physical assault compared with controls. Patient satisfaction increased significantly in the research group (60.5% “very satisfied” vs 51.5% in controls, p = 0.015).ConclusionThe PDCA cycle management model effectively reduces violent behaviour incidence and enhances patient satisfaction in psychiatric wards, offering a systematic approach to violence risk management that strengthens nurse–patient relationships and improves overall care quality.Keywords: schizophrenia, psychiatric nursing, patient safety Introduction Schizophrenia represents one of the most common and severe mental disorders in psychiatric care, characterised by disturbances in cognition, emotion, perception and behaviour. The condition significantly impairs patients’ quality of life and functional capacity.1 Violence and aggressive behaviours in psychiatric inpatient settings constitute major challenges for mental health services, threatening the safety of patients, healthcare workers and the broader therapeutic milieu.2 Research indicates that nurses working in psychiatric wards experience the highest rates of workplace violence among healthcare professionals, with prevalence estimates ranging from 35% to 59% depending on the specific clinical context.3 Violence in psychiatric settings encompasses a spectrum of behaviours, including verbal aggression, property destruction, self-harm and physical assault directed at staff or other patients. The World Health Organization defines violence as the intentional use of physical force or power, threatened or actual, against oneself, another person or a group, which results in or has a high likelihood of resulting in injury, death, psychological harm, maldevelopment or deprivation.4 Although the estimated prevalence of violent behaviour among patients with schizophrenia ranges from 10% to 15%, absolute risks can be as high as 26% over 12 months in first-episode psychosis populations.5,6 The consequences of violence in psychiatric wards extend beyond immediate physical harm. Violent incidents disrupt therapeutic relationships between nurses and patients, create fear and anxiety among staff members, compromise the quality of care delivery and may lead to increased use of restrictive interventions, such as physical restraints and seclusion.7 Furthermore, exposure to workplace violence contributes to burnout, reduced job satisfaction and staff turnover in psychiatric nursing, ultimately affecting the sustainability and effectiveness of mental health services.8 Despite the serious implications of violence in psychiatric care, prediction and prevention remain challenging. Traditional approaches to violence management have often been reactive rather than proactive, focusing on crisis intervention after incidents occur rather than systematic prevention strategies. Structured risk assessment instruments have been developed to identify patients at elevated risk for violence, yet their predictive validity in schizophrenia populations shows considerable variability, with areas under the curve typically ranging from 0.62 to 0.85.9 Moreover, violence risk assessment alone is insufficient without corresponding management strategies that address modifiable risk factors and promote therapeutic engagement. The Plan-Do-Check-Act (PDCA) cycle, originally developed by Walter Shewhart and later refined by W. Edwards Deming, represents a systematic quality improvement framework that has gained widespread application in healthcare management.10 The PDCA cycle consists of four iterative stages: (1) Plan – identifying problems, analysing root causes and developing improvement strategies; (2) Do – implementing the planned interventions; (3) Check – monitoring outcomes and evaluating effectiveness; and (4) Act – standardising successful practices and addressing remaining issues in subsequent cycles.11 This continuous improvement approach aligns well with the complex, multifaceted nature of violence risk management in psychiatric care. The iterative, data-driven nature of the PDCA cycle aligns particularly well with the dynamic and multifactorial nature of violence risk in psychiatric settings. Unlike static protocols, PDCA enables continuous adaptation of prevention strategies in response to emerging risk patterns, staff feedback and environmental changes – features essential for managing behaviours that are often unpredictable and influenced by complex patient, staff and system-level factors. Recent evidence demonstrates the effectiveness of PDCA cycle management in various nursing contexts, including infection control, medication safety and patient care quality improvement.12,13 In mental health settings specifically, PDCA-based interventions have shown promise in reducing relapse rates, improving compliance with clinical protocols and enhancing nursing care standards.14 However, research specifically examining the application of PDCA cycle management to violence prevention in patients with schizophrenia remains limited. Existing studies in mental health settings have largely focused on process indicators – such as protocol adherence or staff compliance – with few investigations employing validated aggression scales or concurrently examining patient-centred outcomes, such as satisfaction with care. Furthermore, evidence from controlled comparative designs in acute psychiatric wards is sparse. Given the critical importance of violence risk management for patient safety and therapeutic outcomes, systematic evaluation of PDCA-based approaches is warranted. The present study addresses this gap by investigating the impact of implementing a comprehensive PDCA cycle management model for violence risk management in inpatients with schizophrenia. We hypothesise that systematic application of the PDCA framework would lead to reduced incidence and severity of violent behaviours through improved risk identification, proactive intervention strategies and continuous quality improvement processes. Additionally, we examine whether PDCA-based violence management would positively impact nurse–patient relationships, as reflected in patient satisfaction with nursing care. By evaluating both safety and relational outcomes within a single framework, this study aims to provide a pragmatic, evidence-based model that psychiatric nursing leaders can adopt to balance risk mitigation with therapeutic engagement. Methods Study Design and Setting This quasi-experimental study employed a non-randomised controlled design with parallel groups. The study was conducted at the psychiatric inpatient units of Tian Shui City Third People’s Hospital, Gansu Province, China, between April 2021 and April 2022 and is registered with the ISRCTN registry under registration number ISRCTN10969704. The hospital is a tertiary psychiatric facility serving the regional population, with four inpatient wards dedicated to the care of patients with severe mental illness. The sample size was calculated based on a pilot study, assuming an effect size of 0.3, a power of 0.80 and an alpha of 0.05. The calculation indicated a minimum of 178 patients per group. To account for a potential 15% attrition rate, 400 patients were recruited. The study protocol was reviewed and approved by the hospital’s ethics committee, and all participants or their legal guardians provided written informed consent prior to enrolment. Participants Four hundred adult patients with schizophrenia admitted to the psychiatric inpatient wards during the study period were enrolled. The research group comprised 200 patients from Ward 1 (female patients) and Ward 6 (male patients), and the control group included 200 patients from Ward 2 (female patients) and Ward 5 (male patients). This allocation strategy was selected to minimise contamination between intervention and control conditions while maintaining comparable ward environments and staffing patterns. The inclusion criteria were as follows: (1) diagnosis of schizophrenia according to the International Classification of Diseases, 10th Revision diagnostic criteria for mental and behavioural disorders; (2) aged 18–65 years; (3) education level of primary school or above; (4) presence of a family caregiver or guardian with a relatively stable social support system; and (5) willingness to participate and provision of written informed consent. The exclusion criteria were as follows: (1) comorbid organic brain disorders or severe physical illnesses; (2) alcohol or substance dependence; and (3) other psychiatric disorders not meeting schizophrenia diagnostic criteria. Termination criteria: patients were withdrawn from the study if they (1) developed severe physical complications requiring transfer to a general hospital, (2) experienced a sudden worsening of psychiatric symptoms requiring emergency electroconvulsive therapy or (3) requested to withdraw consent at any time. Intervention To ensure the quality and consistency of the intervention, the nursing staff in both groups were balanced. A total of 60 registered nurses (30 per group) delivered the care. Statistical analysis showed no significant differences between the two nursing teams regarding age, years of experience or educational background (p > 0.05), minimising staff-related bias. The intervention was implemented over an 8-week period for all patients in the research group, commencing immediately upon enrolment and continuing throughout their study participation. Control Group Management Patients in the control group received standard psychiatric nursing care based on established hospital protocols. Upon admission, nurses collected demographic and clinical information, assessed patients’ history of violent behaviour, conducted safety education for patients and families and completed admission procedures, including informed consent documentation. Throughout hospitalisation, nursing staff monitored patients’ mental status and behaviours, provided medication administration and implemented basic safety precautions. When violent incidents occurred, nurses responded with immediate de-escalation techniques, psychological support and, when necessary, protective restraint measures following physician orders. All incidents were documented and reported during shift handovers in accordance with routine procedures. Research Group Plan-Do-Check-Act Cycle Management The research group received standard care plus a comprehensive PDCA cycle management approach specifically designed for violence risk management. The PDCA intervention was implemented systematically across four phases, as follows: Plan Phase The planning phase began with a retrospective analysis of 120 violent incidents that occurred in the ward over the previous year. A Quality Improvement Committee, comprising the head nurse, senior psychiatrists and charge nurses, conducted brainstorming sessions and root cause analysis (utilising a fishbone diagram) to identify underlying causes. Nursing staff assessed ‘environmental triggers’, which were specifically defined as noise levels >60dB, ward overcrowding (bed occupancy >95%), lack of privacy in shared rooms and inadequate lighting in corridors. Upon admission, nurses collected specific data points, including history of substance use, previous legal convictions for violence and detailed family statements regarding recent aggressive outbursts. Nurses categorised violence risk factors into five domains: (1) patient-related factors, including positive psychiatric symptoms (eg persecutory delusions, command hallucinations), negative emotions, cognitive impairment and physical discomfort; (2) family-related factors, such as inadequate understanding of mental illness, poor treatment adherence and family conflict; (3) staff-related factors, including insufficient vigilance, inconsistent implementation of safety protocols, inadequate legal knowledge and suboptimal communication skills; (4) environmental factors, such as unsafe facility design, inadequate equipment maintenance and hazardous materials accessibility; and (5) management system factors, including insufficient training for new staff, excessive workload and limitations of closed-unit management. Based on this systematic analysis, the nursing team identified direct causes of violence (failure to implement safety protocols, inadequate observation and monitoring, delayed intervention, poor crisis management skills and ineffective communication) and root causes (professional ethics, adherence to nursing standards and communication competencies). Corresponding preventive measures were then developed, including (1) establishing a therapeutic alliance based on respect and empathy to meet patients’ reasonable needs, (2) enhancing nurses’ self-protection capabilities through training in early warning sign recognition, (3) strengthening environmental safety management through regular facility inspections and hazard removal, (4) optimising medication management to ensure treatment adherence and (5) implementing systematic safety audits and quality control mechanisms. Do Phase During the implementation phase, nursing staff executed the planned interventions systematically. This included the following: (1) training – structured training covering three parts (including Initial Intensive Training, Monthly Refresher Workshops and Competency Evaluation) was provided to all caregivers in the research group before and during the intervention, followed by monthly review workshops (see Appendix I); (2) protocols – the “Ward Violence Prevention Manual” was developed, which standardised emergency response codes and restraint procedures; (3) communication – the Situation–Background–Assessment–Recommendation communication tool for shift handovers was implemented to ensure accurate transmission of risk information; (4) high-risk area management – policies were expanded to include restricted access to nursing stations/treatment rooms, a mandatory “two-person rule” when entering seclusion rooms and time-locked doors for pantries during non-meal hours; (5) safety inspections – daily inspections focused on contraband searches (eg lighters, sharps), window lock integrity and functionality of emergency alarm buttons (comprehensive ward audits were conducted twice weekly to check for blind spots in surveillance camera coverage); and (6) maintaining emergency preparedness through daily checks of rescue medications and weekly functionality tests of emergency equipment (eg defibrillators, restraints). Nursing interventions emphasised person-centred care and therapeutic communication using specific de-escalation techniques (eg active listening, maintaining non-threatening body language and verbal redirection). Early intervention was triggered when specific prodromal signs (eg motor restlessness, pacing, increased voice volume, clenched fists) were observed during routine hourly rounds, allowing nurses to intervene before behaviours escalated to physical aggression. Check Phase The Quality Control Group held weekly meetings to analyse data from the Violence Behaviour Recording Forms and Modified Overt Aggression Scale (MOAS) scores. Specifically, “near-miss” events and adherence to the new protocols (eg handover checklist completion rates) were tracked. If the frequency of violent incidents in a specific area (eg bathroom) increased, the team immediately investigated potential environmental causes. Action Phase In the action phase, successful strategies were standardised into institutional policies and procedures, and unresolved issues were incorporated into the next PDCA cycle for continued improvement. This continuous cycle of evaluation and refinement ensured that violence prevention strategies evolved based on evidence and experience. Outcome Measures Modified Overt Aggression Scale The primary outcome was violent behaviour assessed using the MOAS.15 The scale has demonstrated excellent psychometric properties, with an intraclass correlation coefficient of 0.94, indicating high inter-rater reliability.15 This structured rating scale objectively evaluates aggressive behaviours in patients with psychiatric disorders across four domains: (1) verbal aggression (five items ranging from no aggression to repeated/deliberate threats of violence); (2) aggression against property (five items from no aggression to fire-setting or dangerous throwing); (3) auto-aggression or self-harm (five items from no aggression to severe self-injury or suicide attempts); (4) physical aggression against others (five items from no aggression to serious assault causing injury). Each domain uses a 5-point severity rating (0–4), with total scores ranging from 0 to 40 points. Higher scores indicate more severe violent behaviour. The scale is weighted (verbal aggression ×1, property aggression ×2, self-aggression ×3, physical aggression ×4) to reflect the relative seriousness of different violence types. The MOAS is widely used in psychiatric research due to its simplicity, brevity (5-minute administration time), reliability across demographic groups and ability to capture the most severe behaviours during an observation period. Violence Behaviour Recording Form A standardised violence incident documentation form was developed to capture the timing and specific characteristics of violent events. This form recorded the date and time of each incident and the type of aggression (verbal, property, self-directed or physical assault). Nursing staff completed this form immediately following any violent incident, providing a detailed record of violence patterns and frequency throughout the study period. Patient Satisfaction Survey Patient satisfaction with nursing care was assessed using a purpose-designed questionnaire developed specifically for this study, given the absence of a widely validated instrument tailored to the context of violence risk management in Chinese psychiatric wards. The questionnaire was constructed following a systematic process to ensure content validity. First, a comprehensive literature review was conducted to identify domains relevant to patient satisfaction in psychiatric nursing, including nurse–patient communication, perceived safety, respect for autonomy and responsiveness to needs. Second, a preliminary pool of 15 items was generated and reviewed by an expert panel comprising two senior psychiatrists, three head nurses with >15 years of psychiatric experience and a nursing researcher. The panel assessed each item for relevance, clarity and appropriateness using a 4-point Likert scale (1 = not relevant, 4 = highly relevant). Items with an Item-Level Content Validity Index of <0.78 were eliminated or revised, resulting in a final 10-item questionnaire. The overall Scale-Level Content Validity Index was 0.91, indicating excellent content validity. The 10 items assessed the following dimensions: (1) nurses’ professional competence; (2) frequency and quality of nurse–patient communication; (3) perceived respect and dignity; (4) feeling of safety on the ward; (5) timeliness of response to requests; (6) clarity of information provided; (7) involvement in care decisions; (8) emotional support received; (9) management of ward environment; and (10) overall care experience. Each item was rated on a 10-point scale (1 = very dissatisfied, 10 = very satisfied), yielding a total score range of 10 to 100 points. Overall satisfaction was categorised as follows: very satisfied (90–100 points), basically satisfied (70–89 points) or dissatisfied (<70 points). The internal consistency reliability of the questionnaire was assessed in a pilot sample of 30 inpatients with psychiatric disorders (not included in the main study) before formal data collection. Cronbach’s alpha (α) coefficient was 0.87, exceeding the acceptable threshold of 0.70 and indicating good internal reliability. Test–retest reliability was not formally assessed due to the expected clinical instability and rapid changes in mental status characteristic of the acute schizophrenia population; however, the questionnaire was administered at both baseline and week 8 to capture within-patient changes over time. The survey was administered by trained research assistants who were blinded to group allocation to minimise interviewer bias. Data Collection Procedures All research personnel, including nurses and evaluators, participated in comprehensive training on study procedures, assessment instruments, diagnostic criteria and data collection protocols prior to study initiation. Inter-rater reliability testing was conducted, achieving a Kappa coefficient of 0.89, indicating excellent agreement. Baseline assessments were completed during the 1st week following patient enrolment, including demographic information, clinical characteristics, MOAS scores and initial satisfaction ratings. Follow-up assessments were conducted at week 8 (study endpoint) using the same instruments. Violent incidents were documented continuously throughout the 8-week observation period using the standardised recording forms. Patients who completed the full 8-week study period were included in the final analysis. Statistical Analysis All statistical analyses were performed using SPSS version 22.0 software. Continuous variables were expressed as mean ± standard deviation and categorical variables as frequencies and percentages. Between-group comparisons for continuous variables employed independent samples t-tests, and categorical variables were analysed using the chi-square test or Z test. For the main outcome measure (MOAS score), the mean difference between groups and its 95% confidence interval (CI) were calculated, and Cohen’s d was used to evaluate the effect size. This study collected data only at two time points: baseline and week 8, so repeated-measures analysis was not performed. Intention-to-treat (ITT) analysis was performed to minimise attrition bias; missing endpoint data were handled using multiple imputation with five imputed datasets, and sensitivity analyses were conducted to compare complete-case results with ITT-imputed results. The test level was set at α = 0.05 (two-sided). Results Participant Flow and Baseline Characteristics As shown in Figure 1, 331 of the 400 patients initially enrolled completed the 8-week study protocol. The research group experienced 38 dropouts (19% attrition rate), resulting in 162 completers, and the control group had 31 dropouts (15.5% attrition rate), with 169 completers. The difference in attrition rates between groups was not statistically significant (χ2 = 1.03, p = 0.31). The reasons for dropout included early discharge, transfer to other facilities, withdrawal of consent and loss to follow-up (defined as patients who were discharged against medical advice or transferred to another facility before the week 8 assessment could be completed). CONSORT‑style participant flow diagram. Flow diagram showing enrollment, allocation, follow‑up, and analysis of study participants. A total of 400 patients with schizophrenia were allocated to the research group (PDCA cycle management) or control group (standard care). After an 8‑week intervention, 331 patients completed the study and were included in the complete‑case analysis. ITT analysis included all 400 randomized participants. Intention-to-treat (ITT) analysis was performed to minimise attrition bias. Missing endpoint data were handled using multiple imputation (MI) with five imputed datasets, incorporating baseline MOAS scores, age, gender and education level as predictive variables. Sensitivity analyses were conducted to compare complete-case results with ITT-imputed results. Baseline characteristics between dropouts and completers were comparable in both groups (all p > 0.05), indicating that missing data were likely missing at random and unlikely to introduce substantial bias. Baseline demographic and clinical characteristics of the study completers are presented in Table 1. The two groups showed no significant differences in age, gender distribution, educational level or occupational status (all p > 0.05), indicating successful baseline equivalence. Similarly, the nursing staff characteristics were comparable between groups regarding age, years of experience, educational background and professional rank (Table 2), ensuring that staff-level factors would not confound the intervention effects. Table 1Baseline Demographic and Clinical Characteristics of Patient ParticipantsVariableControl Group (n=169)Research Group (n=162)Test StatisticP-valueAge (years, M±SD)37.22±10.0537.80±12.09t=0.4790.632Gender, n (%)χ2=0.2350.628 Male89 (52.66%)81 (50.00%) Female80 (47.34%)81 (50.00%)Education Level, n (%)χ2=6.2780.099 Primary school58 (34.31%)61 (37.65%) Middle school69 (40.83%)46 (28.39%) High school/Technical school24 (14.20%)31 (19.14%) College or above18 (10.65%)24 (14.81%)Occupation, n (%)χ2=10.2650.068 Farmer149 (88.17%)129 (79.63%) Worker4 (2.37%)1 (0.62%) Government employee3 (1.78%)8 (4.94%) Student4 (2.37%)8 (4.94%) Teacher2 (1.17%)6 (3.70%) Unemployed7 (4.14%)10 (6.17%) Baseline Demographic and Clinical Characteristics of Patient Participants Table 2Characteristics of Participating Nursing StaffVariableControl Group (n=29)Research Group (n=28)Test StatisticP-valueAge (years, M±SD)29.93±3.1230.86±3.55t=1.2330.221Years of Experience (M±SD)7.76±3.038.89±3.26t=1.3610.179Education Level, n (%)χ2=0.9830.612 Technical diploma1 (3.44%)0 (0.00%) Associate degree11 (37.94%)11 (39.28%) Bachelor’s degree17 (58.62%)17 (60.72%)Professional Rank, n (%)χ2=2.5600.278 Staff nurse7 (24.14%)5 (17.86%) Nurse practitioner20 (68.97%)17 (60.71%) Senior nurse practitioner2 (6.89%)6 (21.43%) Characteristics of Participating Nursing Staff Violence Severity: Modified Overt Aggression Scale Scores At baseline (week 1), the mean MOAS scores were comparable between the research group (18.03 ± 3.46) and control group (19.02 ± 5.45), with no significant difference (p = 0.051). After 8 weeks of intervention, the MOAS score of the Research group (7.54 ± 2.07) was significantly lower than that of the control group (10.53 ± 3.99). The mean difference between the groups was 2.99 (95% CI: 2.30—3.68). The effect size (Cohen’s d) was 0.94, indicating that the PDCA intervention had a greater effect. This difference is statistically significant (t = −8.49, p < 0.001) (Table 3 and Figure 2). Table 3Modified Overt Aggression Scale Scores (M±SD)GroupWeek 1Week 8t-valuep-valueControl (n=169)19.02±5.4510.53±3.99−16.34<0.001Research (n=162)18.03±3.467.54±2.07−33.11<0.001t-value−1.946−8.49––p-value0.051<0.001––Mean Difference2.99 (95% CI: 2.30—3.68)Cohen’s d0.94 (95% CI: 0.71—1.17) Modified Overt Aggression Scale Scores (M±SD) Comparison of Modified Overt Aggression Scale Scores Between Groups at Baseline and Week 8. Bar graph showing mean MOAS scores (±SD) for control and research groups at Week 1 and Week 8. Research group demonstrates significantly lower scores at Week 8 (**p<0.001). Violence Incident Frequency by Type Detailed analysis of specific violence types revealed that PDCA cycle management significantly reduced the frequency of all four categories of aggressive behaviour (Table 4). For verbal aggression, the research group showed a more favourable distribution, with fewer patients exhibiting frequent or constant verbal aggression compared with controls (χ2 = 16.58, p < 0.001). The research group had 25.9% of patients with no verbal aggression versus 19.5% in the controls, whereas only 9.9% of research group patients showed constant verbal aggression compared with 27.2% in the control group. Table 4Comparison of Violence Incident Frequency by Type Between Groups at week 8Violence TypeControl Group (n=169)Research Group (n=162)χ2p-valueVerbal Aggression16.58<0.001None (0)33 (19.5%)42 (25.9%)Occasional (1–2 times)48 (28.4%)58 (35.8%)Frequent (3–4 times)42 (24.9%)46 (28.4%)Constant (≥5 times)46 (27.2%)16 (9.9%)Property Aggression9.040.029None (0)32 (18.9%)48 (29.6%)Occasional (1–2 times)47 (27.8%)52 (32.1%)Frequent (3–4 times)51 (30.2%)39 (24.1%)Constant (≥5 times)39 (23.1%)23 (14.2%)Self-Directed Aggression8.180.042None (0)32 (18.9%)44 (27.2%)Occasional (1–2 times)50 (29.6%)52 (32.1%)Frequent (3–4 times)42 (24.9%)42 (25.9%)Constant (≥5 times)45 (26.6%)24 (14.8%)Physical Aggression15.200.002None (0)36 (21.3%)59 (36.4%)Occasional (1–2 times)49 (29.0%)52 (32.1%)Frequent (3–4 times)43 (25.4%)32 (19.8%)Constant (≥5 times)41 (24.3%)19 (11.7%) Comparison of Violence Incident Frequency by Type Between Groups at week 8 Similarly, property destruction incidents were significantly less frequent in the research group (χ2 = 9.04, p = 0.029). Self-directed aggressive behaviours, including self-harm and suicidal actions, also occurred less frequently among research group patients (χ2 = 8.18, p = 0.042). Most notably, physical aggression towards others was substantially reduced in the research group, with 36.4% showing no physical violence compared with only 21.3% in the controls, and only 11.7% exhibiting constant physical aggression versus 24.3% in the control group (χ2 = 15.20, p = 0.002). As shown in Figure 3, the research group exhibited a clear shift towards “none/occasional” categories and a reduction in “constant” incident frequency across violence subtypes compared with the control group. Distribution of Violence Incident Frequencies by Type and Group. Stacked bar charts displaying the percentage distribution of violence frequency categories (none, occasional, frequent, constant) for each of the four violence types, comparing control and research groups. Patient Satisfaction with Nursing Care At baseline, patient satisfaction levels were comparable between groups, with no significant difference in the distribution of satisfaction categories (Z = −1.10, p = 0.272). However, following the 8-week intervention, the research group demonstrated significantly higher satisfaction with nursing care compared with the control group (Z = −2.44, p = 0.015) (Table 5). Specifically, in the research group, 60.5% of patients reported being “very satisfied” with nursing services at week 8, compared with 51.5% in the control group. The proportion of dissatisfied patients decreased more substantially in the research group (from 23.4% at baseline to 9.3% at week 8) than in the control group (from 27.9% to 23.7%) (Figure 4). Table 5Patient Satisfaction with Nursing Care, n (%)GroupWeek 1Week 8Very SatisfiedBasically SatisfiedDissatisfiedVery SatisfiedBasically SatisfiedDissatisfiedControl (n=169)67 (39.6%)55 (32.5%)47 (27.9%)87 (51.5%)42 (24.8%)40 (23.7%)Research (n=162)73 (45.1%)51 (31.5%)38 (23.4%)98 (60.5%)49 (30.2%)15 (9.3%)z-value−1.10−2.44p-value0.2720.015 Patient Satisfaction with Nursing Care, n (%) Patient Satisfaction Levels at Baseline and Week 8. Grouped bar chart showing the percentage of patients in each satisfaction category (very satisfied, basically satisfied, dissatisfied) for both groups at Week 1 and Week 8. Asterisk indicates significant improvement in research group at Week 8 (p=0.015). Discussion This quasi-experimental study provides preliminary evidence suggesting that implementing a systematic PDCA cycle management model may significantly reduce violent behaviour and enhances patient satisfaction in inpatients with schizophrenia. The findings demonstrate that a structured, continuous quality improvement approach to violence risk management yields substantial benefits over traditional care models, with implications for both patient safety and therapeutic relationships in mental health settings. The significant reduction in aggressive behaviour observed in the research group aligns with growing evidence supporting systematic quality improvement methodologies in psychiatric nursing. Our results showed a 58.2% reduction in MOAS scores from baseline in the research group compared with 44.6% in controls, with significantly lower endpoint scores in the intervention arm. These findings are consistent with previous research demonstrating that structured management approaches can effectively address behavioural challenges in psychiatric populations. For example, a recent systematic review examining PDCA implementation in nursing management found that this approach consistently improved both patient outcomes and nursing practice quality across diverse clinical contexts.16 The magnitude of improvement in our study suggests that the comprehensive, multifaceted nature of the PDCA intervention – addressing patient assessment, environmental safety, staff training, communication enhancement and continuous monitoring – creates synergistic effects that surpass the impact of individual components. The reduction in violence severity can be attributed to several key mechanisms inherent in the PDCA framework. First, the systematic planning phase enabled early identification of high-risk patients and modifiable risk factors, allowing for proactive rather than reactive interventions. Research has established that although violence in psychiatric settings may appear unpredictable, most incidents have identifiable precursors that can inform prevention strategies when systematically assessed.17 Our intervention equipped nursing staff with structured tools for recognising early warning signs, such as agitation, threatening verbal content and changes in mental status, facilitating timely de-escalation before behaviours escalated to physical violence. Second, the standardisation of protocols during the implementation phase ensured consistent application of evidence-based practices across all shifts and staff members, reducing variability in care quality that may otherwise create opportunities for incidents to occur.18 Furthermore, the continuous monitoring and feedback loops built into the Check and Act phases enabled real-time problem-solving and iterative refinement of violence prevention strategies. When safety audits identified gaps in protocol adherence or emerging risk patterns, corrective actions were implemented promptly and their effectiveness evaluated in subsequent cycles. This adaptive quality improvement process differs fundamentally from static safety protocols, allowing the intervention to evolve based on unit-specific challenges and staff learning.19 The emphasis on staff education and skill development throughout the PDCA process also likely contributed to improved violence management. Enhanced understanding of mental illness, communication techniques for de-escalation and confidence in implementing therapeutic interventions all support more effective nurse–patient interactions that can prevent or defuse potentially violent situations.20 Notably, our analysis of specific violence types revealed that PDCA management reduced all four categories of aggressive behaviour – verbal aggression, property destruction, self-harm and physical assault. This broad-spectrum effectiveness suggests that the intervention addressed fundamental aspects of violence risk rather than merely shifting behaviours from one form to another. The particularly pronounced reduction in physical aggression towards others (11.7% constant frequency in the research group vs 24.1% in controls) holds special clinical significance, as such incidents pose the greatest immediate danger to staff and other patients. These findings support the value of comprehensive risk management frameworks that address multiple violence domains simultaneously rather than focusing narrowly on specific behaviour types.21 The significant improvement in patient satisfaction represents an equally important finding, with implications for therapeutic alliance and overall care quality. The research group demonstrated a shift towards higher satisfaction levels, with 60.5% reporting being “very satisfied” at the endpoint compared with 51.5% in the control group. This enhancement in patient satisfaction likely reflects multiple facets of the PDCA intervention. Patients in the research group received more attentive, individualised care as nurses conducted thorough assessments, developed personalised care plans and maintained regular communication about treatment goals and progress. The emphasis on therapeutic communication, respect for patient autonomy and addressing patients’ concerns created a more positive care experience that fostered trust and collaboration.22 The relationship between violence reduction and improved satisfaction merits particular consideration. As the ward environment became safer and more stable through effective violence prevention, patients likely experienced reduced fear and anxiety, creating conditions more conducive to therapeutic work. Research has demonstrated that violence in psychiatric units negatively affects not only staff but also other patients, who may feel unsafe or re-traumatised by witnessing aggressive incidents.23 By creating a calmer, more predictable milieu, the PDCA intervention may have enhanced patients’ sense of security and their ability to engage constructively with treatment. Additionally, the de-escalation techniques and humanistic approaches emphasised in the intervention – such as validating patients’ emotions, offering choices and using verbal negotiation rather than coercive measures – likely strengthened nurse–patient relationships by demonstrating respect and genuine care.24 The improvement in satisfaction also reflects the quality of nurse–patient interactions fostered by the PDCA approach. Training in therapeutic communication, active listening and person-centred care principles equipped nurses with skills to engage patients more effectively. Studies have shown that perceived nurse caring behaviours strongly predict patient satisfaction in psychiatric settings, with domains such as trust, professional ethics and responsive communication being particularly influential.25 Our intervention explicitly targeted these competencies through comprehensive staff education and regular supervision, likely contributing to the observed satisfaction improvements. Furthermore, the transparency and consistency of care delivery under the PDCA framework may have helped patients understand what to expect and feel more involved in their treatment, enhancing their sense of agency and partnership with the nursing team.26 The finding that systematic violence risk management enhances both safety and satisfaction challenges potential concerns that restrictive safety measures may negatively impact patients’ care experiences. Some critics have argued that heightened security protocols could feel oppressive or punitive to patients, potentially undermining therapeutic relationships. However, our results suggest that when violence prevention is implemented thoughtfully within a framework emphasising therapeutic engagement, early intervention and least restrictive alternatives, patients actually experience improved care quality. This aligns with recent evidence indicating that violence prevention strategies grounded in trauma-informed care principles and recovery-oriented practices can simultaneously promote safety and respect patient dignity.27 The study observed moderate attrition rates (15.5–19%), consistent with real-world psychiatric inpatient research where early discharge and clinical transfer are common. Intention-to-treat analysis with MI was applied to reduce bias from missing data, and sensitivity analyses confirmed that the primary findings (reduced MOAS scores and improved satisfaction) remained consistent between complete-case and ITT populations. Baseline comparisons between dropouts and completers showed no significant imbalances, supporting the assessment that missing data were random and unlikely to alter the core conclusions. However, we acknowledge that unmeasured factors related to early discharge could still influence results, which should be considered when interpreting the generalisability of findings. The study’s methodological approach contributes to the evidence base by examining PDCA implementation in a real-world psychiatric setting with an adequate sample size and prospective data collection. The use of standardised, validated instruments for assessing aggressive behaviour and the inclusion of continuous incident documentation strengthened outcome measurement. However, several limitations warrant consideration when interpreting these findings. First, the use of a non-randomised, quasi-experimental design (allocation by ward) may introduce selection bias and clustering effects. Although the baseline comparison showed that the two groups of patients were balanced in terms of demographic and clinical characteristics, it failed to control for all potential confounding factors (eg ward culture, nursing team dynamics). Future studies should use randomised controlled designs or statistical methods (such as analysis of covariance and mixed-effects models) for more rigorous confounding control. Second, the statistical analysis was limited to the t-test and chi-square test, and repeated measurement analysis was not performed, which is related to the fact that this study only collected data at two time points. Follow-up studies can set multiple follow-up time points and use repeated-measures ANOVA or mixed effects models to evaluate the time effect of the intervention more comprehensively. Third, the relatively brief 8-week intervention period, although sufficient to detect significant effects, may not capture longer-term sustainability of improvements or identify delayed benefits that emerge over an extended implementation.28 In the future, we will extend the follow-up period to evaluate the sustainability and long-term behavioural outcomes of the PDCA intervention. Additionally, the study did not include objective measures of staff competency changes or document specific aspects of the PDCA process that contributed most substantially to outcomes. Future research could employ more detailed process evaluations, examining which PDCA components (eg staff training, environmental modifications, communication protocols) drive the greatest impact on violence reduction. Qualitative exploration of staff and patient perspectives on the intervention would also enrich understanding of how the PDCA approach influences the ward culture and therapeutic relationships. Longer follow-up periods would clarify whether violence reductions and satisfaction improvements persist after the initial implementation phase or require ongoing reinforcement to maintain gains.29 Despite these limitations, this study makes an important contribution by demonstrating that a systematic, evidence-based quality improvement framework can effectively address one of the most challenging aspects of psychiatric nursing. The PDCA cycle may offers a replicable model that other mental health facilities can adapt to their specific contexts, potentially improving safety and care quality on a broader scale. The positive impact on both violence reduction and patient satisfaction underscores that these outcomes are not mutually exclusive but rather synergistic when approached through comprehensive, person-centred management strategies. Conclusions This study provides encouraging but preliminary evidence that implementing a PDCA cycle management model for violence risk management in psychiatric wards significantly reduces aggressive behaviour severity and frequency while simultaneously enhancing patient satisfaction with nursing care. The intervention demonstrated effectiveness across all violence types, including verbal aggression, property destruction, self-harm and physical assault, with particularly pronounced benefits in reducing the most dangerous physical aggression towards others. These findings, pending replication in randomised trials, suggest that systematic violence prevention strengthens rather than undermines therapeutic nurse–patient relationships when implemented with attention to person-centred care principles. The PDCA framework’s iterative approach – emphasising proactive planning, systematic implementation, continuous monitoring and evidence-based refinement – offers psychiatric nurses and administrators a practical methodology for addressing the complex challenge of violence in mental health settings. By integrating risk assessment, staff education, environmental safety optimisation, therapeutic communication enhancement and quality improvement mechanisms, this model may creates comprehensive safeguards that protect patients, staff and the therapeutic milieu. The study findings provide initial support for considering the broader adoption of PDCA-based approaches to violence risk management, with potential benefits extending beyond schizophrenia populations to other diagnostic groups and care settings. Future research should examine the long-term sustainability of PDCA interventions, explore cost-effectiveness considerations, investigate which specific components contribute most to outcomes and evaluate implementation in diverse psychiatric facilities and patient populations. Additionally, studies could assess whether PDCA-based violence management reduces staff burnout, improves retention and enhances organisational safety culture. As psychiatric services increasingly recognise the imperative of creating safe, recovery-oriented environments, systematic quality improvement approaches, such as the PDCA cycle, represent promising pathways towards achieving these goals while maintaining the therapeutic focus essential to mental healthcare. Acknowledgments No funding or sponsorship was received for this study or publication of this article. Funding Statement This research did not receive any funding support. Data Sharing Statement All data generated or analyzed during this study are included in this published article. Ethics Approval and Consent to Participate This study was conducted in accordance with the declaration of Helsinki. This study was conducted with approval from the Ethics Committee of Tian Shui City Third People’s Hospital and all participants or their legal guardians provided written informed consent prior to enrollment. Author Contributions All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work. Disclosure The authors declare that they have no competing interests. References Associated Data Data Availability Statement All data generated or analyzed during this study are included in this published article.