Introduction

Burnout is defined by the World Health Organization (WHO) as “a syndrome conceptualized as resulting from chronic workplace stress that has not been successfully managed” and is characterized by the following three dimensions: high emotional exhaustion, high depersonalisation and low personal accomplishment1. Emotional exhaustion is considered the key dimension and refers to a lack of energy and a depletion of an individual’s emotional resources2. Depersonalisation refers to taking mental distance from work by developing a cynical attitude towards the job and colleagues, whereas personal accomplishment is described as feeling competent about the job2. In 2021, 45–66% of the Europeans experienced or felt on the verge of burnout, which means that they experience persistent high burnout levels but are still able to work3. As employees with high burnout levels show decreased cognitive functioning, lower productivity and effectiveness at work, job dissatisfaction, reduced commitment to the job or the organization, presenteeism and absenteeism2,4, burnout is an increasing concern in today’s workplaces. A recent report on expenses on long-term disability caused by burnout in Belgium showed that half a billion euros were spent in 2019, which is an increase of 22% since 20165. Besides its occupational and financial impact, burnout is also found to be related with several health consequences such as cardiovascular diseases, type 2 diabetes and all-cause mortality6,7.

High recovery need, the feeling of longing for relief from work stressors and the expectation that this relief is inevitable in order to be able to cope with future work stressors, can be regarded as a pre-phase of burnout. Research shows that burnout develops as the result of exposure to work (and life) stressors and insufficient recovery (i.e., the inability to replenish lost resources and gain new ones)8,9. A study by Soderstrom, Jeding10 even discussed that insufficient recovery from work stress, and not the work stress itself, is the most important predictor of the development of burnout. A higher need for recovery thus reflects that resources, such as energy, are lost and reloading is needed for the next work day to prevent a further loss of resources, which may eventually lead to burnout11. Hence, taking into account recovery need when conducting research on burnout can lead to a better understanding what can be done to early prevent this syndrome.

Teaching is known as a highly stressful job, resulting in higher risk of burnout12,13,14. Research namely showed that in the United Kingdom, the teaching profession is one of the most stressful among 26 occupations14, while in Finland it was found that teachers experience stress and burnout more often compared to other professions (12% vs. 8%, respectively)13. In Flanders (Belgium) it was recently found that 21% of the teachers reported burnout symptoms, compared to 13% in the general population12. In addition, a recent meta-analysis showed that especially secondary schoolteachers are at high risk of burnout, showing the highest levels of sick leave due to work-related stress15,16,17,18,. Risk factors of teacher burnout include decreased autonomy, work overload, overcrowded classrooms, conflicts with supervisors, co-workers and parents, misbehaviour of students and limited possibilities regarding promotion and professional development16,17,18,19,20. As high burnout levels in teachers result in lower job performance, increased absenteeism, higher turnover rates, decreased student–teacher interaction, decreased ability in dealing with classroom behaviour problems and decreased student achievement4,20,21,22,23,25, it is an increasing concern for educational institutions.

Given the high prevalence of (teacher) burnout and its severe consequences, it is important to investigate how recovery need and burnout risk can be reduced or even prevented. Physical activity (PA), defined as “any bodily movement produced by skeletal muscles that requires energy expenditure”26, has the potential to reduce recovery need and burnout risk. Several psychological and physiological mechanisms are put forward for the potential negative relationships between PA and recovery need and burnout risk. Psychologically, research shows that PA has immediate effects on recovery from work, such as increased feelings of detachment, relaxation and mastery26,27,28,29, thereby reducing employees’ need for recovery. Regarding physiological working mechanisms, physically fit employees are better able to cope with work stressors, thereby reducing their need for recovery and preventing the probability of developing burnout (i.e., cardiovascular fitness hypothesis)30. Moreover, PA increases cerebral blood flow, upregulates neurotrophic factors (e.g., brain-derived neurotrophic factor; BDNF), and supports cognitive functions such as executive functions (e.g., planning and sequencing), all contributing to employees who are more resilient to unfavorable effects of work stress31,32.

A randomized controlled trial indeed showed an improvement in need for recovery at follow-up (i.e., six and twelve weeks after the intervention) after following a six weeks low-intensity running program33. Furthermore, Coffeng, van Sluijs34 concluded that stair climbing, leisure activities, (physical) detachment at work, relaxation and detachment after work are associated with a lower need for recovery. However, a study by Stevens, Crowley35 demonstrated a cross-sectional positive association between occupational physical activity and need for recovery (showing small effect sizes), with a trend towards a positive association between vigorous behaviours at work and recovery need. This indicates that the role of PA on recovery need may depend of the domain in which it is performed. As regards burnout, two systematic reviews36,37 showed that PA may constitute an effective approach to reduce burnout risk. Following a systematic review by Verhavert, De Martelaer36, positive effects on burnout risk were achieved when being physically active two to five times per week for 20 to 60 min, for six to 18 weeks, showing reductions in burnout risk between 6.9 and 41.3%36. The latter review included studies with a large variety in type, intensity, duration and frequency of the performed PA activities and recommended future research to further unravel the effects of PA intensities and PA domains in the PA-burnout relationship36. Besides the greater physical benefits (e.g., regarding cardiorespiratory functions and mortality) of vigorous-intensity PA (VPA) compared to moderate-intensity PA (MPA)38,39, it is demonstrated that VPA is also more beneficial than MPA regarding mental health, showing lower stress levels and a more favourable sleep pattern, which in turn may positively influence recovery need and burnout40. A study by Elliot, Lang41 in vocational students showed that only VPA was associated with reduced burnout. However, Lindwall, Ljung42 concluded that both employees reporting light-intensity PA (LIPA) and employees reporting moderate-to-vigorous-intensity PA (MVPA) were less likely to show elevated scores on burnout compared to inactive employees, with no differences between LIPA and MVPA. It should be mentioned that Elliot, Lang41 did not take PA domains into account, whereas Lindwall, Ljung42 only assessed PA during leisure-time. Besides PA intensity, it is also important to get more insight into the role of the PA domains (i.e., work-related PA, transport-related PA, leisure-time PA, domestic and garden PA)43. Research shows that particularly engaging in leisure-time PA is beneficial for mental health, and that physical activity in other domains (work, transport, and domestic/garden) are less beneficial or even detrimental for mental health44,45. As regards burnout, it was shown that leisure-time PA may be less effective to reduce burnout risk in employees with a physically demanding job46. Although differentiating PA by domain and intensity is gaining attention in related literature47,48, to the best of our knowledge, no previous studies investigated the relationship of different PA intensity-domain combinations with burnout risk among teachers. Therefore, we aimed to assess the association of different PA intensity-domain combinations with recovery need and burnout risk, respectively, in Flemish secondary schoolteachers.

Materials and methods

Study design and participants

In this cross-sectional study, a non-probability cluster sampling strategy was applied to recruit participants. All secondary schools in Flanders (n = 1075) were contacted through e-mail and telephone in August and September 2019, and were asked to forward a link to the online questionnaire via e-mail to their entire teaching staff. To increase response rate, the Flemish Department of Education (“Vlaams Departement Onderwijs”), as well as all education networks (i.e., schools owned by the Flemish community, subsidized public schools and subsidized free schools), were asked to promote our study through their social media and newsletters. A convenient selection of schools (n = 43) were also visited to promote our study face-to-face. Teachers in sick leave due to illness (except for burnout), teachers not being able to be physically active (due to injuries or physical disabilities) and non-teaching employees working in secondary schools (i.e., administrative staff and principals) were excluded from the sample.

Measures

The last week of September and the first week of October 2019 (i.e., beginning of the school year; in Flanders, school years begin on the 1st of September and last until the 30th of June), teachers were asked to complete a self-report questionnaire (developed by using the web-based software Qualtrics (Qualtrics, Provo, UT)) including questions on recovery need, burnout, PA, alongside socio-demographics (i.e., age, sex, marital status, number of children (if any), ethnicity, diploma), anthropometrics (i.e., self-reported height and weight) and work-related factors (e.g., questions about the weekly amount of teaching and in which school(s) they were working). During the measurement period of two weeks, two reminders were sent to the non-responders, i.e., on the fourth and the eighth day after activation of the online questionnaire.

Recovery need

Recovery need was assessed using the validated Short Inventory to Monitor Psychosocial Hazards (SIMPH)49. Only the ‘recovery need’ part of this particular questionnaire was used, including five items (e.g., “I find it difficult to relax at the end of a working day”) with a two-point dichotomous response scale (i.e., yes (= 1) or no (= 0)). A total score was calculated by summing the item scores. Participants having a score ≥ 3 were classified as having a ‘high need for recovery’. The ‘recovery need’ part of the SIMPH has good psychometric properties49.

Burnout risk

Risk of burnout was assessed using the Utrecht Burnout Scale for teachers (UBOS-L)50, which is based on the validated Maslach Burnout Inventory (MBI)51. The UBOS-L is the Dutch version of the original MBI and was especially developed for teachers, administrators and other staff members working in educational settings. The survey assesses three dimensions of burnout and consists of 22 items: eight items assessing emotional exhaustion (e.g., “I feel mentally exhausted because of my job), seven items assessing depersonalisation (e.g., “I don’t really care what happens to some students”), and seven items assessing personal accomplishment (e.g., “I deal very effectively with the problems of my students”), all presented on a seven-point Likert scale ranging from 0 (i.e., never) to 6 (i.e., every day). The items for each burnout dimension were averaged to obtain an average score with a range from 0 to 6. Higher scores indicate more emotional exhaustion, feeling more depersonalized (i.e., more cynical about their job) and higher levels of personal accomplishment (i.e., feeling more competent). Individuals scoring high on emotional exhaustion (i.e., > 2.5) and low on personal accomplishment (i.e., < 3.56), or high on emotional exhaustion (i.e., > 2.5) and high on depersonalisation (i.e., > 1.43 for men and > 2.00 for women), are considered at risk for burnout based on the UBOS-L norms50. This validated questionnaire has very good to excellent psychometric properties50.

Physical activity

An adapted version of the validated International Physical Activity Questionnaire (IPAQ – Dutch version) was used to estimate PA levels during the last seven days52. This self-report questionnaire assesses three intensities of PA (i.e., walking, MPA and VPA) across four PA domains (i.e., work-related, transport-related, leisure-time, domestic and garden PA) As a meta-analysis by Shiue53 showed that gardening is beneficial for mental health, while a meta-analysis by White, Babic44 did not find a relationship between household PA and mental health, it was decided to split up domestic and garden PA and include those as two independent domains in the statistical models. Based on the assessed intensities and domains, the following ‘PA intensity-domain combinations’ were obtained: work-related walking, work-related MPA, work-related VPA, transport-related walking, transport-related MPA, leisure-time walking, leisure-time MPA, leisure-time VPA, domestic PA at moderate-intensity, and garden PA at moderate-intensity and at vigorous-intensity. Participants were asked to report the number of days they engaged in the particular PA (i.e., particular PA intensity-domain combination) and the average amount of time spent in the particular PA on such a day. For the purpose of this study, PA intensity-domain combinations, expressed in hours/week, were used. Total PA per week was calculated by summing the amount of PA within each PA intensity and domain. Following the IPAQ scoring protocol, several criteria were applied when cleaning the IPAQ data54; only durations of minimum ten minutes of activity were retained, non-relevant observations were excluded (e.g., answering in step counts instead of minutes), and PA levels higher than 960 min/day (i.e., 16 h/day) were excluded due to being unrealistic. More details about the scoring protocol can be found elsewhere54. The IPAQ has fair-to-good psychometric properties55.

Statistical analyses

All statistical analyses were performed using R (R core Team, 2019; R Studio version 3.6.2). Prior to the analyses, all data were screened for outliers and data cleaning was performed. Normality of the outcome measures (i.e., recovery need and burnout dimensions) was checked using histograms and QQ-plots. Representativeness of the sample was assessed by conducting two proportions z-tests, comparing our sample against the population of Flemish secondary schoolteachers on sex, age, and education network56. Preliminary analyses checked if a two-level model was advised by inspecting the amount of variance explained by the school-cluster. Intraclass correlation coefficients (ICCs) were calculated, and following values were found: 0.0500 for recovery need, 0.0004 for emotional exhaustion, 0.0090 for depersonalisation, and 0.0040 for personal accomplishment. As hardly any variance was explained by the school-cluster, it was not taken into account in the statistical models. Multiple linear regression models were applied with the lm() function of the lme4-package57 to identify which PA intensity-domain combination significantly predicted the score on recovery need and on each burnout risk dimension. Since all PA intensity-domain combination could be assumed to be confounders or mediators of the relationship between PA type and recovery need/burnout, and because models should be adjusted for confounders only and not for mediators58, two kinds of models were applied. In the first model (i.e., step-1 model), it was assumed that the relationship between a certain PA intensity-domain combination and recovery need/burnout risk is mediated by the other PA intensity-domain combinations, and thus this model should not be adjusted for the other PA intensity-domain combinations58. In the second model (i.e., step-2 model), it was assumed that the relationship between a certain PA intensity-domain combination and recovery need/burnout risk is confounded by the other PA intensity-domain combinations, and therefore, this model should be adjusted for the other PA intensity-domain combinations58. This means that step-1 models included only one PA intensity-domain combination (i.e., walking at work, work-related MPA, work-related VPA, transport-related walking, transport-related MPA, leisure-time walking, leisure-time MPA, leisure-time VPA, MPA during domestic PA, MPA during gardening and VPA during gardening), in addition to sex, age, being a parent and weekly amount of teaching. The step-2 models included all PA intensity-domain combinations, adjusted for sex, age, being a parent and weekly amount of teaching. The above-described models were applied separately to each of the four outcome measures (i.e., recovery need, emotional exhaustion, depersonalisation and personal accomplishment). The estimates of each model can be interpreted as the average difference in scale points on the outcome variable (e.g., emotional exhaustion) per one-hour difference per week of the specific physical activity. Multicollinearity was checked using the Variation Inflation Factor (VIF), with multicollinearity considered present when VIF > 10. P-values < 0.05 were considered statistically significant. Next to statistical significance, the relevance of the findings based on the magnitude of the estimates was discussed.

Ethics approval

This study was conducted following the principles of the Declaration of Helsinki and its later amendments. Prior to study enrolment, participants signed a written informed consent. Approval for the study was granted by the Medical Ethics Committee of the University Hospital (UZ Brussel, Brussels, Belgium; B.U.N. 143,201,940,533).

Results

In total, 2220 Flemish secondary schoolteachers completed the online questionnaire. After excluding participants based on the predefined exclusion criteria (i.e., not working in secondary education (n = 24), sick leave not due to burnout (n = 23), an adapted PA and/or dietary pattern (n = 175), missing values for IPAQ (n = 31) and being an outlier for total PA (n = 58)), the final sample consisted of 1909 teachers (retention rate: 85.99%).

Sample characteristics

The final study sample consisted of 1909 secondary schoolteachers (77.3% women) with a mean age of 41.1 ± 10.3 years and a mean BMI of 24.9 ± 4.6 kg/m2 (see Table 1).

Table 1 Study sample characteristics.
Full size table

Representativeness of the sample

The sample of this study was compared to the general secondary schoolteacher population in Flanders to assess representativeness56 (see Table 2). In comparison to the general teacher population, our sample included more women (77.3% vs. 64.9%; p < 0.001), less schoolteachers in the age group + 60 (3.2% vs. 5.8%; p < 0.001), more schoolteachers of the Flemish community school network (51.2% vs. 22.5%; p < 0.001), and less schoolteachers of the subsidised free school network (45.4% vs. 68.0%; p < 0.001) and the subsidised public school network (3.4% vs. 9.4%; p < 0.001).

Table 2 Representativeness of the study sample compared to the general secondary teacher population in Flanders.
Full size table

Association of walking with recovery need and burnout risk

The step-1 model regarding walking at work showed a significant positive association with personal accomplishment (see Table 3 – part 2) (p = 0.035), demonstrating that one hour per week more walking at work is associated with on average a 0.004 scale-points higher score on personal accomplishment. Walking at work was not significantly associated with recovery need (see Table 3 – part 1), emotional exhaustion (see Table 3 – part 1), depersonalisation (see Table 3 – part 2) and personal accomplishment (step-2 model only). Regarding walking during transport, only the step-1 model showed a significant positive association with personal accomplishment (p = 0.049), demonstrating that one hour per week more walking during transport is associated with on average a 0.011 scale-points higher score on personal accomplishment. No significant associations were found between walking during transport and recovery need, emotional exhaustion, depersonalisation and personal accomplishment (step-2 model only). Walking during leisure-time was positively associated with depersonalisation in the step-1 model (p = 0.016). More specifically, the step-1 model showed that one additional hour per week of walking during leisure-time was associated with on average a 0.013 scale-points higher score on depersonalisation. No significant associations were found between walking during leisure-time and recovery need, emotional exhaustion, depersonalisation (step-2 model only) and personal accomplishment.

Table 3 Physical activity related correlates of recovery need and/or burnout risk in secondary schoolteachers.
Full size table

Association of moderate-intensity physical activity with recovery need and burnout risk

The step-1 model regarding MPA during domestic activities showed a significant positive association with emotional exhaustion (see Table 3 – part 1) (step-1 model: p = 0.031), demonstrating that one hour per week more MPA during domestic activities is associated with on average a 0.010 scale-points higher score on emotional exhaustion. No significant associations were found between MPA during domestic activities and recovery need (see Table 3 – part 1), emotional exhaustion (step-2 model only), depersonalisation (see Table 3 – part 2) and personal accomplishment (see Table 3 – part 2). Moderate-intensity gardening showed a significant positive association with recovery need in both the step-1 and step-2 models (step-1 model: p = 0.010; step-2 model: p = 0.022). More specifically, the step-2 model showed that one hour per week more gardening at moderate-intensity is associated with on average a 0.022 scale-points higher score on recovery need. Only the step-1 model showed a significant positive association between moderate-intensity gardening and depersonalisation (step-1 model: p = 0.023), whereas MPA during gardening did not show significant associations with emotional exhaustion, depersonalisation (step-2 model only) and personal accomplishment. MPA at work, during transport and during leisure-time were not significantly associated with recovery need, emotional exhaustion, depersonalisation, and personal accomplishment.

Association of vigorous-intensity physical activity with recovery need and burnout risk

VPA during leisure-time was significantly negatively associated with recovery need (see Table 3 – part 1) and emotional exhaustion (see Table 3 – part 1) in both the step-1 and step-2 models (recovery need: step-1 model: p = 0.002; step-2 model: p = 0.008; emotional exhaustion: step-1 model: p = 0.007; step-2 model: p = 0.011). More specifically, the step-2 model showed that one hour per week more leisure-time PA at vigorous-intensity is associated with on average 0.028 scale-points and 0.019 scale-points lower scores on, respectively, recovery need and emotional exhaustion. VPA during leisure-time was not significantly associated with depersonalisation (see Table 3 – part 2) and personal accomplishment (see Table 3 – part 2). Regarding vigorous-intensity gardening, only the step-1 model showed a significant positive association with personal accomplishment (step-1 model: p = 0.042), whereas no significant associations were found with recovery need, emotional exhaustion, depersonalisation, and personal accomplishment (step-2 model only). VPA at work did not show any significant associations with recovery need, emotional exhaustion, depersonalisation and personal accomplishment .

Discussion

The present study examined the associations between different PA intensity-domain combinations, recovery need and burnout risk in secondary schoolteachers in Flanders (Belgium). First of all, the generally high levels of recovery need and burnout risk (55.4% and 25.5%, respectively) in the present sample of secondary schoolteachers, highlight once again the importance of investigating ways to prevent or alleviate these symptoms. Moreover, our findings showed that a distinction between PA intensity-domain combinations is relevant in predicting recovery need and risk of burnout. It was found that more VPA during leisure-time was associated with less recovery need and less emotional exhaustion. However, contrary to the idea that PA is beneficial for recovery need and burnout risk, higher levels of MPA in the garden domain was associated with more recovery need. Due to the cross-sectional study design, we cannot identify which other PA intensity-domain combinations should be considered confounders or mediators when examining the association of one specific PA intensity-domain combination with recovery need and burnout risk. Consequently, we cannot establish whether the results of the step 1 or step 2 models are the most valid. Therefore, below, we will only focus on the predictors showing significant associations in both step-1 and step-2 models.

Even though teachers can have a physically demanding job59 (e.g., in the present study teachers reported to be about 11 h per week physically active at work), and contrary to the finding of de Vries and Bakker46 that leisure-time PA may be less effective in reducing burnout risk in people with physically demanding jobs, our study demonstrated that higher levels of VPA during leisure-time were associated with less recovery need and less emotional exhaustion in secondary schoolteachers. This means that VPA during leisure-time may play a role in early prevention and reduction of recovery need and burnout risk. The contradiction with the results of de Vries and Bakker46 may be explained by the fact that despite the high PA levels at work in our sample, the majority of these physical activities were performed at a low-to-moderate intensity, and thus not jeopardizing the positive effects of VPA during leisure-time. Further, our results may be explained by the fact that VPA during leisure-time is often voluntary and chosen for enjoyment (i.e., autonomous reasons), which is known to positively influence mental wellbeing60,61,62. Moreover, it is shown that these kinds of activities will also improve social interaction, mastery of skills, and detachment and recovery from work, which all can contribute to lower recovery need and risk of burnout46,60,61,63. Elliot, Lang41 also found a negative association between VPA and burnout, however, VPA included both leisure-time and work-related activities. A study by Lindwall, Ljung42 showed beneficial effects of both LIPA and MVPA on burnout, with no differences between both PA intensities. These findings are partly contradictive with our study, as we found that only more VPA during leisure-time is associated with lower recovery need and emotional exhaustion, whereas more moderate-intensity gardening was associated with higher recovery need. It should be mentioned that the study of Lindwall and Ljung42 used the Saltin and Grimby questionnaire64 to assess PA, which also counts transport-related and domestic and garden PA as leisure-time PA, and that burnout was measured by the Shirom-Melamed Burnout Questionnaire65, making no distinction in burnout dimensions. Additionally, the fact that Lindwall and Ljung42 found both LIPA and MVPA to be beneficial, while our study only found higher intensities of PA to be effective, may be due to differences in the study populations. The study by Lindwall and Ljung42 was conducted among healthcare workers and social insurance workers, including both physically active and sedentary employees. In contrast, the teachers in our study are already highly active at work (mostly at low-to-moderate intensity). Therefore, the positive effects of PA on burnout and recovery need may only become evident at higher PA intensities. Lower PA intensities may not provide sufficient physiological stimulation for this highly active study population, suggesting a potential threshold effect of PA intensity on burnout and recovery.

As already mentioned, our study also demonstrated that moderate-intensity gardening (e.g., mowing the lawn) is less beneficial for recovery need. This is in contrast to the findings of a meta-analysis by Shiue53 showing gardening to be beneficial for mental health in adults. Moreover, a study by de Vries and Bakker46 did not show any associations between domestic and garden PA and risk of burnout in a heterogenous sample of full-time workers. It should be mentioned that the latter study did not question recovery need, and domestic and garden PA were not assessed seperately. As proposed by Teychenne, White63, our results may be explained by the fact that moderate-intensity gardening does not provide sufficient distraction from stress, opportunities to boost people’s self-esteem, nor does it provide social interaction63. Another explanation may be that garden activities are more compulsory in nature, and, therefore, these activities may be performed for more controlled reasons (e.g., pressure, guilt, feeling obliged). Controlled behaviours and extrinsic motivation are found to be positively associated with negative affect, anxiety, tension and decreased mental wellbeing60,61,66. However, as these reasons would also imply that domestic activities would be less beneficial regarding recovery need and burnout risk, the present study did not show associations of domestic PA with recovery need and with burnout risk (based on significance in both the step-1 and step-2 models).

It is important to mention that only small effect sizes were found. For example, an increase of one hour per week vigorous-intensity PA during leisure-time is associated with a decrease in 0.019 scale points in emotional exhaustion, and thus larger changes in PA are needed to achieve meaningful changes in burnout. However, compared to the results of our previous study67, which only investigated the role of PA domains regardless of PA intensities, the present study demonstrates larger effect sizes when making a distinction per PA intensity-domain combination. As mentioned in our previous study67, it might be that PA is just one of the many factors contributing to burnout risk and that more relevant effects on burnout risk can be achieved by targeting both individual (e.g., PA) and work-related factors (e.g., workload). A systematic review by Awa, Plaumann68 also concluded that a combination of both individual- and work-directed interventions have longer lasting positive effects compared to individual- or work-directed interventions alone. Moreover, Wang, Liu69 found that tackling individual factors have a more substantial impact regarding personal accomplishment, while work-directed factors played larger roles in predicting emotional exhaustion and depersonalization. Therefore, we recommend to take both kinds of factors into account when targetting burnout risk.

Some PA intensities per PA domain were only found to be significant in the step-1 or step-2 models. As described in our previous cross-sectional paper regarding associations between recovery need, burnout risk and lifestyle67, this may be explained by the fact that some predictors in the models may serve either as confounder,mediator or as a suppressor70,71,72,73. More elaboration on these mechanisms can be found in the study by Verhavert, Deliens67.

Strengths and limitations

This is the first study to provide a better understanding of how PA is related to recovery need and burnout, taking different PA intensities and domains into account. Another strength is that a large sample (n = 1909) of secondary schoolteachers was included, improving statistical power. A third strength is that both step-1 and step-2 models were constructed to take into account possible mediating and confounding effects of the PA predictors, respectively. Lastly, it should be noted that, concerning internal validity, possible effects of sex and age were taken into account by adjusting our models for these possible confounders. In addition, the amount of teaching hours per week was also included as a possible confounder, because research showed that excessive work hours are negatively associated with off-job activities74 and positively associated with burnout75, and thus suggested to take into account when conducting research on this topic46.

A first limitation to this study is that, despite the large sample size, representativeness analysis showed that our sample consisted of more female teachers and more younger teachers compared to the general secondary teaching population in Flanders56. As research showed that women and younger employees tend to have higher burnout risk76,77, there might be a small overestimation of the reported recovery need and burnout risk in the present study. All reported PA levels in the present study may have been under-or overestimated, as research showed that sex differences in PA levels exist, especially when a distinction in PA domains is made78,79. Nevertheless, we do not expect this to have had an influence on the association between PA and recovery need and burnout risk. However, in light of external validity, caution is needed when generalizing prevalence findings to the entire Flemish secondary teacher population. A second limitation is that our results regarding gardening PA should be interpreted with caution, as not all teachers may have a garden, and therefore, may not be able to engage in physical activity within this domain. Third, as earlier mentioned in this discussion, a cross-sectional design was used, and thus we cannot infer cause and effect. It may be, for example, that people with high levels of emotional exhaustion do not have sufficient energy to be physically active. Consequently, a bi-directional relationship may exist between physical activity, recovery need and burnout risk, as also demonstrated in research on the relationship between physical activity and depression80. To gain more insight into the causal effects, future research should prioritize experimental and/or longitudinal designs, including the use of Directed Acyclic Graphs (DAGs). Under the assumption that the DAGs represent the correct underlying causal structure, and the statistical models adjust for the identified confounding pathways, the resulting associations can be interpreted as causal effects81,82,83. A fourth limitation is that there may be possible over- and underestimations of recovery need, burnout risk and PA, as these measures were self-reported, and thus social desirability and recall bias may have been present. Nevertheless, as these measures have been found to be reliable and valid49,55,84, and as we expect the over- and underestimations to be systematic rather than unsystematic, it might have had limited influence on the investigated associations. Lastly, it should be mentioned that data were collected during September–October 2019, prior to the COVID-19 pandemic. Recent literature showed that the pandemic has impacted physical activity levels85, recovery need and burnout risk86 within the teaching profession. However, it is unclear whether these changes may have influenced the associations between both, as observed in our study.

Conclusion

PA intensity may play a role in preventing/reducing recovery need and burnout risk in secondary schoolteachers, but its role may depend on the PA domain in which the activity is performed. Moderate-intensity gardening seem to be unfavourable for recovery need, whereas VPA during leisure-time may be beneficial for recovery need and emotional exhaustion. So, our results suggests that teachers benefit most from engaging in VPA during leisure-time for the management and/or prevention of recovery need and burnout symptoms. Longitudinal and experimental studies are needed to allow for stronger causal inferences.