This is an open-access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on http://www.jmir.org/, as well as this copyright and license information must be included.
Effective interventions to influence workplace sitting are needed, as office-based workers demonstrate high levels of continued sitting, and sitting too much is associated with adverse health effects. Therefore, we developed a theory-driven, Web-based, interactive, computer-tailored intervention aimed at reducing and interrupting sitting at work.
The objective of our study was to investigate the effects of this intervention on objectively measured sitting time, standing time, and breaks from sitting, as well as self-reported context-specific sitting among Flemish employees in a field-based approach.
Employees (n=213) participated in a 3-group randomized controlled trial that assessed outcomes at baseline, 1-month follow-up, and 3-month follow-up through self-reports. A subsample (n=122) were willing to wear an activity monitor (activPAL) from Monday to Friday. The tailored group received an automated Web-based, computer-tailored intervention including personalized feedback and tips on how to reduce or interrupt workplace sitting. The generic group received an automated Web-based generic advice with tips. The control group was a wait-list control condition, initially receiving no intervention. Intervention effects were tested with repeated-measures multivariate analysis of variance.
The tailored intervention was successful in decreasing self-reported total workday sitting (time × group:
Our results point out the significance of computer tailoring for sedentary behavior and its potential use in public health promotion, as the effects of the tailored condition were superior to the generic and control conditions.
Clinicaltrials.gov NCT02672215; http://clinicaltrials.gov/ct2/show/NCT02672215 (Archived by WebCite at http://www.webcitation.org/6glPFBLWv)
In modern societies, adults spend the majority of their waking time in sedentary behaviors, that is, activities in a sitting or reclining posture characterized by a low energy expenditure [
The evidence regarding the effectiveness of workplace interventions focusing on occupational sitting is growing [
A review of sit-stand workstations suggests that they can be effective in reducing occupational sitting time (the pooled effect size of 7 intervention studies was -77 minutes of sitting/8-hour workday), without compromising work performance [
One popular public health promotion method that has been shown to successfully change a variety of health-related behaviors (dietary behaviors, alcohol consumption, smoking habits, and physical activity) is Web-based computer tailoring [
We developed a theory-driven, Web-based, computer-tailored intervention to influence sitting at work and found it to be acceptable in terms of the assessment questioning, interestingness, length, credibility, and relevance of the advice [
The study used a controlled baseline (T0), 1-month follow-up (T1), and 3-month follow-up (T2: T0+3 months) design, with 3 different conditions (
We selected a convenience sample of 2 companies (a university and an environmental agency) in Flanders (ie, the northern, Dutch-speaking part of Belgium), mainly employing desk-based workers, having more than 100 staff members, and each having at least three different worksite locations. Both workplaces had a general health policy following European legislations and were informally committed to health aspects, but they did not yet focus on healthy sedentary behavior at work. We contacted company management by phone and email to inform them about the study, and both companies agreed to participate in this study (response rate 100%). Within the university, 3 departments of the central administration were selected to participate, and within the environmental agency, 3 departments in East Flanders were selected. Within each company, each department was randomly assigned to 1 of the 3 conditions. All selected departments were in different physical locations and the employees had little face-to-face contact with one another, reducing the opportunity for contamination between groups.
Flow chart of a randomized controlled trial of an intervention to reduce and interrupt sitting time at work.
A contact person in each department provided the email addresses of all employees. In October 2014, employees were invited to participate by email. Employees willing to participate were asked to reply to the email within 1 week, indicating whether they wanted to complete the Web-based questionnaires only or whether they wanted to complete the Web-based questionnaires and additionally wear an activity monitor. We sent a reminder email with the invitation to those who had not yet responded, 1 day before the enrollment deadline. The study protocols were approved by the Ethics Committee of the Ghent University Hospital, Belgium.
A researcher emailed a confidential website username and password to all participants. After logging in, participants received a short introduction pop-up screen and were then referred to the home page, inviting them to complete an assessment questionnaire (see the Measures subsection below). After completing the baseline questionnaire, each group received different feedback. Those who were interested in participating but who did not complete the questionnaire within 7 days were sent up to 3 automatic reminders to visit the website.
A member of the research team visited the participants willing to wear an activity monitor at their workplace on a Monday. They were instructed to wear the monitor on the thigh from Monday to Friday. A researcher covered the monitor with a transparent medical tape (Tegaderm, 3M, Diegem, Belgium) before placement, and also attached the monitor itself with this tape. These waterproof attachments allowed for 24-hour wear, including water activities (eg, bathing and swimming). Monitor data from 3 days (Tuesday, Wednesday, and Thursday) were used for this study [
The development of this theory-driven intervention, called
To increase their knowledge, participants first received general information about the importance of sitting behavior to improve health outcomes. This was followed by normative feedback about their own sitting behavior on working and nonworking days, in order to increase awareness of participants’ levels of sitting time. Further, feedback on the frequency of breaks from sitting, information on the importance of these breaks for health, and the suggestion to interrupt prolonged sitting every 30 minutes was provided [
At the end of this advice (section 1), participants were able to request up to 5 other noncommittal sections if they were interested. In line with the self-determination theory [
In the last section, participants who were motivated to change their sitting were invited to create an action plan to convert intentions into specific actions through the specific, measurable, attainable, relevant, and time-bound (SMART) goals and implementation intentions [
Some screenshots of the tool are provided in
In the generic advice condition, after completing the baseline assessment questionnaire, users received generic information on the importance of reducing and interrupting sitting, and generic tips and suggestions on how to interrupt (taking short standing breaks, 6 tips) and reduce (replacing sitting by periods of standing, 8 tips) sitting during work hours, (lunch) breaks, and commuting (topics similar to the tailored group). While the information covered the same topics as in the tailored group, the generic group didn’t receive personalized advice or an action plan, and all information appeared on a single screen page.
The questionnaire consisted of several parts, and all questions were asked at T0, T1, and T2, except for unchangeable variables, such as height.
The following sociodemographic variables were assessed: sex, age, highest educational degree with 5 options dichotomized into low (no diploma, elementary school, secondary school) and high (high school, university) education, height, and weight.
We asked about the number of workdays per week (1–7), average daily amount of time (hours and minutes) spent at the workplace (open-ended), occupational status (blue collar, white collar, management), and employment duration (14 categories ranging from 1–6 months to 55–60 years).
We assessed the level of sitting time in 5 domains using the Workforce Sitting Questionnaire (WSQ) [
The validated International Physical Activity Questionnaire (IPAQ) short version [
Sedentary behavior was measured objectively using the activPAL (PAL Technologies, Glasgow, UK) activity monitor (weight 15 g, dimensions 53 × 35 × 7 mm). This inclinometer, distinguishing periods of sitting or lying from standing and assessing breaks from sitting, has been validated (correlation between activPAL and direct observation:
Participants were requested to complete a day log and record the type of day (workday at home, workday at the workplace, or nonworkday), time of getting up, the start and end time of the working day, and the time of going to sleep.
We collected the number of participants requesting the different sections of the advice from the website administration. Google Analytics provided data on website visiting time [
Data recorded by activPAL were reduced using PAL Technologies software (version 6.4.1). We calculated waking sitting time by subtracting sleep time reported in the day log from the total sitting time recorded by the activPAL device. The percentage of working time spent sitting was calculated as sitting time during work hours/work hours × 100. We used similar formulas to calculate percentage of working time spent standing and number of breaks per working hour. We calculated average values for T0, T1, and T2 from the mean scores of the 3 measurement days.
Within the 5 domains assessed using the WSQ, we truncated values over 12 hours/day to 12 hours to avoid unrealistic values [
Based on the guidelines for data processing and analysis of the IPAQ [
All analyses were conducted in SPSS version 22.0 (IBM Corporation) and significance was set at
To investigate the 1-month and 3-month follow-up effects of the intervention, we conducted 3 repeated-measures multivariate analysis of variance tests with time (T0, T1, or T2) as the within-participants factor, condition (3 groups) as the between-participants factor, and self-reported sitting (workday and nonworkday total sitting; average daily domain-specific sitting) and objectively measured sitting (total waking sitting time, working time spent sitting, working time spent standing, breaks from sitting per work hour) as the dependent variables. When the time (3 levels) × condition (3 levels) effects were significant, we conducted additional post hoc repeated-measures analyses, including 2 times points (T0–T1 or T0–T2) and only 2 conditions, to find out where the differences in changes over time between the conditions occurred. We included the following covariates in the analyses: age, sex, education, hours at work, employment duration, BMI, walking, and moderate and vigorous-intensity physical activity at baseline. Due to the skewed nature of the outcomes, we did the analyses on square root transformations to improve normality, but for reasons of clarity, we report nontransformed average scores in the tables. We executed this approach using both a retained sample analysis (ie, completer analysis) and an intent-to-treat analysis (last value carried forward). Because we found no differences between the 2 analyses, we report results only of the retained sample analysis.
The total sample (N=213) of employees completing the Web-based questionnaire consisted of 31.5% (67/213) men, 81.7% (174/213) with a high level of education, 91.5% (195/213) who were white collar workers, and 69.5% (148/213) with an employment duration of more than 5 years. Participants had a mean age of 40.3 (SD 9.1) years, worked on average 8.0 (SD 0.7) hours/day, and had a mean BMI of 23.9 (SD 3.4) kg/m2.
Baseline characteristics for the 3 study groups.
Variables | Tailored group (n=78) | Generic group (n=84) | Control group (n=51) | Group comparisons | ||
Sociodemographic variables | ||||||
Age in years, mean (SD) | 40.5 (8.6) | 40.7 (9.7) | 39.3 (9.0) | .65 | ||
Males, n (%) | 25 (32.1) | 27 (32.1) | 15 (29.4) | χ22,212=1.9 | .76 | |
High school/university education: n (%) | 58 (75.3) | 70 (83.3) | 46 (90.2) | χ22,212=4.8 | .09 | |
Work-related variables | ||||||
Hours at work, mean (SD) | 8.0 (0.9) | 8.0 (0.6) | 8.0 (0.6) | .70 | ||
White collar occupational status, n (%) | 74 (96.1) | 75 (89.3) | 46 (90.2) | χ22,212=2.8 | .24 | |
Employment duration>5 years, n (%) | 55 (71.4) | 56 (66.7) | 37 (72.5) | χ22,212=0.7 | .72 | |
Health-related variables | ||||||
BMIain kg/m2, mean (SD) | 24.2 (3.1) | 23.6 (3.5) | 23.7 (3.5) | .48 | ||
Walking time in minutes/day, mean (SD) | 18.8 (28.3) | 21.1 (21.6) | 18.6 (19.0) | .74 | ||
Moderate-intensity PAbin minutes/day, mean (SD) | 24.7 (26.9) | 19.3 (20.1) | 18.0 (19.0) | .19 | ||
Vigorous-intensity PA in minutes/day, mean (SD) | 8.4 (11.5) | 11.6 (15.6) | 9.9 (15.6) | .35 |
aBMI: body mass index.
bPA: physical activity.
At baseline, all 78 participants in the tailored group completed section 1 (100%). The average time needed to complete the assessment survey was 16.3 minutes. Time spent on the first advice page was on average 20.1 minutes. A total of 66/78 participants completed section 2 (84.6%), 64/78 completed section 3 (82.1%), 60/78 completed section 4 (76.9%), 59/78 completed section 5 (75.6%), and 54/78 completed an action plan (69.2%).
Mean self-reported sitting at baseline (T0), 1-month follow-up (T1), and 3-month follow-up (T2) for the 3 groups and time × group effects.
Group | T0 | T1 | T2 | |||||
Total sitting in minutes/day, mean (SD) | ||||||||
Total workday sitting | T0–T1–T2: |
<.001*** | ||||||
Tailored (n=36) | 507 (104) | 480 (128) | 425 (110) | T0–T1: |
.24 | |||
Generic (n=64) | 457 (107) | 444 (105) | 437 (95) | T0–T2: |
<.001*** | |||
Control (n=28) | 449 (126) | 434 (131) | 469 (92) | |||||
Total nonworkday sitting | T0–T1–T2: |
.31 | ||||||
Tailored (n=36) | 141 (70) | 139 (69) | 132 (70) | T0–T1: |
.54 | |||
Generic (n=64) | 130 (63) | 131 (67) | 141 (77) | T0–T2: |
.32 | |||
Control (n=28) | 123 (58) | 117 (47) | 134 (55) | |||||
Domain-specific sitting in minutes/day, mean (SD) | ||||||||
Sitting at worka | T0–T1–T2: |
<.001*** | ||||||
Tailored (n=33) | 338 (107) | 279 (92) | 259 (88) | T0–T1: |
<.001*** | |||
Generic (n=61) | 288 (59) | 279 (64) | 280 (69) | T0–T2: |
<.001*** | |||
Control (n=24) | 281 (65) | 280 (50) | 288 (48) | |||||
Sitting during transportb | T0–T1–T2: |
.77 | ||||||
Tailored (n=33) | 78 (84) | 103 (124) | 58 (49) | T0–T1: |
.63 | |||
Generic (n=61) | 66 (79) | 60 (67) | 48 (31) | T0–T2: |
.98 | |||
Control (n=24) | 81 (106) | 74 (88) | 62 (62) | |||||
Television viewingb | T0–T1–T2: |
.23 | ||||||
Tailored (n=33) | 100 (57) | 104 (56) | 106 (61) | T0–T1: |
.29 | |||
Generic (n=61) | 95 (62) | 92 (67) | 102 (68) | T0–T2: |
.29 | |||
Control (n=24) | 91 (68) | 79 (68) | 82 (61) | |||||
Personal computer useb | T0–T1–T2: |
.20 | ||||||
Tailored (n=33) | 50 (46) | 53 (47) | 47 (29) | T0–T1: |
.82 | |||
Generic (n=61) | 51 (40) | 52 (44) | 51 (39) | T0–T2: |
.26 | |||
Control (n=24) | 58 (62) | 59 (71) | 69 (65) | |||||
Other leisure time sittingb | T0–T1–T2: |
.12 | ||||||
Tailored (n=33) | 101 (42) | 90 (44) | 75 (32) | T0–T1: |
.19 | |||
Generic (n=61) | 99 (61) | 98 (61) | 97 (46) | T0–T2: |
.03* | |||
Control (n=24) | 95 (48) | 96 (43) | 102 (64) |
aAverage on workday.
bAverage of workday and nonworkday.
*
Analyses of the domain-specific sitting data showed that changes over time in sitting at work and other leisure time sitting differed significantly between the 3 groups. There was a decrease in sitting at work in the tailored group, which was significantly greater than the changes in the generic group (T0–T1:
Mean objectively measured variables at baseline (T0), 1-month follow-up (T1), and 3-month follow-up (T2) for the 3 groups and time × group effects.
Group | T0 | T1 | T2 | |||
Total sitting time awake in hours/day, mean (SD) | T0–T1–T2: |
.69 | ||||
Tailored (n=35) | 576 (109) | 600 (91) | 607 (117) | T0–T1: |
.60 | |
Generic (n=35) | 578 (101) | 574 (103) | 576 (109) | T0–T2: |
.35 | |
Control (n=23) | 605 (96) | 616 (115) | 623 (100) | |||
Sitting at work in % work hours, mean (SD) | T0–T1–T2: |
.93 | ||||
Tailored (n=35) | 66.8 (15.5) | 71.7 (14.0) | 69.0 (13.7) | T0–T1: |
.76 | |
Generic (n=35) | 69.0 (13.8) | 71.2 (15.1) | 68.8 (15.1) | T0–T2: |
.89 | |
Control (n=23) | 74.3 (15.5) | 78.3 (11.1) | 74.8 (13.5) | |||
Standing at work in % work hours, mean (SD) | T0–T1–T2: |
.98 | ||||
Tailored (n=35) | 24.7 (13.5) | 22.2 (9.0) | 23.6 (11.7) | T0–T1: |
.95 | |
Generic (n=35) | 24.4 (11.3) | 22.7 (15.4) | 24.3 (14.4) | T0–T2: |
.90 | |
Control (n=23) | 16.3 (9.3) | 17.1 (7.9) | 17.8 (9.0) | |||
Breaks at work in no/work hour, mean (SD) | T0–T1–T2: |
.09* | ||||
Tailored (n=35) | 3.8 (1.5) | 3.7 (1.3) | 4.3 (1.6) | T0–T1: |
.40 | |
Generic (n=35) | 3.6 (1.3) | 3.6 (1.4) | 3.5 (1.3) | T0–T2: |
.11 | |
Control (n=23) | 3.0 (1.4) | 3.2 (1.4) | 3.3 (1.6) |
*
To our knowledge, this is the first randomized controlled study evaluating 1-month and 3-month follow-up effects of a theory-driven, Web-based, computer-tailored intervention to reduce or interrupt sitting among employees. Results are promising, with positive intervention effects on self-reported sitting time at work, self-reported sitting time during leisure, and objectively measured breaks at work. For these outcomes, the tailored intervention had superior effects to those of the control and the generic condition, confirming our hypothesis. This suggests the significance of computer tailoring in targeting sedentary behavior, as also seen in Web-based advice for other health-related behaviors, such as physical activity and diet [
It should be noted that the positive findings concerning the decrease in self-reported sitting duration were not reflected in the objective measures, as no effect was found on activPAL-measured total sitting time or sitting time at work. This result emphasizes the importance of combining self-reported and objective measures. The effectiveness study of sit-stand workstations conducted by Chau et al [
Compared with baseline, the self-reported work-related sitting time in the tailored condition was lower at 1-month follow-up (-59 minutes/day) and 3-month follow-up (-79 minutes/day), while this was not the case for the other conditions. These reductions in sitting time are similar to those seen in interventions implementing activity-permissive workstations (-77 minutes/8-hour workday) as shown in the review of Neuhaus et al [
In our subsample wearing the activPAL, we found a positive intervention effect at follow-up for breaks during work (+0.4 breaks/work hour, ~3.2 breaks in an 8-hour workday, an increase from 30 to 34 breaks in an 8-hour workday) in the tailored condition compared with no change in the generic condition. In the review of Martin et al [
It may be surprising that an individual-based intervention such as our computer-tailored intervention resulted in such a relatively high reduction in self-reported sitting time at work. In the case of implementing sit-stand workstations, it is reasonable that sitting time is substantially reduced, as the environment is changed to do so, but without standing desks one could expect the intervention to have less effect on total sitting duration and more on the sitting pattern. Based on the feasibility and acceptability study of our computer-tailored intervention, pointing out that employees perceived interrupting sitting to be more achievable than reducing workplace sitting, this could also be expected to be the case here [
We found no other 1-month follow-up effects on self-reported outcomes (sitting during transport, television viewing, computer use, leisure time sitting) or other objectively measured outcomes (total sitting time, sitting and standing time at work). However, we did find significant and positive 3-month follow-up intervention effects for self-reported total workday sitting and self-reported leisure time sitting. The change over time was more positive in the tailored condition than in the other conditions. The fact that leisure time sitting decreased (-26 minutes/day) from baseline to follow-up was surprising, as the advice mostly focused on work-related aspects: work hours, commuting, and (lunch) breaks. Still, this may mean that employees transferred the information and tips regarding one specific setting (work) to another (leisure). The study of Chau et al [
Our study has several strengths and limitations to take into account. The first strength is the randomized controlled design with a large sample of employees relative to other workplace interventions focusing on sedentary behavior. Second, the use of the activPAL as an objective measure for the outcomes was a strength, as self-reported measures can have recall and social desirability biases. However, as stated earlier, the first limitation is that we used this monitor only in a subsample of employees willing to wear the monitor, which is probably a result of the field-based approach of this study. Further, from a methodological point of view, it would have been more suitable to randomly allocate the monitors within the total sample in order to avoid sampling bias. However, from a compliance point of view, we believed it was better to provide a monitor only to those willing to wear one, in order to limit dropout. The dropout rate was lower (17/122, 13.9%) in the group wearing a monitor (and completing the questionnaires) than in the sample only completing the questionnaires (80/213, 37.6%). A study on the retention rates in physical activity interventions in workplace, health care, and home- or community-based settings revealed a mean retention rate of 78%, with minimal differences between intervention settings [
This study opens perspectives for future research. The effect of this intervention on psychosocial correlates should be tested, including the mediating effect of the change in these factors on the behavioral effects. As it stands, we do not know what the active intervention components are. Further, future research should investigate whether this tailored intervention would be more effective in combination with other (environmental) strategies, for example the use of sit-stand desks. Previous interventions also chose multicomponent programs to tackle the problems of too much sitting [
To our knowledge, this is the first intervention study to describe the effectiveness of a theory-driven, Web-based, interactive computer-tailored intervention aimed at reducing and interrupting sitting at work. The computer-tailored approach showed promising outcomes to address sitting time, as the tailored intervention was successful in decreasing self-reported sitting time at work and during leisure time, and in increasing objectively measured breaks at work compared with the generic and control conditions, which had no significant impact. This suggests that this computer-tailored intervention might have potential to contribute to the health promotion field.
Screenshots of the start to stand tool.
CONSORT-EHEALTH (V 1.6.1) checklist.
body mass index
International Physical Activity Questionnaire
specific, measurable, attainable, relevant, and time-bound
baseline
1-month follow-up
3-month follow-up
Workforce Sitting Questionnaire
KDC conceived the study, participated in its design and coordination, analyzed the data, and drafted the manuscript. IDB, GC, and CV conceived the study, participated in its design, and assisted in drafting the manuscript. All authors read, revised, and approved the final manuscript.
KDC is supported by the Research Foundation Flanders (FWO) (postdoctoral research fellowship: FWO11/PDO/097).
None declared. The authors are the developers of the intervention.