Teacher burnout and wellbeing

1. Vignette: the staff room at three in the afternoon

The scene is not a datum. It joins the workshop and the application to what they do not touch: paperwork, class groups, behaviour that breaks the period, family violence with no channel, a leadership newly arrived and a young teacher who is already looking at the exit. The question, in compulsory education, is whether burnout is a personal failing or a condition of demands and resources. Each claim will be given as an empirical finding, a conceptual framework or a design inference. Nothing in this staff room is a Mexican datum.

2. Problem: from personal deficit to organisational condition

Conceptual framework. Burnout is treated as a psychosocial risk of work, not as a trait. Moens et al. (2026) diagnose the field: teaching is highly sensitive to stress; stress and burnout, as psychosocial risks, are of greater concern; shortage and attrition sharpen them. The question moves from who regulates their emotions to what the post requires.

Population: teachers in compulsory education, classroom teachers and, where the study includes them, those who lead. Concept: burnout and wellbeing in the job demands and resources model. Context: schools, not the clinic. How are those demands and those resources related to burnout and wellbeing, and what is known about teaching, students and staying?

Empirical finding, on the scope of the field. Moens et al. (2026) bring together 366 studies with a repeated map of demands and resources. The detail comes later. Treating burnout as an individual defect is a management decision, not a conclusion of that map. Wartenberg et al. (2026) add that the area grew without a synthesis on absenteeism, interactions, motivation and achievement, although they start from the premise that burnout impairs cognitive, motivational and social functioning. That synthesis is used further on, without advancing figures that the abstract does not give.

Design inference. An organisational condition does not mean that the school, as a block, explains the differences between teachers. It means that demands and resources belong to the work and are distributed unequally within the same building. The nuance lies in Schulze-Hagenest et al. (2023).

3. Framework: job demands and resources in schools

Conceptual framework. This section does not estimate a model of its own. It orders the corpus as demands that consume energy set against resources for teaching and for staying. Burnout appears when there is no counterweight. The scheme does not stand in for the data.

Empirical finding. In 366 studies, Moens et al. (2026) point to workload, student challenges and time pressure as the most reported demands, and to support among colleagues, the teacher–student relationship, collaboration and the link with the leadership as resources. The outcomes are stress, burnout and wellbeing. It is a frequency map: the abstract does not rank effect sizes.

Collie (2023) specifies this with 426 teachers in Australia and controls for gender, experience, qualification and personality. Resources: leadership that supports autonomy, and relatedness with colleagues and students. Demands: leadership that thwarts autonomy, time pressure and disruptive behaviour. Wellbeing: subjective vitality, behavioural engagement and professional growth. Empirical finding. The resources were in general associated with greater wellbeing. Time pressure was associated with less vitality and with greater behavioural engagement. No interaction remained. Design inference. Imbalance means high demands and scarce resources, not a universal interaction. That the association survives personality weakens the thesis of a deficit of character.

Admiraal and Kittelsen Røberg (2023), with TALIS 2018 in 24 European countries, find a safe climate and a collaborative and participative culture as the principal resources, and distress and barriers to professional development as the demands. The pattern was similar across stages; what changed was satisfaction with the school, with the career chosen or with the profession. The abstract does not translate that distress into a lack of resilience.

Wang et al. (2025) follow 645 teachers in China across one semester: time poverty and resilience, then emotional exhaustion, then subjective wellbeing. Empirical finding. Emotional exhaustion mediates an indirect harm from time poverty. Resilience buffers that segment. Gender amplifies the step from exhaustion to wellbeing, with greater vulnerability among women teachers. They call for time and for gender sensitivity, not for resilience alone.

Inabnett and Long (2026) propose secondary traumatic stress as a specific demand: in high-need schools the students arrive with more trauma and the person who teaches is exposed to it at work. Its results are examined in the section on attrition. A disclosure of violence at home is material of the post.

Paudel et al. (2025) separate, with the Maslach Burnout Inventory for Educators, emotional exhaustion, depersonalisation and personal accomplishment. The abstracts do not define those subscales. Klusmann et al. (2023), Schulze-Hagenest et al. (2023) and Wang et al. (2025) report only emotional exhaustion. Saloviita and Pakarinen (2021) cite three subdomains without naming them, and Navy and Yonai (2026) speak of dimensions without a label. Methodological note. Definitions that the abstracts do not provide are not invented.

4. Method: critical narrative review with a screening flow

Methodological note. This is a critical narrative review: there is no meta-analysis of its own and no double coding. The base was the Crossref REST API on 20 October 2026, with 16 queries of 60 rows, journal articles from 2021, on burnout, wellbeing, stress and attrition, and on demands and resources, mindfulness, leadership, resilience and Latin America. The flow is the record of that screening.

Screening step	Records
Crossref queries (rows per query)	16 (60)
Unique verified records	459
Excluded for a year out of range (2027)	1
Records from 2021 to 2026	458
Excluded: corrections, errata or reviews	4
Excluded: journals of doubtful editorial quality	31
Excluded for falling out of scope	42
Eligible after screening	381
Included by manual selection	24
With a usable abstract	23
Cited from the title only	1
Latin American studies included	1

From 459 unique records one from 2027 was set aside. 458 remained. Excluded were 4 corrections, errata or reviews, 31 journals of doubtful quality and 42 out of scope. There were 381 eligible records and 24 entered by hand, with priority for syntheses, longitudinal or multilevel designs, trials and an abstract. Twenty-three have an abstract, one enters by the title only and one is the only Latin American study. Design inference. The filter biases the set towards what is indexed with an abstract. There is no mean effect for the field. Where the abstract is silent on the sign, none is invented. Madigan and Kim (2021) have no abstract and Saloviita and Pakarinen (2021) do not give the sign. Lambert et al. (2023) arrive truncated.

5. Magnitude and trajectories: when and in whom exhaustion accumulates

Empirical finding. Klusmann et al. (2023) observed high levels and an increase in emotional exhaustion in 2157 teachers and 374 school leaders in Germany during the pandemic. Health and workload were related positively; social support, negatively. The open responses described the work as predominantly stressful. There is neither a size of the rise nor a prevalence. Strain also rose among those who lead.

Saloviita and Pakarinen (2021) cross levels with 4567 primary teachers in Finland: 2080 class teachers, 1744 subject teachers, 438 special-class teachers and 305 resource-room teachers. They examined burnout and three subdomains together with teacher category, class size, students with support needs, attitudes towards inclusion and the availability of support. Empirical finding, in outline. There were associations whose sign the abstract does not give and which is not invented here.

Schulze-Hagenest et al. (2023) rule out a comfortable sentence. In two samples (N = 1022 and N = 2886), multilevel regression showed low between-school variance in wellbeing — emotional exhaustion and satisfaction — and substantial variation in social interactions. Wellbeing was related, as a tendency, to differences in how each teacher perceives the climate. Empirical finding. The school as a unit explains little of the difference in wellbeing. Design inference. The thesis rests on demands and resources, not on the school explaining everything: in one and the same building, workload and perceived climate may fail to coincide.

Navy and Yonai (2026) followed for two years 27 teachers of science, technology, engineering and mathematics in the United States, novices of one or two years and advanced beginners of more than two, with three questionnaires and interviews. Empirical finding. The trends changed according to the dimension and the group. Support is presented as a buffer; what is established is that heterogeneity, not a prevalence.

Davis and Park (2025), screening 470 records and including 30 studies (2020 to February 2025) on the first five years, found that workload, isolation, interrupted induction and emotional tension persisted and sustained a risk of burnout and attrition. Empirical finding, from thematic synthesis. Quality mentoring, structured induction, support from colleagues, autonomy and individual coping also persisted. Wang et al. (2025) add, over one semester, that time poverty precedes emotional exhaustion and that emotional exhaustion precedes the deterioration of wellbeing, with resilience on the first segment and greater vulnerability of women on the second.

6. Consequences in the classroom: interactions, engagement and motivation

Association is not causation. An observational meta-analysis says whether the sign appears and whether it depends on who reports. It does not say that any fall in burnout raises achievement.

Empirical finding. Wartenberg et al. (2026) meta-analysed 86 studies. Burnout was associated negatively with interactions — emotional support, classroom management and instructional support — and with motivation and achievement, and positively with teacher absenteeism. The negative associations with interactions and motivation varied according to teacher self-report or external report. The abstract does not extend that moderation to achievement and does not publish coefficients. A person who feels burnt out may see their own class as worse: observation or a report from the students is required before the association is taken as a classroom fact already measured.

Chen et al. (2026) reviewed 15 studies from 1995 to 2025, with 3280 teachers and 15 742 students, drawing on emotional contagion, demands and resources, and self-determination. There was no meta-analysis: few studies, heterogeneous measures and incomplete reporting. Empirical finding. The association with student engagement is predominantly negative, with indications of bidirectionality. The longitudinal evidence suggests that student disengagement may precede the increase in burnout. The link was more marked when engagement was reported by the teacher. Design inference. The disruptive behaviour in Collie (2023) and the student challenges in Moens et al. (2026) enter the circuit; they are not a backdrop. Chen et al. (2026) recommend a dual focus and school-level strategies that lower demands, strengthen relational and instructional resources and look after the climate. This is a recommendation by the authors, not an identified effect. The vignette illustrates the circuit and does not measure it.

7. From burnout to attrition: intentions to leave and risk profiles

Madigan and Kim (2021) published a meta-analysis of burnout, job satisfaction and the intention to quit. Source limit. There is no abstract here. Direction, magnitude and number of studies are not asserted. The paper is cited only for the object of the title. What follows comes from other sources.

Empirical finding. In Collie (2023), the intention to leave was associated negatively with relatedness among colleagues and with vitality, and positively with leadership that thwarts autonomy and with time pressure, over and above gender, experience, qualification and personality, and with no interactions. The abstract speaks of intention, not of a resignation already carried out.

Admiraal (2025), with clusters from TALIS 2018 on 5728 novices in 24 European countries, extracted seven profiles according to demands, individual and school resources, and wellbeing. Low risk: those who are doing well (Thriving). Moderate risk: confident (Confident), school-oriented (School-oriented) and overloaded (Overloaded). High risk: little committed (Uncommitted), insecure (Insecure) and unhappy (Unhappy). Empirical finding. Internal scores that the abstract does not give are not described. The implication is differentiated induction, not a single workshop for an average novice.

Cobb-Clark and Nguyen (2026) reviewed 29 Australian studies from 2014 to 2024 and placed the person and their history at the centre of an ecological framework. Empirical finding, from synthesis. The gaps remain, and leadership is left as a moderator. Predictors that the abstract does not name are not listed, and a change of direction is not trialled.

Empirical finding. Engle et al. (2024), with approximately 8910 rural teachers in the United States in the 2020–2021 survey, find that more teacher leadership — above all the non-instructional kind — is associated with greater satisfaction and less burnout, and that this pair is associated with a lower probability of leaving or of changing school, along an indirect path. It is a survey, not an experiment.

Inabnett and Long (2026) surveyed 104 teachers in high-need rural schools in northern Louisiana. Many are exposed to trauma at work. Around one quarter considered it likely that they would leave the school or the profession in the following year. Secondary traumatic stress contributed significantly, in a logistic regression, to the intention to leave the profession. Empirical finding. There is no coefficient published in the abstract and no national rate. In that context, exposure to the trauma of the students behaves as a demand associated with leaving.

8. Individual interventions: mindfulness, compassion and their limits

These interventions train the person. The criterion is not whether they do something, but whether the outcome is burnout, how long the follow-up lasts and what remains outside: the workload.

Empirical finding. Zareian Jahromi et al. (2026) reviewed 21 experimental or quasi-experimental studies with teachers in compulsory education. There are positive effects on general mindfulness, intrapersonal mindfulness and self-efficacy; not on the interpersonal mindfulness of the small set that measured it, and no association between reported fidelity domains and impacts. Nineteen studies were rated as of good quality, with little power in the nulls. The review was not preregistered. Design inference. An internal resource improves, not the relational facet, and the outcome is not burnout.

Paudel et al. (2025) did measure burnout, and the contrast was null. They started from 218 secondary teachers in 37 schools in Kathmandu; at one month 192 remained. The virtual treatment (n = 102) did not separate from the control (n = 116). Emotional exhaustion changed by −0.93 (95% confidence interval: −2.56 to 0.70) against −1.71 (−3.24 to −0.18). Personal accomplishment rose by 1.04 (−0.21 to 2.30) and by 1.53 (0.35 to 2.70). Depersonalisation fell more in the control (−1.97; −2.78 to −1.16). The treatment interval (−0.12; −0.98 to −0.75) does not contain its own estimate, a probable erratum in the abstract, and is not interpreted. Empirical finding. There were no significant differences. The virtual intervention showed no impact. There were small reductions in both arms. The prevalence of high emotional exhaustion and of low accomplishment fell in both; that of high depersonalisation fell only in the control.

Matos et al. (2024) analysed those who completed eight weeks of compassionate mind training in a two-arm stepped trial (n = 103). Empirical finding. The increase in the three flows of compassion and the reduction in fear of compassion — towards oneself and from others — mediated depression, anxiety, stress and burnout. Calm-safeness emotions and activation emotions mediated positive affect, and a partial serial mediation linked burnout to the fall in those fears and in stress. The authors conclude an improvement in wellbeing along that path. The abstract gives neither a between-arm effect nor a follow-up after the post-test. Part of the change passes through other people and through the climate of the work, without redesigning the workload.

Souza Silva and Lopez (2025) tested four asynchronous weeks of mindfulness and self-compassion, in the Neurosaúde application, with thirty basic-education teachers randomised between the intervention and a waiting list. Empirical finding. The time-by-group ANOVA was large, F(2, 26) = 5.316, p = 0.012, η² = 0.290. In the intervention group anxiety fell from pre-test (mean 3.14; deviation 1.67) to post-test (mean 1.55; deviation 1.49; p = 0.022). The aim included stress, depression and maintenance; what is established is pre–post anxiety. Burnout was not measured.

Janssen et al. (2023) offer the longest follow-up: a cluster trial in five vocational schools in the Netherlands, with mindfulness, mindfulness plus organisational health, and a waiting list. The add-on showed no effects and the arms were merged. The primary outcome was mindfulness. In the short term the total score, observing, non-reactivity and dedication improved. At three and nine months the total score, observing, non-reactivity and non-judging were sustained, together with sleep, negative emotions and work–home interaction. Empirical finding. Performance, competences and perceptions of the work did not improve, and perceived demands did not fall. Organisational commitment fell more in the intervention. Mental health did improve. The idea that mindfulness made it possible to disconnect in free time is an interpretation by the authors, not a tested mediator.

Design inference. There are real and bounded improvements. The three-dimension trial did not separate the programme from the control, several outcomes are not burnout, and the samples of thirty and of one hundred and three weigh. Without a fall in demands, the training is a complement. The workshop and the application in the vignette live here.

9. Organisational interventions: leadership, collective efficacy and workload

An intervention is organisational if the target is the shared work: leadership, norms, collective efficacy, climate or workload. Delivering mindfulness in a school does not make it organisational.

Empirical finding. In Janssen et al. (2023) organisational health added to mindfulness showed no effects, and the arms were merged. There was cluster randomisation, five schools and follow-up at nine months. Perceived demands did not fall and organisational commitment fell more in the intervention. Design inference. Redesigning the post does not rest on a victory for that arm, but on the fact that neither mindfulness nor the add-on touched the demands that the corpus associates with burnout and with leaving.

Sahli Lozano et al. (2025) synthesise collective efficacy, not a programme: 49 studies from 1999 to 2024 and, of those, only ten with multilevel models. In 44 studies and 32 053 teachers, at the individual level, it was related to fewer negative outcomes — burnout and distress, effects from small to medium — and to more positive ones — wellbeing and satisfaction, from small to large. Design inference. It is scarcely modelled at the level at which it is defined. It is not yet a demonstrated school-level lever.

Lambert et al. (2023), in a mixed study whose truncated abstract records 1850 teachers, associate the quality and the stability of leadership with demands, resources, satisfaction, instructional support and vulnerability to stress. Engle et al. (2024) locate the resource in teacher leadership, above all the non-instructional kind, with approximately 8910 rural teachers: it is a survey, not a manual. Ruttledge (2022), exploratory and qualitative in one secondary school, suggests support, greater self-efficacy and less stress, and calls for controls and follow-up. There is no quantified effect.

Admiraal and Kittelsen Røberg (2023) do not trial an intervention: in TALIS 2018 a safe climate and a collaborative culture were the principal resources, with a similar pattern across stages. The abstract does not say how that climate is built.

Type	Level	Example from the corpus	What it shows	What it does not show
Mindfulness	Individual	Zareian Jahromi et al. (2026), 21 studies	Effects on mindfulness and self-efficacy	An interpersonal effect; a burnout outcome
Therapy with mindfulness	Individual	Paudel et al. (2025), 218/192	Small decreases in both arms	A difference between arms
Compassionate mind	Individual	Matos et al. (2024), n = 103	Mediation towards burnout	A between-arm effect and a fall in workload
Online programme	Individual	Souza Silva and Lopez (2025), n = 30	A pre–post fall in anxiety	A burnout outcome
Stress reduction with mindfulness	Individual	Janssen et al. (2023), nine months	Improvements in mindfulness, sleep and dedication	A fall in demands
Added organisational health	Organisational	Janssen et al. (2023), five schools	The add-on showed no effects	Superiority over mindfulness
Collective efficacy	School resource	Sahli Lozano et al. (2025), 32 053 teachers	Less distress and more wellbeing	Multilevel models in the majority; that it is an intervention
Leadership	Organisational	Lambert et al. (2023); Engle et al. (2024)	Quality, stability and teacher leadership associated with less strain	A trial that changes the leadership
Whole school	Organisational	Ruttledge (2022)	Less stress in a qualitative account	A control and follow-up
Climate and collaboration	Resource	Admiraal and Kittelsen Røberg (2023)	Principal resources in TALIS 2018	An intervention that manipulates them

Design inference. Those resources are association. What has been trialled is an add-on with no effect and a single-school programme, without a control. Redesigning the work is not a package ready to be copied.

10. Latin America and Mexico: scarce evidence and cautious transfer

Empirical finding, on the composition of the corpus. The only Latin American study is Souza Silva and Lopez (2025): thirty basic-education teachers, an online programme and a fall in anxiety, not a map of burnout. The query “teacher occupational stress Latin America” contributed no other solid empirical evidence. There are no Mexican data in these 24 sources. The five notes that follow are inferences for schools in Mexico, not findings and not prevalences.

Inference for Mexican schools, not a corpus finding. Administrative workload. Workload and time pressure are among the most reported demands (Moens et al., 2026), and time poverty affects wellbeing by way of emotional exhaustion (Wang et al., 2025). The duplicated entry of attendance, grades and memoranda is a candidate for that demand. Removing redundancy acts on the post; it is not the accessory of an application. Mexican hours are not measured.

Inference for Mexican schools, not a corpus finding. Leadership rotation. Lambert et al. (2023) associate stress with the quality and the stability of leadership, taken together. A handover can preserve competence and lose stability. It is not asserted that leadership rotates a great deal in Mexico: it is asserted that the handover is an event of the work. Leadership that thwarts autonomy was associated with leaving in Australia (Collie, 2023), not in a Mexican sample.

Inference for Mexican schools, not a corpus finding. School technical council. Collaboration and a safe climate are central resources (Admiraal & Kittelsen Røberg, 2023; Moens et al., 2026). The council can be that space if it is protected for teaching and for coexistence, and another demand if it is filled with reports. The collective efficacy attributed to it is poorly studied as a group phenomenon (Sahli Lozano et al., 2025).

Inference for Mexican schools, not a corpus finding. Induction. There are seven profiles of novices (Admiraal, 2025). Workload, isolation and interrupted induction survived the pandemic, and mentoring and structured induction protect (Davis & Park, 2025). The dimensions of burnout do not move together (Navy & Yonai, 2026). An identical course for all new staff does not answer that map. This corpus does not describe Mexican induction.

Inference for Mexican schools, not a corpus finding. Secondary traumatic stress. There is exposure to the trauma of the students, around one quarter who were considering leaving within the year, and a contribution of that stress to leaving the profession (Inabnett & Long, 2026). If a student does not want to go back home, the workshop of the person who listened is not enough: notification, protection and support are required. The vignette illustrates the gap. The Louisiana fraction is not a Mexican rate.

Souza Silva and Lopez (2025) do not turn mindfulness into regional policy by being the only local study: their outcome is anxiety. The individual workshop is not copied for lack of data, and evidence from other countries is not treated as a measurement of the Mexican school.

11. Discussion: redesigning work, not only the teacher

If the answer is a workshop and an application, the teacher is left in charge of regulating what the post produces. There are improvements in attention, self-efficacy, compassion, anxiety, sleep or dedication (Zareian Jahromi et al., 2026; Matos et al., 2024; Souza Silva & Lopez, 2025; Janssen et al., 2023), and a null result when the outcome is burnout (Paudel et al., 2025). None of them shows fewer demands.

Element	Role	Status in this review
Demands: workload, time, students, disruptive behaviour, secondary traumatic stress	Antecedent	Framework, with associations and one mediation
Emotional exhaustion	Intermediate link	It does not exhaust the dimensions of the inventory
Interactions, motivation and engagement	Associated outcomes	A possible reverse path; sensitive to the informant
Intention to leave	Associated outcome	It is not a resignation that has already occurred
Leadership, collaboration, collective efficacy, climate, support	Resources of the work	Association; Collie (2023) did not retain the interaction
Mindfulness, compassion, brief therapy	Complement	They do not replace the redesign

Schulze-Hagenest et al. (2023) found little between-school variance in emotional exhaustion and satisfaction, substantial variation in interactions and a tendential relation with perceived climate. Classifying schools as healthy and as toxic contradicts that datum. What is redesigned is the post — class groups, time, behaviour, secondary trauma, stable leadership and collaboration — not the average of the building.

The second brake is one of evidence: Collie (2023) did not retain interactions, Janssen et al. (2023) found no effect of the organisational add-on, Sahli Lozano et al. (2025) scarcely model collective efficacy as a group, and Ruttledge (2022) is qualitative and from a single school. Design inference. What is modest is to lower redundant workload, protect time, treat disruption and secondary trauma as requirements of the system, and look after leadership, collaboration and climate, with personal training as a complement. Wartenberg et al. (2026) and Chen et al. (2026) leave the students inside the same problem. That is not a Mexican baseline.

12. Limitations

The base is Crossref alone: 16 queries of 60 rows, 459 records, 381 eligible and 24 included by hand, with no double coding and no meta-analysis of its own. Madigan and Kim (2021) have no abstract. Saloviita and Pakarinen (2021) do not give the sign. The abstract of Lambert et al. (2023) arrives truncated. The treatment interval for depersonalisation in Paudel et al. (2025) does not contain its estimate and was not used as a finding. Several intervention outcomes are not burnout and several follow-ups are short.

High-income countries predominate, with exceptions that do not balance the map: China, Nepal and Brazil. The Latin American evidence is almost null and the direct Mexican evidence is null. The five implications above are inferences. Magnitude, consequences and attrition are above all correlational or cross-sectional, even though there are multilevel models, three waves, meta-analyses of associations and a few trials. The vignette is composite. It is not a field case.

13. Conclusions

Teacher burnout is better understood as a condition of the work than as a personal deficit. Workload, time, disruptive behaviour, secondary traumatic stress and leadership that thwarts autonomy are associated with worse wellbeing or with leaving; support, collaboration, stable leadership, climate and collective efficacy, with the opposite side. Schulze-Hagenest et al. (2023) prevent the claim that the school average explains everything.

The strain is associated with poorer interactions, lower motivation and achievement, greater teacher absenteeism and lower engagement. Part of this depends on who reports, and student disengagement may precede burnout. The intention to leave accompanies profiles of novices, leadership and secondary trauma. Madigan and Kim (2021), having no abstract, are not cited for their figures.

Individual interventions improve mindfulness, self-efficacy, anxiety or compassion pathways, and at nine months sleep, emotions and dedication, without lowering demands. The trial on the three dimensions did not separate the programme from the control, and organisational commitment fell more in the treated group. They are a complement, not a substitute. Leadership, climate and collective efficacy stand better as association than as a programme: the organisational add-on contributed no effect and the whole-school study is qualitative and from a single school. Redesigning workload, time, routes in the face of trauma and collaboration is a hypothesis, not a known mean effect.

For Mexico there is no baseline in these 24 sources: the only Latin American study measures anxiety in thirty teachers. Paperwork, rotation, the technical council, induction and secondary traumatic stress are inferences, so that the answer is not a workshop. Confirming them requires local research. Until then the missing figure is not fabricated.

NEXTECH.IA Editorial Lab / Ingeniero Mitre

References

1. Admiraal, W. (2025). Attrition risk profiles of novice teachers based on their job demands, individual and school resources, and well-being. Teaching and Teacher Education, 162, 105070. https://doi.org/10.1016/j.tate.2025.105070
2. Admiraal, W., & Kittelsen Røberg, K.-I. (2023). Teachers’ job demands, resources and their job satisfaction: Satisfaction with school, career choice and teaching profession of teachers in different career stages. Teaching and Teacher Education, 125, 104063. https://doi.org/10.1016/j.tate.2023.104063
3. Chen, X., Zhu, R., & Zhang, L. (2026). The relationship between teacher burnout and student engagement: A systematic review of empirical research (1995–2025). International Journal of Educational Research, 140, 103085. https://doi.org/10.1016/j.ijer.2026.103085
4. Cobb-Clark, M.-A., & Nguyen, H. T. M. (2026). A systematic review of risk and protective factors for teacher attrition and intent to leave the workforce in the Australian context. Teachers and Teaching. https://doi.org/10.1080/13540602.2026.2630394
5. Collie, R. J. (2023). Teacher well‐being and turnover intentions: Investigating the roles of job resources and job demands. British Journal of Educational Psychology, 93(3), 712–726. https://doi.org/10.1111/bjep.12587
6. Davis, T., & Park, E. (2025). A Systematic Review of Early-Career Teacher Wellbeing, Stress, Burnout and Support Mechanisms During and Post COVID-19 Pandemic. Education Sciences, 15(8), 996. https://doi.org/10.3390/educsci15080996
7. Engle, J., Xia, J., & Butler, S. (2024). Teacher Leadership, Wellbeing, and Intent to Leave in US Rural Schools: Evidence from the 2020–21 National Teacher and Principal Survey. Education Sciences, 14(7), 758. https://doi.org/10.3390/educsci14070758
8. Inabnett, M. S., & Long, A. C. J. (2026). Understanding secondary traumatic stress and its impact on teacher retention. School Psychology, 41(4), 480–486. https://doi.org/10.1037/spq0000737
9. Janssen, M., Heerkens, Y., Van der Heijden, B., Korzilius, H., Peters, P., & Engels, J. (2023). Effects of mindfulness-based stress reduction and an organizational health intervention on Dutch teachers’ mental health. Health Promotion International, 38(3), Article daac008. https://doi.org/10.1093/heapro/daac008
10. Klusmann, U., Aldrup, K., Roloff-Bruchmann, J., Carstensen, B., Wartenberg, G., Hansen, J., & Hanewinkel, R. (2023). Teachers’ emotional exhaustion during the COVID-19 pandemic: Levels, changes, and relations to pandemic-specific demands. Teaching and Teacher Education, 121, 103908. https://doi.org/10.1016/j.tate.2022.103908
11. Lambert, R., McCarthy, C., Weppner, C., Malerba, C., Osman, D., Holcomb, T. S., & Bottoms, B. L. (2023). Associations between teacher stress and school leadership: A mixed methods study with implications for school psychologists. School Psychology, 38(6), 370–384. https://doi.org/10.1037/spq0000601
12. Madigan, D. J., & Kim, L. E. (2021). Towards an understanding of teacher attrition: A meta-analysis of burnout, job satisfaction, and teachers’ intentions to quit. Teaching and Teacher Education, 105, 103425. https://doi.org/10.1016/j.tate.2021.103425
13. Matos, M., Galhardo, A., Palmeira, L., Albuquerque, I., Cunha, M., Lima, M. P., Maratos, F. A., & Gilbert, P. (2024). Promoting Teachers’ Wellbeing Using a Compassionate Mind Training Intervention: Exploring Mechanisms of Change. Mindfulness, 15(6), 1346–1361. https://doi.org/10.1007/s12671-024-02360-3
14. Moens, M., Vanblaere, B., Devos, G., & Tuytens, M. (2026). Teacher wellbeing in schools: a systematic review of job demands and job resources. Frontiers in Education, 11, Article 1709029. https://doi.org/10.3389/feduc.2026.1709029
15. Navy, S. L., & Yonai, E. (2026). Early Career STEM Teacher Burnout: Trends and Explanations. Journal of Teacher Education, 77(5), 448–463. https://doi.org/10.1177/00224871261477686
16. Paudel, N. R., K. C., P., Nygård, C.-H., & Neupane, S. (2025). Effect of a mindfulness-based cognitive behavior therapy intervention on occupational burnout among school teachers. Frontiers in Psychiatry, 15, Article 1496205. https://doi.org/10.3389/fpsyt.2024.1496205
17. Ruttledge, R. (2022). A whole school approach to building relationships, promoting positive behaviour and reducing teacher stress in a secondary school. Educational Psychology in Practice, 38(3), 237–258. https://doi.org/10.1080/02667363.2022.2070456
18. Sahli Lozano, C., Wicki, M., Wüthrich, S., & Setz, F. (2025). A systematic review and meta-analysis of Collective Teacher Efficacy's relationships with outcomes in the Job Demands-Resources model. Teaching and Teacher Education, 159, 105006. https://doi.org/10.1016/j.tate.2025.105006
19. Saloviita, T., & Pakarinen, E. (2021). Teacher burnout explained: Teacher-, student-, and organisation-level variables. Teaching and Teacher Education, 97, 103221. https://doi.org/10.1016/j.tate.2020.103221
20. Schulze-Hagenest, T., Carstensen, B., Weber, K., Jansen, T., Meyer, J., Köller, O., & Klusmann, U. (2023). Teachers’ emotional exhaustion and job satisfaction: How much does the school context matter? Teaching and Teacher Education, 136, 104360. https://doi.org/10.1016/j.tate.2023.104360
21. Souza Silva, L. J., & Lopez, L. C. S. (2025). Effects of an online mindfulness intervention on anxiety levels of basic education teachers: Randomized trial. Psicologia - Teoria e Prática, 27(2), 1–21. https://doi.org/10.5935/1980-6906/eptpcp17429.en
22. Wang, K., Huang, T., Lin, X., & Tan, R. (2025). Time poverty and teacher subjective well-being in an accelerated society: a three-wave longitudinal analysis of emotional exhaustion, psychological resilience, and gender. Frontiers in Psychology, 16, Article 1728219. https://doi.org/10.3389/fpsyg.2025.1728219
23. Wartenberg, G., Aldrup, K., Grund, S., & Klusmann, U. (2026). How Strongly Is Teachers’ Burnout Related to Teacher Absenteeism, Teacher–Student Interactions, and Student Motivation and Achievement: A Meta-Analysis and Systematic Review. Review of Educational Research. https://doi.org/10.3102/00346543261455273
24. Zareian Jahromi, A., Brown, J. L., & Jennings, P. A. (2026). The Impacts of Mindfulness-Based Interventions (MBIs) on Teachers’ Mindfulness, Mindful Teaching, and Self-Efficacy: A Systematic Review. Mindfulness, 17(3), 643–666. https://doi.org/10.1007/s12671-025-02749-8