Preliminary Clustering: Exploring the Interplay of Burnout, Stress, Turnover, Psychological Flexibility and Distress in a French Nurse Sample

ABSTRACT

Objective

To identify latent profiles of hospital nurses based on the combination of occupational demands, psychological symptoms and psychological flexibility. Examine how these profiles relate to job turnover intentions.

Design

Cross‐sectional online survey.

Methodology

Registered nurses in France completed an online questionnaire between June and August 2024. The questionnaire covered various aspects of job satisfaction and stress. A non‐hierarchical cluster analysis was then conducted on 12 standardised variables to derive profiles. Next, appropriate group‐comparison tests and sensitivity checks were performed. No abbreviations or references are used here.

Results

Two profiles emerged. One profile showed a higher workload, greater emotional demands, increased stress and psychological distress, burnout and a lower level of psychological flexibility, as well as a higher intention to leave. The other profile showed lower demands and symptoms, higher psychological flexibility and a lower intention to leave. Group differences on core variables were statistically significant and sensitivity analyses indicated a stable solution.

Conclusions

Person‐centred profiles reveal distinct combinations of demands, symptoms and psychological flexibility meaningfully linked to nurses' intention to leave. These findings suggest opportunities for targeted organisational adjustments and brief skills training to strengthen psychological flexibility and retention.

Impact

Health‐service leaders can use brief screening to identify at risk profiles and align interventions. Policymakers can consider staff and scheduling policies to reduce demand in high‐risk units. Educators can incorporate psychological skills into training programmes to promote workforce sustainability.

Summary.

• What is already known

  • ○
    Burnout and psychological distress are consistently associated with nurses' intention to leave.
  • ○
    Person‐centred approaches can reveal subgroups that mean‐based analyses may hide.
  • ○
    Psychological flexibility is a modifiable individual resource that is linked to better well‐being at work.
• What this paper adds

  • ○
    The study, conducted from June to August 2024, identified two distinct nurse profiles: one with higher demands and symptoms, and lower psychological flexibility; and one with lower demands and symptoms, and higher psychological flexibility.
  • ○
    The profile with higher demands and lower flexibility reported a markedly greater intention to leave.
  • ○
    The profile solution was statistically supported and remained stable in sensitivity checks.
• Implications for practice and policy

  • ○
    Brief screening can help services identify high‐risk units or staff and provide targeted support.
  • ○
    Organisational measures that reduce workload and stabilise rosters may encourage staff to adopt the lower‐risk profile.
  • ○
    Training that strengthens psychological flexibility can complement system‐level changes to improve retention.

1. Introduction

The global healthcare sector is facing significant challenges, exacerbated by a growing demand for high‐quality care driven by an aging population and rapid advancements in medical technology (World Health Organization ²⁰²¹). According to the World Health Organization's 2020 State of the World's Nursing report, a global shortage of approximately 5.9 million nurses is projected by 2030 (World Health Organization ²⁰²⁰). This projection is further substantiated by reports from the International Council of Nurses, which in 2023 described the nursing workforce crisis as a ‘global health emergency’ and highlighted the severe stress and burnout threatening the profession's sustainability (International Council of Nurses ²⁰²³).

This significant attrition poses a considerable threat to the sustainability of the nursing profession and, consequently, to the overall quality of healthcare delivery. Among the factors contributing to this crisis, burnout stands out as a major concern requiring immediate attention (Şenol Çelik et al. 2024). Numerous studies have highlighted the stressors to which nurses are exposed, including long working hours, heavy workloads and exposure to potentially traumatic events (Häggman‐Laitila and Romppanen ²⁰¹⁸). These working conditions contribute to elevated stress levels and increased psychological distress among nursing staff.

Beyond the individual consequences for nurses, the challenges associated with recruiting and training new professionals represent a significant financial burden for healthcare institutions, estimated at approximately 1.3 times a nurse's annual salary (Guerrier et al. 2021). This cost exacerbates the difficulties arising from high staff turnover, leading to decreased job satisfaction, reduced productivity and a decline in the quality of care provided to patients (Alfurjani et al. 2024). Despite extensive research on the factors influencing nurse retention, effective solutions remain elusive, underscoring the need for innovative approaches to address this pressing issue (Martin et al. 2023).

While a substantial body of research has focused on the psychosocial risks faced by nursing staff, a notable gap remains regarding the study of these professionals through the lens of typologies. Our study aims to address this gap by identifying robust profiles that consider a range of occupational risks (workload, emotional demands), psychological disorders (stress, psychological distress, burnout) and protective factors such as psychological flexibility (Yatsu and Saeki 2022). Indeed, psychological flexibility can be defined as the capacity to maintain openness to experiences while acting in accordance with one's core values (McCracken ²⁰¹⁴). By exploring these different dimensions, we aim to provide valuable insights that can inform targeted interventions designed to enhance nurses' well‐being and promote their retention within the profession.

2. Conceptual Framework

2.1. Risks Among Nurses

High levels of stress among nurses are often fueled by a combination of factors such as long working hours, heavy workloads and frequent exposure to illness and death. These demanding work environments are characterised by psychosocial risks, which, as defined by the World Health Organization (2020) and Leka and Cox (2010), are features of work design, organization and social context with the potential for psychological or physical harm. For instance, excessive workload, limited scope for decision‐making and poor team relations exemplify such psychosocial risks that contribute significantly to the high‐stress levels experienced by nurses.

Indeed, the World Health Organization defines work‐related stress as a reaction that occurs when there is a mismatch between job demands and an individual's skills and knowledge, coupled with workplace pressures that exceed their capacity to cope (WHO ²⁰²²). When this chronic workplace stress is not successfully managed, it can lead to burnout, which the World Health Organization categorises as an occupational phenomenon. According to Maslach's definition, professional burnout is characterised by three main components: emotional exhaustion (EE), involving physical and psychological fatigue; depersonalization (DP), displaying negative and detached attitudes towards those receiving care; and reduced personal accomplishment (PA), indicated by feelings of ineffectiveness and job dissatisfaction (Maslach and Jackson 1981). The build‐up of these stress and burnout symptoms can result in psychological distress among individuals. In other words, psychological distress is generally described as a response to stressors such as depression, anxiety and burnout, encompassing a sequence of negative psychological cognitions, emotions, behaviours and other psychological manifestations (Kessler et al. 2002).

However, the underlying reasons for nurses leaving the profession remain elusive, and effective solutions for retention continue to be elusive (Hopson et al. 2018). These findings highlight the urgent need for interventions to address the well‐being of nurses and ensure a sustainable healthcare workforce.

2.2. Psychological Flexibility as a Buffer of Stress

Psychological flexibility is of paramount importance in regulating the challenges posed by occupational stress, particularly in high‐stress professions such as nursing. There is a growing body of evidence to suggest that psychological flexibility is a critical factor in mitigating occupational stress and enhancing job satisfaction in nursing. This is evidenced by studies which have demonstrated that psychological flexibility is associated with enhanced mental health, job satisfaction and resilience in demanding work environments (Kinman et al. 2020; Günel and Aydin ²⁰²²). It is noteworthy that research indicates that individuals exhibiting higher levels of psychological flexibility tend to experience lower rates of burnout and emotional exhaustion, as well as improved coping strategies when confronted with workplace adversities (Novaes et al. 2018; Ren et al. ²⁰²²).

In the context of nursing, psychological flexibility is significantly correlated with various positive outcomes, including job satisfaction and overall psychological well‐being (Li et al. 2018; Novaes et al. ²⁰¹⁸). For example, a systematic review indicated that psychological empowerment among nurses is closely associated with job satisfaction, suggesting that the promotion of psychological flexibility may result in enhanced job‐related outcomes (Li et al. 2018). Moreover, the capacity to respond adaptively to stressors is of paramount importance for nurses, who frequently encounter elevated emotional demands and organisational challenges (Ren et al. 2022). The relationship between individual psychological traits and organisational factors is crucial for understanding nurse well‐being. It provides insights into how psychological flexibility can mitigate the adverse effects of workplace stressors (Li et al. 2018; Novaes et al. ²⁰¹⁸). The promotion of an organisational culture that values psychological flexibility has the potential to reduce burnout rates and improve job satisfaction among nursing staff (Larrabee et al. 2010; Sharkey and Caska ²⁰²⁰). For example, the integration of psychological flexibility into training programmes for nurses could serve as a proactive strategy to equip them with the necessary skills to navigate the complexities of their roles effectively (Aydin and Gümüşboğa 2023; Njiru ²⁰²⁰).

2.3. How Chronic Stress May Impair Psychological Flexibility

Indeed, there is a strong correlation between stress and psychological flexibility. In accordance with the prevailing psychophysiological definition, stress may be conceptualised as a pervasive and sustained condition wherein an organism is exposed to risk factors that have the potential to disrupt its homeostatic balance and/or equilibrium. Hayes et al. (2006) define psychological flexibility as ‘the ability to fully engage with the present‐moment and the thoughts and feelings it contains without unnecessary defense and depending on the situation, persisting or changing in behavior in the pursuit of goals and values.’ Moreover, the ability to respond adaptively and flexibly is a vital resource for coping with stress while maintaining homeostasis. This point has been demonstrated in numerous studies (Dawson and Golijani‐Moghaddam ²⁰²⁰; Ryan et al. ²⁰²⁵; McIlvenna et al. ²⁰²⁵).

Experiential avoidance (denoted as EA) is the tendency to alter, suppress, or escape from unwanted private events—thoughts, emotions, bodily sensations—even when such efforts impair long‐term functioning (Hayes et al. 1996). It is considered a central feature of psychological inflexibility and predicts higher levels of anxiety, depression and stress in both clinical and occupational samples (Chawla and Ostafin 2007). Extensive evidence shows that higher experiential avoidance is associated with greater psychological distress—including anxiety, depression and perceived stress—in nurses and other populations (Kashdan et al. 2006; Sairanen et al. ²⁰¹⁸).

Furthermore, EA has been shown to correlate with various health outcomes, including physical health problems and reduced quality of life (Kashdan and Rottenberg 2010). In the field of occupational psychology, a recent study using an intergroup comparison method showed that nurses in the ‘non‐flexible’ group, compared to the ‘flexible’ group, reported significantly higher levels of anxiety, depression, stress and exposure to work stressors. This study also highlights a negative correlation between psychological flexibility and work‐related stressors (Cuenca et al. 2021). These results are consistent with previous studies showing similar findings (Puolakanaho et al. 2020).

Psychological flexibility has been demonstrated to be associated with enhanced work outcomes, including improved mental health, elevated job performance, augmented capacity for skill acquisition and diminished rates of absenteeism (Bond and Flaxman 2006). Additionally, it may serve as a partial mediator in the relationship between job satisfaction and mental well‐being and mental health problems (Chong et al. 2023). A growing body of evidence indicates that individuals with lower psychological flexibility are more likely to turn to psychotropic medication as a coping mechanism during organisational changes, suggesting a preference for avoidance strategies (Brion et al. 2022).

2.4. Aims

While a substantial body of research has been conducted on the challenges encountered by nursing professionals, a significant knowledge gap persists regarding the ways in which these challenges manifest and vary across distinct nurse profiles (Nazeer et al. 2024). The majority of studies concentrate on general trends without delving into the intricate experiences of nurses occupying diverse roles, situated within disparate settings and operating within distinct organisational contexts (Saifan et al. 2021). Moreover, there has been a paucity of attention devoted to the protective function of psychological flexibility as a mitigating factor for occupational stress and burnout (Ibrahim et al. 2023).

This study addresses these shortcomings by delineating typologies of nursing professionals based on psychosocial risks, organisational factors and psychological resources, thereby establishing a foundation for targeted interventions (Harhash et al. 2020).

This study aims to identify distinct profiles of nurses based on their experiences with occupational risks, psychological disorders and protective factors like psychological flexibility.

Our objectives will also be as follows:

• to determine if the studied population can be distinguished into different groups.

• to confirm the consistency of the clusters with the established literature regarding occupational risks and turnover intentions.

• to highlight the impact of psychological flexibility on nurses' symptoms of stress, burnout and psychological distress.

These challenges also contribute to organisational issues such as higher turnover rates and a tendency towards patient dehumanisation. However, research suggests that psychological flexibility can serve as a protective factor, enabling nurses to respond more effectively to workplace stressors. Drawing on these findings and the work of Cuenca et al. (2021), our study aims to identify risk profiles associated with varying levels of psychological flexibility.

Although this study focuses on nurses practicing in France, its findings are relevant internationally. Global assessments by the World Health Organization and the International Council of Nurses highlight a consistent shortage of nurses worldwide and increasing attrition pressures, making the identification of modifiable risk–resource combinations a priority across countries. The mechanisms we examine—workload intensity, emotional demands, burnout, psychological distress and psychological flexibility—map onto the Job Demands–Resources framework, which has been validated in various health system contexts. Furthermore, the instruments used (e.g., MBI‐HSS, PSS‐10, K6 and MPFI‐24) are widely used internationally, enabling benchmarking and replication. Consequently, the person‐centred profiles reported here can inform staffing, retention and well‐being strategies in settings beyond France.

3. Method

3.1. Design

For this study, we employed a cross‐sectional survey methodology using an anonymous survey online between June and August 2024.

3.2. Report Transparency

This manuscript was prepared in accordance with the STROBE Statement guidelines for observational studies (Appendix 2).

3.3. Participants

The participants comprised 267 registered nurses practicing in France, recruited through the social media platforms Facebook, Instagram and LinkedIn. We contacted several nurses via social media to disseminate the questionnaire within their professional networks. Consequently, our study population comprises nurses from multiple hospitals across various regions of France. Participation was restricted to actively employed registered nurses. Of the 499 individuals who accessed the online questionnaire, 267 (53.5%) provided sufficiently complete responses, which were defined as completing all core survey modules, and they were retained for analysis. Fewer than 5% of data points were missing within this analytic sample of 267 participants. Data collection occurred from June to August 2024.

Questionnaires were considered incomplete and excluded from the analysis if an entire scale was missing. However, we included questionnaires where all scales were completed, even if a single response was missing within a scale. Eighty‐eight participants (17.6%) discontinued the questionnaire during the first scale. An additional 15 participants (3%) completed the survey but did not provide their socio‐demographic data. The remaining participants who did not complete the survey (25.8%) discontinued at various points in the middle of the questionnaire. The final dataset for the 267 participants contained less than 5% missing data, which was handled using mean imputation.

3.4. Ethical Considerations

The author asserts that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008. Our study corresponds to a non‐interventional, low‐risk study under French law—comparable to an exempt or minimal risk classification in many international research ethics frameworks. The ethical approval is deemed unnecessary in the French context as the study falls into the ‘Loi Jardé’ RIPH3 for non‐interventional research (Code de la santé publique, Articles L1121‐1 à L1128‐12).

The study received clearance from the University of Rouen's Institutional Data‐Protection Officer, confirming compliance with the General Data Protection Regulation (GDPR No 2016/679). All participants provided written informed consent. This research received no external or internal funding. The authors alone were responsible for the study design, data collection, analysis, interpretation and publication decisions.

3.5. Procedure and Data Collection

A questionnaire was developed using the Qualtrics platform and disseminated via various social media channels including Facebook, Instagram and LinkedIn only in France. Inclusion criteria: All participants were required to be at least 18 years old, in accordance with the legal age of majority in France and actively employed as a nurse in a hospital setting, nursing home, or private practice. Out of the 499 individuals who accessed the questionnaire, 267 provided complete responses. The dataset contained less than 5% missing values for any variable. Pairwise deletion was used in all primary analyses.

3.6. Data Analysis

3.6.1. Preliminary Analyses

The collected data were analysed using the statistical software Jamovi (version 2.5.5) (The Jamovi project ^2025a), which is based on R (version 4.5). Prior to the main analyses, descriptive statistics (means, standard deviations, ranges and frequencies as appropriate) were computed for all study variables to examine their distributions and identify potential outliers. The reliability of the multi‐item scales (work intensity, perceived stress, emotional exhaustion, cynicism, professional accomplishment, psychological distress and psychological flexibility) was assessed using Cronbach's alpha (α).

All variables were inspected for normality using histograms, Q–Q plots and Shapiro–Wilk tests. Skewness and kurtosis values fell within acceptable ranges (±1.0). Univariate outliers, defined as values exceeding ±3 standard deviations from the mean, were identified but represented less than 0.5% of observations for any given variable. We also examined all models for multivariate influence: we calculated Cook's distances and deleted studentised residuals for each case. We found that no observation exceeded the conventional thresholds (Cook's D > 4/n; |studentised residual| > 3). Therefore, no participants were excluded on that basis.

To ensure that all variables contributed equally to the subsequent cluster analysis, all continuous and quasi‐continuous variables were standardised using z‐scores (Table 1). The inclusion of the ‘intention to leave work’ variable, an ordinal scale with three levels (No, In a few years, As soon as another job becomes available), was justified by treating it as a quasi‐continuous measure. This common practice in exploratory research acknowledges the ordered progression of intensity reflected in the numerical values assigned to each category (0, 1, 2), making the distances meaningful for the clustering algorithm (Joye and Dries 2018; Robitzsch ²⁰²⁰). Although z‐score standardisation does not normalise the distribution of variables, the K‐means algorithm is a distance‐based method and is less sensitive to the normality assumption than other parametric tests (Steinley 2006). This standardisation was crucial to prevent variables with larger ranges from disproportionately influencing the clustering outcome.

TABLE 1.

Z‐standardised centroid values for each profile across the twelve clustering variables.

Centroids of clusters table
	Emotional demands	Workload	Perceived stress	Emotional exhaustion	Professional accomplishment	Psychological distress	Psychological flexibility	Age	Seniority	Work hours	Intent to leave
1	−0.433	−0.450	−0.614	−0.702	0.317	−0.704	0.441	0.056	0.036	−0.030	−0.398
2	0.481	0.499	0.681	0.780	−0.352	0.781	−0.490	−0.063	−0.040	0.034	0.442

Note: Values represent the mean z‐score (standardised score) of each variable within each cluster. Positive values indicate above‐average scores relative to the full sample mean (z = 0), and negative values indicate below‐average scores. Values in bold indicate the highest or most defining score for each variable across profiles. For example, a score of +0.780 for emotional exhaustion in the High Burnout profile indicates that, on average, members of this cluster reported emotional exhaustion that was 0.780 standard deviations above the sample mean. (1) Low Burnout/High psychological flexibility (52.6%). (2) High Burnout/Low psychological flexibility (47.3%).

3.6.2. K‐Means Cluster Analysis

A non‐hierarchical K‐means cluster analysis was performed on a sample of 267 individuals. The analysis aimed to uncover naturally occurring constellations of occupational demands, psychological symptoms and resources, using a person‐centred approach. Analyses were conducted in jamovi (Version 2.7; The jamovi project ^2025b) using the snowCluster module (Version 7.5.5; Seol ²⁰²⁵) running on R (Version 4.5; R Core Team ²⁰²⁵); for sensitivity checks with mixed‐type clustering, we additionally used clustMixType (Version 0.4.2; Szepannek and Aschenbruck ²⁰²⁴). We chose K‐means for three key reasons: first, its ability to partition cases based on shared variance between standardised continuous variables makes it well suited for the mixed metrics used here (Hartigan and Wong 1979); second, it performs reliably with our sample size (n = 267) and produces replicable solutions (Spurk et al. 2020); and third, it has a precedent in nursing research for identifying subgroups for targeted interventions (Hillhouse and Adler 1997).

The optimal number of clusters was determined by triangulating the elbow method. We examined the reduction in the within‐cluster sum of squares (inertia) for different cluster solutions. The reduction was substantial when moving from 1 cluster (inertia = 2976) to 2 clusters (inertia = 2317, reduction of 659). The gain became much smaller when moving to 3 clusters (inertia = 2079, reduction of 238) and 4 clusters (inertia = 1931, reduction of 148). This progressively diminishing gain in homogeneity suggests a numerical ‘elbow’ at 2 clusters, supporting our choice of this solution (Figure 1).

FIGURE 1.

Scree plot of the within‐cluster inertia. Scree plot showing the within‐cluster sum of squares (inertia) for solutions with k = 1–6 clusters. The elbow at k = 2 indicates the optimal number of clusters.

3.6.3. Post Hoc Analyses and Cluster Validation

To validate the cluster solution, the homogeneity of variances between the clusters was assessed using Levene's (1960) test (Table 2). The results showed that the assumption of equal variances was met for several variables (p > 0.05): emotional demands (p = 0.945), perceived stress (p = 0.323), emotional exhaustion (p = 0.693), age (p = 0.081), seniority (p = 0.397), full‐time working hours (p = 0.233) and turnover intention (p = 0.669). For these variables, a series of one‐way ANOVAs was conducted to compare the clusters.

TABLE 2.

Test of Homogeneity of Variances (Levene's Test).

	F	ddl1	ddl2	p
Emotional demand	0.00478	1	247	0.945
Workload	5.89311	1	247	0.016
Perceived stress	0.97867	1	247	0.323
Emotional exhaustion	0.15601	1	247	0.693
Depersonalisation	13.39339	1	247	< 0.001***
Professional accomplishment	8.09287	1	247	0.005
Psychological distress	14.91422	1	247	< 0.001***
Psychological flexibility	6.08284	1	247	0.014
Age	3.06948	1	247	0.081
Seniority	0.71980	1	247	0.397
Working hours	1.42874	1	247	0.233
Turnover intentions	0.18298	1	247	0.669

Note: Results of Levene's Test for the Homogeneity of Variances. A significant Levene's test (p < 0.05) indicates that the assumption of homogeneity of variances is violated.

Abbreviations: ddl1, numerator degrees of freedom; ddl2, denominator degrees of freedom; F, F‐statistic; p, probability value.

^*** p < 0.001.

However, the assumption was violated for other variables (p < 0.05): work intensity (p = 0.016), depersonalization of patients (p < 0.001), professional accomplishment (p = 0.005), psychological distress (p < 0.001) and psychological flexibility (p = 0.014). For these specific variables, the clusters were compared using Welch's ANOVA, a more robust test that does not assume equal variances.

3.7. Material

In order to assess psychosocial risks, our first questionnaire is partially based on the INRS grid (Appendix 1). We selected and adapted items related to workload intensity (7 items) and emotional demands (3 items). Cronbach's alpha is a statistic specifically designed to assess the internal consistency and reliability of scales intended to measure a single, underlying psychometric construct. It evaluates the extent to which items within a scale are inter‐correlated and consistently measure the same latent variable. For constructs that are not psychometrically defined or that represent multi‐faceted dimensions rather than a unified psychological concept (such as various aspects of working conditions), calculating Cronbach's alpha is generally inappropriate and can yield misleading results.

3.7.1. The Perceived Stress Scale (PSS‐10, Cronbach's α = 0.91)

Cohen et al. (1983) was employed to measure perceived stress levels. This 10‐item self‐report measure utilises a Likert scale ranging from ‘never’ to ‘very often’ to assess the frequency of stressful experiences during the past month. The scale includes six reverse‐scored items. The PSS‐10 has demonstrated strong psychometric performance across occupational groups, including nurses, with a meta‐analytic mean internal consistency coefficient of 0.83 and mean test–retest reliability of 0.72 over 1 month (Lee 2012). The French adaptation by Rolland (1991) replicates the original two‐factor structure and correlates positively with the General Health Questionnaire (r = 0.58) and negatively with the Life Orientation Test (r = −0.45), supporting convergent and divergent validity. In the present sample, Cronbach's alpha was 0.91 indicating that the scale reliably captures perceived stress in our sample.

3.7.2. Maslach Burnout Inventory‐Human Services Survey (MBI‐HSS)

Burnout was assessed using the 22‐item MBI‐HSS (Maslach et al. 2018). The scale contains three conceptually distinct subscales. Emotional exhaustion captures feelings of being emotionally overwhelmed by one's work; cynicism (i.e., ‘depersonalisation’) reflects a detached, negative reaction to the recipients of one's care; and personal accomplishment measures perceived professional efficacy. Items are scored on a six‐point frequency scale ranging from 0 (never) to 6 (daily). The MBI‐HSS was selected because burnout is the central psychological outcome in the Job Demands‐Resources framework. The scale has demonstrated robust psychometric properties in hospital‐based nursing samples, including a meta‐analytic mean Cronbach's α of 0.89 for exhaustion, 0.79 for cynicism and 0.71 for accomplishment (Rotenstein et al. 2018). Confirmatory factor analyses routinely support the three‐factor structure; the French version replicates this structure and correlates as expected with job satisfaction (r = −0.54) and intent to leave (r = 0.49) (Leiter et al. 2015). In the present sample, the alphas were 0.89 (exhaustion), 0.70 (cynicism) and 0.67 (achievement).

3.7.3. Kessler Psychological Distress Scale (K6, α = 0.84)

General psychological distress was measured by the six‐item K6 (Kessler et al. 2002). Respondents rate how often in the past month they felt ‘nervous’, ‘hopeless’, ‘restless’, ‘depressed’, ‘everything was an effort’ and ‘worthless’ on a four‐point scale from 1 (none of the time) to 4 (all of the time). The K6 complements the burnout measure by capturing non‐specific emotional distress, a key component of the ‘psychological symptoms’ axis in our latent profile model. The instrument has demonstrated unidimensionality, excellent screening accuracy for mood–anxiety disorders (area under the curve = 0.86) and a mean α of 0.84 across 12 countries (Prochaska et al. 2012).

3.7.4. Simplified Multidimensional Psychological Flexibility Inventory (MPFI‐24, Cronbach's ɑ = 0.0.90)

The MPFI‐24, developed by Rolffs et al. (2018) and adapted into French by Grégoire et al. (2020), is a 24‐item questionnaire measuring psychological flexibility (PF) and psychological inflexibility (PI). The PF subscale assesses six dimensions: present‐moment awareness, acceptance, cognitive defusion, self‐as‐context, values and committed action. The PI subscale measures the opposite of these dimensions. Each item is rated on a 6‐point Likert scale from 1 (never) to 6 (always). Higher PF scores indicate greater psychological flexibility.

Intent to Leave the Nursing Profession. This scale assesses nurses' desire to leave the nursing profession and consists of 1 item: ‘Do you intend to leave your job?’ Participants had three response options: ‘No’, ‘As soon as another job becomes available’, ‘In a few years’. The answers were categorised and assigned points as follows: ‘no’ received ‘0’ points, ‘in a few years’ received ‘1’ point and ‘as soon as another job becomes available’ received ‘2’ points. This measure typically includes questions assessing the frequency of thoughts about leaving, active job searching and the anticipated timing of departure from the current position. This concise scale allowed for a focused evaluation of nurses' immediate inclination to leave their employment, consistent with common practice in nursing turnover research. Finally, participants were asked to provide some data on their socio‐demographic variables: age, gender, seniority, work hours, departments in which they work and patient typology.

4. Results

4.1. Participants

Participant ages spanned from 21 to 57 years, with an average age of 33. The sample was predominantly female, with 238 women (89.1%), 22 men (8.2%) and 5 who did not disclose their gender (1.8%). This gender distribution is consistent with the overall gender makeup of the nursing profession (Direction de la recherche, des études, de l'évaluation et des statistiques (DREES) and Santé publique France ²⁰¹⁷). The majority of participants, 235 (88%), worked full‐time, while 31 (12%) were part‐time employees. Years of experience varied from 1 to 35 years among the participants. Table 3 provides a summary of the participants' characteristics.

TABLE 3.

Descriptive analysis.

	%	Mean	SD	Minimum	Maximum
Age		33.142	8.426	21.000	57.00
Female	89.1
Male	8.2
Full‐time job	88
Part‐time job	12
Seniority		8.537	7.646	1.000	35.00
Tension with the public		1.613	0.703	0.000	3.00
Pain management		1.564	0.619	0.000	3.00
Intensity of work		1.887	0.381	0.857	2.71
Stress		2.195	0.531	0.600	3.40
Emotional exhaustion		28.213	11.874	0.000	53.00
Depersonalization		11.004	6.871	0.000	29.00
Professional accomplishment		35.674	6.868	0.000	48.00
Psychological distress		2.694	0.832	1.000	4.83
Flexibility		3.707	0.656	1.910	5.50
Turnover		0.808	0.777	0.000	2.00

4.2. Clustering

Cluster analysis identified two distinct profiles of nurses (Figure 2). Levene's Test (Table 2) and ANOVA (or Welch's ANOVA) (Table 4) confirmed statistically significant differences between these clusters for most variables.

FIGURE 2.

Mean value comparison between two clusters. Mean z‐standardised values for each cluster across the twelve clustering variables. Positive values indicate scores above the sample mean (z = 0), and negative values indicate scores below the sample mean. Cluster 1 (blue) represents the high professional commitment profile, and Cluster 2 (orange) represents the burnout‐experiencing profile. Variables marked with asterisks indicate a statistically significant difference between clusters based on one‐way ANOVA or Welch's ANOVA results. **p < 0.001.

TABLE 4.

One‐Way ANOVA (Welch).

	F	ddl1	ddl2	p
Emotional demand	62.542	1	245	< 0.001***
Workload	66.292	1	246	< 0.001***
Perceived stress	188.042	1	239	< 0.001***
Emotional exhaustion	276.975	1	240	< 0.001***
Depersonalisation	41.489	1	220	< 0.001***
Professional accomplishment	41.696	1	214	< 0.001***
Psychological distress	318.319	1	207	< 0.001***
Psychological flexibility	76.481	1	246	< 0.001***
Age	3.249	1	247	0.073
Seniority	1.466	1	244	0.227
Working hours	0.353	1	235	0.553
Turnover intentions	48.664	1	239	< 0.001***

Note: Results of one‐way analysis of variance (ANOVA) comparing cluster means for each variable.

Abbreviations: ddl1, numerator degrees of freedom; ddl2, denominator degrees of freedom; F, F‐statistic; p, probability value.

^*** p < 0.001 (highly significant).

4.2.1. Cluster 1: Low Burnout/High Psychological Flexibility (52.6% of the Sample)

This cluster was predominantly female (89.31%). No significant differences were observed regarding age (F(1, 247) = 3.25, p = 0.073) or seniority (F(1, 244) = 1.47, p = 0.227) between the two clusters.

This profile is characterised by high engagement, as evidenced by significantly lower scores on measures of emotional demands (F(1, 245) = 62.54, p < 0.001), work intensity (F(1, 246) = 66.29, p < 0.001), perceived stress (F(1, 239) = 188.04, p < 0.001), emotional exhaustion (F(1, 240) = 276.97, p < 0.001), patient depersonalization (F(1, 220) = 41.49, p < 0.001) and psychological distress (F(1, 207) = 318.32, p < 0.001) compared to the other cluster. Additionally, these nurses reported significantly higher levels of professional accomplishment (F(1, 214) = 41.70, p < 0.001) and psychological flexibility (F(1, 246) = 76.48, p < 0.001). Consistent with this profile, this cluster displayed significantly lower turnover intentions (F(1, 239) = 48.66, p < 0.001).

4.2.2. Cluster 2: High Burnout/Low Flexibility (47.3% of the Sample)

This cluster was also predominantly female (93.22%), with a slightly lower proportion of males compared to the first cluster. As previously noted, no significant differences were found for age and seniority.

This profile is distinguished by marked signs of burnout, manifested by significantly higher levels of emotional demands, work intensity, perceived stress, emotional exhaustion and patient depersonalization (all p < 0.001). Conversely, these nurses reported significantly lower levels of professional accomplishment and psychological flexibility (all p < 0.001). Psychological distress was also significantly higher in this group (p < 0.001). Consequently, this cluster exhibited significantly higher turnover intentions (p < 0.001) compared to the first cluster. No significant differences were found between the two clusters regarding working hours.

5. Discussion

The aim of the study was to identify distinct profiles of nurses based on their experiences with occupational risks, psychological disorders, protective factors such as psychological flexibility and their intention to leave the job (turnover). The primary aim of this study was to delineate distinct profiles of nurses based on their experiences with occupational risks, psychological disorders, protective factors such as psychological flexibility and their intention to leave their job (turnover).

In the present study, we identified two distinct clusters within our clinical population. The first cluster (52.6%) exhibited characteristics indicative of adequate working conditions, high psychological flexibility and few indications of stress or burnout. Conversely, the second cluster (47.3%) reported exposure to work‐related risks, lower psychological flexibility and demonstrated signs of psychological distress, including stress, burnout and a lack of professional fulfilment.

5.1. Job Demands

In the present study, our two identified clusters initially align with patterns frequently described in the literature regarding the impact of work‐related stress on nurses' psychological well‐being. Indeed, the psychological suffering of nursing staff is exacerbated by work‐related stress. For instance, work schedules and workload generate a significant proportion of psychological discomfort (Nieuwenhuijsen et al. 2010). Our study confirms the link between indicators of psychological distress (e.g., psychological distress, stress and burnout) and working conditions. Recent person‐centred studies confirm that the configuration of job demands and work environment stressors continues to shape distinct profiles of nurse well‐being. In a four‐profile latent analysis of 2079 US hospital nurses, Rink et al. (2023) found that the subgroup with very high emotional exhaustion and low emotional thriving reported the greatest workload, poorest work‐life integration and highest intention to leave. Similarly, a three‐profile study of 2092 oncology nurses in China showed that nurses classified as ‘high‐pressure adaptive’—characterised by higher workloads and more frequent conflicts with physicians—had significantly lower patient safety competence (Ma et al. 2025). These findings, obtained in very different health care contexts and published within the last 2 years, parallel our high burnout profile in both prevalence (23%–27%) and risk pattern, reinforcing the external validity of the present typology.

Recent evidence confirms robust positive associations between psychological distress, burnout and nurses' intentions to leave their jobs (Bruyneel et al. 2023; Özkan ²⁰²²; Xiao et al. ²⁰²²).

5.2. Individual Differences

Beyond the work‐related factors affecting nursing staff, our results suggest individual differences between members of the two groups, particularly in terms of gender and psychological flexibility.

Individual differences continue to shed light on why some nurses are more vulnerable to distress than others: a large cross‐sectional study in China showed that nurses with a ‘distressed’ personality profile were 4.52 times more likely to meet diagnostic criteria for burnout, whereas those with a ‘resilient’ profile had a 55% lower‐risk (Zhang, Li, et al. ²⁰²⁴; Zhang, Xiao, and Tian ²⁰²⁴); Italian paediatric nurses who scored high on neuroticism or type D personality reported significantly greater emotional exhaustion and secondary traumatic stress than their colleagues (Di Santo et al. 2022); and a meta‐analysis of 27 studies confirmed that high neuroticism is consistently associated with higher burnout in all nursing contexts (Sarafis et al. 2021).

The seniority of nursing staff has also been the subject of research using cluster analysis. This study identified two groups similar in some respects to those obtained in our research. Nurses in group 1 were younger, more educated, had less work experience and had a higher intention to change careers than nurses in group 2. Nurses in group 2 had more work experience, were at a higher level, and were more satisfied with their current employment in terms of peer support, autonomy, career opportunities, schedule and relationships with team members than nurses in group 1 (Chan et al. 2010).

Other studies using cluster analysis have shown results similar to our study in terms of the clustering of psychological disorders and working conditions. Nurses in the first group had a high prevalence of mental disorders, as well as high job demands and conflicts at work. Nurses in the second group exhibited moderate mental strain, chronic illnesses, were significantly older and reported fewer job demands. The lowest proportion of self‐reported mental disorders was observed in the third group, which reported moderate job demands and better physical health compared to the other two groups (p < 0.05) (Tran et al. 2019). Although work‐related factors appear to be important determinants of burnout among nurses, certain individual factors seem to play a crucial role in the development of several disorders, including burnout (Zhang, Li, et al. ²⁰²⁴; Zhang, Xiao, and Tian ²⁰²⁴).

5.3. The Question of Psychological Flexibility Remains to be Explored

In the present study, we found a significant difference in psychological flexibility between the two identified groups. This finding aligns with several other studies examining the role of psychological flexibility in nurse well‐being. This finding aligns with several other studies. Our mean score for psychological flexibility (M = 29.8, SD = 7.4 on the AAQ‐II) is comparable to values reported in more recent European samples of nurses using the same instrument, such as Belgian intensive care nurses (M = 31.2, SD = 7.1; Bruyneel et al. ²⁰²³) and Spanish medical‐surgical nurses (M = 30.6, SD = 6.8; Álvarez‐Gelves et al. ²⁰²²).

Simultaneously, 65.5% of them experience moderate job burnout (M = 59.61). Although 73% express a low to moderate intention to leave their position, a significant negative correlation was established between psychological flexibility and job burnout (r = −0.304) as well as the intention to leave (r = −0.258). These results suggest that psychological flexibility acts as a protective factor against job burnout and the intention to leave, explaining 23.2% and 12.7% of the variance in these two variables, respectively (El‐Ashry et al. ²⁰²⁴).

Other studies linking psychological flexibility to work‐related stress have been conducted in the context of the COVID‐19 pandemic. One study suggests that psychological flexibility tends to support the psychological resilience processes of nursing staff. Moreover, nursing staff with high scores in flexibility and resilience are less likely to experience workplace tension, even in the chaotic context of a pandemic (Orhan et al. 2024).

In a geriatric context, similar results have been observed. A study indicates that geriatric nurses exhibit moderate levels of job burnout and compassion fatigue, and high levels of psychological flexibility. Flexibility seems to act as a stress reliever, which would promote a good measure of compassion (Sarabia‐Cobo et al. ²⁰²¹).

Additionally, the concurrence of several factors such as young age, night shift work and low levels of psychological flexibility may contribute to the deterioration of mental health and well‐being among nursing staff (Li et al. 2024). Another study linking age, working conditions and psychological flexibility indicated that junior nurses exhibited high levels of burnout in various dimensions. Specifically, young nurses reported a lack of support from supervisors and poor relationships with colleagues. They exhibited the lowest levels of psychological flexibility and the highest levels of perceived stress and burnout. Furthermore, this study shows that experiential avoidance, cognitive fusion, perceived stress and burnout were positively associated with each other. Their regression model suggests that psychological flexibility (particularly cognitive fusion) and perceived stress influence burnout among junior nurses. This implies that greater psychological flexibility and lower perceived stress could improve burnout among junior nurses (Zhao et al. 2023).

6. Strengths and Limitations

The cross‐sectional nature of our study, while providing valuable insights into associations between variables, has significant limitations. Cross‐sectional designs cannot establish causal relationships, nor can they capture dynamic changes over time. To fully understand the underlying mechanisms and identify risk factors, longitudinal research is essential.

The response rate of 53.5% (267/499) is consistent with recent web‐based surveys of European nurses, which typically range between 40% and 55% (Bruyneel et al. 2023). The reported 53.5% completion rate reflects the fraction of survey accesses that yielded usable data and does not correspond to a conventional response rate since we cannot determine the total number of nurses who viewed the invitation via email or social media. This uncertainty around the true denominator may introduce self‐selection bias and limit the generalizability of our findings. Although this mitigates concerns about severe non‐response bias, we cannot rule out the possibility that nurses experiencing the highest levels of distress chose not to participate, leading to conservative estimates of burnout prevalence. In terms of sample size, 267 cases met the rule of thumb of at least 20 × k observations (k = 11 clustering variables) (Mardia et al. 1979) recommended for stable K‐means solutions, and internal validation tests confirmed robustness.

Also, the reliance on self‐reported questionnaires in this study introduces the potential for bias. Individuals experiencing psychological stress may be inclined to underreport or exaggerate aspects of their work environment or negative perceptions. To mitigate these limitations, future research should explore alternative, more objective data collection methods, such as direct observation of workplace conditions. Additionally, adopting more rigorous study designs, such as cohort or longitudinal studies, would provide a more comprehensive understanding of the phenomena under investigation. We excluded participants who did not complete all core survey modules, which may have introduced bias. However, we were unable to characterise differences between excluded and included participants due to the anonymity of the survey.

Regarding the robustness of our statistical analysis, while K‐means cluster analysis is generally robust to violations of homoscedasticity, the observed unequal variances for certain variables (work intensity, depersonalization of patients, professional accomplishment, psychological distress and psychological flexibility) represent a limitation of our study. This heteroscedasticity could potentially influence the size and composition of the identified clusters, as well as the significance of the differences observed between them for these specific variables. Therefore, the interpretation of the results concerning these variables should be made with caution, and future research could explore the use of alternative clustering methods or data transformations to mitigate this effect.

While social media may have increased visibility and participation, it may also introduce self‐selection bias by favouring digitally engaged, younger or motivated staff, potentially limiting the representativeness of the sample. Future studies should consider combining digital outreach with face‐to‐face reminders or stratified sampling to improve generalizability.

Despite meeting statistical thresholds for cluster stability and power, the sample of 267 nurses represents only a small fraction of the approximately 600,000 registered nurses in France; therefore, the profile prevalence estimates should be interpreted with caution and should not be assumed to be generalisable to the national workforce.

7. Recommendations for Further Research

Although our study has identified two distinct groups, further investigations are needed to elucidate the relationship between the various variables. For instance, we could explore the specific mechanisms of psychological flexibility (such as acceptance and cognitive fusion) that may impact or prevent chronic stress. Additionally, qualitative studies could be conducted to better understand the underlying mechanisms. This could highlight known phenomena or uncover new avenues of research. Future research could also adopt a longitudinal design to observe causal relationships between variables. Similarly, studies with a test–retest design using mindfulness training or Acceptance and Commitment Therapy (ACT) could be conducted.

Also, external replication—using independent samples or model‐based approaches—would further confirm the robustness of these profiles.

8. Implications for Policy and Practice

Healthcare facilities, or any organization employing nursing staff, would greatly benefit from improving working conditions. Indeed, the link between working conditions and various indicators of distress (stress, burnout, etc.) is well supported in the literature, and we are also aware of the impact on the organization, with a loss of quality of care and high turnover rates. These organisations could also offer programmes to help nursing staff manage difficult aspects of the job, such as emotional demands. For example, interventions aimed at increasing psychological flexibility (acceptance and commitment therapy, mindfulness training, support groups) could be implemented.

For nursing professionals, this research offers a valuable opportunity to gain a deeper understanding of the multi‐faceted nature of workplace distress, as highlighted by the various factors explored in this study. Furthermore, the findings underscore the potential of psychological flexibility as a personal resource that nurses can cultivate to enhance their overall well‐being.

The two data‐driven profiles highlight actionable areas within hospitals. A brief screening of demand and symptom indicators, combined with a short psychological flexibility measure, can help managers identify at‐risk teams. For these units, practical steps may include stabilising rosters. They also include moderating workload intensity. Another step is offering time for debriefing after events. One‐way of doing this is by embedding skills‐based training. This strengthens psychological flexibility. It can be embedded within induction and continuing education. These measures may reduce strain. They can also reduce intention to leave.

Policy‐level workforce planning should treat profiles, not just mean scores, as targets. This allows hospitals and regional authorities to incorporate profile distributions into dashboards, set improvement goals and make resource adjustments and training. Procurement and accreditation processes should prioritise programmes compatible with staffing constraints. Further work should be done to test durability and generalisability. Multi‐site, longitudinal studies are needed to examine whether shifts in profile membership precede changes in turnover and sickness absence. External replication should compare K‐means with model‐based approaches and validate simple assignment rules. Trials evaluating the effects of organisational levers and individual training on psychological flexibility would clarify which combinations yield the greatest retention gains. Given our digital recruitment route and the under‐representation of older nurses, future studies should use mixed‐mode sampling to ensure age diversity and test whether profile membership varies by unit type, shift pattern or career stage. Finally, measurement work examining invariance of the MPFI‐24 across demographic groups would support screening and support.

9. Conclusion

Working conditions in healthcare settings have a significant impact on nursing staff, contributing to burnout, elevated stress levels and psychological distress. Moreover, these conditions also affect the organization itself, leading to high turnover rates and a decline in the quality of patient care. Our findings show a way to act in busy clinical environments. Treating ‘burnout’ as a profile makes visible the co‐occurrence of high demand, high symptom burden and low psychological flexibility in a subgroup of nurses. This invites targeted responses. Simple screening and a quick measure of psychological flexibility can help managers identify teams at risk and trigger support. Adjusting task loads and rosters and conducting brief debriefs are low‐burden solutions. During training, education and shifts, supervisors can reinforce the messages. Combining these solutions will likely shift the profile towards lower strain. Profile distributions are more informative than mean scores for monitoring risk and planning retention strategies. Dashboards can set explicit targets and allocate units where higher‐risk profiles predominate and evaluate changes to staffing and scheduling. Scalability under staffing constraints can reach those most likely to benefit. Future tests should check durability and generalisability. Follow‐up will show if movement out of the higher‐risk profile leads to reduced intention to leave, sickness absence and reports. Replication will clarify transferability, and model‐based comparisons can test the two‐profile solution. Digital recruitment may underrepresent older staff; use mixed‐mode sampling to ensure age diversity. Audits of scale performance across demographic groups will ensure fair screening and targeted support.

Author Contributions

M.C.M. came up with the original idea. M.C.M., O.Z., C.I.B. and L.R. built the design of the study. M.C.M. collected data. M.C.M. and O.Z. did the statistical analysis. M.C.M. wrote a first version of the draft. All authors helped refine it. Every author approved the submitted version.

Ethics Statement

All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. This psychological research complies with the ethical guidelines of the Helsinki Committee.

Consent

Informed consent was obtained from all participants included in the study.

Conflicts of Interest

The authors declare no conflicts of interest.

Appendix 1. Evaluer Les Risques Psychosociaux: Guide RPS/DU (2024)—Assessing Psychosocial Risks: RPS/DU Guide (2024)

Items to assess workload

• –Am I subjected to high work pace demands?
• –Are the set objectives compatible with the means and responsibilities allocated to me to achieve them?
• –Are my objectives clearly defined?
• –Do I receive instructions, orders, or requests that may be contradictory?
• –Am I unexpectedly required to change tasks, positions, or roles to meet immediate demands?
• –Am I frequently interrupted during my work by unforeseen tasks?
• –Do I perform activities that require sustained attention or constant vigilance?

Items to assess emotional demands

• –Does the organization of work generate tensions with the public (patients, residents, families…)?
• –Do I have the means to act effectively when faced with the suffering, distress, or difficulties of the people I am in charge of?
• –In my work, do I have to ‘put on a brave face’ (or ‘present a good image’) in all circumstances?

Appendix 2. The STROBE Checklist

Item 1, Title and abstract—Study design stated in title and in abstract first sentence; informative, balanced summary provided.

Item 2, Background/rationale—Scientific background and rationale explained.

Item 3, Objectives—Aim and four objectives stated explicitly in the Abstract and Introduction.

Item 4, Study design—Key elements of design presented early in Methods.

Item 5, Setting—Description of setting, locations and relevant dates, including period of data collection.

Item 6, Participants—Eligibility criteria and recruitment procedures detailed.

Item 7, Variables—All outcomes, exposures, predictors, potential confounders and effect modifiers defined, including diagnostic criteria where applicable.

Item 8, Data sources/measurement—Sources of data and methods of assessment for each variable described, with reliability/validity evidence.

Item 9, Bias—Efforts to address potential sources of bias discussed.

Item 10, Study size—Sample‐size considerations explained.

Item 11, Quantitative variables—Handling of quantitative variables and any groupings described.

Item 12, Statistical methods—All statistical methods, including cluster derivation, validation checks, handling of missing data and sensitivity analyses, specified.

Item 13, Participants—Numbers at each stage of study and flow diagram noted.

Item 14, Descriptive data—Characteristics of study participants and information on exposures and potential confounders provided.

Item 15, Outcome data—Numbers of outcome events or summary measures reported for each profile.

Item 16, Main results—Unadjusted and adjusted estimates with precision and reference category stated; p values and effect sizes provided.

Item 17, Other analyses—Additional analyses (e.g., sensitivity checks, subgroup comparisons) described in the results section.

Item 18, Key results—Summary of key results with reference to objectives.

Item 19, Limitations—Study limitations, potential bias and imprecision discussed.

Item 20, Interpretation—Overall interpretation of results in context of objectives and relevant evidence.

Item 21, Generalizability—External validity and applicability of findings considered.

Item 22, Funding—Source of funding and role of funders stated.

Mollet, M. C. , Zerhouni O., Bagnis C. I., and Romo L.. 2026. “Preliminary Clustering: Exploring the Interplay of Burnout, Stress, Turnover, Psychological Flexibility and Distress in a French Nurse Sample.” Journal of Advanced Nursing 82, no. 6: 6205–6219. 10.1111/jan.70299.

Data Availability Statement

Data available on request from the authors.