Trained and experienced researchers implemented the same standardized research protocol at each site. Personal information (age, gender and contact details), and body mass index (kg/m2; calculated from researcher assessed height and body mass), were recorded pre-intervention (September at each site). Step counts were also reported (waking-to-bedtime), using a diary and an unsealed pedometer (Yamax SW-200) for five consecutive workdays (Monday-Friday) – employees kept and used their pedometers throughout the duration of the project. Workday sitting times (hours and minutes) were measured using a logbook with question format based on other logbook research . We adapted these questions to specifically refer to the context of the workplace and, to aid recall, employees reported the number of hours and minutes spent sitting, at the end of each morning (including lunch break) and afternoon work period. There is a lack of validity and reliability data for assessing sitting through logbooks, yet they are less dependent on long-term recall, and therefore in all probability more accurate, while providing a practical, less expensive alternative to objective monitors .
Protocols and Interventions
Pre-intervention workday step counts and block stratification were used to randomly and equally assign participants at each site to a waiting list control or one of two intervention groups. Control group participants were asked to maintain their normal behavior over a ten-week period (October-December at each site), while intervention participants were asked to increase their step counts. Employees in the first intervention group were directed to achieve this through brisk, sustained, route-based walking during work breaks. The second intervention group was asked to engage in incidental walking and accumulate step counts during working tasks – this strategy targeted walking and talking to colleagues, rather than sending emails or making telephone calls, and standing and walking in meetings, instead of sitting at desks. Importantly, participants in all groups were instructed not to engage in additional physical activities beyond those usually undertaken and – for route and incidental groups – the walking strategies encouraged in the workplace as part of intervention. Employees were asked to report additional activities or unusual workdays in their pedometer diaries.
Intervention strategies used an ecological approach to facilitate and support changes in walking and sitting behavior. Employees were asked to use their pedometer as a motivational and self-regulatory tool – as a general guide, those above 10,000 daily steps at pre-intervention were encouraged to maintain this level of workday walking and add additional steps where possible. We encouraged relative and progressive, weekly increases of at least 1000 steps/day in those below a 10,000 steps threshold – this approach was based on guidelines for minimal walking bouts , allied with practical considerations around workday time demands.
Detailed instructions on goals and strategies to effect change were provided prior to intervention and then reinforced through weekly group emails – these contained motivational messages and reminders for control employees to maintain normal behaviour. Suggested campus walks, supported by maps, times (10-to-45 minutes) and step counts were provided for the routes group. Incidental employees were encouraged to exploit their office physical environments as a means of increasing steps (i.e. using the toilet at the far end of their building, rather than next door to their office) and to engage managers in providing opportunities for walking in work-tasks (i.e. absenteeism from desks while delivering messages to colleagues).
Measurement Points and Analyses
Workday step counts and reported sitting times were re-assessed for five consecutive days at the beginning (week one), mid- (week five) and endpoint (week ten) of intervention. Three intervention measurement points were selected in an effort not to overload participants with continuous recording, while at the same time enabling assessment of impact at equidistant time points – measurement at week one was specifically included, thereby allowing insight into how acute effects may translate into mid and endpoint assessments.
Pre- and intervention data were inputted into SPSS (Version 15.0; SPSS Inc, Chicago, IL, 2006) by lead researchers at each university, who were instructed not to enter data from workdays that employees reported as unusual or not normal (e.g. when they were sick, on holiday or engaged in physical activities which were not part of their normal routine or intervention goals). Individual sites provided feedback to employees following intervention completion (January to March at each site); on request, control employees were also given access to intervention materials during this period.
Lead researchers forwarded SPSS files electronically to a coordinating researcher where data were pooled and treated in the following ways. Participants with missing step count and/or reported sitting data at pre-intervention were removed, as were those with missing data for two or more intervention measurement points (i.e. step counts for weeks one and five) – those with one missing measurement point were included in analyses, with imputation of the missing data point as an average of the remaining two measurement points. Using recently published guidelines , employees were classified as "highly active (<12,500 steps/day), active (10,000 – 12,499 steps/day), somewhat active (7,500–9,999 steps/day), low active (5,000–7499 steps/day) or inactive (<5000 steps/day)" – these categories were assigned for both pre-intervention and intervention (average of week one, five and ten) step counts. The last of these categories we termed "inactive", rather than the previously used term "sedentary" , in order to avoid confusion with sedentary behavior defined as sitting.
MANOVA was used to analyze interactive effects between gender, site and group, relative to age and BMI. Step count categories were used to contrast the magnitude of change in walking (intervention average – pre-intervention step counts) relative to daily steps at pre-intervention. Steps/day and daily reported sitting times were compared using a mixed factorial ANOVA with timeline (pre; week one; week five; week ten) as the within participant factor and gender (male; female), site (UK; Australia; Spain) and group (control; routes; incidental) as the between participant factors. Group by timeline interactions were used to identify significant intervention effects, with paired sample-t-tests for follow-up, simple effect analyses. Alpha was set at p < 0.05.