Physiological Stress

Relevant outcome variables will be assessed at three time points. The different measurement points are the same for all five intervention groups. The baseline measurement (T₁) is first carried out on the participants directly prior to beginning with the interventions. The post-intervention measure will be performed immediately after the interventions are completed (T₂). After another 6 weeks of no intervention, the sustainability measurement (T₃) will be conducted. For the waiting control group, the first measurement (T₁) will be followed by a period of eight weeks of waiting. The second measurement (T₂) will be conducted after this waiting period. Then participants of the waiting control group will receive one of the interventions and then have their third measurement directly after finishing the respective intervention (T₃). At all three measurement points, the same variables, namely physiological stress, perceived stress, satisfaction with the digital intervention, personality, work-related behavior patterns, and team conflict will be measured.
The respective measurement tools that will be applied are described in the outcomes section including information on reliability and validity. In order to promote data quality, we include evaluated scales that appeared to be reliable and valid in previous studies. We will not conduct duplicate measurements. The collected data will be processed pseudonymously in digital form (from the initial measurement to the sustainability study—approx. 12 weeks).
In order to ensure pseudonymization and simultaneously ensure subsequent deletion of the data, each participant creates a five-digit individual code word immediately after signing the consent form when answering the initial survey, consisting of the first letter of their mother’s first name, the first letter of their favorite color, the first letter of their place of birth, the last digit of their year of birth, and the last digit of their day of birth (e.g., MGF01). This code is used instead of other identifier in all subsequent measurements. There is a coding list on paper that links the name to the code but is only accessible to the investigators and the project manager. The coding list is kept in a lockable cabinet or safe and is destroyed after the data collection is completed. If a respondent wishes to delete their data retrospectively, they can use the five questions mentioned above to reconstruct their code word and thus request deletion of the data. After completion of the sustainability measurement, the code used to pseudonymize the employees will also be replaced by assigning a combination of numbers (e.g., 1647) and thus anonymized. Non-anonymized data sets are deleted by the university and the cooperation partners involved in the data collection after completion of the data collection. Data will only be processed in anonymized form within the framework of the study and for subsequent scientific use as well as for publication of the study results.
The anonymized data sets will be stored on a password-protected project folder of the TUB cloud of the TU Berlin for the duration of the project until the completion of all scientific work and associated data analyses. Subsequently, the data will be transferred to a research data repository of the TU Berlin and stored there for a period of 10 years. Access to the server is granted by assigning passwords to the external project partners. The rights to assign the passwords lie with the project leader PD Dr. Bettina Wollesen. After completion of the project, all raw data collected can be made publicly available in anonymized form in a research data repository for an unlimited period of time as part of the Open Science efforts of the scientific community and in accordance with the Berlin Declaration. Personal information about the participants will be protected by pseudonymization of the data sets described above.

For this study, GSs were defined as recreational areas with vegetation in a public, open and publicly accessible space. These include traditional urban parks and other spaces, such as urban forestry, public gardens, pocket parks or cemeteries, and may include built environmental features.
Exposure to road traffic noise during daytime (L_day; see Section 2.3) and access to public GSs were spatially analyzed in the Geographic Information System (GIS) Esri ArcGIS (version 10.8.1).
Access to public GSs was derived from circular Euclidean buffers with a radius of 300 m around the buildings of the study participants (similar to [17 (link),19 (link)]). Public GSs as identified by land-use classification data were used. With this analysis, buildings could be classified as being in areas without access to GSs (i.e., no GSs within the 300 m circular buffer) vs. with access to GSs (presence of GSs within the buffer). A sensitivity analysis performed in a previous study on the effect of vegetation on noise annoyance revealed that effects were similar for buffer sizes of 150, 300, 500 and 1000 m, and that a 300 m buffer yielded best explained variance [19 (link)]. Buildings that had more than one GS within the buffer were excluded from the analysis, in order to assign participants to a single GS to facilitate the interpretation of results.
The GSs were identified and selected in a stratified sample using land-use classification data of the Federal Swiss Office of Topography (swisstopo; same data as in [19 (link)]). GSs with restricted access (e.g., private/household gardens or playgrounds of schools) or with access requiring payment, namely sport fields (e.g., for soccer and golf), camping grounds, open-air swimming pools as well as the zoo of Zurich, were excluded. Initially, a total number of 124 GSs in the city of Zurich were included in the dataset. These GSs were divided into large (≥10,000 m²) and small (<10,000 m²) and subsequently, into loud and quiet (see details in Section 2.3). Twenty-three loud and 25 quiet GSs were identified; the remaining 76 that matched neither of the two groups were excluded from the study. The final dataset with the four groups of GSs included: (i) loud and large (n = 11), (ii) loud and small (n = 12), (iii) quiet and large (n = 18) and (iv) quiet and small (n = 7).
The vegetation-around-home (VEG-H), i.e., the residential green or greenness within buffers with a radius of 50 m around home locations, was also derived as a proxy for both access to green on the property and view from home on outdoor vegetation. The latter was found to be a crucial parameter for alleviating noise-induced health effects (see, e.g., [19 (link),20 (link)]). To quantify VEG-H, the satellite-based indicator for greenness, Normalized Difference Vegetation Index (NDVI) [21 ], was used. Mean NDVI values for the months of April to October in the years 2019 to 2021 were used (data extracted from ESA [22 ]).
Based on the combinations of different levels of noise exposure and access to GSs, the design included the following study groups: one group with low noise exposure at home with access to quiet and large GSs (LA), one with low noise exposure at home but no GS access (LNA), four groups with high noise exposure at home with access to GSs (specifically to quiet and large (QuLa), quiet and small (QuSm), loud and large (LoLa) and loud and small (LoSm) GSs), and one with high noise exposure at home but no GS access (HNA). This provided seven study groups between which a variation in the stress levels (as measured by perceived and physiological stress) is expected (Figure 1).

To test the hypotheses presented in Section 2.1, different models will be established. The link between the binary variable HAnn with road traffic noise exposure at home (L_den) and access to GSs and/or VEG-H will be explored with logistic regression analysis. The binary variable HAnn will be studied for comparability with previous field surveys (e.g., [24 ,47 (link)]). The association between perceived stress with road noise exposure (L_den) at home and access to GSs will be evaluated through an ordinal or linear regression analysis, treating stress either as an ordinal (5 levels) or continuous variable, respectively. (Ordinal variables may be treated as continuous if they have five or more categories, e.g., [54 (link)].) The approach for physiological stress is similar to that for perceived stress, except it is a purely continuous variable. The models will be adjusted for age, sex, socio-economic status, BMI, physical activity, smoking status and employment situation [55 (link)]. Perceived stress will additionally be adjusted for potentially stress-related factors such as stress in private and work life.
To complement these analyses, models using the NDVI within the 300 m buffer as a general indicator for residential greenness instead of GSs and/or VEG-H will be explored, because [19 (link)] found NDVI to be a particularly strong predictor for (reduced) noise annoyance. Further, structural equation models (SEM) will be implemented to explore the role of mediation. SEM have already been successfully applied to study the effect of transportation noise on annoyance and health-related quality of life [56 (link)]. Details on the modeling approaches will be set during the actual analyses.