To practice work from other physical places than in the conventional office has changed working life [1
]. On the one hand, telework is found to increase individual autonomy, work/life balance, work control and productivity and reduce work related stress [1
]. On the other hand, telework is associated with increased stress, low boundary control (i.e., difficulty separating work and life domains), overtime work, lack of work motivation, and insufficient time for recovery [5
]. Thus, research shows disparate results with respect to the effect of telework on staff health and well-being.
According to a report from Eurofound and the International Labour Office [9
], telework is usually practiced by those with high education and/or organizational position in so-called “knowledge- and information organizations” such as IT-enterprises and universities [1
]. The high adoption of telework among academic staff has been related to expanding work demands caused by digitalization, such as increased demands on staff availability, teaching and research performance [4
]. In order to cope with this demanding work situation, academic staff may exceed normal working hours by compensating with their free time [4
]. Studies on the academic work environment show that staff report high levels of stress and lack of time for relaxation, sleep disturbance, work/life interference and decreased levels of physical activity [4
]. Regarding telework, Currie and Eveline [11
] found that academics experienced difficulties separating work and private life and reported longer working hours and stress-related disorders. Moreover, a study on Swedish academics showed that staff who teleworked frequently (>3 days per week) perceived high stress related to indistinct organization and conflicts [15
]. In contrast, Tustin [23
] found that home-based academic teleworkers perceived lower levels of fatigue, more satisfaction with their work and experienced higher productivity than non-teleworkers.
To the authors’ knowledge, no study has investigated stress during telework among academics, using objective measurements. However, Lundberg and Lindfors [24
] showed that white collar workers had lower cardiovascular reactivity during telework than during work at the conventional office. They also found that men had an elevated blood pressure in the evening after teleworking compared to women. They attributed their findings mostly to more diverged work tasks performed at the conventional office compared to more focused work at home [24
]. Moreover, they suggested that differences in physical activity during telework and work at the conventional office could affect the results [24
In a recent study, Arántzazu et al. [25
] found that self-reported sedentary behavior was associated with psychosocial stress among female professors during telework. However, since self-reports of sedentary behavior may be biased [26
], the effects of telework in physical activity and sedentary behavior should also be investigated using objective measurements. It has been shown that physical activity can reduce psychophysiological reactivity while sedentary behavior can increase it [28
]. Because of this, measures of physical activity are often combined with measures of physiological reactivity (e.g., pulse recordings), to study stress in different work settings [26
]. Several reports and studies emphasize the importance of research on postures and movements to understand health outcomes among teleworkers [1
Studies on the occupational health aspects of telework is of high relevance in these days of rapid technological development and COVID-19 lockdowns. Academics are amongst the groups that practice telework to a high extent, but little is known about staffs’ physical behaviors and psychophysiological responses to such flexible work arrangements. We believe it is important to investigate these factors in order to get a deeper understanding of what risks and/or possibilities telework could entail for academic staffs’ health. In the present study, we therefore, aimed to determine if psychophysiological reactivity, postures and movements differ during telework (i.e., work performed at home) and work performed at the conventional office amongst academic teaching and research staff.
2.1. Study Sample
In a previous questionnaire study [15
], respondents were asked about interest in participating in the present study. In total, 111 employees from six universities in different parts of Sweden reported their interest, and were contacted via e-mail. Eligible participants were employed as junior lecturers, senior lecturers or professors and engaged in teaching and/or research ≥50% of their working time. All participants had to regularly practice telework and do so at least one day during the measurement period. Participants with medical heart conditions and those who had retired were excluded leaving 108 subjects eligible for inclusion. Among those, 23 academic staff agreed to participate. Information about the study aim and procedures were given in writing and verbally during the recruitment and at the time of data collection. Participants gave their informed consent before data collection. The Regional Ethical Review Board in Uppsala, Sweden has approved the study (Reg. No. 2016/494).
2.2. Study Design and Procedures
The study comprised five consecutive workdays of accelerometry measurements, pulse recordings and daily visual analogue scale (VAS) ratings. During at least one of the five workdays, the participants teleworked and during at least one of the days they worked at the conventional office. Salivary samples were collected during one day of telework and one day of work at the conventional office. On each workday, the participants documented place and time for work and leisure activities, as well as work tasks performed, in a diary. The measurement period started on a Monday morning with individual 30-min sessions where one member from the research team (L.W.) informed about the study procedure by handing out and demonstrating the data collection materials, and attaching the accelerometry and pulse recording devices. At Friday night, the participants removed the recording devices. Data collection lasted from August 2018 to June 2019.
2.2.1. Ratings of Stress, Fatigue and Recuperation
Stress, fatigue and recuperation were assessed on a 100 mm VAS scale. It ranged from 0 (= not at all) to 10 (= completely) and higher values indicated more stress, fatigue and recuperation. Participants rated their levels of stress and fatigue before and after each workday, and how recuperated they felt in the morning after each workday.
2.2.2. Salivary Sampling
A self-administrated Salivette active sampling technique [35
] was used to measure cortisol concentration in saliva (ng cortisol/mL saliva). It consists of a standard centrifugation tube containing a small cotton-swab that was actively chewed on for 30 s, or until it contained a sufficient amount of saliva. Thereafter, the cotton-swab is replaced into the sampling tube and stored in a freezer at −18 °C. Salivary samples were collected approximately every 3rd hour, 6 times a day, starting at 7 AM and continuing throughout the day until 10 PM. Participants were instructed to refrain from nicotine, coffee and alcoholic beverages, and vigorous physical activity one hour before providing their salivary samples [35
]. Each sampling tube was marked with a unique number.
2.2.3. Pulse Recordings
Firstbeat Bodyguard 2 pulse recording device (Firstbeat Technologies Ltd., Jyväskylä, Finland) was used to determine psychophysiological reactivity by sampling heart rate (beats per minute (bpm)) and heart rate variability (beat-to-beat intervals (RRI)) continuously at 1000 Hz [36
]. The pulse recording device were attached with electrodes to the participants’ upper right chest area under the collarbone, and to the lower left chest area on the rib cage. Participants were instructed to detach the pulse recording device before coming into contact with water (i.e., shower/bath) and reattach it when dry. Date and time for detach- and reattaching were noted in participants’ diaries.
2.2.4. Accelerometry Measurements
Physical activity and arm elevation were assessed using two AX3 accelerometers (Axivity Ltd., Newcastle, UK) [38
]. They were attached to the skin surface with adhesive tape distal on the deltoid muscle bracket and on mid quadriceps on the dominant side of the body. The accelerometers were then secured by plastic adhesive film. Data were sampled continuously at 25 Hz. In order to synchronize the equipment, participants were asked to perform reference measurements at the start and at the end of the measurement period, as well as once a day. This was done by standing in an upright position with arms alongside the body for 10 s, followed by a jump while remaining in the same position and another 10 s of standing still [38
2.3. Data Processing
Salivary samples were thawed, vortexed and centrifuged at 1500× g
for 15 min in order to remove particle matters that may contaminate and affect the estimated cortisol concentration values. Concentration of salivary cortisol was then analyzed with a Salimetrics Enzyme immunoassay kit according to customary procedures [35
The AX3 OMGUI software (Axivity Ltd., Newcastle, UK) was used for downloading pulse recording data from Firstbeat Bodyguard 2. Data were then exported to the Firstbeat SPORTS 4.7 software (Firstbeat Technologies Ltd., Jyväskylä, Finland) and scanned through an artifact detection filter to remove falsely detected, missed and premature heartbeats [36
]. Pulse recording data were then imported to Spike version 9 for Windows (Cambridge Electronic Ltd., Cambridge, UK) and screened for errors using custom algorithms in Matlab software. The following variables were calculated based on non-overlapping one-minute windows of pulse recordings: heart rate (bpm); heart rate variability in time domains (ms): the standard deviation of RRI (SDNN); the root mean squared successive differences of RRI (rMSSD); heart rate variability in frequency domains (ms2
): low frequency (LF) (0.04–0.15 Hz); high frequency (HF) (0.15–0.40 Hz). Among these variables, rMSSD and HF are indexes of vagal control and reflect parasympathetic reactivity; LF reflects baroreceptor activity influenced by sympathetic reactivity; SDNN reflects the cyclic components responsible for the variability in the period of recordings, i.e., overall heart rate variability. All variables were selected based on recommendations made in previous research [33
Accelerometer data were downloaded (sample rate 25 Hz) with Acti4(a) software [41
] and then exported to Spike version 8 for Windows where it was screened for errors. Subsequently, median arm angle and number of transitions between sitting/lying and standing were calculated. In addition, the average time spent in sedentary behaviors (i.e., sitting, lying) and other behaviors (e.g., standing, walking, running) were computed [26
]. Since time spent in different behaviors are interdependent, compositional data analysis was used [42
]. It involved calculating an isometric log-ratio (ilr) coordinate of the relative information between sedentary behaviors and the other behaviors (i.e., standing/walking/moving) that could be handled using standard statistical methods such as analysis of variance [26
Diary entries were used to distinguish data for each Workplace and Time domain. For each participant, the variables were averaged across workdays spent at the conventional office and workdays teleworking from home, respectively. Workplace had two levels: Office and Telework. Office was defined as work performed during regular workhours at the conventional workplace. Telework was defined as work performed during regular workhours from the participants’ home. For accelerometry measures and pulse recordings, the Time variable was divided into leisure time before workhours (i.e., the time from awakening to the workday begins), work during regular workhours (i.e., 8:00 AM to 17:00 PM excluding commuting time) and leisure time after workhours (i.e., the time from the end of the workday to going to sleep excluding overtime work). For salivary samplings, the Time variable were divided into six hours (7:00, 9:00, 12:00 AM; 15:00, 18:00, 22:00 PM).
2.4. Statistical Analysis
Statistical analyses were performed in IBM SPSS Statistics 22.0 (IBM, Armonk, NY, USA) for Windows. Sample characteristics are presented as proportions, means and standard deviations in Table 1
. Differences in estimated stress and fatigue before and after workhours, and estimated recuperation in the morning after workhours, were compared between Office and Telework in paired sample t-tests. For remaining outcomes, repeated measures analyses of variance (ANOVA) were performed with Workplace (two levels) and Time (three and six levels) as within-subjects effect. The ANOVAs were performed with and without adjustment for commuting time [10
], children living at home [14
] and gender [10
]. When the assumption of sphericity was not met, the Huynh-Feldt correction was used [49
]. Level of significance was set to p
< 0.05 in all tests.
The final sample consisted of 23 participants. There was an equal distribution of men and women, age ranged between 27 and 61 years, and the calculated mean for body mass index (BMI) was 25.4 (Table 1
). Among included professions, 56% were junior lecturers, 35% senior lecturers and 9% professors. According to information provided by the universities’ HR departments, the sample is representative for the population. Their main work tasks were teaching activities such as planning and giving lectures and communicating with students; and administrative and research activities such as reading, responding to emails and participating in meetings. Participants were experienced teleworkers as most of them teleworked several times/month or several times/week. Their average commuting time to work was 38.7 min (data available for n
= 22) and the majority of the sample had children living at home. During the measurement period, participants performed telework on average 1.8 days and worked from their conventional office on average 3.2 days.
3.1. Staff Ratings of Stress, Fatigue and Recuperation
For VAS ratings (shown in mm) of stress, there were marginal differences in ratings before workhours (telework: M = 20; office: M = 27), relative to after workhours (telework: M = 19; office: M = 28) during teleworking days and office days. The differences in ratings of fatigue before (M = 27) relative to after (M = 33) telework, were slightly larger, and similar in size compared to office work (before: M = 35; after: M = 42). Ratings of recuperation in the morning after workhours showed that participants felt fairly recuperated before going to work regardless of where they would work (telework: M = 60; office: M = 52).
Paired sample t-tests showed no significant difference in the effect of Workplace for self-rated stress (p = 0.689) and fatigue (p = 0.842) before compared to after workhours or for recuperation in the morning after workhours (p = 0.151).
3.2. Psychophysiological Reactivity
For the salivary sampling, the repeated measures analysis of variance showed a significant effect of Time on cortisol concentration (p
< 0.001, η2
= 1.000) but no interaction effect of Workplace and Time. Participants’ salivary cortisol showed a normal variation during the day at both workplaces with the highest concentration in the morning and a gradual decrease during the day (Figure 1
The heart rate and heart rate variability measures generated complete data for 20 participants, data from the remaining participants were excluded due to low quality. In the repeated measures analysis of variance (n
= 20), Workplace and Time had a significant interaction effect on heart rate (bpm), which was highest before and after workhours and slightly lower during regular workhours at the office compared to telework. For heart rate variability, a significant Workplace and Time interaction effect were seen in all variables (Table 2
). Figure 2
shows the marginal means for Workplace and Time, for heart rate and each heart rate variability variable. The largest differences in psychophysiological reactivity were seen before workhours, and the smallest differences were seen after workhours (Figure 2
). For teleworking days, the reactivity did not change markedly during the day, while for office days, it peaked during workhours. After adjusting the repeated measure analysis of variance for commuting time, children living at home and gender, the magnitude of the Workplace and Time interaction effect did not change. SDNN was higher and showed more variation for men (telework: M = 130.54 ms; office: M = 121.04 ms) than for women (telework: M = 97.70 ms; office: M = 108.48 ms) throughout the workday regardless of whether work was performed in the conventional office or on teleworking days. Similar results were found for rMSSD and HF.
3.3. Postures and Movements
The accelerometry measures generated complete data for all participants. For two participants, however, all behaviors were not present during all Workplace and Time levels and therefore they were excluded from the ilr transformations. The repeated measures analysis of variance of accelerometry measures with ilr transformed sedentary behavior relative to other behaviors showed no significant interaction effect of Workplace and Time (η2
= 0.013) but a significant effect of Time (η2
= 0.480) (Table 3
). Most time (min) spent in sedentary position was seen during regular workhours and during leisure time after workhours (Table 4
). In total, staff spent most time in different physical behaviors during office days, which is in agreement with previous findings [51
]. There was a significant Workplace and Time interaction effect for the variation in movements, i.e., transitions between sitting and standing (η2
= 0.194), with more transitions being made during teleworking hours (M = 31) than during office hours (M = 25). Arm angles (50th percentile) showed no significant effect for Workplace and Time interaction (η2
= 0.024). For working days performed at the office and home alike, arm elevation increased over the day with lowest elevation during leisure time before workhours (telework: M = 23.93°, SD = 11.38°; office: M = 24.36°; SD = 7.18°) and most elevation after workhours (telework: M = 34.53° SD = 14.72°; office: M = 32.17°; SD = 8.87°). The magnitude of Workplace and Time interaction effects did not change after adjusting for commuting time, children living at home and gender.