A
The mediating role of nurses' social networks between sleep quality and safety behavior: A mixed-methods studyA
JiePeng1✉A
XinQingZhu1✉A
HuaZhenHuang1✉A
XiaolingFeng1✉A
JingLi1✉A
XianZhenLiu1✉A
XiaoChangLi1✉A
HuaQiongDu1✉A
DongLi1✉A
KaiLangHuang1✉YuanQiuHuang1✉Emailhuangyuanqiu@ylsdyrmyy.cn
1The First People’s Hospital of YulinYulinChina
Jie Peng1, XinQing Zhu1, HuaZhen Huang1, Xiaoling Feng1, Jing Li1, XianZhen Liu1, XiaoChang Li1, HuaQiong Du1, Dong Li1, KaiLang Huang1, YuanQiu Huang1,∗
Affiliations:
1The First People's Hospital of Yulin, Yulin, China
∗Corresponding Author:
YuanQiu Huang
The First People's Hospital of Yulin
E-mail: huangyuanqiu@ylsdyrmyy.cn
The mediating role of nurses' social networks between sleep quality and safety behavior: A mixed-methods study
Objective
A
To investigate the pathways through which nurses' social networks and sleep quality influence their safety behaviors, and to provide a theoretical basis for developing targeted interventions.Methods
A
A mixed-methods study was conducted. In June-August 2024, 418 nurses were recruited via convenience sampling to complete a face-to-face survey assessing the Pittsburgh Sleep Quality Index (PSQI), degree centrality of individual social networks, perceived social support, and safety behavior. Path analysis was performed using AMOS 26.0 to test a hypothesized model. In November 2024, 23 nurses were purposively sampled for semi-structured interviews. Thematic analysis was conducted using NVivo 12.6 to explore the influencing factors and pathways related to nurses’ safety behavior.Results: Path analysis revealed that poorer sleep quality directly predicted reduced safety behavior (β = -0.213, P < 0.001) and indirectly predicted it through two significant mediators: lower social network degree centrality (β = -0.098, P < 0.001) and reduced perceived social support (β = -0.058, p = 0.002). A significant serial mediation pathway was identified, wherein sleep quality sequentially affected degree centrality and then perceived social support, ultimately impacting safety behavior (β = -0.087, P = 00.003). This indirect pathway accounted for 18.09% of the total effect. Qualitative findings further elucidated specific job demands and resources that influence nurses' safety behavior.
Conclusion
This study demonstrates that robust social networks and strong perceived social support promote nurses' safety behavior, while poor sleep quality undermines it. The findings highlight a critical pathway through which sleep quality exerts its influence. Nursing managers should prioritize interventions aimed at improving sleep quality and actively foster a supportive social environment to enhance safety performance.
Key words:
Social Network
Nurses' Safety Behavior
Sleep Quality
Job Demands-Resources Model
Mental Health
A
A
A
1.Introduction
Globally, more than 3 million deaths occur annually due to unsafe healthcare practices[1], highlighting a persistent gap between clinical safety behaviors and the goal of ensuring patient safety. The issue of patient safety constitutes a critical global public health priority. As the key caregivers who perform the majority of medical procedures, nurses' safety behaviors are critical to patient safety outcomes. Evidence suggests that 70–80% of adverse medical events are caused by unsafe behaviors, and more than half of these incidents are preventable[2]. Currently, medical safety oversight primarily relies on adverse event monitoring and tracking. However, these monitoring indicators may lack authenticity and have a lagging nature. Consequently, event surveillance is deemed to have a limited impact on safety improvement[3]. Hence, deeply analyzing the pathways influencing nurses' safety behaviors is of paramount importance for improving nursing safety and patient outcomes.
Griffin and Neal (2000) conceptualize safety behavior as a two-dimensional construct. The first dimension, safety compliance, involves employees adhering to established safety rules and procedures (e.g., wearing personal protective equipment). The second dimension, safety participation, encompasses voluntary, discretionary acts that support the safety environment of the organization, such as warning colleagues about hazardous conditions[4]. Nurses' safety behavior is determined by multifaceted factors. Research indicates that professional identity and a positive work environment respectively enhance nurses' safety behaviors[5]. Furthermore, a supportive nursing work environment can positively influence safety behaviors by enhancing nurses’ sense of professional identity. Burnout and job strain have been identified as risk factors for nurses' safety behaviors[6]. Conversely, self-efficacy can mitigate the negative impact of burnout on safety behaviors and has been shown to significantly promote safer practices among nurses[7]. This suggests that the buffering effect between burnout and positive psychological resources protects against the negative impact on nurses' safety behaviors. However, this psychological buffering effect is under significant threat.
A large U.S. cohort study on nurse health revealed that the nature of nursing shift work inevitably adversely affects nurses' physical and mental health[8]. One prominent consequence is that the decline in sleep quality among nurses may disrupt this psychological buffering effect, which ultimately threatens their safety behaviors. A systematic review revealed that the global prevalence of sleep disorders among nurses is as high as 61%[9]. Declining sleep quality leads to impaired mental health and reduced emotional self-control[10]. Yao L et al.[11] demonstrated that nurses with higher sleep satisfaction exhibit better safety behavior performance. On the other hand, a correspondingly prominent impact is that the combined effect of shift work and poor sleep quality induces emotional exhaustion in nurses, consequently leading to a decline in the quality of care and patient safety. Sleep deprivation or reduced sleep duration diminishes an individual's empathic response to the suffering of others through underlying neurobiological mechanisms[12]. Studies indicate that the prevalence of empathy exhaustion among nurses is strikingly high, reaching up to 70% [13]. Therefore, it is imperative to identify crucial psychological protective resources for nurses in the context of widespread sleep and mental health problems.
Social support is a critical psychological protective resource that positively influences safety behaviors. Studies have indicated that perceived organizational support[11] and family support positively predict nurses' safety performance[14]. Furthermore, peer relationships are recognized as a conduit for value transmission[15]. Therefore, interpersonal relationships among nurses possess significant potential for enhancing safety behaviors. Social networks represent the structural characteristics of these relationships and can quantify resource support embedded in social connections[16], while, social support is a manifestation of the function of social relationships[17]. However, the current state and positive effects of nurses' interpersonal relationships remain inadequately explored.
In 2001, Demerouti E[18] proposed the Job Demands-Resources (JD-R) model, which classifies the characteristics influencing job burnout into job demands and job resources. The model suggests that job demands lead to burnout by depleting an individual's physical and psychological resources, while job resources help alleviate burnout by providing support and motivation.The JD-R model reveals three mechanisms through which job demands and job resources influence work engagement. The first is the health impairment process triggered by job demands, that is excessive or sustained job demands lead to physical and mental exhaustion, thereby reducing an individual's work engagement. The second is the motivational process driven by job resources, that is job resources stimulate an individual's work motivation, thereby enhancing their work engagement. The third is the buffering effect of the interaction between job demands and job resources, it is said that when individuals perceive adequate support from job resources, their resilience to job demands increases, thus promoting work engagement. The model has been continuously studied and refined, and it has been demonstrated across various professions and cultural contexts to effectively explain the interplay between job demands, job resources, and work-related emotions and behaviors [19]. In nursing, current researches show that job demands primarily involve factors such as work intensity, workload, frequency of night shifts, and working hours, whereas resources largely encompass organizational support, individual psychological resilience, and self-efficacy[20]. However, the relationship between nurses' sleep and social resources, as well as their pathways of influence on nurses' safety behavior, remains unexplored.
Thus, our study aimed to examined the impact of social network resources and sleep quality on safety behaviors from the perspective of individual nurses. We further investigate the mediating pathways through which social support resources affect the relationships between sleep quality, empathy, and nurses' safety behaviors by using Structural Equation Modeling (SEM). Furthermore, to gain an in-depth understanding of nurses' psychological perceptions of their social networks and sleep quality, as well as the impact of these factors on their safety behaviors, our study employed an explanatory sequential mixed-methods design. We ultimately aim to provide empirical evidence for the supportive role of social network resources and to establish a reliable theoretical framework for developing effective interventions to enhance the management of nurses' safety behaviors.
2 Materials and Methods
2.1 Study Design
We employed an explanatory sequential mixed-methods design. Prior to the formal study, a pilot survey was conducted online with 70 nurses. Of these, 21(30%) questionnaires were incomplete. Thus, the survey instrument was adjusted to finalize, and the data collection method was modified to face-to-face interviews in the formal study. Quantitative data collection was carried out from June to December 2024, followed by semi-structured descriptive qualitative research from January to February 2025. Our study's report refers to the Good Report of A Mixed Methods Study (GRAMMS).
2.2 Participants
2.2.1 Quantitative Study
Nurses were recruited from a Grade A tertiary hospitals in Guangxi, China, using a convenience sampling method between June and November 2024. Inclusion criteria were (1) possession of a valid nurse practice certificate with at least one year of clinical experience; (2) voluntary participation in the study. Exclusion criteria were (1) nurses who were on leave for more than one consecutive month during the investigation period (e.g., for further training or maternity leave); (2) pregnancy or a diagnosed psychiatric disorder. According to the recommendations by Bentler et al. [21], the sample size should be 10 to 20 times the number of observed variables. This study included 17 variables in the structural equation model. Accounting for a 20% rate of invalid questionnaires, the calculated theoretical sample size ranged from 213 to 425 participants.
2.2.1 Qualitative Study
Nurses were categorized into high-, medium-, and low-level safety behavior groups based on the 66th (P66) and 33rd (P33) percentiles of their safety behavior scores. Using purposive sampling, a number of participants from each group were selected for semi-structured interviews in November 2024 and and January 2025. The final sample size was determined according to the principle of theoretical saturation.
2.3 Research Tools
2.3.1 Quantitative Study
(1) Demographic Characteristics. 1) Personal characteristics, including age, sex, education level, marital status, parenting status, monthly income, and self-rated health status. 2) Work-related characteristics, including years of working experience, professional title, department, teaching responsibilities (answering with “Yes” or “No” ), number of night shifts per month, satisfaction with night shift frequency (evaluating with “Satisfied” ”Neutral” and ”Dissatisfied”), and experience of medical errors (answering with “Yes” or “No” ).
(2)The Nurse Safety Behavior Questionnaire (NSBQ), originally translated and culturally adapted into Chinese by Rong Yanfu[22], was used to assess nurses’ safety behaviors. This 12-item instrument employs a 5-point Likert scale ranging from 1 (“never”) to 5 (“always”), yielding a total score between 12 and 60. Higher total scores indicate better safety behavior performance. In our study, the questionnaire demonstrated high internal consistency, with Cronbach's α coefficients of 0.909 in the pilot test and 0.940 in the formal investigation.
(3) Egocentric Social Network Questionnaire. Social networks are generally categorized into two primary types-whole networks and egocentric (personal) networks. Our study specifically investigated egocentric social networks among nurses. Utilizing the name interpreter approach, we developed a customized egocentric social network questionnaire to quantify nurses' personal social ties and resource access patterns, which items included 1) When experiencing work-related concerns, whom do you typically consult for discussion? Please list the initials of these individuals (up to five may be listed). 2) What is your frequency of contact with each of these individuals? Responses were measured on a 5-point Likert scale ranging from "Very rarely" to "Daily", with higher scores indicating stronger relationship strength. These two items measure network size and tie strength, respectively. Based on network size (n) and tie strength (Sij), degree centrality (Degreei) was calculated using the formula (
). Degree centrality serves as an indicator of an individual's level of social engagement within the network [23]. This metric was incorporated into the structural equation modeling (SEM) framework.
(4) The Perceived Social Support Scale (PSSS), developed by Zimet[24], was used to assess social support. This instrument comprises three dimensions: other support (from colleagues or supervisors, etc.), family support, and friend support, with four items per dimension. Items are rated on a 7-point Likert scale ranging from 1 ("strongly disagree") to 7 ("strongly agree"), yielding a total score between 12 and 84. Higher total scores indicate greater perceived social support. In our study, the scale demonstrated excellent internal consistency, with Cronbach's α coefficients of 0.965 in the pilot survey and 0.970 in the formal survey.
(5) The Pittsburgh Sleep Quality Index (PSQI) is a widely used instrument for measuring sleep quality, particularly applicable for assessing healthcare workers' sleep patterns over a one-month period[25]. It comprises seven components, including subjective sleep quality, sleep latency, sleep duration, sleep efficiency, sleep disturbances, use of sleep medication, and daytime dysfunction. The global PSQI score ranges from 0 to 21, with higher total scores indicating poorer sleep quality. In ourstudy, the PSQI demonstrated acceptable internal consistency, with Cronbach's α coefficients of 0.740 in the pilot survey and 0.761 in the formal survey.
A
2.3.2 Qualitative Interview ProtocolA
Based on the relationships between variables identified in the quantitative study, our research team developed a semi-structured interview protocol guided by JD-R model. (1) Could you describe which nursing job demands might undermine safety behaviors, using specific examples from your clinical practice? (2) What supportive resources in enhance your safety behaviors? (3) Please share a concrete example of how your social network has influenced your safety practices in clinical settings.2.4 Data Collection
2.4.1 Quantitative Study
A
Before survey, we obtained consent from all participants. The purpose and significance of the study were explained in detail, along with instructions for completing the questionnaire. Emphasis was placed on the anonymity and confidentiality of responses to ensure the authenticity of the data provided. Following data collection, two researchers independently reviewed all questionnaires. Those with obviously patterned, identical, or incomplete responses were excluded. A total of 489 nurses were surveyed, and 71 questionnaires with patterned responses were excluded, resulting in 418 valid surveys (85.48%).2.4.2 Qualitative Study
Before interviewed, the purpose, main content, privacy protection measures, and the need for audio recording were explained to the participants. The concepts of job demands, resource support factors, and social networks related to the interview themes were also clarified.
A
The interview commenced only after the participant had provided informed consent and confirmed their understanding of the procedure. A
The time and location of the interviews were arranged according to the participants’ convenience, with priority given to quiet and comfortable settings free from interruptions to facilitate open communication. A
During the interviews, any questions from the participants were addressed promptly. Probing and follow-up questions were used to explore responses in greater depth and to identify additional influencing factors. Each interview lasted approximately 15 to 30 minutes. Within 48 hours after the interview, the audio recordings were transcribed verbatim. The transcripts were then returned to the participants for validation to ensure accuracy. Data collection was terminated when thematic saturation was reached, indicated by the recurrence of similar responses and no emergence of new information.2.5 Data Analysis
2.5.1 Statistical analysis
Statistical description was performed using SPSS 27.0 software. Continuous variables were presented as mean ± standard deviation ( 
±SD) or median (P25, P75). Normality testing for variables included in the structural equation model was conducted using skewness and kurtosis (absolute values of skewness < 3 and kurtosis < 8 for all data indicated that the variables approximately followed a normal distribution[26]). Univariate analysis was carried out using one-way analysis of variance (ANOVA) and Pearson correlation analysis. Confirmatory factor analysis and model fit were tested using maximum likelihood estimation in AMOS 26.0. Mediation effects were examined using the Bootstrap test. The significance level was set at α = 0.05 (two-tailed).
2.5.2 Data Analysis and Integration of Results
A
The interview recordings were transcribed and imported into NVIVO 12.6 software for analysis. A directed content analysis approach was adopted as the following steps (1) Repeatedly reviewing the transcripts to gain an in-depth understanding and a holistic sense of the data; (2) Identifying meaningful statements relevant to the research questions and performing open coding; (3) Aggregating all codes, comparing and categorizing them iteratively, and grouping codes with similar attributes under broader categories (axial coding); (4) Developing themes and sub-themes based on the JD-R model and the objectives of the interview (selective coding); (5) Iterative reading and comparison to integrate findings from both quantitative and qualitative analyses. To ensure reliability, two researchers independently analyzed and extracted the data. Any discrepancies were resolved through group discussion within the research team until a consensus was reached.3.Results
3.1 Quantitative Findings
3.1.1 General Characteristics of the Survey Participants
A total of 418 nurses were included in this study, with an average age of (31.11 ± 7.16) years and a median work experience of 7 years (4 to 12 years). Detailed demographic characteristics are presented in Table 1.
Variables | n (%) | Variables | n (%) | |||
|---|---|---|---|---|---|---|
Gender | Female | 388(92.82) | Professional Title | Deputy Chief Nurse or above | 32(7.66) | |
Male | 30(7.18) | Charge Nurse | 103(24.64) | |||
Education | Bachelor's or higher | 298(71.29) | Nurse Practitioner | 183(43.78) | ||
Associate's or below | 120(28.71) | Staff Nurse | 100(23.92) | |||
Marital Status | Married | 227(54.31) | Teaching Responsibilities | Yes | 148(35.41) | |
Unmarried | 191(45.69) | No | 270(64.59) | |||
Number of Children | 0 | 196(46.89) | Department | Internal Medicine | 182(43.54) | |
1 | 96(22.97) | Surgical | 97(23.21) | |||
≥ 2 | 126(30.14) | Intensive Care Unit | 55(13.16) | |||
Monthly income | >9000 | 87(20.81) | Emergency | 41(9.81) | ||
6001–9000 | 205(49.04) | Pediatrics | 22(5.26) | |||
<6000 | 116(27.75) | Others | 21(5.02) | |||
Working Years | ≥ 15 | 79(18.90) | Night Shift Satisfaction | Satisfied | 117(27.99) | |
10–14 | 75(17.94) | Moderate | 253(60.53) | |||
5–9 | 145(34.69) | Dissatisfied | 48(11.48) | |||
<5 | 119(28.47) | Self-Rated Health Status | Very good | 62(14.83) | ||
Night Shifts /Month | 0 | 92(22.01) | Good | 132(31.58) | ||
≤ 3 | 40(9.57) | Fair | 196(46.89) | |||
4–5 | 184(44.02) | Poor | 28(6.70) | |||
≥ 6个 | 102(24.40) | Personality | Introverted | 93(22.25) | ||
Extroverted | 84(20.10) | |||||
Ambiverted | 241(57.65) | |||||
3.1.2 Analysis of degree centrality, PSSS, PSQI, and NBSQ scores using univariate analysis
The total score of NBSQ score in OUR study was (52.29 ± 7.85), with an average item score of ( 4.35 ± 1.05). Results of the one-way ANOVA revealed that NSBQ scores differed significantly based on professional title (F = 2.658, P = 0.048) and self-perceived health status (F = 4.232, P = 0.005). Results of the Pearson correlation analysis demonstrated that NBSQ score was significantly positively correlated with degree centrality and PSSS (P < 0.001), and significantly negatively correlated with the PSQI (P < 0.001). Specific scores for each variable and detailed analysis results are presented in Table 2.
Variable | Score(  | Degree Centrality | PSSS | PSQI | NSBQ |
|---|---|---|---|---|---|
Degree Centrality | 4.25 ± 0.80 | 1.000 | |||
PSSS | 66.42 ± 13.98 | 0.232** | 1.000 | ||
PSQI | 6.81 ± 3.56 | -0.273** | -0.310** | 1.000 | |
NSBQ | 52.29 ± 7.85 | 0.411** | 0.315** | -0.330** | 1.000 |
| ** The correlation was significant at the 0.01 level (2-tailed). | |||||
3.1.3 Structural Equation Modeling, Path Analysis, and Chain Mediation Analysis
The structural equation model was fitted and modified employing the maximum likelihood method with AMOS 26.0 software. The model demonstrated a good fit, with the following fit indices meeting the standard criteria: χ²/df = 2.581 (< 3), RMSEA = 0.062 (< 0.08), GFI = 0.892 (> 0.8), AGFI = 0.865 (> 0.8), CFI = 0.947 (> 0.8), TLI = 0.937 (> 0.8). The model is presented in Fig. 1. Specifically, degree centrality positively associated with PSSS. Both degree centrality and PSSS (as resource factors) positively associated with NBSQ. Conversely, the PSQI (as a job demand factor) negatively predicted NBSQ. The corresponding path coefficients are provided in Table 3.
 |
|---|
A Figure 1. The structural model of nurses' social networks, perceived social support, sleep quality, and nurse safety behavior. |
Path | Unstandardized coefficients | S.E. | C.R | P | Standardized coefficients |
|---|---|---|---|---|---|
Degree Centrality<---PSQI | -0.416 | 0.078 | -5.308 | <0.001 | -0.303 |
PSSS<---Degree Centrality | 0.795 | 0.269 | 2.957 | 0.003 | 0.151 |
PSSS<---PSQI | -2.263 | 0.445 | -5.084 | <0.001 | -0.312 |
NSBQ<---PSQI | -0.213 | 0.055 | -3.886 | <0.001 | -0.226 |
NSBQ<---Degree Centrality | 0.207 | 0.033 | 6.219 | <0.001 | 0.302 |
NSBQ<---PSSS | 0.022 | 0.007 | 3.371 | <0.001 | 0.171 |
The chain mediation effect of degree centrality and PSSS was examined using the Bootstrap sampling method (5000 iterations, 95% CI). The model fit indices for the mediation effect model were as follows: χ²/df = 2.780, RMSEA = 0.065, GFI = 0.880, AGFI = 0.852, CFI = 0.940, TLI = 0.932, indicating a good model fit. The results revealed a significant chain mediation effect of degree centrality and PSSS between PSQI and NSBQ, accounting for 18.09% of the total effect. The path coefficients for the chain mediation pathways are presented in Table 4.
Effect Type | β | SE | P | 95%CI | Proportion(%) | |
|---|---|---|---|---|---|---|
Lower | Upper | |||||
Total effect | -0.481 | 0.047 | < 0.001 | -0.651 | -0.355 | - |
Direct effect | -0.224 | 0.061 | < 0.001 | -0.364 | -0.116 | 46.57 |
Indirect effect | ||||||
PSQI→Degree Centrality→NBSQ | -0.098 | 0.026 | < 0.001 | -0.160 | -0.055 | 20.37 |
PSQI→PSSS→NBSQ | -0.058 | 0.023 | 0.002 | -0.116 | -0.023 | 12.06 |
PSQI→Degree Centrality→PSSS→NBSQ | -0.087 | 0.041 | 0.003 | -0.087 | -0.087 | 18.09 |
3.2 Qualitative Research Results
3.2.1 General Characteristics of Qualitative Interview Participants
Based on NSBQ scores, 23 nurses were categorized into three groups: low-level (n = 9, 39.14%), medium-level (n = 7, 30.43%), and high-level (n = 7, 30.43%). There were 21 females (91.30%) and 2 males (9.70%). The age was (31.57 ± 6.71) years. The years of working experience were (9.04 ± 5.50) years, with a range of 1 to 21 years. Regarding professional titles, there were 8 Staff Nurses (34.78%), 7 Senior Nurses (30.43%), 5 Charge Nurses (21.74%), and 3 Associate Chief Nurses (13.04%).
3.2.2 Interview Results
(1)Job Demands Related to Nurse Safety Behavior
1) Heavy Workload. Approximately 50% of interviewees indicated that a heavy workload negatively influenced their adherence to safety behaviors. 5 nurses specifically reported that overload led to physical or psychological fatigue, resulting in suboptimal safety performance.
N9: A sudden influx of patients or emergency resuscitations can trigger a sense of urgency, leading to deviations from standard operating procedures.
N10: During peak workloads, there is insufficient time to attend to details, increasing the likelihood of oversights.
N16: Excessive nursing tasks make it impractical to complete all duties in strict accordance with protocol.
N13: Work intensity often induces mental and physical exhaustion, which naturally contributes to negligence.
N22: Heavy workload, combined with night-shift fatigue and stress, readily leads to ineffective patient communication and potential safety risks.
2)High Task or Overload Time Demands. Four interviewees expressed that leadership’s excessively meticulous job requirements often led to the neglect of critical patient safety issues during nursing care. With such tedious work demands, nurses were required to extend their working hours, which induced negative psychological states and consequently compromised care safety.
N6: When handover is delayed too long, I become rushed to leave and tend to forget some important safety procedures.
N16: The demands are overly perfectionistic. We end up focusing on minor issues instead of prioritizing prominent ones that could prevent unsafe practices.
N20: Some things don’t need to be so heavily scrutinized. Overemphasis on details can be counterproductive—the top priority should always be ensuring patient safety.
3) Poor Physical Work Environment. During nursing practice, a noisy and disorganized work environment can induce psychological discomfort, impede effective communication with patients, and potentially compromise safety behaviors.
N7: The large number of patients and family members creates a noisy environment, which also undermines nursing safety.
N12: I currently feel that the unit environment is rather poor and chaotic, which contributes to patient dissatisfaction and hinders effective safety-related communication.
N16: The moment I step into the workplace, hearing loud noises from various medical devices and seeing a cluttered environment, I immediately experience an unpleasant psychological response.
3)Sleep Disturbances and Negative Emotions. The majority of interviewees indicated that personal or work-related emotional and negative psychological states—including negative occupational mentality—were significant contributors to unsafe behaviors. Additionally, eight respondents reported that insufficient sleep or poor sleep quality resulted in suboptimal work performance and reduced engagement in safety behaviors.
N1: Lack of sleep brings a general sense of lethargy, and one cannot perform effectively at work.
N8: When in a negative mood, it becomes difficult to fully focus on the patient's condition, increasing the likelihood of unsafe practices.
N6: Frequently experiencing poor sleep often leads to a detached and indifferent attitude when arriving in the unit.
N16: One’s emotional state greatly influences safety behavior. Doing the same tasks day after day, year after year, eventually leads to feelings of weariness—and naturally, less emphasis on safety procedures.
(2) Job Resource Related to Nurse Safety Behavior
1) Collaborative and Supportive Work Environment. The majority of interviewees indicated that a mutually supportive work environment among colleagues contributes to the promotion of nurses' safety behaviors. Moreover, such mutual assistance helps alleviate frustration associated with excessive workloads.
N4: The relationships among colleagues in our department are highly harmonious, and the working atmosphere is very positive. This enables us to remind each other about potential safety incidents.
N14: During an exceptionally busy period, I experienced significant frustration. However, with everyone’s support, we managed to maintain safety despite the heavy workload.
N16: When colleagues have strong relationships and assist each other during shifts, the occurrence of unsafe behaviors is likely to be significantly reduced.
N22: If collaboration is inadequate, it becomes difficult to implement safety practices effectively, as we function collectively as a team.
2) Receiving Care and Support Form Leaders and Colleagues. 7 nurses said that care and support from both head nurse and colleagues facilitated nurses’ active engagement in safety behaviors.
N8 : When I was in a negative emotional state, my colleagues would counsel me, which helped me maintain a positive mindset during work.
N14: The practice of humanistic nursing is highly valuable. When I encountered upsetting situations, colleagues showed concern for me, which also reminded me to be more attentive to patients and deliver more meticulous and safe care.
N19: When I experienced emotional difficulties, the head nurse would also show concern and ask whether I needed time off, otherwise, persisting working might lead to potential risks.
3) Emotional Support from Family and Friends. Emotional support from families and friends, can alleviate work-related psychological distress, enhance sense of nursing professional identity, and thereby facilitate greater engagement in safety behaviors.
N6: My families strongly support my career as a nurse and consistently reassures me that nursing is a meaningful profession. My friends also express respect for our work, which helps me maintain a positive attitude toward my job.
N8: Whenever I encounter difficulties, my friends are always there and counsel me, which allows me to move past.
N11: When I made a mistake at work, my mother criticized me, which motivated me to improve my performance.
4) Effective Safety Training and Priority Management. Effective training and warning in nursing safety practice by leadership, can help compensate for a lack of experience, raise awareness of safety behaviors, and foster the development of sound safety-related nursing habits.
N2: Both newly recruited nurses and their preceptors require strengthened training to develop habitual safety practices.
N5: Occasional quality inspections have limited effect, as the attitudes and behaviors displayed during inspections often differ from those in daily practice.
N9: If head nurse could consistently emphasize safety, we would pay more attention to safety behaviors.
2.3 Integration of Quantitative and Qualitative Findings
A narrative integration approach was adopted to combine the quantitative and qualitative results. Details are presented in Table 5.
Table 5 Influencing factors of nurses’ safety behavior and integration of action path results | ||||
|---|---|---|---|---|
Dimensions | Theme Description | Quantitative findings | Qualitative Findings | Inferences |
Job Demands | Sleep Quality–Psychological | PSQI (β=-0.159, P = 0.004) | Sleep disturbances or negative emotions (-) | Consistency: Poor sleep quality negatively impacts nurses' safety behavior. Extension: Qualitative study supplements that the negative pathway through which sleep affects nurses' safety behavior, indicating it is a response based on negative psychological changes. |
Workload– Physical/ Mental Fatigue | Self-rated health (F = 4.232, P = 0.005) | High workload leads to physical or mental fatigue (-) | Consistency: Poor psychological or physical health status negatively impacts nurses' safety behavior. Extension: Qualitative study explains that high workload declining nurses' physical or mental health status, further detailing the negative pathway to safety behavior. | |
Task or Time Demands | - | High task or time demands (-) |
| |
Physical Work Environment | - | Poor physical work environment (-) |
| |
Job Resources | Interpersonal Relationships –Positive Psychology | (1) Degree centrality (β = 0.207, P<0.001); (2) PSSS (β = 0.022, P<0.001). | ༈1༉ Cohesive and supportive work atmosphere (+); ༈2༉ Care and support from colleagues or supervisors (+) ; ༈3༉ Emotional support from family or friends (+). | Consistency: Positive interpersonal interactions positively influence nurses' safety behavior. Extension: Qualitative study explains that interpersonal relationships can also alleviate nurses' negative emotions and enhance professional identity, thereby promoting their safety behavior. |
Safety Management | - | Effective safety training and safety-prioritized management (+) | Extension: Qualitative study supplements that appropriate safety reminders and training from leadership promote nurses' safety behavior. | |
| Note: *Data Source: No relevant quantitative data; attribute findings from qualitative study. (+) Positively influences nurses' safety behavior. (-) Negatively influences nurses' safety behavior. | ||||
3. Discussion
3.1 Nurses' sleep quality negatively predicts Nurses’ safety behavior
Sleep disorders among nurses represent a global concern, with a reported overall prevalence of 61.03% worldwide[27]. The proportion with 73.68% in our study which is notably higher than this global estimate. However, it remains consistent with the prevalence reported among psychiatric nurses in China (71.51%)[28], suggesting a generally poorer sleep quality within the Chinese nursing population. Both quantitative and qualitative findings indicated that sleep deprivation, as measured by the PSQI, negatively impacted safety behavior through the mediation of adverse work states and emotional responses. Research indicates that poor sleep quality contributes to occupational stress in nurses[29], which maybe a key pathway through which it further undermines their safety behavior. Thus, the adverse effect of sleep on nurses' safety behaviors could be a consequence of psychological alterations. Furthermore, sleep deprivation impairs emotion regulation[7], which reducing safety behaviors. This occurs through two primary pathways: first, sleep disruption elevates the secretion of stress hormones, thereby increasing perceived work pressure[30]; second, sleep disorders adversely affect mental health, which subsequently undermines safety practices[31]. Therefore, nurses' sleep quality must be optimized. Simulation studies suggest that forward-rotating shift schedules (i.e., morning-evening-night) are beneficial in reducing sleep disturbances and work-life imbalance, and thus, backward-rotation should be avoided[32]. Furthermore, incorporating short naps (15–20 minutes) during shifts, as recommended by the Sleep Health Foundation, can enhance alertness among shift workers[33]. This practice not only improves nursing safety but also mitigates post-night-shift fatigue and subsequent sleep disorders. Additionally, the physical work environment should be optimized, for instance, by employing specific-wavelength lighting (e.g., LED) to help regulate circadian rhythms[34].
3.2 Nurses' social networks positively predict Nurses’ safety behavior.
In our study, nurses had a moderate network size and a moderately tie strength. Path analysis demonstrated that degree centrality positively predicted safety behavior, suggesting that established interpersonal relationships within the nursing context play a beneficial role and facilitate engagement in safety practices. This is likely because nurses with higher degree centrality are more active in social interactions, enabling them to garner greater emotional support and share perspectives on safety, thereby enhancing their access to valuable social resources.
A
According to the qualitative data, interpersonal support—from leaders for psychological backing and from colleagues for sharing burdens—was pivotal in helping nurses manage negative emotions, thus encouraging safer practices. As well as conversations with family and friends, which were found to improve adherence to safety behaviors by reinforcing professional identity. A study conducted in China[35] revealed that nurses' degree centrality in social networks positively influences their organizational citizenship behavior. This finding further supports the conclusion that social connect among nurses contributes to enhanced safety behavior. Thus, nursing managers should gain insights into the structure of these networks and implement effective strategies to optimize this resource. Peer support contributes to nurses' positive social psychology, while supportive family relationships improve work safety by reducing work-family imbalance[36]. Consequently, it is imperative for nurse managers to establish tailored communication platforms and support robust nurse-family connections.3.3 The Serial Mediating Effect of Social Networks and Perceived Social Support between Sleep Quality and Nurses’ Safety Behavior
Our findings revealed that sleep quality predicted safety behavior both directly and indirectly. The indirect pathways were mediated by two parallel factors: the degree centrality of social networks and perceived social support, both of which were diminished by poorer sleep, thereby negatively impacting safety behavior. The following two mechanisms may explain how sleep quality compromises social interaction. Firstly, sleep deprivation may decrease the secretion of key neurotransmitters essential for social behavior, thus reducing social interaction[37]. Secondly, poor sleep quality increases sleep need, which in turn diminishes the time and motivation for social connections, resulting in lower degree centrality. The consequent reduction in emotional support and psychological resources ultimately compromises safety behavior, establishing degree centrality as a partial mediator in this relationship. Research indicates that sleep deprivation compromises emotional regulation[38], leading to a heightened sensitivity to negative emotions. This increased sensitivity predisposes individuals to focus more on the negative aspects of social interactions, thereby resulting in a diminished perception of social support. Since reduced self-regulatory capacity undermines safety behavior[7], as well as poor sleep quality adversely affects safety performance by decreasing perceived social support. Ours study identified a serial mediation model in which sleep quality influences safety behavior through the sequential pathway of social network centrality and perceived social support. To mitigate this risk, nursing managers should focus on improving sleep environments, enhancing team building, and promoting peer interaction and support, thereby fostering greater safety compliance and elevating overall care quality.
4. Conclusion
Based on the JD-R model and employing a mixed-methods approach, our study provides an in-depth exploration of the pathways through which social networks, perceived social support, and sleep quality either motivate or deplete nurses' safety behavior. It confirms the serial mediating roles of social network centrality and perceived social support in the relationship between sleep quality and nurses' safety behavior. The findings underscore that initiatives aimed at improving nurses' psychological well-being must prioritize sleep quality and interpersonal relationships as critical entry points to enhance safety performance. However, there are some limitations. Firstly, the mechanisms of different types of social networks was not well explained. Secondly, its cross-sectional design precludes the identification of dynamic causal relationships among the variables. Future research should incorporate longitudinal designs, expand sample sizes, and further investigate the causal relationships between various social network types and safety behavior to gain a deeper understanding of the supportive mechanisms within nurses' social networks.
.
Declare Conflict of Interest
The authors declare that there are no conflicts of interest regarding the publication of this paper. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.
Ethics approval and consent to participate
A
This study was performed in line with the principles of the Declaration of Helsinki. A
Approval was granted by the Ethics Committee of The Second Affiliated Hospital of Guangxi Medical University (Approval No.: 2023-KY(0943)). A
Informed consent was obtained from all individual participants included in the study. All participants were informed about the purpose of the study and that their anonymity would be preserved.A
Funding
A
This work was supported by The 2nd Research Development Program in Humanistic Nursing of the Humanistic Nursing Committee, China Association for Life Care(RW2024PY29).A
A
Author Contribution
J.P. and X.L.F. wrote the main manuscript text ， handling data processing. X.Q.Z. and H.Z.H. provided support for data collection. J.L. and X.Z.L. prepared figures and tables. Y.Q.H guided study design and reviewed the manuscript. All authors collected the data and reviewed the manuscript.
Reference
1.
Organization WH. Patient safety[EB/OL]. (2023-09-11)[2025-02-10]. https://www.who.int/news-room/fact-sheets/detail/patient-safety
2.
Hodkinson A, Tyler N, Ashcroft DM, et al. Preventable medication harm across health care settings: a systematic review and meta-analysis.[J]. BMC Med. 2020;18(1):313.
3.
Sanni-Anibire MO, Mahmoud AS, Hassanain MA, et al. A risk assessment approach for enhancing construction safety performance[J]. Saf Sci. 2020;121:15–29.
4.
Griffin MA, Neal A. Perceptions of safety at work: a framework for linking safety climate to safety performance, knowledge, and motivation.[J]. J Occup Health Psychol. 2000;5(3):347–58.
5.
Seo JK, Lee SE. Spillover effects of organizational support for patient and workplace safety on safety outcomes: The mediating role of safety compliance.[J]. Journal of advanced nursing; 2024.
6.
Su MY, Zhang SL, Chen YL, et al. Correlation between occupational stress and safety behavior among newly graduated nurses in standardized operating room training. China J Health Psychol. 2021;29(12):1854–8.
7.
Liu M, Liu L, Lv Z, et al. Effects of burnout and work engagement in the relationship between self-efficacy and safety behaviours-A chained mediation modelling analysis.[J]. J Adv Nurs. 2024;80(4):1473–83.
8.
Shi H, Huang T, Schernhammer ES, et al. Rotating Night Shift Work and Healthy Aging After 24 Years of Follow-up in the Nurses' Health Study.[J]. JAMA Netw open. 2022;5(5):e2210450.
9.
Matricciani L, Clarke J, Wiley S, et al. Sleep of Nurses: A Comprehensive Scoping Review.[J]. Journal of advanced nursing; 2024.
10.
Olthuis SG, Hinsenveld WH, Pinckaers FM, et al. Association between type of intervention center and outcomes after endovascular treatment for acute ischemic stroke: Results from the MR CLEAN Registry[J]. Eur Stroke J. 2023;8(1):224–30.
11.
Yao L, Zhao Q, Bai D, et al. Path analysis of factors influencing registered nurses' safety behavior in China: A mediation effect model.[J]. Japan J Nurs science: JJNS. 2025;22(1):e12638.
12.
Gerace A, Rigney G. Considering the relationship between sleep and empathy and compassion in mental health nurses: It's time.[J]. Int J Ment Health Nurs. 2020;29(5):1002–10.
13.
Xie W, Chen L, Feng F, et al. The prevalence of compassion satisfaction and compassion fatigue among nurses: A systematic review and meta-analysis.[J]. Int J Nurs Stud. 2021;120:103973.
14.
DePasquale N. Family-supportive supervisor behaviour positively affects work behaviour and nonwork well-being among men in long-term care.[J]. J Nurs Adm Manag. 2020;28(7):1504–14.
15.
Mathis Schulte, N A C K. The Coevolution of Network Ties and Perceptions of Team Psychological Safety | Organization Science [J]. Organ Sci. 2010;23(2):299–596.
16.
Chen C, Zhu Y, Sun Y, et al. The relationship between social support and interpersonal self-efficacy among higher vocational college students: parallel mediation effects of anxiety and loneliness.[J]. BMC Psychol. 2025;13(1):102.
17.
Langford CP, Bowsher J, Maloney JP, et al. Social support: a conceptual analysis.[J]. J Adv Nurs. 1997;25(1):95–100.
18.
Demerouti E, Bakker AB, Nachreiner F, et al. The job demands-resources model of burnout.[J]. J Appl Psychol. 2001;86(3):499–512.
19.
Bakker AB, Demerouti E. Job demands-resources theory: Taking stock and looking forward.[J]. J Occup Health Psychol. 2017;22(3):273–85.
20.
Cheng H, Yang H. Advances in the application of the Job Demands-Resources theory in the field of nursing. J Evidence-Based Nurs. 2021;7(16):2181–5.
21.
Chou PMBC. Practical Issues in Structural Equation Modeling[J]. Volume 16. Sociological Methods & Research; 1987. 1.
22.
Rong YF. Exploring the Relationship Between Patient Safety Culture and Safety Behavior [Dissertation]. Hualien: Tzu Chi University, Graduate Institute of Public Health; 2009.
23.
MARTIN G E P S P. D. Some centrality results new and old [J]. J Mathemat Soc. 2004;28(4):215–27.
24.
Zimet GD, Dahlem NW, Zimet SG, et al. The Multidimensional Scale of Perceived Social Support[J]. J Pers Assess. 1988;52(1):30–41.
25.
Liu ZX, Shi QP, Yang HX, et al. Evaluation of reliability and validity of the Pittsburgh Sleep Quality Index in nursing populations. J Shantou Univ Med Coll. 2020;33(3):176–80.
26.
McDonald RP, Ho MR. Principles and practice in reporting structural equation analyses.[J]. Psychol Methods. 2002;7(1):64–82.
27.
Imes CC, Barthel NJ, Chasens ER, et al. Shift work organization on nurse injuries: A scoping review.[J]. Int J Nurs Stud. 2023;138:104395.
28.
Lu L, Lok KI, Zhang Q, et al. Sleep disturbance and its association with quality of life among psychiatric nurses in China[J]. PeerJ. 2021;9:e10659.
29.
Yuan MZ, Fang Q. Latent class analysis of the sleep quality of night shift nurses and impact of shift-related factors on the occupational stress and anxiety[J]. J Adv Nurs. 2024;80(7):2772–84.
30.
Nollet M, Wisden W, Franks NP. Sleep deprivation and stress: a reciprocal relationship.[J]. Interface focus. 2020;10(3):20190092.
31.
Chueh K, Chen K, Lin Y. Psychological Distress and Sleep Disturbance Among Female Nurses: Anxiety or Depression?[J]. J transcultural nursing: official J Transcultural Nurs Soc. 2021;32(1):14–20.
32.
Di Muzio M, Diella G, Di Simone E, et al. Comparison of Sleep and Attention Metrics Among Nurses Working Shifts on a Forward- vs Backward-Rotating Schedule[J]. JAMA Netw Open. 2021;4(10):e2129906.
33.
Foundation SH. Healthy Sleep Guidelines for Shift Workers[EB/OL]. [2025-04-10]. https://sleepcarepro.com/sleep-hygiene/sleep-hygiene-for-shift-workers-strategies-and-schedules/
34.
Zhong C, Masters M, Donzella SM, et al. Electronic Screen Use and Sleep Duration and Timing in Adults.[J]. JAMA Netw open. 2025;8(3):e252493.
35.
Liu ZY, Wei WH, Zhang HM, et al. The influence of social network degree centrality and job engagement on nurses' organizational citizenship behavior. Mod Prev Med. 2018;45(6):1039–42.
36.
Hurtado DA, Dumet LM, Greenspan SA, et al. Social Network Analysis of peer-specific safety support and ergonomic behaviors: An application to safe patient handling.[J]. Appl Ergon. 2018;68:132–7.
37.
Zhu J, Chen C, Wu J, et al. Effects of propofol and sevoflurane on social and anxiety-related behaviours in sleep-deprived rats.[J]. Br J Anaesth. 2023;131(3):531–41.
38.
Sullivan EC, James E, Henderson LM, et al. The influence of emotion regulation strategies and sleep quality on depression and anxiety[J]. Cortex. 2023;166:286–305.