Mental Health, Coping Strategies, and Work Engagement: Interrelationships Among Brazilian Military Police Officers

Abstract

Background: Military police officers are exposed to occupational stressors associated with mental health, coping strategies, and work engagement. This study examined mental health indicators and their associations with coping strategies and work engagement among military police officers in the pre-pandemic period. Methods: A quantitative, cross-sectional, descriptive, and correlational study was conducted in 2018 with 773 Brazilian military police officers from São Paulo (n = 506) and Paraná (n = 267). Participants completed the Work Stress Scale (WSS), Maslach Burnout Inventory (MBI-HSS), Utrecht Work Engagement Scale (UWES), and the Scale of Problem Coping Modes (EMEP). Results: The prevalence of occupational stress was 30.2%, with high proportions of Emotional Exhaustion and Depersonalization. Burnout was interpreted dimensionally (MBI-HSS subscales), with 17.6% (n = 134) joint prevalence of the high Emotional Exhaustion + high Depersonalization + low Personal Accomplishment profile, alongside frequent mixed profiles (e.g., 38.3% with high Depersonalization + low Personal Accomplishment). In the multivariable model, the 6 h shift was associated with higher odds of stress (OR = 7.76; 95% CI: 1.02–58.79), while the absence of self-reported health/quality-of-life issues was associated with lower odds (OR = 0.60; 95% CI: 0.39–0.94), along with Emotional Exhaustion (OR = 1.15; 95% CI: 1.10–1.20) and Depersonalization (OR = 1.12; 95% CI: 1.04–1.20). In sensitivity analysis, work shift was not associated with stress (aOR = 1.20; 95% CI: 0.66–2.21). Stress and burnout dimensions were negatively correlated with work engagement (r = −0.52), problem-focused coping, and social support and positively correlated with emotion-focused coping. São Paulo officers reported higher engagement and greater use of problem-focused coping and social support, whereas those in Paraná reported greater reliance on emotion-focused coping. Conclusions: Stress and burnout dimensions may coexist with high engagement, supporting the need for integrated institutional strategies that address organizational stressors (e.g., workload schedules) and strengthen potentially protective coping repertoires, while accounting for contextual differences between units. The high prevalence of burnout profiles underscores the urgency of preventive interventions to mitigate syndromic manifestations in high-stress occupations.

1. Introduction

Military police work is widely recognized as a high-psychosocial-risk occupation, exposing professionals to intense and repeated stressors such as lethal violence, life-threatening situations, operational pressure, and frequent contact with potentially traumatic events [1,2,3,4,5]. Evidence suggests that these demands elicit acute physiological responses, even in controlled training environments [6], and are associated with relevant mental health and well-being outcomes. These outcomes, in turn, are linked to variations in occupational functioning, decision-making under pressure, and the quality of service provided to the population, making police mental health a strategic issue for public safety and collective health, especially in Brazil, where structural and social vulnerabilities can intensify such challenges [7,8].

Prolonged exposure to critical demands and events is frequently associated with significant psycho-emotional distress, including symptoms and conditions compatible with Post-Traumatic Stress Disorder (PTSD) [9,10]. In addition, anxiety, depression, somatic comorbidities, and burnout syndrome are recurrently mentioned in the literature as relevant problems in this occupational group [7,11,12,13,14]. The association between these exposures and health outcomes depends not only on the intensity of the stressors but also on a set of individual and organizational factors (e.g., job characteristics, available resources, organizational climate), which may be associated with how stress is experienced and reported [15]. Convergently, the perception of social support and the quality of the work environment are consistently associated with better or worse mental health indicators in police officers [7,16,17].

In occupational health, burnout, often operationalized as emotional exhaustion, depersonalization, and reduced personal accomplishment, has been associated with poorer quality of life and greater psychological distress in security professionals [13,14,16]. Conversely, work engagement, characterized by vigor, dedication, and absorption, is a positive state related to greater energy, involvement, and persistence at work [3,17,18,19]. Literature suggests engagement is associated with better indicators of well-being and greater availability of personal and organizational resources, such as emotional support and affective commitment, with potential relevance for resilience and performance [7,17,20].

While burnout and work engagement are often considered inversely related, contemporary perspectives treat them as distinct constructs that may partially overlap, especially in high-demand environments where workers can remain dedicated and energetic while simultaneously experiencing emotional exhaustion and cynicism [21]. Therefore, interpreting engagement solely as the absence of burnout may obscure mixed profiles that carry different implications for health and performance [22,23]. This theoretical framework is particularly relevant to policing, where sustained dedication and absorption may reflect professional identity and institutional expectations, while exhaustion and depersonalization may reflect cumulative strain and resource depletion, implying that engagement should not be treated as a standalone marker of psychosocial safety.

Furthermore, coping strategies are mobilized to deal with work demands and may be associated with different psychosocial outcomes [24,25]. Strategies focused on the problem and social support tend to be related to better adaptation in various occupational contexts, while strategies predominantly focused on emotion may be associated with greater psychological distress, depending on the type of stressor, chronicity, and available resources [24,25,26,27]. In this context, studies also indicate that higher levels of engagement may be associated with a better perception of quality of life, functioning as a psychosocial resource in the face of the pressures inherent to the profession [7,28].

Despite the topic’s relevance, a significant gap remains in understanding how mental health, coping, and engagement specifically interrelate among Brazilian military police officers, a population exposed to extreme levels of urban violence [10,13,24]. Although recent international evidence has explored patterns of occupational stress in police forces [29] and the link between work-related stress and psychological distress [30], these constructs remain largely investigated in isolation. Contemporary research has begun to propose more systemic models, such as the “exhaustion triangle,” which integrates psychosocial risks, engagement, and burnout [31], and has explored the complex coexistence of work engagement and cynicism in industrial settings [32]. However, in the context of high-risk public safety occupations in Brazil, such an integrative approach remains scarce.

Although prior Brazilian studies, including some from our group, have examined occupational stress, burnout, or engagement [7,13,14,24,28], they focused on partial combinations or isolated dimensions. This study advances current knowledge by integrating occupational stress (WSS), burnout dimensions (MBI-HSS), work engagement (UWES), and coping modes (EMEP) within a single analytic framework. Rather than assuming engagement is simply the opposite of burnout, we examine whether positive (engagement) and negative (stress and burnout symptoms) states can co-occur in high-risk occupations, testing a “mixed-profile” perspective. This approach is critical to avoid the “engagement trap”, the false assumption that high dedication necessarily implies low psychosocial risk.

Utilizing a large pre-pandemic sample (n = 773) from two distinct state contexts (São Paulo and Paraná), this study addresses a distinct research question regarding the coexistence of motivational resources and strain indicators. While this study utilizes a broader database from which previous partial analyses were derived [7,13,14,24,28], the current manuscript represents a distinct integrative analysis, modeling the full set of variables to address the theoretical question of psychological sustainability versus high-performance exhaustion. Thus, this study aims to analyze mental health indicators and their association with coping strategies and work engagement among military police officers, providing a baseline reference for institutional policies and practices.

Based on this framework, this study tests the following hypotheses: (H1) emotion-focused coping strategies are positively associated with occupational stress and burnout dimensions; and (H2) higher levels of emotional exhaustion and depersonalization, alongside poorer self-reported health, are positively associated with the occurrence of occupational stress. These hypotheses are addressed through both bivariate associations and multivariable logistic regression modeling, contributing to a more integrated understanding of how organizational stressors, burnout, engagement, and coping co-occur in this occupational context.

2. Materials and Methods

2.1. Type and Location of the Study

A quantitative, cross-sectional study with a descriptive and correlational design was conducted with military police officers from the Interior Policing Command—5th Region of the State of São Paulo (CPI-5/SP) and the 3rd Military Police Battalion of the State of Paraná (3rd BPM/PR), Brazil.

The CPI-5/SP, headquartered in São José do Rio Preto, covers 96 municipalities and comprises approximately 2200 police officers distributed across four battalions. The 3rd BPM/PR, headquartered in Pato Branco, serves 16 municipalities with approximately 312 police officers distributed across three companies.

Site selection was based on operational feasibility and institutional access for data collection, subject to the requisite authorizations. The study sought to examine the interrelationships among occupational stress, burnout, coping strategies, and work engagement in two different operational contexts. The results should be interpreted as representative of the study participants and not as generalized estimates for the entire Brazilian military police population.

2.2. Population and Sample

The target population consisted of 2512 military police officers (2200 assigned to CPI-5/SP and 312 to the 3rd BPM/PR). The inclusion criteria were: (i) being an active-duty military police officer, (ii) being assigned to one of the participating units, and (iii) being on duty during the data collection period. Police officers on vacation, medical or maternity leave, or other leaves of absence that precluded participation during the collection period were excluded to ensure the sample characterized personnel in regular activity.

The minimum sample size was estimated separately for each organization, considering a margin of error (e) of 5% and a confidence level (z) of 95%, using the formula for estimating a population proportion:

n = N⋅z²⋅p⋅(1−p)
e²⋅(N−1) + z²⋅p⋅(1−p)

where n is the sample size, N is the population size, z is the z-score (1.96 for 95% confidence), e is the margin of error (0.05), and p is the population proportion (assumed to be 0.5 to maximize the sample size). To account for potential data loss (missing values), 10% was added to the estimate. Consequently, the estimated minimum sample size was 554 military police officers (361 from CPI-5/SP and 193 from the 3rd BPM/PR).

The non-probabilistic convenience sample comprised 773 military police officers who volunteered to participate after an in-person invitation, with institutional support ensuring logistical feasibility. The invitation emphasized voluntary participation, response anonymity, and the right to withdraw at any time without prejudice.

Among the 773 participants who completed the instruments in full, 506 (65.5%) belonged to CPI-5/SP and 267 (34.5%) to the 3rd BPM/PR, resulting in an overall response rate of approximately 30.8% (773/2512). The number of non-respondents was 1739 police officers (1694 in CPI-5/SP and 45 in the 3rd BPM/PR).

Although the final sample (n = 773) exceeded the estimated minimum (n = 554), the use of convenience sampling and the resulting response rate may introduce selection bias, limiting the generalizability of the findings. This bias may systematically affect the results; for instance, officers experiencing severe psychological distress might have been less likely to participate (a potential healthy-worker effect), or conversely, those highly motivated to report organizational issues might be overrepresented. Consequently, prevalence estimates and the magnitude of observed associations should be interpreted as context-specific patterns rather than universal parameters for the entire Brazilian military police population. Future studies should explore potential non-response biases and probability-based sampling strategies.

2.3. Data Collection Instruments

Validated instruments adapted for the Brazilian context were employed.

2.3.1. Sociodemographic and Professional Questionnaire

A questionnaire developed by the researchers was used to collect sociodemographic (age, gender, marital status, and education) and professional (unit/battalion, position, type of activity, workload, work shift, length of service) variables, as well as lifestyle aspects (physical activity, other paid activity) and self-reported history of disciplinary offenses, health problems, and quality of life (QoL). Self-reported health and quality of life issues were assessed through a single-item question: “Are there any problems currently affecting your quality of life?”. Responses were collected on a binary scale (Yes/No) referring to current issues.

2.3.2. Work Stress Scale (WSS)

The WSS, developed and validated by Paschoal and Tamayo [33], comprises 23 negative statements rated on a five-point Likert scale (“strongly disagree” to “strongly agree”). The scale assesses occupational stress with an emphasis on organizational factors and has satisfactory psychometric properties [33].

In this study, the internal consistency of the WSS was excellent (Cronbach’s alpha = 0.94). The occupational stress score was calculated as the average of the items, resulting in a composite indicator ranging from 1 to 5. As recommended by the authors, values ≥ 2.5 were considered indicative of relevant occupational stress [33], and this categorization was used to define the outcome in the logistic regression model.

The threshold used to define relevant occupational stress (≥2.5) follows prior recommendations for interpretability in organizational stress research [33]. However, it is important to acknowledge that this threshold represents a pragmatic simplification of a continuous construct, and prevalence estimates are inherently sensitive to the chosen cutoff. A lower threshold would increase sensitivity but decrease specificity, potentially overestimating the prevalence of clinical-level stress. Therefore, direct comparisons with studies using alternative operational definitions (e.g., different cut-offs or percentile-based classifications) should be made with caution, prioritizing the comparison of association patterns over absolute prevalence values.

2.3.3. Maslach Burnout Inventory—Human Services Survey (MBI-HSS)

The MBI-HSS, standardized in the MBI Manual [34], was translated and adapted into Portuguese by Benevides-Pereira [35], with psychometric properties investigated in Brazilian samples by Carlotto and Câmara [36]. It is a self-administered 22-item scale with responses on a five-points Likert scale (1 = never to 5 = every day), which assesses three dimensions: Emotional Exhaustion (EE), Depersonalization (DP), and Personal Accomplishment (PA) [36,37].

In the present study, internal consistency was adequate for EE (alpha = 0.91), DP (alpha = 0.75), and PA (alpha = 0.84). Subscale scores were calculated by summing the items for each dimension: EE (items 1, 2, 3, 6, 8, 13, 14, 16, and 20), DP (items 5, 10, 11, 15, and 22), and PA (items 4, 7, 9, 12, 17, 18, 19, and 21). Scores were classified as low, moderate, or high according to the parameters presented in

Table 1. Parameters for categorizing MBI-HSS scores.

Subscales	Burnout Levels
	Low	Moderate	High
Emotional Exhaustion	≤16	17–26	≥27
Depersonalization	≤6	7–12	≥13
Personal Accomplishment	≥39	38–32	≤31

[34].

For descriptive purposes and comparison with the literature, a profile of high scores in EE and DP and low scores in PA was considered indicative of a higher risk of burnout [14,37]. This operationalization was applied with interpretive caution, given the cross-sectional design.

2.3.4. Scale of Problem Coping Modes (EMEP)

The EMEP, adapted and validated by Seidl et al. [38], assesses thoughts and actions mobilized to deal with stressful events. It consists of 45 items and assesses four coping modes: (i) problem-focused, (ii) emotion-focused, (iii) religious practices/fantasy thinking, and (iv) seeking social support.

In the present study, Cronbach’s alpha coefficients were adequate for problem-focused coping (alpha = 0.90), emotion-focused coping (alpha = 0.75), seeking social support (alpha = 0.69), and religious practices/fantasy thinking (alpha = 0.70). Coping modes were analyzed using the mean scores of the items for each factor [38].

2.3.5. Utrecht Work Engagement Scale (UWES)

The UWES (17 items) measures work engagement across three dimensions: vigor, dedication, and absorption [39,40]. In the present study, internal consistency was good for vigor (alpha = 0.88), dedication (alpha = 0.88), and absorption (alpha = 0.80).

Scores for the dimensions were calculated as the arithmetic means of the corresponding items. The overall score was calculated as the mean of all items [39,40]. Scores were classified as: very low (0 to 0.99), low (1 to 1.99), medium (2 to 3.99), high (4 to 4.99), and very high (5 to 6) [41,42].

2.4. Data Collection Procedures

Data collection was conducted in person between January and December 2018, after Institutional Review Board approval and formal authorization from the respective organizational commanders.

The researchers contacted the commanders of the participating units, presented the study objectives, and, after obtaining signed Informed Consent Forms (ICF), made the instruments available to the participants. The ICF informed participants of the objectives, procedures, risks and benefits, the guarantee of confidentiality and anonymity, and the right to refuse or withdraw at any time without penalty.

The instruments were completed individually at a location of the participant’s choice and returned in sealed envelopes. To minimize potential peer influence and hierarchical constraints, participants were instructed to respond without the presence of superiors and to avoid discussing responses with colleagues.

2.5. Statistical Analysis

Data were entered into a Microsoft Excel^® spreadsheet for preliminary data cleaning and consistency checks. Analyses were performed using SPSS version 25.0 (IBM SPSS Statistics^®, New York, NY, USA) following sequential phases.

Analyses were conducted using an available-case analysis. For descriptive statistics and group comparisons, the analytic sample comprised participants with non-missing data for the variables involved in each analysis. Pearson correlations were computed using pairwise deletion, whereas the multivariate logistic regression model was estimated using a complete-case (listwise deletion) approach, resulting in a final analytic sample of 600 participants with non-missing data on the outcome and all predictors. Binary variables (e.g., relevant occupational stress, health/quality-of-life problems) were coded as 0 (No) and 1 (Yes) for analysis.

Descriptive and bivariate analysis. Sociodemographic and professional variables were described using absolute and relative frequencies. Continuous variables (occupational stress, burnout, engagement, and coping scores) were summarized using means and standard deviations. Hypotheses H1-H2 were primarily examined through Pearson correlation analyses among WSS, MBI-HSS dimensions, UWES dimensions, and EMEP coping modes, with False Discovery Rate (FDR) correction for multiple comparisons using the Benjamini–Hochberg method [43]. Specifically, H1 was evaluated through the correlation matrix linking emotion-focused coping (EMEP) to organizational stress (WSS) and burnout dimensions (EE and DP). H2 was examined through both bivariate associations and the multivariable logistic regression model. Additionally, joint burnout profiles (co-occurrence of high EE, high DP, and low PA) were calculated to characterize mixed clinical manifestations.

Comparative and stratified analyses. To compare coping and engagement scores between independent groups, data normality was assessed using the Shapiro–Wilk test and visual inspection of histograms. When assumptions were met, the independent-samples t-test was used; otherwise, the Wilcoxon rank-sum test was applied. For comparisons involving more than two groups, the Kruskal–Wallis test was used, followed by Dunn’s post hoc test when necessary. Correction for multiple comparisons was performed by adjusting the FDR using the Benjamini–Hochberg method [43]. Effect sizes were reported to support the interpretation of practical significance (Cohen’s d for mean differences or rank-biserial correlations for non-parametric comparisons).

Univariate analysis and preliminary variable selection. The association between relevant occupational stress (WSS ≥ 2.5) and independent variables was evaluated using univariate analyses. To ensure interpretability and meet reviewer recommendations, univariate logistic regression was used for both categorical and numerical predictors to estimate univariate odds ratios (ORs). To construct the multivariate model, purposive variable selection [44] was adopted, considering p < 0.20 in the univariate analysis as a preliminary criterion, in addition to theoretical plausibility.

Multivariate logistic regression. To identify factors associated with the presence of relevant occupational stress (WSS ≥ 2.5), a binary logistic regression model was estimated. To improve model stability and reduce sparse-cell instability, working hours were recoded into broader categories using the modal category as the reference group. We explicitly acknowledge that the 6 h workload category remained small (n = 10), requiring cautious interpretation of its associated estimates. Additionally, education level was recoded into a binary variable (0 = lower education; 1 = higher education) due to the low frequency of participants in the elementary school category, which previously generated unstable estimates. Variables in the final model were selected using purposive selection, including predictors that reached p < 0.20 in the univariate analysis or demonstrated theoretical plausibility. The final multivariate model was fitted using the ‘enter’ method to estimate the adjusted effect of the full set of selected predictors. The model specifically tested H2 by predicting relevant occupational stress from burnout dimensions and the health/quality-of-life covariate. Model adequacy was assessed using the Hosmer-Lemeshow test [44], and Nagelkerke’s pseudo-R² was used as a descriptive measure of fit [45]. Results were expressed as odds ratios (ORs) with 95% CIs and p-values, adopting a significance level of 5% (p < 0.05).

Sensitivity analysis (work shift). Because some categories of the original work-shift variable were sparse, we conducted a sensitivity analysis by collapsing work shift into two broader groups to improve estimate stability: fixed shift (morning or afternoon, morning/afternoon, and afternoon/night) versus rotating/operational schedule (shifts). We then re-estimated the fully adjusted logistic regression model using this collapsed parameterization to mitigate sparse-cell issues that can inflate standard errors and confidence intervals.

3. Results

A total of 773 military police officers participated in the study, of whom 506 (65.5%) belonged to CPI-5/SP and 267 (34.5%) to the 3rd BPM/PR. Ages ranged from 19 to 54 years, with a mean of 34.5 (SD = 7.8) and a median of 34 years. Most participants were between 21 and 40 years old (72%).

There was a predominance of male participants (87.2%) and married police officers (67.0%). Regarding education, 56.2% had completed high school and 41.9% had completed higher education. Most held the rank of soldier (53.7%), and 63.8% performed operational duties. The most frequent work schedule was 12/24 + 12/48 h (40.2%), followed by 8 h per day (30.3%), and the most common shift was afternoon/night (53.8%). Regarding length of service, 30.7% had between three and 10 years of service. In addition, 12.5% reported having another paid job, and 26.9% reported problems that compromise their health and quality of life (

Table 2. Sociodemographic and professional characteristics of the sample of military police officers.

Variables	n (%)
Battalion
Paraná	267 (34.5)
São Paulo	506 (65.5)
Age group
Up to 20 years old	14 (1.8)
21–30 years old	250 (32.3)
31–40 years old	307 (39.7)
41 years old or older	200 (25.9)
Not specified	2 (0.3)
Gender
Male	674 (87.2)
Female	98 (12.7)
Not specified	1 (0.1)
Marital status
Married	518 (67.0)
Single	202 (26.1)
Separated	44 (5.7)
Widowed	7 (0.9)
Not specified	2 (0.3)
Education
Elementary school	7 (0.9)
High school	434 (56.2)
Higher education	324 (41.9)
Not specified	8 (1.0)
Rank
Soldier	415 (53.7)
Corporal	175 (22.6)
Sergeant	59 (7.6)
Lieutenant	12 (1.6)
Captain	8 (1.0)
Sub-lieutenant	7 (0.9)
Cadet	7 (0.9)
Major	2 (0.3)
Not specified	88 (11.4)
Type of Activity
Operational	493 (63.8)
Administrative	197 (25.5)
Not specified	83 (10.7)
Workload
6 h	10 (1.3)
8 h	234 (30.3)
12/36 h	60 (7.8)
24/48 h	137 (17.7)
12/24 + 12/48 h	311 (40.2)
Not specified	21 (2.7)
Work shift
Afternoon/evening	416 (53.8)
Shifts	217 (28.1)
Morning/afternoon	80 (10.3)
Morning or afternoon	11 (1.4)
Not specified	49 (6.3)
Length of service as a Military Police Officer
Up to three years	183 (23.7)
Three to 10 years	237 (30.7)
10 to 20 years	212 (27.4)
More than 20 years	134 (17.3)
Not specified	7 (0.9)
Has another paid job
No	672 (86.9)
Yes	96 (12.5)
Not specified	5 (0.6)
Answered for disciplinary offenses
Yes	177 (22.9)
No	588 (76.1)
Not specified	8 (1.0)
Physical activity
Yes	591 (76.5)
No	178 (23.0)
Not specified	4 (0.5)
Health and quality of life issues
Yes	208 (26.9)
No	533 (69.0)
Not specified	32 (4.1)

Total sample n = 773. Percentages are based on non-missing data for each variable. “Not specified” indicates missing responses.

).

Because some variables had missing values, the effective sample size varied across analyses. Descriptive statistics were computed using all available observations for each variable (maximum n = 773). Group comparisons and bivariate tests used available-case denominators, whereas the multivariate logistic regression model was estimated using complete cases (listwise deletion), resulting in a final model sample of n = 600. Therefore, the Ns reported in Table 2, Table 3, Table 4, Table 5, Table 6, Table 7, Table 8, Table 9 and Table 10 refer to the number of participants with non-missing data for each specific analysis.

The prevalence of relevant occupational stress (WSS ≥ 2.5) was 30.2%. Regarding burnout dimensions, 24.1% of police officers showed high levels of Emotional Exhaustion (EE), 46.2% showed high levels of Depersonalization (DP), and 67.5% presented low Personal Accomplishment (PA; ≤31), which represents the impairment component of this subscale (Table 3).

Table 3. Distribution of burnout levels by dimension (MBI-HSS) and mean scores among military police officers.

Subscales	Average Score (SD)	Burnout Levels *
		Low (%)	Moderate (%)	High (%)
Emotional Exhaustion	21.7 (7.0)	23.5	52.7	23.8
Depersonalization	12.2 (4.0)	8.5	45.6	45.9
Personal Accomplishment	29.4 (5.2)	5.3	27.6	67.1

SD: standard deviation. * Because PA is inversely scored (lower values indicate greater burnout impairment), the ‘Low PA’ category corresponds to higher burnout-related risk.

Table 3. Distribution of burnout levels by dimension (MBI-HSS) and mean scores among military police officers.

Subscales	Average Score (SD)	Burnout Levels *
		Low (%)	Moderate (%)	High (%)
Emotional Exhaustion	21.7 (7.0)	23.5	52.7	23.8
Depersonalization	12.2 (4.0)	8.5	45.6	45.9
Personal Accomplishment	29.4 (5.2)	5.3	27.6	67.1

SD: standard deviation. * Because PA is inversely scored (lower values indicate greater burnout impairment), the ‘Low PA’ category corresponds to higher burnout-related risk.

The analysis of joint profiles further revealed that 17.6% (n = 134) of participants met the criteria for the full high-risk burnout profile (High EE + High DP + Low PA). In addition, mixed profiles were frequent, with 19.1% presenting High EE + High DP and 38.3% presenting High DP + Low PA, highlighting substantial co-occurrence across dimensions and clinically relevant heterogeneity in burnout manifestations (Table 4).

In this study, low scores in Personal Accomplishment are consistently treated as indicative of greater burnout impairment (reduced professional efficacy).

Table 4. Distribution of Burnout dimensions and joint profiles among military police officers (n = 725).

Burnout Profiles	n (%)
Isolated Dimensions
High Emotional Exhaustion (EE)	175 (24.1)
High Depersonalization (DP)	335 (46.2)
Low Personal Accomplishment (PA)	489 (67.5)
Joint Profiles (Co-occurrence)
High EE + High DP	145 (19.1)
High EE + Low PA	159 (21.2)
High DP + Low PA	289 (38.3)
Full Burnout Syndrome
High EE + High DP + Low PA	134 (17.6)

Note: Categorization based on MBI-HSS cut-off points: EE ≥ 27; DP ≥ 13; PA ≤ 31. Percentages refer to valid cases for each combination.

Table 4. Distribution of Burnout dimensions and joint profiles among military police officers (n = 725).

Burnout Profiles	n (%)
Isolated Dimensions
High Emotional Exhaustion (EE)	175 (24.1)
High Depersonalization (DP)	335 (46.2)
Low Personal Accomplishment (PA)	489 (67.5)
Joint Profiles (Co-occurrence)
High EE + High DP	145 (19.1)
High EE + Low PA	159 (21.2)
High DP + Low PA	289 (38.3)
Full Burnout Syndrome
High EE + High DP + Low PA	134 (17.6)

Note: Categorization based on MBI-HSS cut-off points: EE ≥ 27; DP ≥ 13; PA ≤ 31. Percentages refer to valid cases for each combination.

In the bivariate analysis, relevant occupational stress was associated with the battalion of origin (p < 0.001), with a higher frequency in the 3rd BPM/PR compared to CPI-5/SP, and with age group (p = 0.019), with a higher prevalence between 31 and 40 years (Table 5). No significant association was observed with gender (p = 0.500).

Table 5. Analysis of the association between occupational stress and sociodemographic characteristics of military police officers.

Variables	With Stress (n = 234)	Without Stress (n = 483)	p-Value a	Cramér’s V
Battalion	n (%)	n (%)	<0.001	0.232
Paraná	125 (53.4)	141 (29.2)
São Paulo	109 (46.6)	342 (70.8)
Age group			0.019	0.118
Up to 20 years old	2 (0.9)	9 (1.9)
21–30 years old	75 (32.1)	159 (32.9)
31–40 years old	111 (47.4)	178 (36.9)
41 years old or older	46 (19.7)	136 (28.2)
Not specified	-	1 (0.2)
Gender			0.500	0.021
Male	210 (89.7)	424 (87.8)
Female	24 (10.3)	58 (12.0)
Not specified	-	1 (0.2)

^a Chi-square test of independence.

Table 5. Analysis of the association between occupational stress and sociodemographic characteristics of military police officers.

Variables	With Stress (n = 234)	Without Stress (n = 483)	p-Value a	Cramér’s V
Battalion	n (%)	n (%)	<0.001	0.232
Paraná	125 (53.4)	141 (29.2)
São Paulo	109 (46.6)	342 (70.8)
Age group			0.019	0.118
Up to 20 years old	2 (0.9)	9 (1.9)
21–30 years old	75 (32.1)	159 (32.9)
31–40 years old	111 (47.4)	178 (36.9)
41 years old or older	46 (19.7)	136 (28.2)
Not specified	-	1 (0.2)
Gender			0.500	0.021
Male	210 (89.7)	424 (87.8)
Female	24 (10.3)	58 (12.0)
Not specified	-	1 (0.2)

^a Chi-square test of independence.

Univariate logistic regression analyses (Table 6) identified several factors associated with relevant occupational stress. Officers from the Paraná battalion (OR = 2.78; p < 0.001) and those in the 31–40 age group (OR = 1.84; p = 0.003) showed higher odds of stress. Regarding professional variables, a 6 h workload (OR = 7.59; p = 0.013) and 24/48 h shifts (OR = 1.82; p = 0.005) were significant predictors, as was the rotating/operational work shift (OR = 1.32; p = 0.018). Additionally, self-reported quality of life problems (OR = 2.54; p < 0.001) and a history of disciplinary offenses (OR = 2.44; p < 0.001) more than doubled the odds of stress. All three burnout dimensions were significantly associated with the outcome in the univariate models (p < 0.001).

Table 6. Univariate logistic regression for factors associated with relevant occupational stress (n = 717).

Variable	n (%)	Univariate OR	95% CI	p-Value
Battalion
São Paulo (Ref)	451 (62.9)	1.00	—	—
Paraná	266 (37.1)	2.78	2.01–3.84	<0.001
Age Group
41–60 years (Ref)	182 (25.4)	1.00	—	—
18–25 years	11 (1.5)	0.66	0.14–3.15	0.600
26–30 years	234 (32.7)	1.40	0.91–2.15	0.132
31–40 years	289 (40.4)	1.84	1.22–2.78	0.003
Education
Higher education (Ref)	307 (43.3)	1.00	—	—
Elementary/High school	402 (56.7)	1.16	0.84–1.59	0.366
Workload
12/24 + 12/48 h (Ref)	282 (40.1)	1.00	—	—
6 h	9 (1.3)	7.59	1.55–37.27	0.013
8 h	221 (31.4)	0.70	0.47–1.04	0.079
12/36 h	55 (7.8)	1.06	0.57–1.96	0.865
24/48 h	136 (19.3)	1.82	1.19–2.77	0.005
Work shift
Fixed (Ref)	507 (70.0)	1.00	—	—
Rotating/Operational	217 (30.0)	1.32	1.05–1.67	0.018
QoL Problems
No (Ref)	496 (71.6)	1.00	—	—
Yes	197 (28.4)	2.54	1.80–3.58	<0.001
Disciplinary Offenses
No (Ref)	539 (76.0)	1.00	—	—
Yes	170 (24.0)	2.44	1.71–3.48	<0.001
Burnout (Continuous)
Emotional Exhaustion	—	1.20	1.16–1.24	<0.001
Depersonalization	—	1.29	1.22–1.35	<0.001
Personal Accomplishment	—	0.88	0.85–0.91	<0.001

Note: OR = Odds Ratio; CI = Confidence Interval; “—” indicates the reference category (OR = 1). Univariate logistic regression models were fitted for each predictor independently. Reference categories (Ref) were selected based on the lowest expected risk or modal frequency.

Table 6. Univariate logistic regression for factors associated with relevant occupational stress (n = 717).

Variable	n (%)	Univariate OR	95% CI	p-Value
Battalion
São Paulo (Ref)	451 (62.9)	1.00	—	—
Paraná	266 (37.1)	2.78	2.01–3.84	<0.001
Age Group
41–60 years (Ref)	182 (25.4)	1.00	—	—
18–25 years	11 (1.5)	0.66	0.14–3.15	0.600
26–30 years	234 (32.7)	1.40	0.91–2.15	0.132
31–40 years	289 (40.4)	1.84	1.22–2.78	0.003
Education
Higher education (Ref)	307 (43.3)	1.00	—	—
Elementary/High school	402 (56.7)	1.16	0.84–1.59	0.366
Workload
12/24 + 12/48 h (Ref)	282 (40.1)	1.00	—	—
6 h	9 (1.3)	7.59	1.55–37.27	0.013
8 h	221 (31.4)	0.70	0.47–1.04	0.079
12/36 h	55 (7.8)	1.06	0.57–1.96	0.865
24/48 h	136 (19.3)	1.82	1.19–2.77	0.005
Work shift
Fixed (Ref)	507 (70.0)	1.00	—	—
Rotating/Operational	217 (30.0)	1.32	1.05–1.67	0.018
QoL Problems
No (Ref)	496 (71.6)	1.00	—	—
Yes	197 (28.4)	2.54	1.80–3.58	<0.001
Disciplinary Offenses
No (Ref)	539 (76.0)	1.00	—	—
Yes	170 (24.0)	2.44	1.71–3.48	<0.001
Burnout (Continuous)
Emotional Exhaustion	—	1.20	1.16–1.24	<0.001
Depersonalization	—	1.29	1.22–1.35	<0.001
Personal Accomplishment	—	0.88	0.85–0.91	<0.001

Note: OR = Odds Ratio; CI = Confidence Interval; “—” indicates the reference category (OR = 1). Univariate logistic regression models were fitted for each predictor independently. Reference categories (Ref) were selected based on the lowest expected risk or modal frequency.

In the final multivariate model (n = 600), battalion, education level, work shift, and history of disciplinary infractions were no longer significantly associated with occupational stress after adjustment, but were retained to ensure optimal model specification. The model yielded a Nagelkerke pseudo-R² of 0.366, indicating a satisfactory overall fit (Table 7).

Regarding workload, using the modal category (12/24 + 12/48 h) as the reference, the 6 h schedule was associated with significantly higher odds of relevant occupational stress (OR = 7.76; 95% CI: 1.02–58.79; p = 0.047). However, this specific finding should be interpreted with caution due to the small sample size (n = 10) and the resulting wide confidence interval. Other workload schedules did not show statistically significant differences compared to the modal group. Furthermore, the absence of self-reported health and quality of life problems was associated with lower odds of stress (OR = 0.62; 95% CI: 0.40–0.95; p = 0.029). Finally, Emotional Exhaustion (OR = 1.14; 95% CI: 1.09–1.19; p < 0.001) and Depersonalization (OR = 1.12; 95% CI: 1.04–1.20; p = 0.001) remained the primary factors significantly associated with higher odds of occupational stress in the adjusted model (Table 7).

In the sensitivity analysis using the collapsed work-shift variable to mitigate sparse-cell instability, work shift was not associated with stress (adjusted OR = 1.20; 95% CI: 0.65–2.21; p = 0.550), supporting the conclusion that the original fine-grained shift parameterization was sensitive to sparse categories.

Table 7. Multivariate logistic regression model for the presence of occupational stress among military police officers (complete-case analysis; n = 600).

Characteristics	OR	95% CI	p-Value
Battalion
São Paulo	—	—
Paraná	0.69	0.37–1.30	0.256
Education
Lower education	—	—
Higher education	0.79	0.52–1.18	0.253
Workload
12/24 + 12/48 h	—	—
6 h	7.76	1.02–58.79	0.047
8 h	0.57	0.32–1.03	0.063
12/36 h	0.75	0.31–1.80	0.524
24/48 h	0.59	0.28–1.25	0.167
Work shift
Afternoon/evening	—	—
Shifts	0.59	0.03–13.55	0.742
Morning/afternoon	0.83	0.13–5.20	0.844
Morning or afternoon	0.56	0.11–2.83	0.483
Disciplinary offenses
Yes	—	—
No	0.64	0.39–1.03	0.065
Health and quality of life issues
Yes	—	—
No	0.62	0.40–0.95	0.029
Burnout Dimensions (Continuous)
Emotional Exhaustion	1.14	1.09–1.19	<0.001
Depersonalization	1.12	1.04–1.20	0.001
Personal Accomplishment	1.00	0.95–1.05	0.875

Note: Model n = 600 (listwise deletion). OR = odds ratio; CI = confidence interval; “—” indicates the reference category (OR = 1). Model estimated with binary logistic regression (enter method). Outcome coded as occupational stress (WSS ≥ 2.5; 0 = no, 1 = yes). Reference categories: Battalion = São Paulo; Education = Lower education; Workload = 12/24 + 12/48 h (modal); Work shift = Afternoon/evening; Disciplinary offenses = Yes; Health/QoL issues = Yes. Continuous predictors (EE, DP, PA) were entered as raw MBI-HSS subscale scores. Model fit: Nagelkerke R² = 0.366; Hosmer–Lemeshow χ²(8) = 11.671, p = 0.167.

Table 7. Multivariate logistic regression model for the presence of occupational stress among military police officers (complete-case analysis; n = 600).

Characteristics	OR	95% CI	p-Value
Battalion
São Paulo	—	—
Paraná	0.69	0.37–1.30	0.256
Education
Lower education	—	—
Higher education	0.79	0.52–1.18	0.253
Workload
12/24 + 12/48 h	—	—
6 h	7.76	1.02–58.79	0.047
8 h	0.57	0.32–1.03	0.063
12/36 h	0.75	0.31–1.80	0.524
24/48 h	0.59	0.28–1.25	0.167
Work shift
Afternoon/evening	—	—
Shifts	0.59	0.03–13.55	0.742
Morning/afternoon	0.83	0.13–5.20	0.844
Morning or afternoon	0.56	0.11–2.83	0.483
Disciplinary offenses
Yes	—	—
No	0.64	0.39–1.03	0.065
Health and quality of life issues
Yes	—	—
No	0.62	0.40–0.95	0.029
Burnout Dimensions (Continuous)
Emotional Exhaustion	1.14	1.09–1.19	<0.001
Depersonalization	1.12	1.04–1.20	0.001
Personal Accomplishment	1.00	0.95–1.05	0.875

Note: Model n = 600 (listwise deletion). OR = odds ratio; CI = confidence interval; “—” indicates the reference category (OR = 1). Model estimated with binary logistic regression (enter method). Outcome coded as occupational stress (WSS ≥ 2.5; 0 = no, 1 = yes). Reference categories: Battalion = São Paulo; Education = Lower education; Workload = 12/24 + 12/48 h (modal); Work shift = Afternoon/evening; Disciplinary offenses = Yes; Health/QoL issues = Yes. Continuous predictors (EE, DP, PA) were entered as raw MBI-HSS subscale scores. Model fit: Nagelkerke R² = 0.366; Hosmer–Lemeshow χ²(8) = 11.671, p = 0.167.

Regarding work engagement, high levels of dedication, vigor, and overall score were observed, along with an average level of absorption. Among coping strategies, the highest average was identified in problem-focused coping, while the lowest average occurred in emotion-focused coping (Table 8).

In comparisons between battalions, CPI-5/SP police officers had higher mean engagement scores (dedication, absorption, vigor, and overall score) than the 3rd BPM/PR, with statistically significant differences after FDR correction (p and q < 0.001). Regarding coping, CPI-5/SP had higher mean scores for problem-focused coping (p < 0.001), whereas no significant difference was observed for seeking social support (p = 0.067) after adjustment. The 3rd BPM/PR had a higher mean score for emotion-focused coping (p < 0.001) (Table 8).

Table 8. Distribution of mean scores for work engagement (UWES) and problem coping strategies (EMEP), according to battalion of origin (n = 773).

Variables	Paraná n = 267	São Paulo n = 506	p-Value	Effect Size (d)
Work Engagement	Mean ± SD	Mean ± SD
Dedication	4.0 ± 1.3	4.9 ± 1.0	<0.001 1	0.75 (Medium/Large)
Absorption	3.4 ± 1.2	4.1 ± 1.2	<0.001 1	0.60 (Medium)
Vigor	3.8 ± 1.2	4.6 ± 1.0	<0.001 1	0.71 (Medium/Large)
Overall score	3.7 ± 1.1	4.5 ± 1.0	<0.001 1	0.73 (Medium/Large)
Coping strategies
Problem-focused	3.6 ± 0.6	3.8 ± 0.6	<0.001 1	0.39 (Small/Medium)
Emotion-focused	2.5 ± 0.6	2.3 ± 0.6	<0.001 1	−0.37 (Small/Medium)
Based on seeking religious practices	2.9 ± 0.7	3.1 ± 0.7	0.003 1	0.15 (Negligible)
Based on seeking social support	2.7 ± 0.8	2.8 ± 0.9	0.067 1	0.22 (Small)

SD: standard deviation. ¹ Wilcoxon rank sum test. Values are presented as mean ± SD for each battalion. p-values refer to between-group comparisons (Paraná vs. São Paulo). Effect sizes are reported as Cohen’s d.

Table 8. Distribution of mean scores for work engagement (UWES) and problem coping strategies (EMEP), according to battalion of origin (n = 773).

Variables	Paraná n = 267	São Paulo n = 506	p-Value	Effect Size (d)
Work Engagement	Mean ± SD	Mean ± SD
Dedication	4.0 ± 1.3	4.9 ± 1.0	<0.001 1	0.75 (Medium/Large)
Absorption	3.4 ± 1.2	4.1 ± 1.2	<0.001 1	0.60 (Medium)
Vigor	3.8 ± 1.2	4.6 ± 1.0	<0.001 1	0.71 (Medium/Large)
Overall score	3.7 ± 1.1	4.5 ± 1.0	<0.001 1	0.73 (Medium/Large)
Coping strategies
Problem-focused	3.6 ± 0.6	3.8 ± 0.6	<0.001 1	0.39 (Small/Medium)
Emotion-focused	2.5 ± 0.6	2.3 ± 0.6	<0.001 1	−0.37 (Small/Medium)
Based on seeking religious practices	2.9 ± 0.7	3.1 ± 0.7	0.003 1	0.15 (Negligible)
Based on seeking social support	2.7 ± 0.8	2.8 ± 0.9	0.067 1	0.22 (Small)

SD: standard deviation. ¹ Wilcoxon rank sum test. Values are presented as mean ± SD for each battalion. p-values refer to between-group comparisons (Paraná vs. São Paulo). Effect sizes are reported as Cohen’s d.

In the comparison by gender, differences in engagement were initially observed for some dimensions (p < 0.05), but did not remain after FDR correction. For coping, no statistically significant differences were observed after correction (Table 9).

Table 9. Distribution of mean scores for work engagement (UWES) and problem coping strategies (EMEP) according to gender in the assessment of military police officers.

Variables	Male n = 674	Female n = 98	Total n = 772	p-Value	Value q 1
Work Engagement	Mean ± SD	Mean ± SD	Mean ± SD
Dedication	4.6 ± 1.2	4.4 ± 1.1	4.6 ± 1.2	0.023 2	0.091
Absorption	3.8 ± 1.2	3.7 ± 1.2	3.8 ± 1.2	0.300 2	0.300
Vigor	4.4 ± 1.2	4.1 ± 1.1	4.3 ± 1.2	0.023 2	0.091
Overall score	4.3 ± 1.1	4.1 ± 1.1	4.2 ± 1.1	0.046 2	0.091
Coping strategies
Problem-focused	3.7 ± 0.7	3.8 ± 0.5	3.7 ± 0.6	0.500 2	0.500
Emotion-focused	2.3 ± 0.6	2.5 ± 0.7	2.3 ± 0.6	0.039 3	0.091
Based on seeking religious practices	3.0 ± 0.7	3.1 ± 0.7	3.0 ± 0.7	0.069 2	0.110
Based on seeking social support	2.7 ± 0.8	2.9 ± 0.9	2.7 ± 0.9	0.150 2	0.200

SD: standard deviation. ¹ False discovery rate correction for multiple tests. ² Wilcoxon rank sum test. ³ Welch’s t-test.

Table 9. Distribution of mean scores for work engagement (UWES) and problem coping strategies (EMEP) according to gender in the assessment of military police officers.

Variables	Male n = 674	Female n = 98	Total n = 772	p-Value	Value q 1
Work Engagement	Mean ± SD	Mean ± SD	Mean ± SD
Dedication	4.6 ± 1.2	4.4 ± 1.1	4.6 ± 1.2	0.023 2	0.091
Absorption	3.8 ± 1.2	3.7 ± 1.2	3.8 ± 1.2	0.300 2	0.300
Vigor	4.4 ± 1.2	4.1 ± 1.1	4.3 ± 1.2	0.023 2	0.091
Overall score	4.3 ± 1.1	4.1 ± 1.1	4.2 ± 1.1	0.046 2	0.091
Coping strategies
Problem-focused	3.7 ± 0.7	3.8 ± 0.5	3.7 ± 0.6	0.500 2	0.500
Emotion-focused	2.3 ± 0.6	2.5 ± 0.7	2.3 ± 0.6	0.039 3	0.091
Based on seeking religious practices	3.0 ± 0.7	3.1 ± 0.7	3.0 ± 0.7	0.069 2	0.110
Based on seeking social support	2.7 ± 0.8	2.9 ± 0.9	2.7 ± 0.9	0.150 2	0.200

SD: standard deviation. ¹ False discovery rate correction for multiple tests. ² Wilcoxon rank sum test. ³ Welch’s t-test.

In the comparison by age group, engagement showed significant variation between groups, with lower means concentrated in the 31–40 age group; similar patterns were observed for some coping strategies, with differences that remained significant after correction for multiple comparisons (Table 10).

Table 10. Distribution of mean scores for work engagement (UWES) and problem coping strategies (EMEP) according to age group in the assessment of military police officers.

Variables	Up to 20 Years Old n = 14	21 to 30 Years Old n = 250	31 to 40 Years Old n = 307	41 Years Old or Older n = 200	p-Value 1	Value q 2
Work Engagement	Mean ± SD	Mean ± SD	Mean ± SD	Mean ± SD
Dedication	5.1 ± 0.8 ab	4.7 ± 1.2 a	4.4 ± 1.2 b	4.7 ± 1.2 a	<0.001	<0.001
Absorption	4.1 ± 1.0 ab	3.9 ± 1.1 a	3.6 ± 1.3 b	4.0 ± 1.2 a	0.011	0.014
Vigor	4.7 ± 0.8 ab	4.4 ± 1.2 a	4.1 ± 1.2 b	4.5 ± 1.1 a	0.003	0.004
Overall score	4.6 ± 0.8 ab	4.3 ± 1.1 a	4.0 ± 1.1 b	4.4 ± 1.1 a	0.001	0.002
Coping strategies
Problem-focused	3.9 ± 0.5 ab	3.7 ± 0.7 ab	3.7 ± 0.6 a	3.9 ± 0.6 b	0.015	0.018
Emotion-focused	2.1 ± 0.5 ab	2.2 ± 0.6 a	2.4 ± 0.6 b	2.3 ± 0.6 ab	<0.001	<0.001
Based on seeking religious practices	3.2 ± 0.7 ab	2.8 ± 0.7 a	3.0 ± 0.7 b	3.2 ± 0.7 b	<0.001	<0.001
Based on seeking social support	3.1 ± 0.8 a	2.8 ± 0.8 a	2.7 ± 0.8 a	2.8 ± 0.9 a	0.032	0.032

SD: standard deviation. ¹ Kruskal–Wallis test. ² False discovery rate correction for multiple tests. Groups sharing the same letter are not significantly different from each other (p > 0.05) based on Dunn’s test with Benjamini–Hochberg correction. Groups with different letters are significantly different (p < 0.05).

Table 10. Distribution of mean scores for work engagement (UWES) and problem coping strategies (EMEP) according to age group in the assessment of military police officers.

Variables	Up to 20 Years Old n = 14	21 to 30 Years Old n = 250	31 to 40 Years Old n = 307	41 Years Old or Older n = 200	p-Value 1	Value q 2
Work Engagement	Mean ± SD	Mean ± SD	Mean ± SD	Mean ± SD
Dedication	5.1 ± 0.8 ab	4.7 ± 1.2 a	4.4 ± 1.2 b	4.7 ± 1.2 a	<0.001	<0.001
Absorption	4.1 ± 1.0 ab	3.9 ± 1.1 a	3.6 ± 1.3 b	4.0 ± 1.2 a	0.011	0.014
Vigor	4.7 ± 0.8 ab	4.4 ± 1.2 a	4.1 ± 1.2 b	4.5 ± 1.1 a	0.003	0.004
Overall score	4.6 ± 0.8 ab	4.3 ± 1.1 a	4.0 ± 1.1 b	4.4 ± 1.1 a	0.001	0.002
Coping strategies
Problem-focused	3.9 ± 0.5 ab	3.7 ± 0.7 ab	3.7 ± 0.6 a	3.9 ± 0.6 b	0.015	0.018
Emotion-focused	2.1 ± 0.5 ab	2.2 ± 0.6 a	2.4 ± 0.6 b	2.3 ± 0.6 ab	<0.001	<0.001
Based on seeking religious practices	3.2 ± 0.7 ab	2.8 ± 0.7 a	3.0 ± 0.7 b	3.2 ± 0.7 b	<0.001	<0.001
Based on seeking social support	3.1 ± 0.8 a	2.8 ± 0.8 a	2.7 ± 0.8 a	2.8 ± 0.9 a	0.032	0.032

SD: standard deviation. ¹ Kruskal–Wallis test. ² False discovery rate correction for multiple tests. Groups sharing the same letter are not significantly different from each other (p > 0.05) based on Dunn’s test with Benjamini–Hochberg correction. Groups with different letters are significantly different (p < 0.05).

In correlation analyses, occupational stress, emotional exhaustion, and depersonalization showed inverse and predominantly moderate correlations with dimensions of work engagement. In addition, these variables were inversely correlated (generally low magnitude) with problem-focused coping and seeking social support, and positively correlated (predominantly weak to moderate) with emotion-focused coping. Correlations involving religious coping showed low magnitude, and cautious interpretation is recommended (

Table 11. Analysis of correlations between occupational stress and Burnout subscales with dimensions of work engagement and subscales of problem-coping modes in military police officers.

Variables	Occupational Stress	Burnout
		Emotional Exhaustion	Depersonalization	Personal Accomplishment
Work Engagement
Dedication
Correlation (r)	−0.502 b	−0.652 b	−0.471 b	0.616 b
p-value	<0.001 **	<0.001 **	<0.001 **	<0.001 **
Absorption
Correlation (r)	−0.418 b	−0.549 b	−0.414 b	0.468 b
p-value	<0.001 **	<0.001 **	<0.001 **	<0.001 **
Vigor
Correlation (r)	−0.547 b	−0.701 b	−0.483 b	0.624 b
p-value	<0.001 **	<0.001 **	<0.001 **	<0.001 **
Overall score
Correlation (r)	−0.522 b	−0.676 b	−0.486 b	0.606 b
p-value	<0.001 **	<0.001 **	<0.001 **	<0.001 **
Coping strategies
Problem-focused
Correlation (r)	−0.231 a	−0.300 a	−0.300 a	0.469 b
p-value	<0.001 **	<0.001 **	<0.001 **	<0.001 **
Emotion-focused
Correlation (r)	0.336 a	0.401 b	0.372 a	−0.207 a
p-value	<0.001 **	<0.001 **	<0.001 **	<0.001 **
Based on seeking religious practices
Correlation (r)	0.085 a	0.132 a	0.012 a	0.061 a
p-value	0.019 *	<0.001 **	0.736 **	0.092 **
Based on seeking social support
Correlation (r)	−0.224 a	−0.237 a	−0.233 a	0.326 a
p-value	<0.001 **	<0.001 **	<0.001 **	<0.001 **

Note: ^a (n varies between 690–710); ^b (n varies between 760–773) due to pairwise missing data handling. * Significant values at p < 0.05; ** Significant values at p < 0.01.

).

4. Discussion

This study examined the interrelationships among mental health indicators, coping strategies, and work engagement in Brazilian military police officers (n = 773) across two institutional contexts (São Paulo and Paraná). As expected, engagement was negatively associated with occupational stress and burnout indicators; however, the overall pattern also suggests that high engagement may coexist with substantial emotional exhaustion and depersonalization. This supports the conception of engagement as a motivational state sustained by professional identity, the meaning attributed to work, and institutional demands [46,47], while burnout symptoms reflect cumulative stress and the costs associated with coping, which develop through partially distinct pathways [48,49,50]. Taken together, the results of the present study are consistent with evidence and the reviewed theoretical frameworks [7,28,51,52,53]. This interpretation is more compatible with a dual-process approach than with a simplified “opposites” model, implying that interventions aimed at promoting well-being in high-demand occupations should consider differentiated mechanisms for strengthening engagement and preventing burnout [7,28].

Given the cross-sectional design, causal inferences cannot be drawn and the temporal ordering of stress, burnout, engagement, and coping cannot be established. This limitation is particularly salient for trauma-related outcomes, for which occupational and organizational factors may function both as correlates and as consequences (e.g., sickness absence and reassignment), as suggested by evidence of meaningful PTSD prevalence among military police officers in Rio de Janeiro [9].

The prevalence of relevant occupational stress (30.2%) is consistent with literature describing police work as characterized by recurrent threats, exposure to violence, organizational pressure, and sustained emotional demands, with cumulative implications for quality of life and clinical vulnerability [54,55,56]. In a Brazilian study of military police officers in Paraná, an even higher prevalence (46.7%) was reported and was associated with organizational stressors such as limited career advancement, training perceived as insufficient, discrimination/favoritism, and long working hours, suggesting that a substantial share of risk is produced by working conditions [7].

Moreover, national evidence indicates that exposure to critical incidents may be associated with considerable clinical burden, including full and partial PTSD, reinforcing the importance of institutional policies that integrate primary prevention (reducing avoidable stressors) with secondary and tertiary prevention (early detection and care). This approach is consistent with the Job Demands-Resources framework, under which high demands not offset by resources tend to sustain burnout trajectories, even when engagement remains high [9,28].

With respect to burnout, the full syndrome was observed in 17.6% (n = 134) of officers under the operational criteria adopted, alongside frequent mixed profiles (e.g., 38.3% with high Depersonalization + low Personal Accomplishment). Although the dimensional approach highlights heterogeneity in burnout manifestations, the high prevalence of syndromic and mixed profiles underscores a significant clinical burden, potentially unsustainable over time and indicative of early syndromic consolidation. In the literature, exhaustion has been linked to functional impairment and increased vulnerability to illness, whereas depersonalization is often conceptualized as a distancing response to emotionally intense demands [57,58].

The coexistence of high Personal Accomplishment with elevated stress and exhaustion warrants a nuanced interpretation: officers may perceive themselves as effective and productive, yet this perceived efficacy may be maintained at the expense of psychological sustainability. This “high-performance exhaustion” profile is likely unsustainable and may precede clinical deterioration despite preserved operational performance [59,60,61].

Importantly, our dimensional focus on MBI-HSS subscale scores (EE, DP, and PA) reveals that mixed profiles (e.g., high engagement alongside high stress) serve as critical warnings for occupational health governance, indicating that monitoring must look beyond maintained operational performance to capture early signs of psychological strain.

When interpreting the multivariable model, adjusted associations should be understood as conditional on the set of covariates included, rather than as evidence of mechanisms. Notably, the 6 h shift was associated with higher odds of stress compared with the modal schedule (12/24 + 12/48 h), whereas the absence of self-reported health/quality-of-life problems was associated with lower odds of stress. In sensitivity analysis, work shift was not associated with stress (aOR = 1.20; 95% CI: 0.66–2.21), supporting that sparse categories may inflate instability in the original parameterization. This pattern has practical relevance because it suggests that key intervention targets may lie less in individual-level attributes and more in job design and organizational conditions [7,13,14].

Differences between contexts suggest that part of the observed risk may reflect territorial and institutional factors [62,63]. Although bivariate analyses indicated a higher prevalence of stress in Paraná than in São Paulo, this difference did not persist after adjustment for workload, health/QoL issues, and burnout dimensions, suggesting that contextual variation may be partly mediated by these factors. Evidence on structured support initiatives and operational predictability also points to potentially protective institutional components that may be leveraged in prevention strategies [64,65,66].

In line with the Job Demands-Resources model, our findings are consistent with the concurrent operation of a strain process driven by high demands and a motivational process supported by organizational resources. From a management perspective, higher-yield interventions are often upstream: redesigning schedules and shifts, strengthening training and capacity-building governance, improving communication, promoting organizational justice, and implementing recognition practices, rather than relying predominantly on individual coping efforts [19].

Regarding coping, the predominance of problem-focused strategies is broadly consistent with adaptive repertoires [67,68]. Nevertheless, lower emotion-focused coping scores may reflect occupational norms that discourage acknowledging vulnerability, potentially constraining early help-seeking. Correlational patterns in this sample indicate that higher stress and burnout indicators are associated with lower engagement and reduced use of problem-focused coping and social support, alongside greater reliance on emotion-focused coping [69,70,71,72]. This pattern supports the value of continuous monitoring approaches focused on early identification of dimensional risk profiles before syndromic conditions consolidate.

Limitations and Contributions

Interpretation of these findings should account for several limitations. First, the study’s reliance on a non-probabilistic convenience sample introduces potential selection bias. Participation likely depended on individual availability, willingness, and local organizational conditions; thus, officers experiencing severe distress may have been less inclined to participate (a potential “healthy worker” effect) or, conversely, those more motivated to report organizational grievances may be overrepresented. As the direction of this bias cannot be definitively determined, prevalence estimates and group comparisons should be interpreted as associational patterns within these specific units rather than as population-level parameters for all Brazilian military police.

The use of self-reported measures also carries a risk of response bias, such as social desirability and underreporting, which is particularly salient in organizational cultures that may discourage the expression of vulnerability. Additionally, because exposures and outcomes were measured concurrently using similar instruments, common method bias and residual confounding cannot be ruled out. Differences observed between São Paulo and Paraná may reflect multiple institutional factors, such as organizational climate, leadership styles, and operational protocols, not fully captured by the variables analyzed.

Despite these limitations, the study provides actionable insights for institutional practice by identifying modifiable correlates, such as coping repertoires and burnout dimensions, that can serve as direct targets for preventive programming. By mapping how these factors interrelate within specific units, the findings offer a baseline for institutional screening and risk assessment, enabling managers to prioritize resources where maladaptive strategies or high exhaustion are most prevalent. Specifically, the identification of “mixed profiles” (high engagement alongside high stress) serves as a critical warning for occupational health governance, indicating that monitoring must look beyond maintained operational performance to capture early signs of psychological strain.

Even without causal inference, the model serves as a risk map for occupational health surveillance. Analytically, by integrating stress, burnout, engagement, and coping within a single framework, the study supports a more systemic understanding of occupational adaptation. From an applied perspective, the results support the planning of stratified, context-sensitive interventions, helping to avoid generic responses that fail to address local specificities. Furthermore, by providing a pre-pandemic baseline of mental health indicators, this study offers a relevant benchmark for future investigations into the long-term impacts of global health crises on police forces.

As a future agenda, we recommend the use of longitudinal designs and mixed methods, incorporating more granular measures of work organization, such as perceived institutional support, psychological safety climate, and constructs from the demand–control and effort–reward models, to expand the translational utility and explanatory power of subsequent investigations.

5. Conclusions

This study indicates that, among the 773 officers investigated, there was a significant prevalence of occupational stress and important proportions of high depersonalization and emotional exhaustion, characterizing a relevant psychosocial risk profile. A considerable percentage met criteria for the full burnout syndrome, and mixed profiles (e.g., high depersonalization combined with low personal accomplishment) were frequent, even within a context of high personal accomplishment and elevated engagement.

Although engagement levels were high and problem-focused coping predominated, higher stress and burnout indicators were associated with lower engagement and reduced use of social support. Health and quality-of-life problems emerged as key factors associated with stress, whereas work shift was not significantly associated after adjustment, suggesting that workload and individual vulnerabilities may play a more central role than scheduling characteristics per se.

Despite bivariate differences between states, these did not persist in multivariate analyses. Overall, the coexistence of burnout dimensions and high engagement underscores the complexity of psychosocial functioning in military police settings and highlights the need for integrated institutional strategies that reduce organizational stressors, strengthen support systems, and foster adaptive coping, particularly among subgroups at greater psychosocial risk.