Trajectories and Predictors of Emotional Exhaustion in Clinical Nurses in the Context of Healthcare Crisis: A Case Study in Xi'an, China

Background In response to the diminishing toxicity and fatality of the novel coronavirus, China implemented a policy shift at the end of 2022 to relax its control measures pertaining to the COVID-19 pandemic. Consequently, a rapid surge in community-level infections ensued, exerting a pronounced strain on the medical and healthcare systems and posing signi�cant challenges and pressures for healthcare workers. Objective This study investigated the growth trajectory of emotional exhaustion and its predictors in clinical nurses in the context of healthcare crisis.


Introduction
The novel coronavirus pneumonia, as the most widespread and impactful public health emergency [1], has posed a serious threat to the lives, health, and property safety of people worldwide [2,3].Since the outbreak of COVID-19 pandemic, China has taken strict control and prevention measures and followed the dynamic zero-COVID policy [4].It was not until late 2022, when the Omicron variant with the waning virulence became prevalent, that China chose to "coexist" with the virus, thus avoiding a massive epidemic-related death [5].The relaxation of epidemic control measures led to a large-scale infection within a short period of time.Hospitals experienced different degrees of medical overcrowding, especially during the rst two weeks, with a high patient volume in the emergency departments and a signi cant lack of beds in the medical wards, particularly in the respiratory intensive care unit and emergency treatment [6].As frontline healthcare workers, they faced high work pressure and the risk associated with the healthcare system's openness and overcrowding, which increased their vulnerability to both physical and psychological stress.The combination of intense work demands and the risk of infection elevated the risk of developing mental health disorders [7,8].
The occurrence of emotional exhaustion among nursing staff, who were the main body of healthcare workers and directly in contact with patients, was not optimistic under the impact of healthcare system overload [9,10]..They not only needed to overcome their own fears and risks of being infected with infectious diseases but also had to deal with the transmission of negative emotions from patients who were affected by the infection, such as anxiety and irritability [11].In this situation, nursing staff were prone to experiencing feelings of stress and intense work pressure, which resulted in a more pronounced issue of emotional exhaustion among them [12].
Emotional exhaustion referred to an extreme state of emotional fatigue and depletion that occurred when individuals experienced excessive emotional demands and over-extension in their work.It re ected a sense of helplessness and psychological burnout in the individual's professional role [13].The study revealed a close association between emotional exhaustion and negative emotions such as insomnia and impaired attention [14].Emotional exhaustion was also considered a risk factor for cardiovascular diseases, gastrointestinal disorders, and other related illnesses [15,16].Once nursing staff experienced emotional exhaustion, it not only affected their physical and mental wellbeing but also impacted their treatment to the patients, leading to a decrease in work e ciency.This had serious implications for the nursing staff themselves, the hospital, and society as a whole [17].Moreover, emotional exhaustion, as a core component of occupational burnout, was considered an initial stage in the process, which only developed into full-blown burnout when individuals did not cope effectively with work-related stressors [18,19].Therefore, preventing emotional exhaustion was crucial for maintaining the physical and mental well-being of individuals as well as preventing occupational burnout [20].
Psychological capital referred to a positive psychological state developed by individuals during their growth and developmental process.It encompassed four core components: hope, self-e cacy, resilience, and optimism [21,22] Previous studies have indicated that psychological capital had a signi cant positive impact on individuals' attitudes, behaviors, and performance.It also demonstrated a bene cial buffering effect on negative emotions [23].The research showed that psychological capital attenuated the workload-exhaustion relationship [24].Therefore, we speculated that it also had a certain buffering effect on individuals' emotional exhaustion.
It was well known that individual emotions were not stable and unchanging, but rather dynamically uctuated over time [25].Therefore, in the context of healthcare system overload, emotional exhaustion among nursing staff was not a stable and unchanging state but rather a potentially modi able and dynamic developmental process.Therefore, we hypothesized that emotional exhaustion in different clinical nurses may develop different trajectories over time.
Despite scholars conducting relevant studies on the emotional exhaustion of nursing staff, most of the studies were mainly cross-sectional.As a result, these research methods were unable to fully re ect the trajectory of changes in emotional exhaustion in clinical nurses [9,20,26], and overlook the dynamic temporal variations in emotional exhaustion experienced by nursing staff over time.Besides, previous studies on the changes in emotional exhaustion often treated the psychological states of nurses as uniformly distributed, overlooking the issue of individual differences.There was a lack of longitudinal exploration regarding the differential impact factors among different developmental trajectories.Therefore, there is an urgent need for a new approach to address the shortcomings in previous studies, and the growth mixture model (GMM) offers a promising approach to address these de ciencies [27].GMM is a common statistical method that groups heterogeneous populations according to their development trajectories.It can provide methods to identify which subset of a variable based on the trajectory distinction is more susceptible to the in uence of another variable and to clarify the nonlinear relationship between these variables [28].
The results of this research are expected to provide important theoretical references and practical guidance for the protection against emotional exhaustion and the adjustment of psychological well-being among nursing staff in the post-pandemic era.Additionally, it will also have theoretical implications and practical signi cance for the future maintenance of psychological health among all healthcare workers in responding to major outbreaks of infectious diseases in public health emergencies.The ndings can serve as empirical evidence for preventing occupational burnout among nursing staff, reducing attrition rates, ensuring a stable workforce, improving the quality of clinical nursing, and enhancing team effectiveness.

Participants
In this study, convenience sampling was used to choose the clinical frontline nurses after the relaxation of epidemic control policies.The inclusion criteria were as follows: nursing personnel who have obtained nursing quali cation certi cates and are engaged in frontline clinical nursing work, voluntarily participate in the research and ensure their continuous participation in the study.Exclusion criteria: Those who have recently experienced signi cant life events.This prospective study investigated eligible clinical nurses at three time points.The calculation of the sample size refers to the table used for sample size estimation in the design of single-group repeated measurements.We selected with a mean correlation coe cient of r = 0.5, f = 0.14 (weak effect), α = 0.05, and a sample size of 142 nurses was required under conditions that ensured 1-β = 0.8.Considering a potential dropout rate of 20%, the minimum sample size for this study is 178 participants.Methods

General Information Survey Form
The questionnaire was designed by the researcher, including demographic characteristics such as gender, age, professional title, marital status, years of work experience, department, involvement in emergency public health event rescue, and educational level.

Maslach Burnout Inventory
Maslach Burnout Inventory included three dimensions: Emotional Exhaustion, Depersonalization, and Personal Accomplishment.For this research, the emotional exhaustion subscale was selected, consisting of ve items, including "I feel very exhausted," "I worry that work affects my mood," "I often feel worn out," "At the end of the day, I feel extremely fatigued," and "I have been feeling somewhat depressed lately."The scale was rated on a 7-point Likert scale, ranging from "completely not applicable" to "extremely applicable," with scores ranging from 1 to 7. Higher scores indicated a higher level of emotional exhaustion in individuals.The scale is widely used among the nursing population in China [17,29], and had demonstrated good reliability and validity.The Cronbach's α coe cients of the scales used in this study were 0.823.

Psychological Capital Scale
The Psychological Capital Scale consisted of 24 items, including four dimensions: self-e cacy, hope, resilience, and optimism.Each dimension contained six items, rated on a 6-point Likert scale, ranging from "strongly disagree" to "strongly agree," with scores ranging from 1 to 5. The overall level of psychological capital was determined by the total scores of each dimension, with higher scores indicating higher levels of psychological capital in the respondents.The scale was widely used among the nursing population in China [30,31], and had demonstrated good reliability and validity.

Procedure
This study involved frontline clinical nursing staff who were surveyed at three time points with a 2-week interval: the rst week after the relaxation of COVID-19 control policies in China (Wave 1), the third week after the relaxation of COVID-19 control policies in China (Wave 2), and the fth week after the relaxation of COVID-19 control policies in China (Wave 3).The researchers initiated communication and informed the eligible clinical nursing staff about the purpose of the study and the subsequent follow-up survey times, ensuring that they had the option to terminate or withdraw from the study at any time.After obtaining informed consent from the participants, baseline data were collected.To ensure the accuracy and reliability of the follow-up survey data, the time points for each follow-up survey were xed within a three-day interval.In total, 422 clinical nurses were included in the study, and ultimately, 358 (response rate: 84.83%) completed all the surveys.Among them, 57 (13.51%) clinical nurses were on leave due to COVID-19 infection, while 7 (1.66%)clinical nurses explicitly stated their lack of interest in continuing the study and withdrew from participation.

Statistical Analyses
The data were analyzed with SPSS 26.0 and Mplus 8.3 software.The unconditioned Latent Class Growth Model (LCGM) and Gaussian Mixture Model (GMM) were used to judge the trajectory categories with Mplus 8.3 software.The baseline model was a single-category model, setting the variance within the category to 0, increasing the number of categories in the model one by one, and then comparing the tting indexes between the models.The optimal model was determined by combining the practical signi cance and statistical indices.Fit indicators included the akaike information criterion (AIC), bayesian information criterion (BIC), adjusted BIC (aBIC), entropy, likelihood ratio test (LRT), and bootstrapped likelihood ratio test (BLRT).The smaller the AIC, BIC and aBIC values are, the better the model t.Entropy represented the accuracy of classi cation, and the higher the entropy value was, the more accurate the classi cation was.LRT and BLRT were commonly used to compare t differences between K-1 and K-category models, with P < 0.05 indicating that the K-category model was superior to the K-1-category model.The category to which an individual belonged was then determined based on a posterior probability.According to the tting results of the individual category model, the above indexes were comprehensively evaluated, the best tting model was selected, and the nurses were divided into different categories.Then, SPSS 26.0 was used for analysis.Data that followed a normal distribution were expressed as the mean and standard deviation, and analysis of variance was used to compare multiple groups.Count data were expressed by frequency and constituent ratio, and the chi-square test was used to compare multiple groups.Logistic regression analysis was used to explore the in uence of age, years of working experience, self-e cacy, hope, gender, past participation in emergency public health event rescue, sleep quality on the class of emotional exhaustion.There was a signi cant difference at P < 0.05.

Ethical considerations
This study has been reviewed and approved by the Ethics Committee of the First A liated Hospital of Air Force Medical University (KY20224143-1).Prior to the implementation of the study, strict adherence to the principle of informed consent was followed.The participants were introduced to the purpose, methods, and signi cance of the study and were informed about the principles of voluntary participation and strict con dentiality of their information.
Their consent was obtained, and they signed an informed consent form.All data and information pertaining to the study subjects are strictly used for the purpose of this research only.

Identi cation of the Trajectories of Emotional Exhaustion in Clinical Nurses in the Context of Healthcare Crisis
Taking the emotional exhaustion score of clinical nurses at 3 time points as an observation index, data from 358 clinical nurses were included in the model analysis.First, LCGM was used and set as the free estimation of the time parameter, and 1 to 5 categories were extracted in turn.When the number of potential classes increased from 1 to 5, AIC, BIC and aBIC all decreased.However, when the number of categories increased from 3 to 4, entropy decreased, and LRT and BLRT became insigni cant, suggesting that the best t was to retain 3 categories.
To further judge the optimal model, the model was set to GMM, AIC, BIC, aBIC values of the model of Category 2 to 5 were less than the values of LCGM, suggesting that the model was optimized.Following the selection criteria in LCGM, in GMM, the best t was obtained by retaining 3 categories.Based on the above information, the classi cation probability of the model and the interpretability of the results, the 3 categories of GMM were retained.The results of model tting are shown in Table 1.Each class was named based on the emotional exhaustion score and its changing trend.Class 1 (9.78%) had consistently higher scores, and was named the "High-level and Continuously Rising Emotional Exhaustion Group".In Class 2 (83.52%), emotional exhaustion was at a moderate level and gradually decreased over time, so this group was named the "Moderate-level and Continuously Decreasing Emotional Exhaustion Group".The score of Class 3 (6.70%)was the lowest, and the scores decreased continuously during follow-up, and this group was named the "Lowlevel and Continuous-Decreasing Emotional Exhaustion Group".The three trajectories of emotional exhaustion from the model t are shown in Fig. 1.

Healthcare Crisis
Single factor analysis was used to identify the possible predictive factors according to the different categories determined by GMM, with P < 0.05 indicating a signi cant difference.The general data and psychological capital were included in the single factor analysis.The results showed that age (χ 2 = 24.348,P < 0.001), years of working experience (χ 2 = 12.301, P = 0.015), gender (χ 2 = 69.709,P < 0.001), past participation in emergency public health event rescue(χ 2 = 60.967,P < 0.001) and sleep quality (χ 2 = 24.699,P < 0.001), were signi cantly different among the three groups.The difference between the other variables was not signi cant.The results are shown in Table 2.The mean scores of psychological capital of the three groups were (101.66 ± 16.12), (103.57± 14.94), (112.79 ± 15.01), with F = 4.636, P = 0.010.Further pairwise comparisons within the group showed that the comparison of class1 and class3, class2 and class3 were signi cant (P < 0.01).The variance analysis of the four dimensions of psychological capital showed that there were statistically signi cant differences in scores of self-e cacy and hope dimensions among the three groups of emotionally exhausted individuals.Further pairwise comparisons revealed statistically signi cant differences in self-e cacy and hope scores between Group 1 and Group 3, as well as between Group 2 and Group 3. As shown in Table 3.The emotional exhaustion trajectories of clinical nurses determined by GMM were used as the dependent variable, and Class 3 was the control group.The independent variables were assigned as follows: Gender: 0 = Male, 1 = Female; Have you ever participated in the rescue of public health emergencies: 0 = Yes, 1 = No; Sleep quality: 0 = Poor sleep, 1 = As usual, 2 = Good sleep.Logistic regression analysis of emotional exhaustion trajectory categories in clinical nurses was shown in Table 4.After the relaxation of epidemic control policies, in the context of healthcare crisis, the majority of frontline clinical nursing staff experienced a continuous decline in emotional exhaustion, which was classi ed as the "the Moderatelevel and Continuously Decreasing Emotional Exhaustion Group", accounting for 83.52%.This indicated that emotional exhaustion was commonly present among nursing staff when dealing with healthcare crisis, which was consistent with the research ndings of Crippa et al [32].This required attention from hospital administrators and nursing staff [33].In the context of healthcare crisis, a large number of patients were ooding into hospitals, resulting in overloaded operations.Nursing staff were facing the risk of infection while working under immense pressure.In this situation, they were prone to experience emotional exhaustion [34].The state of emotional exhaustion could have a serious impact on the physical and mental health of nursing staff, as well as their work e ciency and the quality of nursing care [35].. Paying attention to the emotional state of nurses could improve the care conditions of patients, and nursing managers should actively work towards improving the emotional exhaustion state of nurses in different situations and crises.In our study, we observed that the emotional exhaustion in Class 2 gradually decreased.This may be attributed to the nursing staff gradually adapting and regulating themselves, as well as the proactive attention from hospital managers towards the physical and mental health of the nursing staff.It also correlated with the increasing support and concern from various sectors of society towards healthcare workers.This suggested that in the early stage of healthcare crisis, it was crucial to pay high attention to and address the emotional well-being of this group of nursing staff.
The condition of nursing staff in the High-level and Continuously Rising Emotional Exhaustion Group was worthy of active attention.This group accounted for 9.78% in our study, and their emotional exhaustion levels showed a continuous increase.This suggested that individuals in the high emotional exhaustion state were less likely to regulate themselves effectively.Therefore, it was important to actively address the work and emotional well-being of this group in clinical practice.
The Predictive Indicators of Trajectories of Emotional Exhaustion in Clinical Nurses in the Context of Healthcare Crisis The results showed that age, year of working experience, gender, past participation in emergency public health event rescue, sleep quality, psychological capital were signi cantly different among the three groups.Further logistic regression analysis demonstrated that male nursing staff and those who had participated in public emergency health event rescue were less likely to experience emotional exhaustion.This nding was consistent with the research results of Clari et al [36] According to the study ndings, it was revealed that female nurses exhibited elevated levels of emotional exhaustion compared to their male counterparts.This disparity could be attributed to the additional social responsibilities that females often shoulder in relation to household and family obligations, alongside their professional duties.The challenges associated with striking a harmonious equilibrium between managing job-related demands and ful lling child-rearing responsibilities could signi cantly contribute to heightened stress levels among female nurses [37,38].In contrast, males exhibited stronger psychological resilience compared to females [39].
Therefore, it was essential to provide female nurses with enhanced care and support to help them cope with the stress of healthcare crisis.Compared to nursing staff who had not participated in public emergency health event rescue, those who had experience in crisis management during such events exhibited lower levels of emotional exhaustion in the context of healthcare crisis.This insight suggested that nursing staff should actively engage in the summarization and sharing of their experiences in responding to public health emergencies, as it could prove valuable for their future work.Compared to Class 3, nursing staff with poor sleep quality were more likely to be categorized into the high-level emotional exhaustion group.The study demonstrated a strong positive correlation between emotional exhaustion and sleep condition [35].During the COVID-19 pandemic, the risk of psychological distress among healthcare professionals had been increased [40] and particularly frontline nurses had been shown to face enormous mental health challenges [41].Intense psychological pressure may lead to poor sleep quality, further exacerbating the condition of emotional exhaustion.Therefore, managers should pay active attention to female nursing staff and those without rescue experience, providing them with support and encouragement to help them better cope with the impact of medical crises.

Limitations
Firstly, this study utilized convenience sampling and only focused on a sample of tertiary hospitals in Xi'an city.The pandemic situation in Xi'an may not represent the overall pandemic situation in China, which could introduce potential biases in the results.Secondly, due to the intense medical pressure and busy clinical work after the relaxation of pandemic control policies, it was challenging to collect a large sample size for this study.Additionally, there were limited time points for data collection and follow-up.Thirdly, the causal relationship between poor sleep quality and high emotional exhaustion required further examination and veri cation.

Conclusion
Our study proved the heterogeneity of emotional exhaustion in clinical nurses in the context of healthcare crisis.By examining the in uence of demographic factors and psychological capital, we have identi ed a subgroup at higher risk of developing a high emotional exhaustion pattern, characterized by being female, having no previous experience in public emergency health event rescue, and reporting poor sleep quality.Future studies should focus on developing timely interventions targeted at these speci c subpopulations. Figures

Table 2
Single-factor analysis of in uencing factors of emotional exhaustion trajectories (N = 358) Variance Analysis of Psychological Capital of Emotional Exhaustion Trajectories in Clinical Nurses in the Context of Healthcare Crisis

Table 3
Variance analysis of psychological capital of emotional exhaustion trajectories (N = 358) a : Comparison of the rst and third items (P < 0.05) b : Comparison of the second and third items (P < 0.05) Logistic Regression Analysis of In uencing Factors of Emotional Exhaustion Trajectory Categories in Clinical Nurses in the Context of Healthcare Crisis

Table 4
Logistic regression analysis of emotional exhaustion trajectory categories in clinical nurses in the context of medical shock (N = 358)