Moral injury severity profiles and their pathways to negative emotions in medical staff: the differential mediating role of emotion regulation Authors: Runkang Zhao, Kai Zhang, Yiming Zhou, Kuiliang Li, Yuting Zhang, Yifei Wang, Lei Ren Published: 2026 ABSTRACT BackgroundMoral injury (MI) and negative emotions (NE) are core psychological issues that severely impair the mental health and professional competence of medical staff, while emotion regulation serves as a crucial psychological resource for coping with occupational psychological distress. However, the latent severity profiles of MI among medical staff, as well as the differential mediating effects of emotion regulation strategies, i.e., cognitive reappraisal (CR) and expressive suppression (ES), between MI and negative emotion (NE) remain unexplored.ObjectiveThis study aimed to identify the latent subtypes of MI among medical staff via latent profile analysis and to explore the mediating role of CR and ES between MI and NE, so as to provide empirical evidence for formulating targeted psychological intervention strategies for medical staff.MethodsThis was a cross-sectional survey. 1001 medical staff were enrolled via a convenience sampling method. Questionnaires were used to assess MI, anxiety, depression and emotion regulation among medical staff, with latent profile analysis applied for subgroup identification. The relationships among variables were analyzed using Pearson correlation analysis and mediation analysis.ResultsThree profiles of MI were identified and designated as the low injury profile (62.84%), emotion injury profile (25.27%), and value injury profile (11.89%). MI correlated positively with NE (r = 0.38–0.41) and negatively with CR (r = -0.19). Overall, CR had mediated MI-NE associations, while ES did not. Critically, profile‑specific mediation showed that ES had mediated effects in value injury profile, whereas CR partially mediated effects in emotion injury profile.ConclusionTo mitigate negative emotions in medical staff, administrators should account for individual differences in MI profiles. For the emotion injury profile, prioritize CR to restructure self-blame. For value injury profile, focus on reducing ES and fostering healthy emotional expression. Tailoring emotion regulation strategies to specific MI profiles is essential to safeguard staff mental health and care quality. Introduction Moral injury (MI) among medical staff is defined as psychological and spiritual impairment arising from encounters with actions that transgress their deeply held moral beliefs-whether by committing, witnessing, or failing to prevent such actions [1, 2]. Empirical research has indicated that core manifestations of MI include intense feelings of guilt, shame, and anger [3–5]. Theoretical frameworks postulate that these psychological disturbances can be chronic, as impaired meaning-making, self-reproach, and loss of trust disrupt the processing and integration of emotions, thereby perpetuating psychological distress [6–8]. Although MI is not a formally diagnosed mental disorder, numerous studies have established a robust link between MI and adverse mental health outcomes such as post-traumatic stress disorder (PTSD), depression, and suicidal idea [3, 9, 10]. Distinct from other forms of psychological distress, MI is uniquely rooted in the transgression of personal moral tenets, which often co-occurs with psychological commodities and functional impairments, rendering it particularly persistent and refractory to intervention [11]. As front-line professionals tasked with the sacred duty of saving lives and relieving suffering, medical staff frequently encounter intricate ethical dilemmas in daily clinical practice. These scenarios include making difficult triage decisions amid scarce medical resources, allocating critical care resources such as ventilators and beds, witnessing unavoidable patient mortality despite exhaustive clinical efforts, navigating conflicts between professional duties and personal moral principles, and even being compelled to implement medical interventions that conflict with their core values. Collectively, these experiences render medical staff highly susceptible to MI. Studies have demonstrated that the prevalence of MI among medical staff remains consistently high, particularly amid public health emergencies (e.g., the COVID-19 pandemic) and high-pressure clinical settings characterized by prolonged working hours and heavy workloads [12, 13]. MI exerts far-reaching consequences, manifesting in both psychological commodities and functional impairments. Psychologically, it frequently coexists with depression, anxiety, PTSD, and professional burnout, reflecting profound emotional distress and guilt stemming from morally challenging experiences. Beyond individual psychological symptoms, MI also imposes significant functional impacts [14], including diminished job satisfaction, impaired teamwork, and compromised professional performance and collaboration. Additionally, it can strain personal relationships due to emotional withdrawal or increased irritability in affected individuals [12, 15]. Organizational factors such as inadequate resource allocation and poor leadership can exacerbate these adverse effects, further straining both the personal and professional lives of medical staff [16]. In summary, MI exerts a multifaceted impact on medical staff, influencing not only their mental health but also interpersonal dynamics and professional competence. Thus, investigating the characteristics and impacts of MI among medical staff is of great significance for enhancing their mental health and fostering the stable development of the medical industry. Negative emotion (NE) refers to a spectrum of unpleasant affective states elicited by adverse life events or interpersonal interactions, with anxiety and depression being the most common manifestations [17]. Medical staff regularly encounter a variety of potentially morally injurious events (PMIEs), such as the helpless observation of patient death [18, 19] and treatment delays [20]. These events induce varying degrees of MI [21], making medical staff particularly prone to experiencing a range of negative emotional states [22–24]. Anxiety is often accompanied by excessive worry regarding work performance, patient safety, and personal health, which may lead to restlessness, impaired concentration, and sleep disturbances. Depression, on the other hand, is characterized by persistent low mood, diminished interest in work and daily activities, and feelings of helplessness and hopelessness [25]. These NE directly compromised the physical and mental health of medical staff, leading to symptoms such as depressed mood, anhedonia, guilt, feelings of worthlessness, sleep and appetite disturbances, fatigue, and poor concentration [26]. In turn, these impairments further exacerbate the burden of clinical work, reduce work efficiency, and hinder the sustainable development of the medical industry [26, 27]. Given the high prevalence and severe adverse impacts of NE among medical staff, exploring its influencing factors and underlying mechanisms is imperative for clinical practice and healthcare management. In recent years, a growing body of empirical evidence has substantiated the association between MI and NE in professional populations. For instance, front-line health and social care workers during crises are often forced to make morally challenging decisions with limited knowledge and resources; in one such study, 33.6% of participants met the diagnostic criteria for major depressive disorder, 21.5% for generalized anxiety disorder, and 19.1% reported comorbidity depression and anxiety [28]. Conceptually, PMIEs have been posited to profoundly undermine the moral frameworks of veterans, challenging their fundamental perceptions of right and wrong [29]. Veterans who fail to reconcile the psychological discomfort caused by their actions are at an increased risk of social condemnation or rejection, which may subsequently lead to depressive symptoms and other adverse emotional experiences [30]. However, the majority of existing studies adopt a variable-centered approach, which focuses on the overall correlation between the average levels of MI and NE while neglecting the potential quantitative differences of MI among medical staff [31, 32]. In other words, medical staff may exhibit distinct MI profiles with varying severity and symptom characteristics, such as low, moderate, and high MI subgroups. The nature of the relationship between MI and NE may differ substantially across these subgroups-a nuance that cannot be captured by traditional variable-centered analytical approaches. Emotion regulation (ER) is defined as the conscious or unconscious cognitive and behavioral processes through which individuals modulate the intensity, duration, and expression of their emotional experiences to adapt to external environmental demands and internal psychological needs [33, 34]. It serves as a critical psychological resource for individuals to cope with NE and psychological distress [35]. Two well-documented ER strategies in the literature are cognitive reappraisal (CR) and expressive suppression (ES), which exert distinct effects on emotional states [36] and are closely associated with NE [37–39]. CR is an adaptive and proactive ER strategy that involves reinterpreting the cognitive meaning of emotional events to alter their emotional impact; compared with ES, it is consistently associated with fewer negative emotional experiences and more favorable mental health outcomes [40]. Meta-analytic evidence further indicates a moderate negative correlation between CR and symptoms of anxiety and depression, suggesting that frequent use of CR can effectively reduce the risk of these affective disorders [41]. In contrast, ES is a maladaptive ER strategy characterized by the conscious inhibition of emotional expression without modifying the underlying emotional experience-it is often linked to increased negative emotional experiences, heightened psychological distress, and even the development of chronic mental health problems [42], as it blocks the healthy release of negative feelings and leads to rumination that amplifies NE. Furthermore, the association between ER and MI has been widely validated in previous research [43]. MI arises when individuals (e.g., medical staff) perceive a violation of their core moral values (e.g., facing ethical dilemmas or feeling powerless to provide optimal care), inducing intense psychological distress that depletes the cognitive resources required for effective ER. This impairment disrupts individuals’ ability to use adaptive CR (which demands substantial cognitive effort for event reinterpretation) and promotes over-reliance on maladaptive ES (a more automatic, less cognitively demanding response), leading to the maladaptive use of ER strategies (e.g., over-reliance on ES instead of CR), which in turn exacerbates both MI symptoms and NE. Conversely, effective ER can help individuals mitigate the psychological impact of MI, alleviate associated distress, and reduce the likelihood of developing NE. Based on these mechanisms-the differential effects of CR/ES on NE and the impairment of ER processes by MI, we hypothesize that ER (including its two core sub-components, CR and ES) plays a mediating role in the relationship between MI and NE (i.e., anxiety and depression) among medical staff. Latent profile analysis (LPA) is an individual-centered analytical method that identifies distinct subgroups of individuals based on patterns of variation across multiple indicator variables. It facilitates the identification of high-risk populations and the development of targeted intervention strategies [44]. To date, few studies have systematically explored the latent profiles of MI and their differential associations with negative emotional outcomes; existing research in this area has primarily focused on populations such as police officers and military veterans with PTSD [45]. Additionally, the underlying psychological mechanism linking MI to NE from the perspective of ER remains poorly understood. This research gap restricts our comprehensive understanding of the psychological pathways of MI and impedes the development of targeted and effective intervention measures for affected medical staff. To the best of our knowledge, the present study is the first to systematically examine the mediating role of ER in the relationship between MI and NE among medical staff, while simultaneously accounting for both the distinct latent profiles of MI and the differential effects of specific ER strategies (CR and ES). This study addresses the aforementioned research gaps by adopting a person-centered approach to identify latent MI profiles in a sample of medical staff and exploring the mediating effects of different ER strategies on the relationship between MI profiles and NE. This research design not only helps to elucidate the complex relationship between MI and NE but also provides a theoretical foundation for developing targeted mental health interventions for medical staff with different MI profiles. Based on the aforementioned empirical findings and theoretical analyses, we propose the following three research hypotheses: H1: There are correlations among MI, anxiety, depression, and ER in medical staff.H2: The quantitative differences of MI among medical staff can be identified through LPA.H3: ER (i.e., CR and ES) plays a mediating role in the relationship between MI and NE (i.e., anxiety and depression). H1: There are correlations among MI, anxiety, depression, and ER in medical staff. H1: There are correlations among MI, anxiety, depression, and ER in medical staff. H2: The quantitative differences of MI among medical staff can be identified through LPA. H2: The quantitative differences of MI among medical staff can be identified through LPA. H3: ER (i.e., CR and ES) plays a mediating role in the relationship between MI and NE (i.e., anxiety and depression). H3: ER (i.e., CR and ES) plays a mediating role in the relationship between MI and NE (i.e., anxiety and depression). The conceptual model of this study, which illustrates the hypothesized relationships between MI, ER (CR and ES), and NE (anxiety and depression), is shown in the Fig. 1. Fig. 1The conceptual model. Note: MI is the independent variable, CR and ES are the mediating variables, and NE (anxiety and depression) are the dependent variables The conceptual model. Note: MI is the independent variable, CR and ES are the mediating variables, and NE (anxiety and depression) are the dependent variables Methods Participants Survey data were collected via Wenjuanxing (www.wjx.cn) from April 10 to 21, 2025, among 1,425 healthcare practitioners from three top-tier Grade III Class A general hospitals. Questionnaires were distributed and notified to eligible medical staff via department heads, who served only as messengers for notification and delivery without the authority to screen or select participants; all respondents completed questionnaires anonymously and voluntarily online, with administrators unable to access any individual responses. Only participants with written informed consent were included; 424 questionnaires were excluded after data cleaning (31 for incomplete/inaccurate demographics, 393 for failing two attention-check questions). 1,001 valid questionnaires were recovered, yielding an effective recovery rate of 70.25%. Measures General information questionnaire The self-designed questionnaire included demographic characteristics such as gender (male = 1, female = 2), identity (doctor = 1, nurse = 2), age (continuous variable), working years (continuous variable), marital status (married = 1, single or divorced = 2), educational level (technical secondary school = 1, college = 2, undergraduate = 3, postgraduate = 4), title (none or Junior = 1, intermediate = 2, senior = 3). The chinese version of moral injury symptom scale The Moral Injury Symptom Scale for Health Professionals (MISS-HP) is a specialized tool for assessing MI symptoms in medical practitioners, with 10 evaluation dimensions (e.g., betrayal, guilt, shame, loss of trust) [46]. It consists of 10 items rated on a 10-point scale, with total scores ranging from 10 to 100 (higher scores indicate more severe MI symptoms) [47]. Wang et al. translated it into Chinese, which showed acceptable reliability and validity among mainland Chinese healthcare professionals (Cronbach’s α = 0.70) [48]. In this study, the MISS-HP also had acceptable internal consistency (Cronbach’s α = 0.74). Patient health questionnaire-9 (PHQ-9) and generalized anxiety disorder scale (GAD-7) Depressive symptoms were measured via the 9-item PHQ-9 [49]. Its items cover typical depressive symptoms, scored 0–3, with total scores ranging 0–27 and higher scores reflecting more severe depression. The PHQ-9 has sound psychometric properties for Chinese medical workers with a reported Cronbach’s α of 0.820 [50], and its α value was 0.88 in this study, showing good internal consistency. Anxiety symptoms were assessed using the 7-item GAD-7 [51]. Each item is rated 0–3, with total scores from 0 to 21 positively linked to anxiety severity. The GAD-7 is reliable and valid for anxiety screening among Chinese medical staff [52], and its Cronbach’s α was 0.91 in the present research. Emotion Regulation Questionnaire (ERQ) Originally developed by Gross & John [53], the ERQ was revised for Chinese samples [54]. It includes 10 items divided into 6-item CR and 4-item ES, rated on a 7-point scale with higher scores indicating more frequent emotion regulation use. Validated for Chinese medical workers, its reported Cronbach’s α were 0.827 (CR) and 0.714 (ES) [55]. In this study, the Cronbach’s α was 0.86 for CR and 0.78 for ES, showing favorable internal consistency. Data Analysis All statistical analyses were performed with SPSS (version 26.0), and Mplus (version 8.3) software. First, descriptive statistics were carried out to outline the demographic and occupational attributes of the study sample, and Pearson’s correlation analysis was employed to explore the pairwise correlations among MI, ER and its sub-scales (CR, ES), as well as NE (anxiety, depression). Second, LPA was utilized to identify distinct latent subgroups of medical staff according to their MI symptom scores, with sequential testing of 1 to 4-category profile models for this analysis. Model fit was assessed through a set of metrics: lower values of the Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC) and sample-adjusted Bayesian Information Criterion (aBIC) represented better model fit; the entropy value was used to measure classification accuracy; the Lo-Mendell-Rubin (LMR) test was adopted to compare the fit of k and k-1 category models, where a significant p-value indicated the k-category model was more optimal. Finally, mediation analysis with the Bootstrap approach was conducted to examine the mediating effects of ER and its two sub-scales (CR, ES) on the association between MI latent profile membership and NE (anxiety, depression), with the low MI subgroup designated as the reference group. The statistical significance of all mediating effects was judged by the 95% Bootstrap confidence interval, with an effect deemed significant if zero was not included in the interval. Results Sample Demographic Characteristics The final analytical sample comprised 439 physicians (240 females) and 562 nurses (527 females), with participants’ ages ranging from 19 to 75 years (M = 34.86, SD = 7.94) and professional working experience spanning 0 to 52 years (M = 12.11, SD = 8.42). Detailed descriptive statistics for the study sample are presented in Table 1. Table 1Demographic characteristics (N = 1001, %)Variables N PercentageVariables N PercentageGenderMale23423.38Marital StatusMarried65165.03Female76776.62Single or divorced35034.97IdentityDoctor43943.86Educational backgroundTechnical secondary school90.90Nurse56256.14Associate degree26926.87Age(years)≤ 3032432.37Bachelor’s degree60160.0331–5064464.34Postgraduate12212.20≥ 51333.30Professional TitleNone or Junior51851.75Working Years≤ 1049149.05Intermediate44944.8511–2033633.57Senior343.40≥ 2117417.38 Table 1Demographic characteristics (N = 1001, %)Variables N PercentageVariables N PercentageGenderMale23423.38Marital StatusMarried65165.03Female76776.62Single or divorced35034.97IdentityDoctor43943.86Educational backgroundTechnical secondary school90.90Nurse56256.14Associate degree26926.87Age(years)≤ 3032432.37Bachelor’s degree60160.0331–5064464.34Postgraduate12212.20≥ 51333.30Professional TitleNone or Junior51851.75Working Years≤ 1049149.05Intermediate44944.8511–2033633.57Senior343.40≥ 2117417.38 Demographic characteristics (N = 1001, %) Correlation analysis The correlation analysis revealed that MI was significantly positively correlated with depression (r = 0.38, 95% CI [0.32, 0.44], P < 0.01) and anxiety (r = 0.41, 95% CI [0.36, 0.47], P < 0.01), but significantly negatively correlated with CR (r = -0.19, 95% CI [-0.26, -0.13], P < 0.01). Detailed results are shown in Table 2. Table 2The level and association of medical staff’s MI with ER and NEVariablesMeanSD123451.MI26.4813.4012.CR31.617.07− 0.19** [-0.26, − 0.13]13.ES15.845.410.02 [-0.05, 0.08]0.45** [0.39, 0.51]14.Depression3.033.560.38** [0.32, 0.44]− 0.15**[-0.23, − 0.09]0.06 [-0.01, 0.12]15.Anxiety1.842.970.41** [0.36, 0.47]− 0.13** [-0.20, − 0.06]0.08* [0.02, 0.14]0.81** [0.77, 0.85]1*P < 0.05, **P < 0.01 Table 2The level and association of medical staff’s MI with ER and NEVariablesMeanSD123451.MI26.4813.4012.CR31.617.07− 0.19** [-0.26, − 0.13]13.ES15.845.410.02 [-0.05, 0.08]0.45** [0.39, 0.51]14.Depression3.033.560.38** [0.32, 0.44]− 0.15**[-0.23, − 0.09]0.06 [-0.01, 0.12]15.Anxiety1.842.970.41** [0.36, 0.47]− 0.13** [-0.20, − 0.06]0.08* [0.02, 0.14]0.81** [0.77, 0.85]1*P < 0.05, **P < 0.01 The level and association of medical staff’s MI with ER and NE *P < 0.05, **P < 0.01 LPA of MI In this study, LPA was employed to explore the latent classes of MI. Models with 1 to 4 profiles were sequentially fitted, and multiple statistical indicators were integrated to determine the optimal number of profiles. The information criteria (AIC, BIC, aBIC) decreased consistently as the number of profiles increased: the largest reduction occurred from 1 to 2 profiles, followed by a moderate decrease from 2 to 3 profiles, while the decline from 3 to 4 profiles was slightly smaller than in the previous two stages. Model interpret-ability and clinical utility were prioritized as the primary principles. The additional class in the 4-profile model accounted for only 5.69% of the sample, falling below the 8% threshold for clinical acceptability [56]. Moreover, its pattern of scores closely resembled that of an existing class, differing only quantitatively rather than qualitatively. No distinct subgroup with clear theoretical or clinical significance was identified, indicating a risk of overfitting. Taken together, considering multiple statistical indices, class proportions, theoretical interpretability, and parsimony, the 3-profile model was ultimately selected as the most robust and clinically meaningful solution for the MI data. Detailed results of LPA are presented in Table 3. Table 3Fitting index and group size of LPA modelsIndices1-profile2-profile3-profile4-profileLL-22449.645-21340.543-20718.722-20172.192AIC44939.28942743.08541521.44440450.384BIC45037.46442895.25641727.61140710.548aBIC44973.94342796.79941594.21740542.217Entropy-0.9590.9530.961BLRT-< 0.001< 0.001< 0.001LMR-< 0.0010.05830.0109Proportion C1100%72.53%62.84%61.74% C2-27.47%11.89%10.49% C3--25.27%22.08% C4---5.69% Table 3Fitting index and group size of LPA modelsIndices1-profile2-profile3-profile4-profileLL-22449.645-21340.543-20718.722-20172.192AIC44939.28942743.08541521.44440450.384BIC45037.46442895.25641727.61140710.548aBIC44973.94342796.79941594.21740542.217Entropy-0.9590.9530.961BLRT-< 0.001< 0.001< 0.001LMR-< 0.0010.05830.0109Proportion C1100%72.53%62.84%61.74% C2-27.47%11.89%10.49% C3--25.27%22.08% C4---5.69% Fitting index and group size of LPA models LPA identified three heterogeneous subgroups (Fig. 2). Profile 1 (62.84%) was labeled low injury profile, characterized by consistently low scores on all negative MI items. Profile 2 (11.89%) represented value injury profile, exhibiting the most severe MI, including loss of professional meaning, self-denial, and fatalistic attribution, while retaining some capacity for trust and belief restoration. Profile 3 (25.27%) was defined as emotion injury profile, marked by intense guilt and self-blame over professional failures, without generalized devaluation of professional identity. Fig. 2Mean scores characteristics for the three-profile of MI Mean scores characteristics for the three-profile of MI Meditation Analysis of ER Between MI Profiles and NE Considering potential confounding effects, demographic variables including gender, identity, age, working years, marital status, education and professional title were set as covariates in mediation analysis. A parallel mediation model was established, taking MI as the independent variable, depression and anxiety as dependent variables, and CR and ES as mediators (Fig. 3). Fig. 3The mediating effect of ER (including its two sub-scales) on MI. Note: **P < 0.01; ***P < 0.001. Dashed line indicates no statistically significant difference, P ≥ 0.05 The mediating effect of ER (including its two sub-scales) on MI. Note: **P < 0.01; ***P < 0.001. Dashed line indicates no statistically significant difference, P ≥ 0.05 In the depression pathway, the total effect of MI was 0.373. Its direct effect was 0.342 (91.7% of the total effect), and the total indirect effect was 0.031 (8.3%, 95% CI [4.2%, 13.6%]), among which the indirect effect via CR was 0.028 (7.5% of the total effect). In the anxiety pathway, the standardized total effect of MI was 0.413. The direct effect was 0.389 (94.3% of the total effect), and the total indirect effect was 0.024 (5.7%, 95% CI [2.2%, 10.1%]), with the CR-mediated indirect effect being 0.021 (5.1% of the total effect). The indirect effect through ES was 0.003, showing no statistical significance for both outcomes. Path analysis was performed based on LPA, with Profile 1 (low injury profile) as the reference group to explore the impacts of Profile 2 (value injury profile) and Profile 3 (emotion injury profile) on depression and anxiety (Table 4). Consistent results were found for both indicators. Relative to Profile 1, Profile 2 and Profile 3 significantly elevated depressive and anxious risks, with significant total and direct effects (95% CIs excluding zero), and both presented notable partial mediating effects. For Profile 2, the total indirect effects on depression and anxiety were 0.116 (95% CI: 0.040–0.227) and 0.095 (95% CI: 0.036–0.183), accounting for 9.8% and 7.1% of total effects respectively. Such indirect effects were solely mediated by ES, while the CR pathway was insignificant. For Profile 3, the total indirect effects on depression and anxiety were 0.296 (95% CI: 0.107–0.554) and 0.184 (95% CI: 0.034–0.381), with mediated proportions of 8.3% and 5.9%. Differently, its indirect effects were only mediated via CR, whereas the ES pathway showed no statistical significance. Table 4The total, direct, and indirect effects of ER (categorical variable) on NEPath / Effect (95%CI)1 vs. 21 vs. 3Depression Total effect1.188 (0.700, 1.706)3.549 (2.665, 4.602) Direct effect1.072 (0.586, 1.569)3.253 (2.379, 4.196) Total indirect effect0.116 (0.040, 0.227)0.296 (0.107, 0.554) → via CR0.006 (-0.073, 0.084)0.383 (0.192, 0.678) → via ES0.110 (0.038, 0.219)-0.088 (-0.240, 0.010) Proportion mediated9.8% (3.4%, 21.3%)8.3% (3.1%, 14.9%)Anxiety Total effect1.343 (0.940, 1.792)3.137 (2.312, 4.043) Direct effect1.247 (0.846, 1.671)2.953 (2.182, 3.828) Total indirect effect0.095 (0.036, 0.183)0.184 (0.034, 0.381) → via CR0.004 (-0.049, 0.058)0.257 (0.111, 0.471) → via ES0.091 (0.031, 0.176)-0.073 (-0.193, 0.008) Proportion mediated7.1% (2.8%, 14.2%)5.9% (1.1%, 11.6%) Table 4The total, direct, and indirect effects of ER (categorical variable) on NEPath / Effect (95%CI)1 vs. 21 vs. 3Depression Total effect1.188 (0.700, 1.706)3.549 (2.665, 4.602) Direct effect1.072 (0.586, 1.569)3.253 (2.379, 4.196) Total indirect effect0.116 (0.040, 0.227)0.296 (0.107, 0.554) → via CR0.006 (-0.073, 0.084)0.383 (0.192, 0.678) → via ES0.110 (0.038, 0.219)-0.088 (-0.240, 0.010) Proportion mediated9.8% (3.4%, 21.3%)8.3% (3.1%, 14.9%)Anxiety Total effect1.343 (0.940, 1.792)3.137 (2.312, 4.043) Direct effect1.247 (0.846, 1.671)2.953 (2.182, 3.828) Total indirect effect0.095 (0.036, 0.183)0.184 (0.034, 0.381) → via CR0.004 (-0.049, 0.058)0.257 (0.111, 0.471) → via ES0.091 (0.031, 0.176)-0.073 (-0.193, 0.008) Proportion mediated7.1% (2.8%, 14.2%)5.9% (1.1%, 11.6%) The total, direct, and indirect effects of ER (categorical variable) on NE Discussion The present study systematically explored the relationships among MI, ER, and NE in medical staff. The results confirmed all three research hypotheses, revealing significant correlations among the core variables, distinct latent subgroups of MI (low, value, and emotion injury profile) with specific proportional characteristics, and qualitatively different mediating effects of CR and ES across subgroups-with ES mediating the value injury profile and CR mediating emotion injury profile. These findings deepen the understanding of the psychological mechanisms underlying MI-induced NE in medical staff and provide a targeted, subgroup-specific theoretical basis for clinical psychological intervention and human resource management in the medical field. Associations among MI, anxiety, depression, and ER The results of correlation analysis in this study supported Hypothesis 1, with the results as follows: MI was significantly positively correlated with both depression and anxiety, and significantly negatively correlated with CR. In contrast, ES had no significant association with MI, only a weak positive correlation with anxiety, and no significant correlation with depression, which is partially consistent with the conclusions of existing empirical studies on medical staff [57]. The moderate positive correlation between MI and anxiety/depression indicates that medical staff with higher MI levels experience more severe negative emotional symptoms. The core reason maybe that morally injurious events violate the core moral convictions of medical staff to save lives and relieve pain, triggering intense guilt, shame, and self-blame [2, 5], which in turn disrupt emotional regulation homeostasis and ultimately manifest as clinical symptoms of anxiety and depression [28, 58]. This result is consistent with previous studies on healthcare workers during the COVID-19 pandemic and emergency front-line workers [59, 60], highlighting the unique role of MI as an important risk factor for negative emotional symptoms in medical staff-medical work is closely related to life and well-being, and the moral conflicts caused by MI are more intense, which also explains the moderate correlation observed in this study. The differential associations of the two sub-dimensions of emotional regulation are of great significance: CR was significantly negatively correlated with both MI and NE, confirming its adaptive role [61, 62]. Specifically, medical staff who effectively use CR may alleviate NE and reduce the risk of anxiety and depression by reinterpreting morally injurious events (e.g., attributing clinical failures to systemic limitations rather than personal moral deficiencies). However, the results that ES had no significant association with MI and only a weak correlation with NE deviated from the classic findings of Gross & John in the general population [53]. This deviation can be explained by the professional characteristics of medical staff: in clinical work, medical staff are required to maintain professional calm in scenarios such as patient death and ethical dilemmas, and ES may be regarded as a necessary professional skill rather than a maladaptive strategy, thereby weakening its negative effects. In summary, the core results of this study provide empirical evidence for MI-related psychological interventions in medical staff: improving CR skills may be more effective in alleviating MI-related NE, while interventions targeting ES should rationally view its role in combination with the professional context. Future longitudinal studies are needed to further clarify the causal relationships among various variables. Three latent subgroups of MI in medical staff via LPA The present study results are in line with Hypothesis 2. From a person-centered perspective, LPA was adopted to categorize medical staff into three latent subgroups based on MI, namely the low injury profile (62.84%, n = 629), emotion injury profile (25.27%, n = 253), and value injury profile (11.89%, n = 119). The findings suggest obvious individual disparities in MI levels among medical staff, which is consistent with the conclusions of prior studies exploring occupational psychological characteristics via person-centered approaches [63]. Different from conventional variable-oriented studies that merely focus on the overall average level of MI [64], this study further indicates that medical staff tend to experience MI with varying severity degrees and diverse symptom combinations in clinical settings. Notably, MI severity among participants fails to show obvious bi-polarization, and participants with low MI constitute the dominant part of the whole sample. The distinct proportional distribution of MI subgroups may be jointly shaped by individual and organizational factors. On the individual level, medical staff could receive systematic professional ethics education and clinical skill training throughout their career development [64, 65], which may help most of them build adequate psychological resilience to adjust their mental state and cope with common clinical moral injury [66]. On the organizational perspective, sound resource allocation mechanisms and ethical consultation services in medical institutions may help reduce the occurrence of severe MI events, such as unavoidable patient death caused by extreme medical resource shortages. In clinical practice, most medical staff are mainly exposed to relatively mild stressors, including slight delays in treatment implementation, temporary resource shortages, and minor conflicts between professional responsibilities and personal morality, which may partly explain why individuals with low MI account for the largest proportion [67]. In contrast, members of the emotion injury profile are prone to generate strong guilty feelings and self-blame over certain professional mistakes, while they seldom form negative evaluations of their overall professional identity. As the smallest subgroup, the value injury profile manifests diversified and profound psychological manifestations including weakened professional purpose, self-deprecation and fatalistic attribution, which are mainly triggered by rare high-impact events such as extreme medical resource shortage and exposure to colleagues’ unethical conducts [68], though they still have potential to rebuild inner trust and positive beliefs. The subgroup distribution of MI among medical staff appears to differ considerably from that of other high-risk occupational groups. Relevant evidence has shown that the proportion of individuals affected by moral injury reaches 33.18% among military veterans and trauma-exposed police officers [45], whereas the proportion of participants in the value injury profile is only 11.89% in this study. When combining the emotion injury profile and value injury profile, this proportion rises to 37.16%. Although this proportion is close to that observed among military veterans, there exist apparent differences in their internal psychological features. Such discrepancies may be partially attributed to the accessible regular psychological training and relatively complete clinical support systems available for medical staff [69]. By contrast, military veterans and police officers are more likely to encounter extreme traumatic events and have less access to sustained on-the-job psychological assistance. Cross-occupational comparative analysis can help clarify the unique characteristics of MI among medical staff, and also enrich person-centered research on MI across different professional populations. MI subgroup distribution provides empirical evidence for hierarchical mental health management of medical staff. Dominant low-MI individuals can be taken as the benchmark for routine hospital psychological screening. Though free from acute NE, they need regular psychological monitoring due to sustained work stress, which needs further longitudinal verification. The emotion injury profile should be prioritized for targeted cognitive interventions. This subgroup scores significantly higher on guilt, shame, and self-condemnation items, driven primarily by excessive self-attribution. Interventions should focus on correcting irrational attribution styles and guiding balanced perspectives on occupational setbacks to reduce remorse. In contrast, the value injury profile shows elevated scores on loss of meaning, trust, and faith, reflecting profound disruption to core beliefs. This subgroup requires comprehensive, multi-dimensional psychological interventions to address these deeper identity-related wounds. This study may partly make up for the deficiency that traditional variable-centered studies tend to ignore individual heterogeneity of moral injury [70, 71]. From the person-centered perspective, it preliminarily explores the heterogeneous associations between MI and NE, and offers some references for targeted psychological interventions among different subgroups. Further longitudinal research is still required to confirm the stability of these latent subgroups. Mediation results of ER between MI and NE Supported by this study’s data, the distinct mediating effects of two ER strategies on the association between MI and NE were clarified. In the overall sample, only CR exerted a significant protective indirect effect (accounting for 7.5% of the total effect), while the indirect effect of ES was not statistically significant. Latent profile-based path analysis further confirmed obvious subgroup specificity in these parallel mediating pathways, which was also supported by the data. The non-significant overall mediating effect of ES can be explained by the essential attributes of the two emotion regulation strategies, based on data support and reasonable speculation. CR reshapes cognitive appraisal before emotional arousal, enabling medical staff to attribute moral injury to institutional or situational factors rather than personal faults, thereby alleviating guilt and shame [72]. In contrast, long-term use of ES consumes cognitive resources and exacerbates intrusive negative thoughts [73]. We speculate that medical staff rarely use ES method to cope with MI, or its adverse impacts are offset in the overall sample, though this overall effect conceals significant subgroup heterogeneity confirmed by the data. Study data showed the protective mediating effect of CR was only significant in the emotion injury profile (25.27% of participants; Table 4), where only CR-mediated pathways were significant. Combined with the negative correlation between MI and CR in the total sample, data confirmed higher MI was associated with lower adaptive CR use, and this deficiency mediated the MI-NE relationship. We speculate individuals in this subgroup experience intense guilt from specific professional mistakes (without denying professional identity), leading to biased CR. CR’s efficacy depends on reappraisal content, and inappropriate application of CR may also worsen NE [74]. Accordingly, interventions targeting this subgroup may focus on adjusting CR content rather than simply increasing its application, which remains to be surveyed in further longitudinal studies. According to study data, the maladaptive mediating effect of ES was only significant in the value injury profile (11.89% of participants; Table 4), where only ES-mediated pathways were significant (95% CI for CR-mediated pathways included zero). We speculate repeated severe MI exhausts their psychological resources, leading to loss of professional meaning, self-denial, fatalistic attribution, and learned helplessness, with emotional suppression as the default coping method. The medical industry’s emphasis on professional rationality may further promote ES use. Long-term ES causes emotional numbness or outbursts, exacerbating anxiety and depression; data confirmed ES is a key mechanistic pathway linking to MI, consistent with previous study [75]. Integrating mediation analysis results, MI subgroup characteristics, and physician-nurse professional differences, this study constructed a three-level hierarchical psychological intervention framework based on existing data; its effectiveness requires longitudinal validation. Universal prevention: Regular mental health education and basic coping training are required for individuals with low MI and adequate coping resources [76]. Selective intervention: For the emotion injury profile, use cognitive restructuring as the core to help abandon self-punitive CR and form rational reappraisal, accompanied by regular mental health screening. Indicated intensive intervention: For the value injury profile, reduce ES use and guide reasonable emotional expression to break the cycle of emotional distress (including anxiety and depression). Prioritizing the reduction of ES frequency and the establishment of adaptive emotional expression, combined with intervention approaches such as cognitive behavioral therapy and mindfulness training, is essential for synergistic interventions [77, 78]. Limitations This study has several inherent limitations that should be acknowledged. First, the cross-sectional design only permits correlational and mediational analyses, without establishing causal or temporal relationships. Longitudinal studies are therefore needed to verify the causal pathways and stability of the model. Second, only two classic emotion regulation strategies were included, while other adaptive strategies were not examined. Future studies may explore more ER strategies to clarify their roles in the MI-NE relationship. Third, all variables in this study were assessed exclusively via self-report scales, which are prone to self-report bias and social desirability bias. Meanwhile, the lack of objective clinical diagnostic instruments and structured interviews may reduce the accuracy of symptom identification. Future studies should adopt multi-source assessments and integrate clinical diagnostic tools to improve measurement validity and assessment precision. Fourth, this study used convenience sampling, which may restrict sample representativeness and generalizability. Future studies are recommended to employ random sampling to improve representativeness. Fifth, despite using the culturally validated Chinese MI scale, its core theories and original items were developed in Western contexts. Cultural disparities may restrict cross-cultural generalizability of our findings. Further cross-cultural research may be needed to refine culture-specific MI assessment tools. Conclusion This study explored the links between MI, ER (CR and ES) and NE (anxiety and depression) via LPA and mediation analysis. Three latent MI subgroups were identified, confirming the quantitative differences of MI in this population, and the study further clarified the significant correlations among the core variables as well as the differential mediating effects of CR and ES in the MI-NE association. The findings enrich person-centered research on MI in medical staff and expand the application of the ER process model in this field by verifying the distinct mediating roles of CR and ES. They confirm MI as a critical risk factor for medical staff’s NE and ER strategy selection as a core coping element, providing targeted guidance for their mental health intervention and medical human resource management. They confirm MI as a critical risk factor for medical staff’s NE and ER strategy selection as a relatively useful coping element, which may provide guidance for their mental health intervention and medical human resource management.