The effectiveness of sit-stand workstations for changing office workers’ sitting time: results from the Stand@Work randomized controlled trial pilot

Participants

Participants were staff from a non-government health agency in New South Wales, Australia, aged over 18 years, employed at least three days per week, and who had sufficient English language proficiency to undertake study tasks. The project was advertised to staff as part of their workplace wellness program via internal mail, staff meetings and information fliers in the office. Staff members who were interested in participating, contacted the research team and received additional project information and an expression of interest form. They could then join the study ballot by returning the expression of interest form.

Design

This pilot study had a randomised controlled trial design with crossover and rolling recruitment and a waitlist control condition (Figure 1). Eligible staff members who returned an expression of interest form were randomly drawn from a ballot by a researcher in the presence of potential participants and other researchers, and were included in the study after providing written informed consent. The first four participants drawn from the ballot were allocated to the intervention group to use a sit-stand workstation for four weeks (Figure 1, study group 1), the next four participants drawn from the ballot served as the control group (Figure 1, study group 2). The remaining participants were assigned to the waitlist control condition and were placed on the waiting list in seven groups (four to five people per group). After the initial four weeks, the previous control group (study group 2) received the intervention with the next group from the ballot draw serving as their controls (study group 3). This was repeated until all nine groups had received the intervention. This study design was used to maximise the evaluation sample size taking into account the five available sit-stand workstations. Allocation concealment was not possible due to the open plan nature of the study office environment. Data collection ran from September 2011 to July 2012.

Intervention

The intervention involved giving participants a sit-stand workstation (Ergotron Workfit S) to use for four weeks for all workdays (3 to 5 workdays per week depending on full-time or part-time employment status). The sit-stand workstation allowed office workers to vary their posture throughout the workday between sitting and standing. Prior to receiving the sit-stand workstation, participants received brief training on how to use the sit-stand workstation, including how to adjust workstation height and manufacturer specified ergonomic set ups while sitting or standing. Participants also completed an ergonomic self-assessment prior to use and were advised to increase their standing time gradually to prevent injury and fatigue.

Outcome assessment

Assessments, consisting of objective monitoring and self-report questionnaires, took place at three time points scheduled four weeks apart. Assessment 1 was six weeks pre-intervention, assessment 2 at two weeks pre-intervention and assessment 3 in the third week of the intervention (Figure 1). Changes between assessments 1 and 2 were treated as the control condition, changes between assessments 2 and 3 as the intervention condition. The exception was the first intervention group, which only participated in assessments 2 and 3, because they could not serve as a time synchronized control condition.

During all three assessments participants wore an ActivPAL activity monitor (model ActivPAL3; PAL Technologies Ltd., Glasgow, UK) for the working week during work time and kept a monitoring log book to note the times they wore the ActivPAL each day, the days they worked, and the times they started and finished work on each of those days. We used the ActivPAL to objectively measure time spent sitting, standing and stepping at work. The ActivPAL is a small activity monitor that is worn on the front mid thigh [31],[32]. The ActivPAL identifies episodes of sitting/lying, standing or stepping based on the inclination of the thigh. It has demonstrated reliability and validity for measuring postures and activities of daily living (sitting, standing, stepping) [31]: good to excellent interdevice reliability (intraclass correlation coefficients ranged from 0.79 to 0.99); good device-observer agreement for sitting, standing and stepping (mean percentage difference between ActivPAL and observer of 0.19%, -1.4% and -2.0%, respectively, on Bland-Altman plots). ActivPAL has also been shown to be responsive to change [33]. ActivPAL data were first processed using proprietary software (ActivPAL v6.1.2.17, PAL Technologies Ltd., Glasgow, UK) which classified the data into sitting, standing and stepping categories. We further analysed the data using custom software (HSC analysis software v2.19, Philippa Dall and Malcolm Granat, Glasgow Caledonian University) which allowed us to isolate participants’ work time data based on their work start and finish times reported in their monitoring log.

Additionally, participants completed questionnaires at each assessment. We used the Occupational Sitting and Physical Activity Questionnaire (OSPAQ) [34] to measure time spent sitting, standing, walking, and doing more physically demanding tasks at work. It has demonstrated excellent test-retest reliability for assessing sitting, standing, walking and doing physically demanding tasks at work (ICC = 0.54-0.89) [34]. Validity correlations for occupational sitting, standing and walking measures were 0.52, 0.49 and 0.27, respectively, against Actigraph accelerometers [34].

Sitting time over the whole day was measured with the Workforce Sitting Questionnaire (WSQ) [35], which measures total and domain-specific sitting based on work and non-workdays in working adults. The WSQ asked participants to report their time spent sitting (1) while travelling to and from places; (2) while at work; (3) while watching TV; (4) while using a computer at home; and (5) while doing other leisure activities on a workday and a non-workday in the last 7 days. It has acceptable test-retest reliability (ICC = 0.76) and validity against Actigraph accelerometers (r = 0.40) for assessing total sitting time based on work and non-workdays [35]. The WSQ is also a reliable (ICC = 0.63) and valid (r = 0.45) measure of domain-specific sitting at work on a workday [35].

Participants also provided information about their sex, age, height, weight, employment status (full time or part time, number of days worked, and hours worked per week), type of office arrangement (own office or open plan) and highest level of education. Additionally, Body Mass Index (BMI) was calculated as kg.m^-2.

Statistical analyses

Statistical analyses were carried out with SAS/STAT software, Version 9.3 of the SAS system for Windows with statistical significance set at p < 0.05 (two-tailed). We used linear mixed models to estimate changes in self-reported and objectively assessed time spent sitting, standing and walking/stepping before and after the use of a sit-stand workstation. We used the SAS procedure, PROC MIXED, to estimate the effect of the intervention, allowing adjustments for measurement time, potential confounders, clustering of participants within groups, and repeated measures on the same participant. The model included the fixed effects of assessment time (assessment 1, assessment 2, assessment 3), as well as the average group measurement time, a blocking variable used to account for the effect of time. Estimates were adjusted for the following potential confounding variables: age, BMI (continuous), office type (open plan or own office), highest level of education completed (high school; trade, technical certificate or diploma; university; or other), and full time or part time work. Participant, nested within group number (order in which group received intervention), was included as a random component to adjust for within-subject correlation. For the control condition, we defined change in outcome as the difference in the adjusted estimates of outcome measures at assessment 1 and assessment 2 (assessment 2 minus 1); for the intervention condition, this change was defined as the difference in the adjusted estimates of outcome measures at assessment 2 and assessment 3 (assessment 3 minus 2). ActivPAL data were considered valid when the participant wore the device for at least 75% of their time at work. Analyses of ActivPAL outcomes involved participants with valid data, and who worked full time with at least 2 days of wear time at work or worked part time with at least 1 day of wear time at work.

Changes in time spent sitting at work

Initial analyses with only adjustment for study group and measurement time (Additional file 1: Table S1) produced the same pattern of results, except objectively measured stepping time at work increased significantly in the intervention condition (estimate, 13 min/day (95%CI: 1, 25), p = 0.035) but not in the control condition (estimate, 3 min/day (95%CI: -9,15), p = 0.645); however, the comparison between intervention and control was not significant (p = 0.316).

Changes in time spent sitting over the whole day

For self-reported sitting time on a work day, we found significant reductions in total sitting time in the intervention condition but not in the control condition, although the difference between the two conditions was not significant (p = 0.179) (Table 3). There were also significant changes in self-reported time spent watching TV on a work day in both groups: we observed an increase in workday TV-viewing in the control condition and a reduction in workday TV-viewing in the intervention condition (between condition comparison, p = 0.007). Time spent watching TV on a non-workday was also significantly different between the intervention and control conditions (p = 0.019) with a decrease observed in the intervention condition and no change found in the control condition. There were no changes in self-reported sitting time in other domains on workdays or non-workdays. Preliminary analyses only adjusting for study group and measurement time yielded the same pattern of results (Additional file 1: Table S1).

Strengths and limitations

The strengths of this study include having a randomized controlled cross over design, using objective measurement of sitting and standing, and having strong staff and management support for evaluation. The novel study design made use of limited resources to inform organisational decision making related to workplace wellness practices. Limitations include short term follow-up and a lack of blinding to study allocation, which was not possible due to the open plan office setup. Additional follow up would have strengthened results, but this was not possible as participants could only trial the sit-stand workstations for four weeks and we did not have the opportunity to re-assess their behaviours at 12 months. Furthermore, Stand@Work involved a convenience sample of participants working in a health-related field who were mostly female and had tertiary education levels, similar to previous sit-stand desk intervention studies [23],[25]-[27]. Future exploration is needed to see whether these results would be found in other office/desk-based worker groups, such as those with non-tertiary education levels, in conjunction with formative work to determine the feasibility and acceptability of the sit-stand workstation as a strategy for other non-health related workplaces. For example, it has been reported that uptake of sit-stand workstation use varied depending on whether the devices were electronically or manually operated [28], while the `hot’ desk strategy only had partial uptake [24], even though people who used the sit-stand `hot’ desk only used it in the standing position.

Given the increasing awareness about the potential health risks from extended occupational sitting, and the substantial focus on workplace wellness, these study findings can potentially inform employer decisions about environmental approaches to reducing their employees’ sitting time at work. While the evidence on sitting, health and occupational outcomes is not yet sufficient to develop a business case for wide scale investment in sit-stand workstations; specifically, there is a need for ongoing intervention evidence, conducted in a variety of office settings, to inform current policy and practice. Up front cost is a likely barrier for many workplaces to roll out this type of intervention on a larger scale, so demonstrating longer term effectiveness is important. Evaluating other lower-cost behavioural strategies aimed at reducing extended sitting at work is also warranted, to determine whether similar reductions in sitting time can be achieved. Investigating the effects of changing sitting behaviour on work-related outcomes, such as productivity, is required as well. From a policy perspective, the message to reduce occupational sitting time is additional to the existing evidence that supports ongoing investment in workplace physical activity promotion.