3.1. Ethics

This study will be reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement [17]. Ethical approval has been granted by Ethics Commission of the Medical Faculty of the University of Cologne based on written informed consent and in accordance with the Declaration of Helsinki (Project Reference Number 24–1224). The participants were provided with an information document (approved by the above-mentioned Ethics Commission) and provided written informed consent after having read through the document and been given time to ask questions of the study coordinator. All individual data will be de-identified; only the principal investigator will have access to identifiable information. Participation is voluntary and participants are able to withdraw their consent at any time. The protocol has been registered on the Open Science Framework https://doi.org/10.17605/OSF.IO/U4R5M.

3.2. Recruitment

Secondary schools will be contacted and asked to participate in the study by non-probability (“convenience”) sampling. For secondary schools in this region, students typically enter at ~10 years old and compulsory education is until ~15 years old. Students who complete the lower secondary level may prepare for university-entry qualifications in the upper secondary level, finishing at ~19 years old. We do not include vocational schools (a form of upper secondary school geared toward preparing students to enter the workplace with specific skills). If a school’s representatives give positive feedback, all teachers will be invited to participate via the school’s internal general messaging service. With permission from the given school, appointments will be scheduled and data collected on site following participant informed consent.

To be as inclusive as possible, there are no restrictions beyond being currently employed and actively working as a secondary school teacher in the Cologne-Bonn region. We have no restrictions by part-time vs full time work nor by time-of-year of recruitment with the exception of school holidays. In other words, some teachers may participate in more or less busy/stressful times of year and the week of participation is not systematically chosen. We do not limit data collections to on site in schools – teachers can contact us independently with their interest to participate and we can schedule briefing and data collection for a non-school location. We anticipate that advertisement of our study will spread to other schools in our catchment area by word-of-mouth (friends who have studied together, professional development trainings, etc). We aim to include as many schools as possible in terms of different secondary levels within the German academic system, urban and rural, and different social index locations.

We are limited in terms of how many teachers can provide data at a given time due to the number of available light and activity sensors. We exclude primary and tertiary level educators to increase the homogeneity of our population. We expect primary and tertiary level educators to face different work requirements and workloads since they work with young children or adults, in comparison to secondary school teachers, who predominantly work with adolescents. Additionally, more than half of the teachers in Germany work in secondary education [4].

3.3. Data collection

Teachers will be met individually or in small groups in a suitable (quiet) location (e.g., during free periods in schools). First the teachers will receive study information and will be given time to read it thoroughly and ask questions. If they provide consent to participate, they will be provided with questionnaires that take ~30 mins to complete. Following questionnaire completion, we will provide them with an activity and light sensor and a diary and instruct them on how to use them (Fig 1). After one week, we shall collect sensors and diaries.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Example Diary Day.

The diary entry is in German as it will be used in a German speaking population. Briefly, “Schlaf” = Sleep; “Aufgestanden” = Got out of bed; “Arbeitsweg” = Journey to Work; “Arbeit in der Schule” = Work in School; “Nahrungsaufnahme” = Nutrition Activities; “Körperliche Activität” = Physical Activity, “Schlafenszeit (Licht aus, im Bett)” = Sleep time (Lights out, in bed).

https://doi.org/10.1371/journal.pone.0348121.g001

3.4. Tools

All of the following questionnaires are validated for use in German-speaking populations and use simple continuous scale scoring methods. The Munich Chronotype Questionnaire (MCTQ) focuses on sleep timings on free days and work days to evaluate chronotype [18]. The Insomnia Severity Index (ISI) measures insomnia characteristics and symptoms, like subjective sleep disturbances, worries concerning disturbances, and impairments in daily life [19]. The Pittsburgh Sleep Quality Index (PSQI) evaluates sleep quality during the previous month, and is a broader construct than the insomnia severity evaluated in the ISI [20]. The Beck Depression Inventory (BDI) is used to assess the severity of depression symptoms [21]. The General Anxiety Disorder 7-Item scale (GAD-7) is used to assess the severity of anxiety symptoms [22]. The Maslach Burnout Inventory (MBI) is the most commonly used tool to assess burnout and its sub-components of emotional exhaustion, depersonalisation, and accomplishment.

A one-page questionnaire to assess demographic (e.g., age, sex, ethnicity) and work characteristics (e.g., subjects taught, years of experience, size of classes) will also be provided.

The provided diaries have sections for each of the 7 days to report the following: working times and task types, work locations, meal timings and contents, activity timings and contents, and sleep timing. An example for a completed diary entry is provided in Fig 1.

Light and activity data will be collected with use of the MotionWatch8 (CamNTech, UK). This is a medical grade, wrist-worn, combined light and activity sensor, similar in size, weight, and shape to a typical wristwatch. The activity component is a tri-axial accelerometer that provides an output as counts per selected time epoch (from 1–60 s). The light sensor provides an output of average lux per epoch, with a range of 0–64,000 lux. Participants will be instructed to wear the sensor continuously on their non-dominant wrist, just removing it in case of fully submerging the hand into water, e.g., when taking a shower or swimming. Data will be recorded in 30 s epochs, and the tri-axial mode 3 will be used for activity. The MotionWare software (Version 1.4.20, CamNTech, UK) will be used for data analyses.

3.5. Power analysis

Our power calculation estimates what effect size differences in emotional exhaustion severity (a key component of burnout) would be detectable for different sample population sizes between an evenly split assumed “exposed” group and an “unexposed” group (e.g., unexposed defined as ≤ median intensity exposure and exposed defined as > median exposure), if the unexposed group is assigned a mean population based emotional exhaustion score of 2.4 ± 1.3 (mean ± standard deviation) [23], a common standard deviation is used across evenly split groups, α = 0.05, and β = 0.2. Sample sizes of 100, 200 and 300 participants would yield detectable effect size differences of 0.74 (moderate-large), 0.52, and 0.42 (moderate-small), respectively. The control group will be an overestimate as it based on a mean population score [23]. A moderate effect size difference of 0.5 (corresponding to ~0.65 points of emotional exhaustion) would be relevant given the sizeable teacher population, and that a decrease of this size would mean shifting to lower severity categories for many. Thus, our aim is recruit ~200 participants. Mean and standard deviations of anxiety and depression scores have lower spreads, so even smaller effect size differences would be statistically detectable with ~200 participants.

3.6. Data analysis

Analyses will be performed with Stata/SE-17 (StataCorp LLC, USA). Descriptive statistics for demographics, light exposure, activity, and questionnaire scores/results will be generated. Comparisons of mean and medians of LAMAS variables and outcomes by demographic and work characteristics will be conducted. Associations between LAMAS variables and outcome variables (i.e., burnout, anxiety and depression severity scores) will be explored in multivariable regression models. Joint analyses will also be conducted.

Diurnal distributions of LAMAS variables associations with outcomes will be explored by several means. These will include: (i) assessing features of multi-component cosinor models with nested generalised linear model approaches and predictions at shifted diurnal profiles that are identified from k-means cluster analyses; (ii) assessing exposure and relative exposure volumes during school times, after school times, public holidays and weekends, and ratios between these time periods; (iii) assessing theoretical exposure volume diurnal redistributions to different times of day using quasi-compositional data (iso-temporal substitution) analyses. The latter can be achieved by partitioning mean total daily exposure into 4-hour time blocks of the 24-hour day (starting with, e.g., 04:00 or with time of awakening), expressing proportions as totals of exposure and using regression models that include mean total daily exposure and omitting one time block; thus, coefficients for each bin will represent the association with reallocation of exposure from the reference bin (rescaled predictors will be used to allow assessing percentage of reallocation rather than total time block reallocation). Theoretical dose responses will be assessed using increasing percentage reallocations. Sensitivity analyses will include using different time-block durations.

In terms of analytical prioritisation, our primary analysis involves the joint effects of the diurnal distributions of LAMAS (as primary exposures) on associations with burnout severity scores (as primary outcome). We expect that diurnal distributions of LAMAS and work exposures that lean toward higher intensities or proportions later in the day relative to earlier in the day (with the exception of sleep) will be associated with higher burnout symptom severities. As our analyses are exploratory, the analytical approaches are complementary, and we do not want for potentially meaningful patterns to be obscured, we shall not make adjustments for multiple tests.

Diary timing data will be checked against data from activity and light sensors. Misclassified diary timing data will be corrected in accordance with sensor data. Missing activity sensor data (e.g., such as removal for swimming that corresponds with swimming time indicated in the diary) will be imputed based on context (e.g., using plausible counts based on within-subject data from a similar physical activity). Missing physical activity data > 10 minutes and corresponding with no indicated activity will be deemed non-wear time. Unusual light exposures (e.g., low lux accompanying high physical activity from the sensor and diary indicated outdoor activity due to, for instance, covering of the sensor) will also be imputed based on context (e.g., appropriate lux levels for the given time of day and location). Results will be reported for sensor data with and without imputations.

Confounding and effect modification will be explored using age, sex, ethnicity, chronotype, subjects taught, school form, years of experience, class sizes, official working hours per week, presence of chronic disease, taking medication, smoking and alcohol consumption, recent jet-travel across time zones in the previous two weeks.

3.7. Timeline

Regarding recruitment, the first participant was recruited on 05 March 2025. Forty-four participants have been recruited to date. Recruitment is expected to continue through to 31 December 2026. This is based on the number of sensors we have at our disposable, the individual monitoring periods, assumes we can collect data from new participants every 14 days during the school terms and for all sensors to be in use, and allows a buffer for unexpected pauses or difficulty in recruiting for all sensors to be in use. No data has yet been analysed. Completed study results are expected in 2027.

3.8. Key assumptions of the study

This study assumes that the short-term measurements capture stable patterns sufficiently to detect meaningful associations without assuming directionality. Regarding population, a systematic non-response bias among the working teacher participants is not assumed. Accuracy of sensors and honest and accurate responses to questionnaires and in diary entries are assumed where necessary (i.e., checks will be made wherein sensor and self-reported data overlap). Context-dependent assumptions including that week-specific external events did not systematically bias results and that work demands were typical though time-of-year differences will be explored to this end if and where this is possible.