The effect of job stress, organizational support and self-compassion on nurses’turnover intention: the mediating role of compassion fatigue and compassion satisfaction

Abstract

Background

The turnover rate of nurses is a global issue. In China, nurses are experiencing high job-related burdens due to a shortage of nursing human resources. Reducing turnover intention is crucial since nurses’ intention to leave their jobs harms patient safety and health outcomes.

Purpose

To verify a hypothesized model of the relationships between job stress, organizational support, self-compassion and turnover intention, compassion fatigue and compassion satisfaction are introduced as mediating variables.

Methods

This cross-sectional study included 349 nurses from 3 comprehensive hospitals in Anhui Province, China. They completed questionnaires on job stress, organizational support, self-compassion, compassion fatigue, compassion satisfaction, and turnover intention. The data were analyzed via SPSS 27.0 and AMOS 23.0, and a structural equation model was constructed.

Results

The final model showed a desirable fit with significant paths. Job stress, compassion fatigue, and compassion satisfaction directly affect turnover intentions. Job stress and organizational support were significantly and indirectly related to turnover intention and the mediating role of compassion fatigue. In contrast, the mediating role of compassion satisfaction on turnover intention was insignificant. Moreover, compassion fatigue and compassion satisfaction have multiple mediating effects.

Conclusion

Job stress and organizational support are antecedents of turnover intention among nurses in comprehensive hospitals, and they affect turnover intention through compassion fatigue and compassion satisfaction. Therefore, it is important to find ways to improve organizational support and reduce job stress to alleviate compassion fatigue and thus reduce turnover intention. Intensifying support for nurses and developing intervention strategies from the perspective of compassion are crucial for nursing managers.

Introduction

Nurses, who account for 59% of healthcare workers [1], play a central role in providing healthcare services and are a key component of patient-centered healthcare systems [2]. The phenomenon of a worsening global shortage of nurses is becoming increasingly pronounced. Nursing staff shortages intensify nurses’ physical and emotional workloads, directly undermining the quality of care, eroding patients’ trust in hospitals, and impeding the healthcare industry’ s progress [3].

The number of registered nurses per 1,000 people in China has more than doubled over the past 10 years, from 2.36 in 2015 to 4 in 2025, as a result of the government’ s efforts to strengthen nursing endeavors and workforce development [4]. However, compared with the 2021 average of 9.2 registered nurses per 1,000 people among the Organization for Economic Cooperation and Development (OECD) member countries, there is still indicates a severe shortage [2]. Although the number of nurses in China has increased, there has been a recent upward trend in turnover intention, increasing from 35.2% [5] to 51.6% [6]. This level is very high even when compared to the turnover intention of nurses in Germany [7] and South Korea [8], which is approximately 30%. This suggests that merely implementing quantitative expansion strategies for nurses, such as increasing the nursing workforce, cannot fundamentally resolve the shortage of nursing personnel.

A high turnover tendency among nurses leads to workforce instability and a shortage of nursing resources, thereby impacting the quality of nursing services [9]. Addressing nurses’ turnover intentions is crucial—not only by expanding staffing numbers but also by attending to the deeper emotional and psychological strains that drive nurses away from the profession. Therefore, this study aims to identify the structural pathways of turnover intention based on the extended job demands-resources (JD-R) model among nurses, and to explore strategic approaches for sustainable nursing workforce management.

Background

Numerous countries are scaling up enrollment in nursing education programs to prioritize the cultivating of nursing talent to replace retiring nurses and address the ongoing shortage of the nursing workforce [10]. However, persistent challenges in alleviating the deficit in nursing labour remain [2]. Nursing staff shortages exacerbate workloads, leading to a decline in professionals’ quality of life and ultimately increasing their intention to leave the profession [11]. Turnover intention refers to considering a career transition and is a critical determinant of nurse attrition [12]. Therefore, reducing turnover intention has emerged as a key strategic priority for ensuring workforce sustainability.

Nursing scholars have identified various factors associated with nurses’ turnover intention [5–8]. However, most studies have been limited to examining individual relationships between variables, and there is a lack of an integrated explanation of the psychological pathways through which these factors lead to turnover intention. To address these limitations, researchers in the nursing field have adopted the Job Demands-Resources (JD-R) model to identify factors and processes that can mitigate nurses’ turnover intentions [13–15]. It categorizes all work characteristics into two significant types: job demands and job resources [16]. To address the limitations of the original model, the expanded JD-R model incorporates the concept of personal resources, thereby enhancing its explanatory power and practical applicability [17]. Job demands, job resources and personal resources can generate two psychological processes: stress and motivation [17]. Even when the work environment constantly poses physical and psychological job demands, if sufficient personal resources and job resources are available to cope with these demands, employees experience high motivation levels, leading to positive organizational outcomes. Conversely, if inadequate job and personal resources are available to address job demands, job strain will be affected, resulting in adverse organizational outcomes. However, despite these theoretical strengths, studies that have empirically examined nurses’ turnover intention by focusing on integrated mediating pathways within the expanded JD-R model remain limited.

Job stress is one of the representative job demands. In the modern healthcare environment, nurses’ job stress has increased compared to the past due to expanded responsibilities such as infection control and safety management, patient advocacy, and quality improvement (QI) activities [6]. Heavy workloads, patient conflicts, and poor working environments contribute to excessive job stress among nurses, increasing job demands and leading to tension, anxiety, sleep disturbances, job burnout, and poor quality of life [18]. Researchs [19, 20] indicates that when they experience high job stress, they adopt stress-avoidant behaviors and increase their turnover intention. Moreover, as job stress intensifies, the degree of compassion fatigue increases, whereas the level of compassion satisfaction diminishes [21]. Moreover, as job stress intensifies, it has been shown that the professional quality of life deteriorates, with the degree of compassion fatigue increasing while the level of compassion satisfaction diminishes.

To buffer the negative effects of job demands, the role of job resources that can moderate these effects is crucial. Organizational support, including emotional and instrumental support such as compensation, work environment, and staffing support [22], is a work resource that reduces negative experiences related to job demands and stimulates work motivation, thereby facilitating personal growth and development [13]. Organizational support directly and indirectly influences nurses’ turnover intention [23]. Research has shown that lack of organizational support can negatively impact nurses’ physical and mental health and their professional quality of life, leading to increased turnover intention [24].

Since nurses’ turnover intention is shaped not only by job demands and job resources but also by the interaction with individual internal resources, it should be considered in an integrated manner. While nursing work emphasizes compassion and empathy toward those being cared for, the importance of self-compassion for nurses is often overlooked. Self-compassion involves being open to one’s suffering, wanting to alleviate it, and treating one self-compassionately [25]. Self-compassion fosters an attitude of kindness, gentleness, and cooperation with oneself, including the desire to ease or heal one’s pain. Unlike generic personal resources such as self-efficacy, self-compassion shares a conceptual root with the very emotional processes central to nursing. It promotes coping and adaptability, alleviates compassion fatigue [26], and enhances compassion satisfaction among nurses [27]. Additionally, self-compassion helps nurses cope with workplace challenges and reduces turnover intentions [26]. Therefore, self-compassion can serve as a personal resource that enables individuals to cope with their surrounding environment successfully.

Nursing practice is rooted in compassion and caring, and nurses often choose this profession to alleviate patients’ suffering [28]. However, beyond personal motives, the sociocultural context significantly shapes their emotional experiences. In contrast to the individualism prevalent in Western societies, Asia (particularly China), influenced by Confucianism, emphasizes collectivism and altruism [6]. This culture highly praises nurses’ dedication but tends to neglect their emotional needs, as individual affective dimensions are often overshadowed by collective goals. Therefore, nurses may experience conflicting psychological states, namely compassion satisfaction derived from helping others as a motivational factor and compassion fatigue resulting from emotional investment without sufficient attention and support as a source of stress.

Compassion satisfaction, also known as the “reward of love”, refers to the fulfillment and pleasure derived from helping others and performing their duties well. It provides a positive outlook and enhances the ability to care for patients [29]. When nurses experience higher compassion satisfaction at work, they tend to be more positive and engaged in their job [30]. compassion fatigue is a negative outcome that arises when helping those who have experienced trauma and pain. A systematic literature review and meta-analysis revealed that nurses experience a greater level of compassion fatigue and a lower level of compassion satisfaction than other healthcare professionals [31]. Currently, nurses in Asian regions, with China as a representative, report the lowest level of compassion satisfaction compared with their counterparts in other regions (the Americas, Europe, and Australia) [32]. In contrast, they exhibit the highest level of compassion fatigue [32]. Wells-English et al. [33] discovered that compassion fatigue is an essential predictor of turnover intention, whereas compassion satisfaction is a protective factor against turnover intention.

This study aims to identify the structural pathways through which job demands, job resources, and personal resources lead to turnover intention, in order to ensure the sustainability of the nursing workforce. By applying the expanded JD-R model, the study focuses on nurses in Chinese general hospitals and examines the pathways among job stress, organizational support, self-compassion, compassion fatigue, compassion satisfaction, and turnover intention. It seeks to contribute to the development of practical and integrated management strategies for reducing nurses’ turnover intention. To achieve this, we proposed the following hypothesis:

Hypothesis 1

Organizational support, self-compassion, and job stress directly affect turnover intention.

Hypothesis 2

A mediating role of compassion fatigue exists between job stress, organizational support, self-compassion and turnover intention.

Hypothesis 3

A mediating role of compassion satisfaction exists between job stress, organizational support, self-compassion and turnover intention.

Hypothesis 4

Multiple mediating roles of compassion fatigue and compassion satisfaction exist between job stress, organizational support, self-compassion and turnover intention. A theoretical model illustrating these relationships is presented in Fig. 1.

Fig. 1.

A hypothesized model

Methods

Study design and participants

This study is designed as a descriptive cross-sectional survey. Based on the recommendation of 20 observations per parameter for model testing [34], and given that the hypothesized model involved 17 parameters, an initial minimum sample size of 340 participants was calculated. To account for an anticipated 10% non-response rate, the final target sample size was established at 378.

This study was conducted from 2022.11 ~ 2023.2. The participants were recruited via convenience sampling. The inclusion criteria were as follows: (1) possession of a valid nursing practice license; (2) providing voluntary informed consent; (3) having at least six months of cumulative clinical work experience (i.e., the standard probationary adaptation period). The exclusion criteria were as follows: (1) being on sick leave, maternity leave, or other long-term leave during the survey period; (2) having a past or present diagnosis of a psychological disorder or mental illness; (3) being administrative nurses or nursing managers not directly involved in patient care; (4) being intern or training nurses.

A total of 380 nurses from three general hospitals in Anhui Province, China, were recruited for this study, of whom 356 ultimately participated. After seven incomplete responses were excluded, 349 valid questionnaires were analyzed, yielding an effective response rate of 91.8%. The survey took approximately 15–20 min to complete. The study was approved by the Institutional Review Board (IRB) of the CUP (Approval No.CUPIRB-2022-049). Upon identifying eligible respondents, full disclosure was provided before written informed consent was obtained.

Measurements

Demographic and employment-related characteristics

A questionnaire was designed to collect participants’ demographic and employment-related characteristics, including gender, age, educational level, marital status, years of work experience, professional title, and department.

Job stress

The Chinese Nurse Stressors Scale (CNSS) was used to measure nurses’ work-related stress. This scale was developed by Li et al. [35] in 2001 and aims to assess the level and primary sources of occupational stress among Chinese nurses. The scale originally consists of 35 items, categorized into five subscales. Based on expert input, two subscales (work environment and resources, organization and interpersonal relationships) were found to overlap with the content of the organizational support tool; therefore, only three subscales (nursing profession and work, workload and time allocation, patient care) were adopted, retaining a total of 23 items. A 4-point Likert scale was used for scoring (1 = slight pressure to 4 = serious pressure). The total score for each subarea is converted to a standard scale of 100 via the following formula: [(actual score - number of items) / (highest possible score - number of items)] × 100. The scale is evaluated continuously, where higher scores indicate increased job stress. The Cronbach’s alpha of the dimensions in this study ranged from 0.894 to 0.922.

Organization support

The Perceived Organizational Support Scale (POSS) was used to assess nurses’ perceived organizational support. This scale was developed by Zuo et al. [36] in 2009 and aims to measure the level of perceived organizational support among occupational groups such as nurses. The scale consists of 13 items, rated on a 5-point Likert scale(1 = strongly disagree, 2 = disagree, 3 = neutral, 4 = agree, 5 = strongly agree), and is classified into two subscales: affective support (10 items) and material support (3 items). The total score ranges from 13 to 65, with all items positively scored. Higher scores indicate a stronger perception of organizational support. The scale demonstrated high reliability, with Cronbach’s alpha for each dimension ranging from 0.901 to 0.923 in this study.

Self-compassion

The Self-Compassion Scale (SCS) was used to assess nurses’ level of self-compassion. This scale was originally developed and validated by Neff et al. [37] in 2003 to measure an individual’s ability to treat themselves with kindness and understanding when facing personal inadequacies, failures, or suffering. The scale was subsequently translated and validated into Chinese by Chen et al. [38] in 2011. The SCS consists of 26 items, rated on a 5-point Likert scale(1 = almost never, 2 = rarely, 3 = sometimes, 4 = often, 5 = almost always). The scale is classified into six subscales: self-kindness (5 items), self-judgment (5 items), common humanity (4 items), isolation (4 items), mindfulness (4 items), and over-identification (4 items), the total score ranges from 26 to 130. All items from the three negative subscales (self-judgment, isolation, and over-identification) are reverse-scored, while the remaining items are positively scored. Higher total scores indicate a higher subjective level of self-compassion. The scale demonstrated high internal consistency, with Cronbach’s alpha for each subscale ranging from 0.855 to 0.894 in this study.

Compassion fatigue and compassion satisfaction

The Professional Quality of Life Scale (ProQOL) was used to assess nurses’ compassion fatigue and compassion satisfaction. This scale was originally developed by Stamm et al. [39] in 2005 to measure the levels of compassion fatigue and compassion satisfaction experienced by helping professionals. The scale was later translated and validated into Chinese by Shen et al. [40] in 2015. The Chinese version of the ProQOL consists of 30 items, rated on a 5-point Likert scale(1 = almost never, 2 = rarely, 3 = sometimes, 4 = often, 5 = almost always). The scale comprises two independently scored subscales: compassion fatigue (20 items) and compassion satisfaction (10 items). The score range is 20 to 100 for compassion fatigue, and 10 to 50 for compassion satisfaction.Items 1, 4, 15, 17, and 29 are reverse-scored, while the remaining items are positively scored, with higher scores indicating higher levels of the respective dimension. In this study, the Cronbach’s alpha for each dimension ranged from 0.913 to 0.937.

Turnover intention

The Turnover Intention Scale (TIS) was used to assess nurses’ turnover intention. This scale was originally developed by Michaels et al. [41] in 1982 and subsequently translated and validated into Chinese by Li et al. [42] in 2000. The scale aims to measure an individual’s intention to voluntarily leave their current organization. The TIS consists of six items, categorized into three subscales: likelihood of leaving (2 items), motivation to seek a job (2 items), and perceived possibility of finding a new job (2 items). Items are rated on a 4-point Likert scale (1 = almost never, 2 = rarely, 3 = often, 4 = almost always), with all items positively scored. The total score ranges from 6 to 24, with higher scores indicating stronger turnover intention. In this study, the Cronbach’s alpha for the overall scale was 0.841.

Procedures

Before data collection, the researcher visited the study hospitals, explained the purpose and methods of the research to the directors of the nursing departments, and obtained permission to collect data. Research assistants then visited various nursing units, explained the research purpose and content to the unit heads and nurses, and posted recruitment notices on workplace bulletin boards for 14 days. Nurses who voluntarily wished to participate in the study were required to complete a consent form for participation and data use via an online application (SoJump). During the data collection process, it was clearly communicated to all participating nurses that their participation was entirely voluntary and that they could discontinue their involvement at any point without facing any form of negative repercussions.

All the data were collected and processed by the researchers to ensure the authenticity and validity of the study. All the questionnaires used in this study had good reliability and validity. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines were followed and reported.

Data analysis

Data analysis was conducted via IBM SPSS Statistics (version 27.0) to compute descriptive statistics and Pearson’s correlations. To observe the effects of job stress, organizational support, self-compassion, compassion fatigue and compassion satisfaction on turnover intention, the hypotheses were tested via robust weighted least squares structural equation modeling (SEM) with IBM SPSS AMOS 23.0. Path analysis can simultaneously examine the direct and indirect effects of the variables and produce models. Model fit was evaluated via the following indices: normed chi-square(χ²/df), RMSEA, SRMR, GFI, AGFI, NFI, CFI and TLI. Standardized regression weights (SRW) along with their critical ratios and P-values were examined to assess the significance of the path coefficients within the hypothesized model. Furthermore, squared multiple correlations (SMCs) quantify the proportion of variance explained in the endogenous variables. The statistical significance of the model’s direct, indirect, and total effects was tested by bootstrapping (2000 resamples) with 95% bias-corrected CIs, with P < 0.05 considered significant.

Results

Sample characteristics

The vast majority of the 349 study participants were female (n = 343, 98.3%), with an average age of 37.24 years (SD = 8.33). A total of 81.7% of the participants held a bachelor’s degree, and 86.8% were married. The average length of service was 15.47 years (SD = 9.64), and over half of the participants (n = 184, 52.7%) held midlevel professional titles. The departments included surgical ward (n = 68, 19.5%), medical ward (n = 55, 15.8%), emergency departments (n = 20, 5.8%), critical care departments (n = 21, 6.0%), obstetrics and gynecology departments (n = 57, 16.3%), pediatrics departments (n = 25, 7.2%), and other departments (n = 103, 29.5%).

Correlation analysis

The correlation matrix was examined before the hypothesized model was tested (Table 1). The Pearson analysis revealed that compassion fatigue was significantly associated with job stress(r = 0.385, P < 0.001), organizational support (r=-0.281, P < 0.001), self-compassion (r=-0.047, P = 0.378) and compassion satisfaction(r=-0.452, P < 0.001); compassion satisfaction was significantly associated with job stress(r=-0.437, P < 0.001), organizational support (r = 0.460, P < 0.001) and self-compassion (r = 0.152, P = 0.005); turnover intention was significantly associated with job stress(r = 0.341, P < 0.001), organizational support (r=-0.225, P < 0.001), self-compassion (r=-0.074, P = 0.167), compassion fatigue(r = 0.432, P < 0.001) and compassion satisfaction(r=-0.437, P < 0.001).

Table 1.

Level of study variables and correlations of related variables(N = 349)

Variables	M(± SD)	skewness	kurtosis	VIF	1	2	3	4	5	6
1.JS	2.71 ± 0.51	-0.30	0.88	1.320	1
2.OS	3.15 ± 0.67	0.09	0.89	1.293	-0.291**	1
3.SC	3.29 ± 0.58	0.01	0.25	1.026	-0.052	0.030	1
4.CS	3.60 ± 0.73	-0.03	0.31	1.629	-0.437**	0.460**	0.152**	1
5.CF	2.53 ± 0.71	0.08	-0.26	1.336	0.385**	-0.281**	-0.047	-0.452**	1
6.TI	1.99 ± 0.56	0.06	-0.59	-	0.341**	-0.225**	-0.074	-0.437**	0.432**	1

M = Mean; SD = standard deviation; VIF=variance inflation factor; JS = job stress; OS = organizational support; SC = self-compassion; CF = compassion fatigue; CS = compassion satisfaction; TI = turnover intention

^**:P < 0.01; ^*:P < 0.05

Path analysis

Path analysis commenced with evaluating a hypothesized model comprising six variables, wherein all proposed paths were empirically tested. In an iterative process of model refinement, paths that did not achieve statistical significance were eliminated from the model, including the paths between self-compassion and compassion fatigue(β=-0.030, P = 0.614), organizational support and turnover intention(β = 0.016, P = 0.865), and self-compassion and turnover intention(β=-0.036, P = 0.541). The final model, presented in Fig. 2, retained only the statistically significant paths and demonstrated an acceptable model fit. (χ²/df = 2.516, RMSEA = 0.066, SRMR = 0.06, GFI = 0.916, AGFI = 0.881, NFI = 0.865, CFI = 0.913, TLI = 0.891). The indices for the present model exhibited adequate data fitting.

Fig. 2.

Final model of path analysis with standardized regression weights

All path coefficients in the final model were statistically significant. The specific parameter estimates for the relationships between variables are provided in Fig. 2. Together, job stress and organizational support jointly predicted 30.1% of the variance in compassion fatigue; job stress, organizational support and self-compassion collectively explained 53.3% of the total variance in compassion satisfaction; job stress, compassion fatigue and compassion satisfaction collectively explained 34.2% of the total variance in turnover intention. Table 2 summarizes the standardized total, direct, and indirect effects for all variables within the model.

Table 2.

Standardized total, direct and indirect effects of the final model

Dependent Variable	Independent Variable	Standardized total effects(P)	Standardized direct effects(P)	Standardized indirect effects(P)
CS	JS	-0.292(0.006)	-0.184(0.092)	-0.108(0.002)
	OS	0.477(0.001)	0.362(0.001)	0.115(0.001)
	SC	0.109(0.020)	0.109(0.020)	-
	CF	-0.343(0.001)	-0.343(0.001)	-
CF	JS	0.316(0.001)	0.316(0.001)	-
	OS	-0.336(0.001)	-0.336(0.001)	-
TI	JS	0.346(0.002)	0.188(0.045)	0.158(0.002)
	OS	-0.202(0.001)	-	-0.202(0.001)
	SC	-0.022(0.072)	-	-0.022(0.072)
	CS	-0.203(0.080)	-0.203(0.080)	-
	CF	0.383(0.002)	0.313(0.004)	0.069(0.070)

Abbreviations: JS, job stress; OS, organizational support; SC, self-compassion; CS, compassion satisfaction; CF, compassion fatigue; TI, turnover intention

Table 3 presents the significance levels for all paths in the final path analysis model. Bootstrapping results indicate a mediating effect if the upper and lower bounds of the 95% confidence interval do not contain zeros and no effect if they contain zeros [30]. Job stress (B = 0.031, P = 0.004) and organizational support (B=-0.020, P = 0.002) were significantly and indirectly related to turnover intention, and the mediating role of compassion fatigue, whereas the mediating role of compassion satisfaction on turnover intention was not significant. In addition, compassion fatigue and compassion satisfaction significantly multiple mediated the effects of job stress (B = 0.007, P = 0.044) and organizational support (B=-0.004, P = 0.048) on turnover intention.

Table 3.

Significance of the paths in the final model of path analysis

path	BΒ	SE	Bias-correctes 95%CI
			Lower	Upper	P
JS→CF→TI	0.031	0.021	0.006	0.088	0.004
OS→CF→TI	-0.020	0.013	-0.055	-0.005	0.002
JS→CS→TI	0.012	0.011	-0.001	0.046	0.099
OS→CS→TI	-0.014	0.009	-0.034	0.000	0.062
SC→CS→TI	-0.009	0.007	-0.027	0.000	0.072
JS→CF→CS→TI	0.007	0.005	0.000	0.020	0.044
OS→CF→CS →TI	-0.004	0.003	-0.011	0.000	0.048

JS: job stress; OS: organizational support; SC: self-compassion; CS: compassion satisfaction;

CF: compassion fatigue; TI: turnover intention

Discussion

On the basis of the expanded JD-R model, this study explored the compassion-centric pathways linking organizational support, self-compassion, job stress, and nurses’ turnover intention. By validating hypotheses 1, 2, and 4. Specifically, while compassion satisfaction does not mediate the relationships directly on its own, it significantly influences turnover intention through multiple mediating effects with compassion fatigue.

Direct effects of compassion fatigue, compassion satisfaction and job stress on turnover intention

Compassion fatigue, compassion satisfaction and job stress account for 38.7% of the variance in turnover intention, indicating their significant effects on turnover intention among nurses in Chinese general hospitals. As a result, in this study, compassion fatigue and compassion satisfaction are predictors of turnover intention among nurses in Chinese general hospitals. These findings are consistent with previous research [43, 44]. Nurses, like psychologists and social workers, assist others and may experience emotional transfers toward their clients during their work, resulting in significant physical and emotional energy expenditure. Such emotional transfer can lead to the accumulation of compassion fatigue, subsequently contributing to turnover intention [44]. However, when nurses experience high levels of compassion satisfaction, they can transform the stress of empathy arising from patient interactions into positive emotions and derive a sense of achievement from their work [45]. Therefore, while compassion fatigue increases turnover intention, compassion satisfaction serves as a factor that reduces it, suggesting that these constitute key psychological mechanisms explaining nurses’ turnover intention.

Job stress also directly impacts turnover intention, which aligns with previous research findings [6] and indicates that higher job stress is associated with increased turnover intention. When nurses experience considerable job stress, they may experience negative emotions, such as anxiety, tension, and fatigue, leading to decreased job satisfaction, and increased turnover intention [20]. Therefore, nursing managers need to make special efforts to create a positive work environment, such as through effective work design and staffing, to alleviate nurses’ workloads.

Mediating effects of compassion fatigue and compassion satisfaction

The study revealed that job stress can influence nurses’ turnover intentions through the mediating effect of compassion fatigue. This aligns with the theoretical predictions of the expanded JD-R model and is consistent with the nurse turnover models proposed by Kim et al. [45] and Yu et al. [44]. Excessive workload and continuous emotional labor exacerbate job stress [46, 47]. When this emotion continuously accumulates and is not effectively relieved, it significantly weakens practitioners’ empathy ability and induces compassion fatigue [46], ultimately leading to turnover [47]. In essence, job stress influences turnover intention through the stress pathway (compassion fatigue). Consequently, nursing organizations should prioritize the systematic identification of stressors and focus on mitigating workloads and optimizing staff allocation to reduce excessive job demands. Furthermore, it is essential to implement psychological support interventions, such as unit-based self-care programs [48] and compassion fatigue resilience programs [49], to alleviate job stress. On an individual level, nurses should be encouraged to regulate their stress through proactive measures, including regular physical examinations, yoga, exercise, and mindfulness practices [50].

The study found that organizational support affects nurses’ turnover intention through compassion fatigue, consistent with Yu et al. [44] and Galanis P et al. [23]. According to social exchange theory, members evaluate their relationship with an organization based on the balance of received support and rewards for their contributions [51]. In a supportive, team-oriented culture, nurses gain emotional support from leaders and colleagues, reducing individual emotional burden. Conversely, insufficient perceived organizational support hinders nurses’ ability to cope with negative emotions, leading to helplessness, despair, and ultimately turnover [51]. Therefore, nurse managers need to allocate nursing tasks reasonably and establish a supportive and inclusive work environment [6]. Furthermore, beyond mere task-oriented support, it is necessary to enhance perceived organizational support by simultaneously providing support for nurses’ career development and emotional well-being [50].

While the expanded JD-R model underscores the critical role of personal resources, this study found that self-compassion exerted no significant direct or indirect influence on turnover intention. This divergence from previous findings in Western [26, 27]and Japanese contexts [52] suggests that the efficacy of personal resources contingent may upon organizational and cultural contexts. In hierarchical healthcare environments, the protective buffering of personal resources may be overshadowed by dominant job demands.Future research should examine the generalizability of personal resources’ effects on turnover intention among Chinese nurses, and further explore self-compassion’s role across contexts. Nevertheless, this study verified that self-compassion directly impacts compassion satisfaction. Consistent with recent research among Chinese nurses [53], our findings indicate that while self-compassion’ s role in preventing turnover may be constrained, it contributes substantially to alleviating negative affect and fostering work engagement. To foster this resource, ‘self-briefing’ and ‘self-reflection’ should be utilized as coping strategies to help nurses rationalize emotions objectively [26]. Furthermore, Mindfulness-Based Stress Reduction (MBSR) facilitates non-judgmental self-awareness and reflection on personal needs, thereby enhancing self-compassion and compassion satisfaction [54]. To maximize the positive role of self-compassion identified in this study, it is recommended that organizations implement mindfulness-based intervention programs to support the emotional well-being of nurses.

Contrary to existing literature, this study found that job stress, organizational support, and self-compassion did not influence turnover intention through the mediating effect of compassion satisfaction. While previous research suggests that higher compassion satisfaction helps nurses navigate emotional challenges [55] and enhances professional identity [56], these findings were not replicated in the present context. This discrepancy may be rooted in the organizational and institutional framework of Chinese hospitals. Specifically, current performance evaluation scales predominantly emphasize clinical knowledge and technical skills, often overlooking the emotional and spiritual dimensions of nursing labor [57]. Consequently, nurses may find it difficult to perceive the inherent value of their emotional efforts, preventing them from deriving sufficient positive experiences or compassion satisfaction [56]. These results suggest that the theoretical protective function of compassion satisfaction—which should ideally offset negative emotional impacts—may be constrained by specific organizational and cultural contexts.

Multiple mediating effects of compassion satisfaction and compassion fatigue

The multiple mediating effects show that organizational support and job stress can indirectly affect compassion satisfaction through compassion fatigue and subsequently affect nurses’ turnover intention. These results can be explained by the Conservation of Resources (COR) theory. According to COR theory, individuals tend to acquire and protect valued resources [58]. Organizational support serves both resource gain and buffering functions: on one hand, it directly provides external resources (e.g., benefits, training) to reduce resource depletion; on the other hand, it buffers work stress, preventing excessive depletion of emotional and cognitive resources. Conversely, lack of support combined with chronic high stress leads to continuous resource loss, triggering and exacerbating compassion fatigue. According to the COR theory, when individuals are in a state of resource scarcity, their ability to acquire new resources is also impaired. Thus, compassion fatigue, as a state of resource depletion, weakens nurses’ empathic behavior and reduces compassion satisfaction [6]. Given that resources are limited and losses persist without replenishment, nurses tend to voluntarily withdraw from stressful situations to avoid further losses, thereby developing turnover intentions or even engaging in actual turnover behaviors [6]. The findings of this study suggest that reducing turnover intention requires moving beyond simple human resource management and developing psychological resource-based organizational strategies, including structural alleviation of job stress, enhancement of organizational support, and thereby reducing compassion fatigue and promoting compassion satisfaction.

The applicability of the expanded JD-R model in nurses’ turnover intention

This study applied the extended JD-R model to analyze the determinants of turnover intention among Chinese nurses. According to the findings of this study, as articulated by the expanded JD-R model, job stress as a job demand and organizational support as a job resource directly or indirectly influence nurses’ turnover intentions through the mediating psychological process of compassion fatigue. Concurrently, the research confirms that satisfaction mitigates and protects against compassion fatigue.

Based on the above findings, future efforts may systematically develop intervention strategies using an extended JD-R model as the theoretical framework. Our results underscore the necessity of alleviating job demands and bolstering organizational resources to counteract the developmental trajectory of compassion fatigue and its subsequent impact on turnover intention. First, from the perspective of job demands, managers should optimize staffing allocation and regulate emotional labor through reasonable job rotation, thereby alleviating occupational stress. Second, to enhance nurses’ job resources, nursing organizations can foster a humane organizational climate, recognize nurses’ emotional efforts, and improve their perceived organizational support. In addition, emotional support resources should be organized to help nurses manage negative emotions. Lastly, to enhance compassion satisfaction through interventions aimed at strengthening personal resources such as self-compassion, management should incorporate mental health training programs into the continuing education system. For example, the Mindful Self-Compassion (MSC) training program developed based on Neff’ s theoretical framework—to cultivate nurses’ personal resources [25].

Limitations

This study has several limitations. First, it omitted specific contextual and demographic variables, including salary, benefits, and employment type. The absence of socio-economic data may introduce unmeasured confounding effects regarding nurses’ intention to resign. Future research must control for these diverse demographic and environmental factors. Second, the reasons for the multimodal effects, rather than the mediation effects of compassion satisfaction in the structural model, remain unclear due to the lack of prior research. Third, since this study only tested the hypothetical model of self-compassion as a personal resource, and given the differences between this model and the expanded JD-R model, the role of personal resources remains unconfirmed. Future research should incorporate additional types of personal resource-related variables to fully demonstrate the explanatory power and applicability of the expanded JD-R model within the nursing population.

Conclusion

The findings of this study indicate that the theoretical framework constructed on the basis of the expanded JD-R model demonstrates specific applicability in exploring nurses’ turnover intentions through a compassion-centric lens. Organizational support and job stress are critical antecedent variables that influence turnover intention primarily through the mediating pathways of compassion fatigue and satisfaction. Therefore, for nursing managers, it is crucial to move beyond fragmented approaches and develop systematic intervention strategies that protect the emotional ecosystem and sustain the professional resilience of the nursing workforce.

Abbreviation

JD-R model

Job Demands-Resources model

Author contributions

Conceptualization, methodology, and formal analysis: Myoungju Jo, Miaomiao Li; Data collection: Miaomiao Li, Hongyan Jin; writing—original draft: Miaomiao Li; writing—review and editing: Myoungju Jo, Miaomiao Li; Data curation: Hongyan Jin.

Funding

The study was supported by the Grant Talent Support Fund of Anhui University of Chinese Medicine (grant/ award number: DT2400000517).

Data availability

The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

This study was conducted in accordance with the principles of the Declaration of Helsinki and the study plan and process were approved by the Clinical Ethics Committee of Catholic University of Pusan (Approval No. CUPIRB-2022-049). All participants were informed about the study details before participation, including its purpose, procedures, potential risks, benefits, confidentiality measures, and voluntary involvement. Informed consent was obtained from all participants, and the survey was conducted following relevant procedural guidelines.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.