Abstract

Background:

Perceived stress is a significant risk factor for depression among healthcare professionals. While this fundamental relationship is well-documented, the potential mediating mechanisms—specifically the role of psychological flexibility and its distinct components—in buffering the impact of perceived stress on depression within the nursing community remain poorly understood.

Methods:

This cross-sectional study recruited 3,920 nurses from three tertiary Grade A public hospitals in Xi’an, China, using convenience sampling, of whom 3,611 were included in the final analysis. Data were collected online using the Chinese versions of Cohen’s Perceived Stress Scale and the Comprehensive Assessment of Acceptance and Commitment Therapy Processes, as well as the Patient Health Questionnaire-9. Pearson correlation analyzes and mediation analysis were conducted to examine the associations among perceived stress, psychological flexibility, and depression.

Results:

Depression was positively correlated with perceived stress (r = 0.63, p < 0.01). Mediation analysis indicated that perceived stress was significantly associated with depression both directly (b = 0.47, 95% CI: 0.45 to 0.49) and indirectly via psychological flexibility (b = 0.12, 95% CI: 0.11 to 0.13) and its components, including acceptance and cognitive defusion (b = 0.14, 95% CI: 0.13 to 0.15), and values and committed action (b = 0.06, 95% CI: 0.04 to 0.08). The indirect association through mindfulness and self-as-context was not significant (b = 0.01, 95% CI: 0.00 to 0.02).

Conclusions:

Psychological flexibility and some of its components were involved in partial indirect associations between perceived stress and depression among Chinese nurses. These findings suggest that psychological flexibility may help understand how perceived stress is statistically associated with depression in this population. Given the cross-sectional design, however, the results should be interpreted as statistical associations rather than evidence of causal relationships.

1 Introduction

Nurses are frontline healthcare professionals who experience substantial stress and a considerable burden of depression (1–4). A China-led multicenter cross-sectional study reported that the prevalence of depression among nurses was 26% (5). Depression among nurses warrants attention not only because it affects their own physical and mental health, but also because it has been associated with poorer quality of care, reduced patient safety, and instability in the nursing workforce. Accordingly, addressing depression among nurses has become an increasingly important priority in nursing research and practice.

Perceived stress has long been recognized as an important correlate of depression (6, 7). Among nurses, higher levels of perceived stress have been linked to greater depression, and some studies suggest that perceived stress explains a substantial proportion of the variance in depression (8). Perceived stress tends to increase when individuals appraise external demands as exceeding their available coping resources (9). Even so, the psychological processes through which perceived stress is associated with depression remain insufficiently understood. In this context, examining potential mediating mechanisms between perceived stress and depression among nurses is important, as a better understanding of this association may help identify potential intervention targets and inform the selection of appropriate preventive and supportive strategies.

Psychological flexibility refers to the ability to fully engage in the present moment as a conscious individual and to alter or maintain behavior flexibly to pursue valued goals, especially when faced with unexpected life events (10, 11). It comprises three components: acceptance and cognitive defusion, mindfulness and self-as-context, and value and committed action. Numerous studies have identified that psychological flexibility and its components are correlated with both perceived stress and depression (12–15). Because psychological flexibility is a multidimensional construct (16), its components may not relate to perceived stress and depression in the same way or to the same extent. Examining its individual components may therefore help clarify whether specific aspects of psychological flexibility also show distinct indirect associations in the relationship between perceived stress and depression. Therefore, psychological flexibility and its components may be involved in the indirect associations between perceived stress and depression.

Correlated studies have shown that perceived stress is significantly negatively correlated with psychological flexibility and significantly positively correlated with psychological inflexibility (12, 17). From the perspective of Lazarus’ Stress and Coping Theory (18), individuals under higher perceived stress may be more likely to adopt maladaptive coping responses, which may in turn be related to lower psychological flexibility. Meanwhile, psychological flexibility has been consistently reported to be negatively correlated with depression (15, 19, 20). Higher psychological flexibility has also been linked to more adaptive cognitive-emotional coping, whereas lower psychological flexibility has been associated with greater emotional distress, including depression (21, 22). Together, these findings provide a rationale for examining psychological flexibility as a potential mediator in the association between perceived stress and depression.

The Job Demands–Resources (JD-R) model is a helpful framework for understanding health outcomes in contexts characterized by high job demands (23). The model proposes that high job demands can impose sustained physical and psychological strain, whereas resources can buffer these pressures and facilitate adaptation and recovery (24, 25). In nursing, heavy workloads, professional competency requirements, and complex interpersonal interactions constitute substantial job demands, which may contribute to elevated perceived stress and poorer mental health. Within this framework, psychological flexibility may function as a personal resource relevant to understanding the association between perceived stress and depression. Examining its role may therefore help identify potential intervention targets for nurses.

Based on prior evidence that perceived stress is negatively associated with psychological flexibility and that psychological flexibility is negatively associated with depression, this study examined a mediation model in which psychological flexibility was tested as a potential mediator of the association between perceived stress and depression among nurses. Because psychological flexibility is a multidimensional construct and its components may differ in their associations with perceived stress and depression, its three components—acceptance and cognitive defusion, mindfulness and self-as-context, and values and committed action—were also examined. In addition, gender, age, years of work experience, marital status, number of children, economic status, educational level, and professional title were included as covariates because previous studies have shown that these variables may be associated with depression among nurses and could therefore confound the associations of interest (26–33). We hypothesized that perceived stress would be negatively associated with psychological flexibility and its three components, and that psychological flexibility and these components would be negatively associated with depression. We further hypothesized that psychological flexibility and its three components would show significant indirect effects in the association between perceived stress and depression.

2 Methods2.1 Participants and study design

This cross-sectional study was conducted in Xi’an, China, between December 2023 and January 2024. Using convenience sampling, we recruited 3,920 nurses from three tertiary Grade A public general hospitals in Xi’an. All three hospitals were large institutions that integrated clinical care, teaching, and research. The inclusion criteria were as follows: (1) nurses with professional qualification certificates; (2) nurses who provided informed consent and participated voluntarily in the survey. The exclusion criteria were: (1) nurses on leave during the survey period; (2) visiting or rotating nurses; (3) intern nurses. Before the formal survey, the researchers contacted the nursing departments of the three hospitals, introduced the purpose of the study to the directors of nursing, and obtained their cooperation. The head nurses of each department then assisted in informing eligible nurses about the voluntary, unpaid online questionnaire survey. Investigators recorded the number of nurses who voluntarily agreed to participate and met the inclusion criteria, and eligible nurses were invited by the investigators and head nurses to join the corresponding WeChat survey groups for each department in each hospital.

The study was approved by the Human Sciences Ethics Committee of the Xi’an Jiaotong University Health Science Centre (approval number: 2023-1336). The research objectives, the principle of information confidentiality, and the principle of voluntary participation were fully explained to the participants, and electronic informed consent was obtained.

2.2 Measurement process

Two trained members of the research team coordinated the online data collection procedure and provided participant support when needed. Before completing the questionnaire, participants were informed of the study purpose, confidentiality principles, and the voluntary nature of participation. After informed consent had been obtained, standardized instructions for completing the questionnaire were provided to participants through their departmental WeChat groups. Subsequently, a Wenjuanxing (a professional online survey platform in China) link was distributed in the corresponding departmental WeChat survey groups of each hospital, and participants completed the questionnaire independently to reduce potential response bias. During the survey period, the investigators were available through WeChat to answer participants’ questions and provide clarification when needed. The estimated time to complete the questionnaire was approximately 20 minutes.

2.3 Measures

The research team developed a General Information Questionnaire in accordance with the study objectives. It included two sections: a demographic survey and an occupational factors survey. The demographic survey covered six items: gender, age, marital status, number of children, economic status, and education. The occupational factors survey included two items: working years and professional title.

The Chinese Version of Cohen’s Perceived Stress Scale (CPSS) was designed by Professor Cohen from the Department of Psychology at Carnegie Mellon University based on Lazarus’ stress and coping theory (6). The Chinese version was localized and revised by Chinese scholars, including Yang Tingzhong. It comprised 14 items in two dimensions, tension and loss of control, with seven items in each dimension. Items were rated on a 5-point Likert scale, with response options ranging from never to always. The scale included both forward-scored and reverse-scored items. Items 1, 2, 3, 8, 11, 12, and 14 were scored in the forward direction, whereas items 4, 5, 6, 7, 9, 10, and 13 were reverse scored. The total score was obtained by summing all items and ranged from 0 to 56. Respondents with scores of 25 or above were considered to be under health-risk stress. Internal consistency was assessed using Cronbach’s α, and the overall scale showed a Cronbach’s α of 0.78. The scale had been widely used in China and had demonstrated good validity. It was used to measure perceived stress (predictor) in the present study. The Cronbach’s α coefficient of this scale was 0.91 in the present study.

The Chinese version of the Comprehensive Assessment of Acceptance and Commitment Therapy Processes (CompACT) was initially developed by Francis based on Relational Frame Theory (11). It was later translated and validated by Professor Zhuo-Hong Zhu’s team and colleagues, including Ming Wang, at the Institute of Psychology, Chinese Academy of Sciences. The questionnaire comprised 15 items across three dimensions: acceptance and cognitive defusion, mindfulness and self-as-context, and values and committed action. Internal consistency was evaluated using Cronbach’s α, and the overall scale showed a Cronbach’s α of 0.87, indicating good internal consistency. The scale used a 7-point Likert response format, with options ranging from never to always. In the original scoring procedure, items 4–7 and 9–15 were reverse scored, and the total score was calculated by summing all item scores, with higher total scores indicating lower psychological flexibility. For ease of interpretation, in the present study, items 1, 2, 3, and 8 were reverse scored instead. Specifically, these four items were recoded so that original scores of 0, 1, 2, 3, 4, 5, and 6 were reassigned to 6, 5, 4, 3, 2, 1, and 0, respectively. The recoded scores were then summed with the scores of the remaining items to yield a total score ranging from 0 to 90, such that higher total scores indicated greater psychological flexibility. The CompACT was used to measure psychological flexibility as the mediating variable in the present study. In this study, Cronbach’s α for the scale was 0.84.

The Patient Health Questionnaire-9 (PHQ-9) was developed by Spitzer and colleagues on the basis of the diagnostic criteria in the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV). It was a self-administered depression module from the full PHQ (34). The questionnaire consisted of nine items, each rated on a 4-point Likert scale. The response categories were not at all, several days, more than half the days, and nearly every day. All items were scored positively. The total score was calculated by summing all item scores and ranged from 0 to 27. A score of 0 ~ 4 indicated no depression, 5 ~ 9 indicated mild depression, 10 ~ 14 indicated moderate depression, 15 ~ 19 indicated moderately severe depression, and 20 ~ 27 indicated severe depression. The questionnaire showed good internal consistency, with a Cronbach’s α of 0.84. It had been widely used internationally and was recognized as a valid instrument. The PHQ-9 was used to measure depression (outcome) in the present study. In this study, the Cronbach’s α coefficient of the scale was 0.96.

2.4 Data analysis

The data were collected and downloaded through the Wen Juan Xing online platform. All items were set as required, and incomplete questionnaires could not be submitted. Consequently, there were no missing item data, and no additional missing data handling was performed. Invalid responses were defined as careless responding, including obvious repetitive response patterns and uniform responses across all items. These cases were excluded during the data cleaning stage. To preliminarily assess potential common method bias, Harman’s single-factor test was conducted. SPSS 24.0 was employed to perform statistical analysis on the data. The descriptive data of the main variables were reported using the mean and standard deviation (M ± SD). The independent-samples t tests and one-way ANOVA were used to analyze differences in depression scores across demographic and occupational characteristics among Chinese nurses. For these comparisons, effect sizes were calculated using Cohen’s d or η² to assess the magnitude of the differences. Formal adjustments for multiple comparisons were not implemented for the independent samples t-tests, as these demographic analyzes were fundamentally exploratory and preliminary. Their primary purpose was to provide a descriptive profile of the sample, rather than to serve as tests of the study’s central hypotheses. Pearson correlation was used for the correlation analysis among variables. To test the mediation effects, the bias-corrected bootstrap method using the SPSS Process macro (Model 4) was utilized. Given the cross-sectional design, the temporal sequence among variables was based on theoretical assumptions rather than empirical verification; therefore, the findings were interpreted as statistical, rather than causal, indirect effects. A total of 5,000 bootstrap samples with replacement were drawn to obtain a 95% confidence interval of the statistical mediating effect value. If the upper and lower limits of the interval did not include 0, the statistical mediating effect was considered statistically significant. Finally, a sensitivity analysis was conducted to assess the robustness of the statistical mediation model. Specifically, the unadjusted models were compared with models adjusted for gender, age, working years, marital status, number of children, economic status, education, and professional title.

3 Results

A total of 3,920 questionnaires were distributed online, and 3,611 valid questionnaires were eventually collected, with an effective response rate of 92.12%. The main reasons for nonparticipation included incomplete or incorrectly filled questionnaires. The general demographic characteristics of 3,611 nurses are shown in Table 1.

Variablen (%)Gender Male120 (3.30) Female3,491 (96.70)Age ≤301442 (39.90) 30-401631 (45.20) >40538 (14.90)Working years ≤52805 (77.70) 5-10400 (11.10) 10-20295 (8.20) >20111 (3.10)Marital status Unmarried971 (26.90) Married2566 (71.10) Divorced74 (2.00)Number of children 01348 (37.30) 11889 (52.30) ≥2374 (10.40)Economic status (RMB) ≤50001057 (29.30) 5000-80001395 (39.60) 8000-10000574 (15.90) 10000-15000391 (10.80) >15000194 (5.40)Education Junior college degree or below545 (15.10) Bachelor degree or above3066 (84.90)Professional title Others3,497 (96.80) Chief nurse114 (3.20)

Demographic and occupational information of participants (n = 3,611).

3.1 Common method bias

In our study, Harman’s single-factor test was employed to examine common method bias. All the items of variables were subjected to exploratory factor analysis. The analysis extracted eight common factors, and the first principal component accounted for 35.96% of the variance, below the critical threshold of 40%.

3.2 Descriptive statistics of the study variables

The mean depression score was 12.05 ± 6.60. A total of 3,076 Chinese nurses (85.18%) scored above the cutoff point between minor depressive symptoms and mild depression (> 4) on the PHQ-9. Among those who screened positive for depression, 908 (29.52%) had mild depression (5 ~ 9 points), 873 (28.38%) had moderate depression (10 ~ 14 points), 869 (28.25%) had moderately severe depression (15 ~ 19 points), and 426 (13.85%) had severe depression (20 ~ 27 points).

Depression scores differed significantly across nurses’ gender (p < 0.01), age (p < 0.01), working years (p < 0.01), marital status (p < 0.01), number of children (p < 0.01), economic status (p < 0.01), and professional title (p < 0.01). Details are presented in Table 2.

VariableDepression
M ± SDt/F (p)Cohen d/η2*95% confidence intervalLowerUpperGender Male10.63 ± 7.50-2.11 (< 0.01)-0.21-0.40-0.04 Female12.10 ± 6.57Age ≤3010.85 ± 6.5541.70 (< 0.01)0.020.010.03 30-4012.96 ± 6.60 >4012.49 ± 6.26Working years ≤513.64 ± 6.02307.70 (< 0.01)0.200.180.23 5-106.75 ± 5.45 10-206.40 ± 5.53 >205.79 ± 5.27Marital status Unmarried11.27 ± 6.5913.53 (< 0.01)0.000.000.01 Married12.27 ± 6.59 Divorced14.57 ± 6.23Number of children 011.55 ± 6.606.72 (< 0.01)0.000.000.01 112.41 ± 6.58 ≥212.01 ± 6.74Economic status (RMB) ≤500011.10 ± 7.0312.85 (< 0.01)0.010.010.02 5000-800011.92 ± 6.62 8000-1000012.89 ± 6.14 10000-1500013.31 ± 5.77 >1500013.07 ± 6.13Education Junior college degree or below11.77 ± 6.571.09 (= 0.30)0.000.000.01 Bachelor degree or above12.10 ± 6.61Professional title Others 12.11 ± 6.613.61 (< 0.01)0.320.130.50 Chief nurse10.03 ± 6.06

Comparisons of depression scores across demographic and occupational characteristics (n = 3,611).

*Cohen’s d was reported for binary comparisons, and η² was reported for variables with more than two groups. The 95% confidence interval refers to Cohen’s d or η².

3.3 Correlation analysis

Table 3 presents the correlations among the study variables. Depression was significantly and positively correlated with perceived stress and significantly negatively correlated with psychological flexibility. Moreover, psychological flexibility was significantly negatively correlated with perceived stress.

VariablePSPFDPS1.00PF-0.60**1.00D0.63**-0.17**1.00

Inter-correlations among measures (n = 3,611).

**p < 0.01. PS, perceived stress; PF, psychological flexibility; D, depression.

3.4 Mediation analysis

We examined the mediating role of psychological flexibility in the association between perceived stress and depression. Sensitivity analyzes indicated that the overall pattern of results was materially unchanged after adjustment for covariates. In the unadjusted models, perceived stress scores were significantly negatively associated with psychological flexibility and all of its component scores. Psychological flexibility and its components, including acceptance and cognitive defusion, and values and committed action, were significantly and negatively associated with depression scores. However, the association between mindfulness and self-as-context and depression was not statistically significant (Table 4). After adjustment for covariates, the overall pattern of associations remained largely unchanged, and the previously significant associations remained statistically significant (Figures 1, 2; Table 5).

PathwayEstimateEstimate (standardized)95% confidence intervalLowerUpperPS-PF-D0.120.190.110.13PS-AC-D0.140.220.130.15PS-MS-D0.010.010.000.02PS-VC-D0.060.100.050.08

Indirect effects and 95% confidence intervals without covariate adjustment.

PS, perceived stress; PF, psychological flexibility; AC, acceptance and cognitive defusion; MS, mindfulness and self-as-context; VC, values and committed action; D, depression.

Mediation model exploring the indirect effects of psychological flexibility between perceived stress and depression after adjusting for covariates. ***P < 0.001. PS, perceived stress; PF, psychological flexibility; D, depression.

Mediation model exploring the indirect effects of the components of psychological flexibility between perceived stress and depression after adjusting for covariates. ***P < 0.001. PS, perceived stress; PF, psychological flexibility; D, depression; AC, acceptance and cognitive defusion; MS, mindfulness and self-as-context; VC, values and committed action.

PathwayEstimateEstimate (standardized)95% confidence intervalLowerUpperPS-PF-D0.120.190.110.13PS-AC-D0.140.220.130.15PS-MS-D0.010.010.000.02PS-VC-D0.060.090.040.08

Indirect effects and 95% confidence intervals with covariate adjustment.

PS, perceived stress; PF, psychological flexibility; AC, acceptance and cognitive defusion; MS, mindfulness and self-as-context; VC, values and committed action; D, depression.

As shown in Figure 1, the unstandardized direct effect of perceived stress on depression was 0.47. The path coefficient for the association between perceived stress and psychological flexibility was −0.77, whereas that for the association between psychological flexibility and depression was −0.15. The resulting unstandardized indirect effect via psychological flexibility was approximately 0.12 [(−0.77)×(−0.15)≈0.12]. Given a total effect of 0.59, the indirect effect accounted for approximately 20.34% of the total effect. These findings suggest that the indirect association through psychological flexibility accounted for approximately one-fifth of the overall association between perceived stress and depression. We further examined the unstandardized indirect effects of the three components of psychological flexibility in the association between perceived stress and depression after adjustment for covariates. As shown in Figure 2, perceived stress was significantly negatively associated with acceptance and cognitive defusion (β = −0.20), mindfulness and self-as-context (β = −0.21), and values and committed action (β = −0.77). In turn, acceptance and cognitive defusion (β = −0.70) and values and committed action (β = −0.08) were significantly and negatively associated with depression, whereas mindfulness and self-as-context was not significantly associated with depression (β = −0.03).

As presented in Table 5, both the unstandardized and standardized indirect effects of psychological flexibility, acceptance and cognitive defusion, and values and committed action were statistically significant after adjustment for covariates. In contrast, neither the unstandardized nor the standardized indirect effect of mindfulness and self-as-context was statistically significant. Moreover, as shown in Figure 1 and 2, both the unstandardized and standardized direct effects of perceived stress on depression remained statistically significant after the mediators were included in the model. Taken together, these findings suggest that psychological flexibility, acceptance and cognitive defusion, and values and committed action were involved in significant partial indirect associations between perceived stress and depression, whereas mindfulness and self-as-context was not.

4 Discussion

The principal findings of this study indicated that perceived stress was associated with depression b