A workplace-based intervention to increase levels of daily physical activity: the Travel to Work cluster RCT

Aim and objectives

The current cluster RCT incorporated lessons learned from the Walk to Work feasibility study [National Institute for Health Research (NIHR) Public Health Research (PHR) project number 10/3001/04]. 49 The aim of the feasibility study was to build on existing knowledge and resources to develop an employer-led scheme to increase walking to work and to test the feasibility of implementing and evaluating it in a full-scale cluster RCT. The objectives were to (1) explore with employees and employers the barriers to, and facilitators of, employer-led schemes to promote walking to work, (2) use existing resources and websites to develop a Walk to Work information pack to train work-based Walk to Work promoters, and (3) conduct an exploratory RCT of the intervention to pilot workplace and employee recruitment procedures, examine retention rates, pilot cost and outcome measures and inform a sample size calculation for a full RCT.

Study design

The feasibility study comprised two phases of the Medical Research Council’s framework for evaluating complex interventions. 50 During phase I, a review of resources that promote walking (and in particular the benefits of walking to work) was undertaken. In addition, three focus groups were conducted with employees, and interviews were conducted with three employers, in one small, one medium-sized and one large workplace outside Bristol to finalise the intervention design. Phase II comprised an exploratory randomised trial incorporating process evaluation and an assessment of costs. A cluster trial was required because randomisation of individual employees would risk contamination of the control group: the intervention was to be delivered within workplaces with the potential for employees to share information about the intervention and in which employers would be encouraged to support walking to work.

Recruitment

Workplaces were approached through Bristol Chambers of Commerce for initial expressions of interest. Fifty-five workplaces expressed an interest and were asked to complete a short questionnaire about the size and type of the business. Because the intervention initially aimed to focus on employees within ‘walking distance’ of their workplace, employers were also asked to identify how many of their employees lived within 2 miles of the workplace. This process was aided by the research team supplying the first four digits of postcodes likely to contain employees living within the required range, and an instruction leaflet of how to calculate distance using the website walkit.com (accessed 28 February 2019). Nevertheless, some workplaces found this burdensome and it may have affected recruitment. Of the 55 workplaces initially expressing an interest, 19 were recruited and 17 completed the study. Two workplaces left the study after randomisation to the intervention group: one because of downsizing and one because of heavy workload.

Within participating workplaces, employees living within 2 miles of the workplace were given information about the study and invited to participate. As the study progressed, it was felt that this was too restrictive, and a second round of recuitment was undertaken to include people who lived further away and might be willing to incorporate some walking as part of a mixed-mode commute. A total of 187 participants were recruited to the study: 147 living within 2 miles of the workplace and 40 living further away. In the intervention group, study participants were asked to sign an additional consent form before receiving the intervention. This was also considered to be restrictive and was not thought necessary for a future full-scale trial.

It was felt that naming the study Walk to Work may have restricted interest among some employers and may have encouraged the control group to consider walking to work. Therefore, it was decided to name the full-scale trial the Travel to Work study and to name the intervention the Walk to Work intervention.

Data collection

At baseline, all participating employees were asked to complete a questionnaire giving basic personal data and providing information relating to travel behaviour, costs and health. Participants were also asked to wear accelerometers during waking hours for 7 days to provide an objective measurement of physical activity, and to carry a personal GPS (Global Positioning System) receiver during the commute to confirm the duration of the journey and quantify its contribution to overall physical activity. Post intervention, questionnaires were administered again to explore views and experiences of walking to work, and additional questions about the acceptability of the intervention were included for the intervention group only. The questionnaires, accelerometers and GPS receivers were administered again across the intervention and the control groups (as per the baseline protocol) at a 12-month follow-up data collection point. To examine key issues in more depth, baseline and post-intervention interviews were conducted with employers, Walk to Work promoters and a purposive sample of employees.

Aim

The focus of the trial was to evaluate the effectiveness of a workplace-based intervention to increase walking during the commute for adults working in urban and suburban areas.

Primary outcome

Does the intervention lead to an increase in the daily number of minutes of MVPA after 12 months compared with the control group?

Secondary outcomes

As well as the primary outcomes, there were several secondary objectives relating to physical activity and travel mode. These were:

1. Does the intervention lead to an increase in overall physical activity compared with the control group?

2. Does the intervention decrease the daily number of minutes of sedentary time compared with the control group?

3. Does the intervention lead to an increased number of journeys in which walking to work is the major mode of travel compared with the control group?

4. Does the intervention increase MVPA attributable to walking on the commute compared with the control group?

Economic outcomes

There were three key economic outcomes of interest:

1. What are the intervention costs to participating employers and employees?

2. Does the intervention lead to increased or decreased costs in terms of health-care use, commuting costs and productivity losses?

3. Does the intervention lead to improved well-being, as measured by the ICEpop CAPability measure for Adults (ICECAP-A) questionnaire?52

Process outcomes

The purpose of the process evaluation was to examine the context of, delivery of and response to the intervention from the perspectives of employers, Walk to Work promoters and employees. There were two main outcomes of interest:

1. What were the barriers to, and facilitators of, walking during the daily commute?

2. Was there evidence of any social patterning in the uptake of the intervention, particularly in relation to socioeconomic status, age and gender?

Sample size and justification

Using the findings from the feasibility study, the sample size for the full-scale trial was based on an average cluster size of eight, an ICC of 0.15 and participant attrition of 25%. The calculation needed to allow equal numbers of workplaces in the intervention and the control groups. We calculated that we needed 339 participants per study group to detect a 15% difference in MVPA levels (equal to a difference of 0.36 SDs) with 80% power at the 5% significance level. Therefore, 678 employees were required from 84 workplaces.

South Gloucestershire

South Gloucestershire, in the south-west of England, comprises multiple towns and population centres to the north and east of the city of Bristol. In 2016, the population of South Gloucestershire was estimated to be 277,600, of whom 174,700 people (62.9%) were aged 16–64 years and 146,700 (52.8%) were economically active. 53 Key employers are local and national government departments, engineering and manufacturing industries and large insurance companies. 54 Many employers are located between the northern edge of Bristol and the M5 motorway. This area includes a large regional shopping centre and surrounding retail and business parks. South Gloucestershire contains a network of roads serving the industries, distribution centres and retail centres in the area. The railway network in South Gloucestershire is connected to major cities and towns across the UK.

Bath

Bath is a city in the south-west of England. In 2016, the population aged 16–64 years in the parliamentary constituency of Bath (which includes the city and surrounding suburbs) was estimated to be 64,900, of whom 49,300 people were economically active. 55 The city has strong software, publishing and service-oriented industries. Other important economic sectors in Bath are education, health, retail, tourism and leisure and business and professional services. 56 Major employers are the NHS, the city’s two universities and Bath and North East Somerset council. In an attempt to reduce car use in the historical centre of Bath, park-and-ride schemes have been introduced through which car drivers are encouraged to use car parks on the edge of the city and travel into the centre by bus. Nevertheless, underground city-centre parking was also provided for a recent large shopping centre development. Bath is served by a main railway station providing connections to major cities, as well as some suburban railway services and a network of bus routes to surrounding towns and cities.

Swansea

Swansea is a coastal city and county in south Wales and has the second highest population of the 22 Welsh local authorities. 57 In 2016, the population of the Swansea local authority area was 244,500, of whom 155,300 people (63.5%) were aged 16–64 years and 113,500 were economically active. 57 The Swansea economy has a proportionately large share of jobs in the public administration, health, hospitality, financial services and retail sectors. Of the people in employment, an estimated 87.5% are employed in the service sectors, with 28.2% working within the public sector. Other main business activities are in the construction and scientific and technical sectors. The M4 motorway and several major trunk roads link Swansea to other cities in Wales and England. Bus services include smaller bus and coach operators, a road-based rapid transport route and two park-and-ride services. 58 There is also a main railway station and three smaller suburban stations.

Bristol

Bristol is a city and county in the south-west of England. In 2016, the population of the city of Bristol was 454,200, of whom 309,900 people (68.2%) were aged 16–64 years and 255,400 were economically active. 59 The main employment opportunities are in wholesale and retail, health and social work, administrative and support services and professional, scientific and technical activities. Bristol is connected to London and other major UK cities by two motorways and connecting major roads, as well as through its main railway station. Sustrans (www.sustrans.org.uk; accessed 19 December 2017), a charity promoting sustainable transport, was founded in Bristol. In 2015, Bristol won the European Union’s European Green Capital award (www.bristol2015.co.uk; accessed 19 December 2017), in which sustainable transport was an important focus.

Swindon

Swindon is a large town in the south-west region of England. In 2016, the Swindon area had a population of 217,900, of whom 139,800 people (64.2%) were aged 16–64 years and 120,600 were economically active. 60 The majority of employment opportunities are in manufacturing (including car production plants), wholesale and retail, administrative and support services, health and social work, and finance. Swindon is on the main railway line linking London with the south west of England and south Wales. The town can be accessed by a strategic road network including two junctions along the M4 motorway. 61 The recent transport plan for Swindon indicates that the development of fast and efficient public transport has ‘lagged behind’ because of the relative ease of car use. 61

Neath Port Talbot

Neath Port Talbot is a County Borough and Unitary Authority in central south Wales. The area stretches from the south coast to the borders of the Brecon Beacons National Park. The majority of the population lives in the principal towns of Neath and Port Talbot. In 2016, the population of Neath Port Talbot was recorded as 141,600, of whom 88,000 people (62.1%) were aged 16–64 years and 67,300 were economically active. 62 Regeneration ‘to make Neath Port Talbot a place that is better connected, better for business and a better place to live’ is an important theme for the county borough council (reproduced with permission from Neath Port Talbot Council63). The main employment opportunities are in manufacturing (including steelworks), wholesale and retail trades, and human health and social work. The towns of Neath and Port Talbot both have railway stations, connecting them to major cities in Wales and England, and are also served by a network of buses. The M4 motorway cuts through Port Talbot, linking it to towns and cities along the M4 corridor. The Port Talbot docks complex is used mostly for the import of iron ore and coal for use by the nearby steelworks.

Newport

Newport is a city and unitary authority area in south-east Wales. In 2016, the population of the city was 149,100, of whom 93,100 people (62.4%) were aged 16–64 years and 73,100 were economically active. 64 Newport was once Wales’ largest coal-exporting port, but the docks declined in importance during the 20th century. The main employment opportunities are now in manufacturing, the wholesale and retail trade, and human health and social work. Newport lies within the M4 corridor and is accessed by a network of major roads. A railway line passes through the centre of the city and a network of buses also serves the city. 65

Ethics committee approval

As the study was not a clinical trial, and did not involve patients or users of the NHS, it was not necessary to apply for NHS ethics approval. All protocols and relevant paperwork were submitted to the Faculty of Medicine and Dentistry Research Ethics Committee at the University of Bristol in February 2014 and ethics approval was granted on 20 April 2014.

Adverse events

The Walk to Work intervention was considered low risk, involving generally healthy adults, and no adverse events had been reported during the feasibility study. However, there are potential risks to pedestrians in relation to road traffic safety and personal safety. Furthermore, it was possible that people with low activity and no history of walking might suffer initial muscle stiffness. In most cases, this would be mild and a normal consequence of increased physical activity, but participants were made aware that some symptoms might require medical attention, for example when underlying joint weakness is exposed. Such incidents were monitored throughout the trial.

Participating employers, Walk to Work promoters and employees were provided with guidance about adverse events and how to report them. Because available adverse events forms for health research tended to relate to clinical trials rather than low-risk public health interventions, bespoke forms were designed for the current study and agreed with the University of Bristol Research Governance and Ethics officer (see Appendix 1).

It was agreed that adverse events would be recorded by the key researcher for each site and collated by the study manager and principal investigator. When appropriate, adverse events would be reported to the University of Bristol Research Ethics Committee and the chairperson of the Trial Steering Committee. If adverse events were attributable to the intervention, relevant participants would be informed immediately (e.g. if an incident had happened on a particular route, work colleagues using the same route would be provided with relevant information).

It was also acknowledged that Walk to Work promoters might experience difficulties due to disruption to usual working relationships or employers’ concerns about time taken out of usual work activities. The intervention activities might also present problems for employers and employees, such as disruption to work routines as a result of permitting elements of the intervention during working hours. These issues were considered through the qualitative research undertaken as part of the process evaluation.

Participant recompense

A small amount of recompense, a £10 gift voucher, was given to study participants who returned accelerometers and GPS monitors at the baseline and the 12-month data collections, in recognition of their contribution to the research. Interview participants were also given a £10 gift voucher. This was handled discretely by providing relevant individuals with a plain envelope containing the gift voucher thanking them for their help with the study.

Data storage

All data relating to workplaces and research participants are stored at the University of Bristol in accordance with the Data Protection Act 201866 and University of Bristol research governance requirements. Information collected from the paper questionnaires was transferred onto the study database, which is held on secure file storage at the University of Bristol and protected by a combination of user accounts and file access control lists, limiting access to agreed members of the research team. The data set will be kept, with limited access by agreed members of the research team, for 10 years from the end of the study. Following the transcription of interview recordings, all potentially identifiable personal information was removed to ensure anonymity. Personal information required for routine contact was stored on a separate database that could be linked using the unique participant identification number. The hard copies of the questionnaires and consent forms will be kept at the University of Bristol in a locked filing cabinet for at least 3 years after termination of the study.

Trial management and scrutiny

An overview of the study management structure is provided in Figure 1.

FIGURE 1.

Travel to Work study management structure.

Trial Steering Committee

The Trial Steering Committee met with key members of the research team twice a year throughout the study and comprised an independent chairperson, statistician and health economist plus additional independent experts in travel behaviour, research evaluation, physical activity and health promotion.

Study co-applicant group

The research co-applicant group met on a quarterly basis throughout the study and was chaired by the principal investigator. The group comprised co-applicants from each of the research sites (Bristol, Bath and Swansea), the study manager, a representative of the Bristol Randomised Trials Collaboration (BRTC) and co-applicants with expertise in physical activity measurement, statistical analysis, health economics and sustainable transport.

Research team

A research co-applicant and researcher were based at each research centre (Bristol, Bath and Swansea) and the team met on a quarterly basis. There was frequent and ongoing communication, by e-mail or telephone, between the research sites throughout the trial.

Study registration

The study was listed on the ISRCTN (International Standard Randomised Controlled Trial Number) registry with trial number ISRCTN15009100. 67

Lessons from the feasibility study

Following the feasibility study,49 it was felt that the workplace recruitment process should be simplified so that employers were not required to calculate the number of employees who lived within 2 miles of the workplace before being recruited to the study. This requirement had appeared to be onerous for some workplaces that did not sign up for the study after an initial expression of interest. It was also thought that participant eligibility should be expanded to include people living further away from the workplace who might be willing to incoporate walking as part of a mixed-mode commute. Nevertheless, information about the length of the commute was considered valuable to assess how many participants were within a reasonable walking distance (defined as 2 km for the full-scale trial) and might be encouraged to walk the full distance, and how many might be more suited to a mixed-mode commute. The survey questionnaires, following participant consent, were considered to be the most suitable method of collecting postcodes and enabling study researchers to calculate the distances between home and workplace.

Attrition during the feasibility study (29% at the 12-month follow-up) compared well with other workplace-based physical activity interventions. However, no attempt had been made during the feasibility study to contact employees who left the workplace for other employment between baseline and follow-up. To reduce attrition during the main trial, it was agreed that researchers would ask consenting participants to provide contact details if they left the workplace before the 12-month follow-up. This would enable data-collection packs to be sent to them and returned to the researchers by post.

Eligibility criteria

The following exclusion criteria applied at the workplace level: (1) workplaces with a large proportion of staff on short-term or zero-hours contracts, or workplaces with plans to significantly downsize or relocate during the study period, as follow-up data might not be achievable, and (2) workplaces with fewer than five employees, as there was limited potential to recruit a sufficient number of participating employees into a workplace cluster. All employees within participating workplaces were eligible to take part unless they met any of the following exclusion criteria: (1) they already always walked or cycled to work, (2) they were disabled in relation to walking, (3) they were due to retire before the 12-month follow-up data collection or (4) daily driving was a key part of their role.

Workplace recruitment and consent

The aim was to recruit 84 workplaces of different sizes and industrial classifications. Workplace recruitment took place in two phases during May to July 2015 and March to May 2016. The initial intention was to recruit across three urban areas in south-west England and south Wales: South Gloucestershire, Bath and Swansea (Table 2).

For the 2015 recruitment round, lists of employers were obtained from relevant Chambers of Commerce and local authorities. The BRTC generated a list of random numbers (blinded to workplace addresses and contact details) for half of the workplaces to be invited to express an interest in the study, with the intention of sending information out to the remaining workplaces in 2016. However, recruitment in 2015 was insufficient to reach the required sample size. Following discussions within the co-applicant group, and subsequently with the Trial Steering Committee and the funders (the NIHR PHR programme), it was agreed that four additional areas would be included in the second recruitment round. In 2016, employers in the other half of the lists for South Gloucestershire, Bath and Swansea were sent information about the study, in addition to all workplaces on available lists of employers for Bristol, Swindon, Neath Port Talbot and Newport.

Workplace recruitment packs

Workplaces were sent a letter of invitation together with an information leaflet and a short form to return to the research team to express interest and provide basic information about the workplace (see Appendix 2).

Following expressions of interest, eligible workplaces were contacted by telephone or e-mail and a meeting was arranged with one member of the research team at which the trial was explained in more detail and written consent was sought for participation in the study.

Participant recruitment and consent

Employers within participating workplaces were provided with an information leaflet (see Appendix 3), describing the study and eligibility criteria, to distribute to all their employees. All eligible employees were invited to participate in the study and given consent forms for their individual participation. This consent was provided before the baseline data collection and subsequent randomisation.

Baseline characteristics

Study participants were asked to complete baseline questionnaires, which included questions about their sociodemographic characteristics, mode of travel to work and occupation.

Workplace recruitment

Approximately 9800 invitation packs were sent out over the two recruitment phases (Table 3). Because of time and resource constraints, it was not possible to check whether or not all the workplaces on the lists were still in existence, or whether or not the packs reached an appropriate person with the authority to decide about workplace participation. For example, of 1892 invitations packs sent out in 2015, 114 (6%) were returned undelivered.

Workplaces expressing an interest were checked for eligibility by the research team and those meeting the criteria were contacted and provided with further information about the study. As a result of this process, only 29 workplaces were recruited in the first year (Table 4). The number of areas was increased, and a further 58 workplaces were recruited in the second year, making a total of 87 workplaces. This was in line with the target of recruiting 84 workplaces, but there was not an even spread of workplaces across the different areas and the numbers ranged from 35 in Bristol to three in Neath Port Talbot (see Table 4).

Reasons for workplace withdrawal following expression of interest

Overall, 271 workplaces expressed an interest in the study. These were checked for eligibility and provided with an opportunity to discuss study participation in more detail. At this stage, 162 workplaces did not continue with the study. Reasons for workplaces not consenting to participation after expressing interest are summarised in Table 5: 28% were deemed ineligible by the research team because there were fewer than five employees, 19% reported that their employees already always walk or cycle, 18% felt unable to prioritise the study activities when they were explained in more detail, 12% had consulted their staff and found a lack of interest in participating and 11% indicated that their employees needed to drive as part of their job.

Following workplace consent, a further 22 workplaces withdrew before the baseline data collection (Table 6). The main reason for workplaces withdrawing at this stage of the study related to lack of interest among the workforce, despite their employer having an interest in participating.

Workplace characteristics

Workplaces were diverse in relation to their function and included public administration, professional and scientific organisations, retail, services and manufacturing (Table 7). 68 The workplaces varied in size: 10 (11.5%) were micro-sized (< 10 employees), 35 (40.2%) were small (10–49 employees), 22 (25.3%) were medium-sized (50–249 employees) and 20 (23.0%) were large (≥ 250 employees). Table 7 shows a good balance of workplace characteristics between the two groups following randomisation.

Participant recruitment and characteristics

Across the 87 workplaces, 654 participants were recruited, with a mean cluster size of approximately 8 (as was the case during the feasibility study). 49 The number of participants in each workplace at baseline ranged from 1 to 28. The average age was 40 years and the majority of participants (455, 65.9%) lived in a household with an income of > £30,000 per year (Table 8). There was a slight balance in favour of females (n = 371, 56.7%) and being educated to at least degree level (377, 57.7%). A large majority of participants, 557 of 626 who gave this information (89%), lived > 2 km from their workplace, and two-thirds travelled to work by car at baseline. Table 8 shows a good balance of key participant characteristics in the intervention and the control groups following randomisation.

Data collection

Study participants were asked to complete travel diaries, to wear accelerometers (ActiGraph GT3X+) for 7 days during waking hours and to carry a personal GPS receiver (Qstarz BT-1000X), set to record positional data at 10-second intervals, during their commute. Participants were provided with instructions about how to use the monitors (see Appendix 4) and those who returned the equipment were provided with a £10 gift voucher to acknowledge their contribution to the study.

Objectively measured physical activity and main mode of travel during the commute

Raw accelerometer data were downloaded using ActiLife software (version 6.11.8; ActiGraph, Pensacola, FL, USA) and reintegrated to 10-second epochs for analysis. Reintegrated accelerometer data were processed using KineSoft data reduction software (version 3.3.80; KinesSoft, Loughborough, UK) to generate outcome variables. Continuous periods of ≥ 60 minutes of zero values were considered ‘non-wear’ time and removed. To be included in the analysis of daily physical activity and sedentary behaviours, participants were required to provide ≥ 3 days of valid accelerometer data of ≥ 600 minutes in duration. In relation to mode and physical activity during the commute, participants were required to provide at least 1 valid day of combined accelerometer and GPS (accGPS) data on a working day. Days on which cycling was identified as the main mode of travel to work were excluded owing to the inability of waist-worn accelerometers to accurately record physical activity during cycling. 71

Accelerometer and GPS data were combined for every 10-second epoch (accGPS) based on the timestamp of the ActiGraph data. The participant’s workplace and home were geocoded using the full postcode and imported into a GIS (ArcMap version 10.2.2; Environmental Systems Research Institute, Inc., Redlands, CA, USA). The merged accGPS files were imported into ArcMap and participants’ journeys to and from work were visually identified and segmented from other accGPS data using the ‘identify’ tool. Journeys were identified as a continuous sequence of GPS locations between the participant’s home and workplace, and therefore may include trips to other destinations (e.g. shopping).

Mode of travel (walking, cycling, using public transport or driving) for the outward and return journeys over the measurement week was derived from visual analysis using the following variables: counts per 10 seconds (sustained counts per epoch of < 17 were bus, train, car; sustained counts per epoch of > 325 were walking and cycling);72 changes to the sum of the signal-to-noise ratio (approximate threshold of a drop to < 250 was employed to indicate movement from an indoor to an outdoor environment);73 maximum speed of the journey (walking – not > 10 km/hour; cycling – not > 40 km/hour; bus – 10 to 50 km/hour; train and car speeds of > 50 km/hour);74 and GIS location for each epoch.

For participants who used a mixed mode of travel (e.g. walking and travelling by train), the mode of transport covering the greatest distance was considered the mode for that journey. MVPA accrued from walking during the journey was captured in a separate variable. When an outward/return journey was missing, it was assumed to be the same mode of travel as the outward/return journey on the same day. Any remaining missing data were replaced with the corresponding travel diary mode when available. The most frequent mode of travel during data collection was used to derive an overall mode of travel for each participant.

Time spent being sedentary and in MVPA were defined using validated thresholds [sedentary, < 100 c.p.m.; MVPA, ≥ 1952 c.p.m.]. 72 To examine the proportion of participants who met current physical activity guidelines,1 we calculated the total MVPA accumulated in bouts of ≥ 10 minutes over the data-collection week (by multiplying the mean daily number of bouts of MVPA by 7). In line with another study,75 participants were classified as ‘active’ or ‘inactive’ during the commute if their mean daily MVPA accrued during the commute was ≥ 10 minutes or < 10 minutes, respectively.

Variables

Analysis

Initially, descriptive analyses comprising counts, percentages, medians and interquartile ranges, were conducted. Differences in physical activity variables (overall and during the commute) were analysed by main mode of travel (car users, public transport users and walkers) using analysis of variance (ANOVA) and chi-squared statistics. Data related to participants classified as cyclists are not presented because of the inability of waist-worn accelerometers to accurately record physical activity during cycling.

Associations with objectively measured physical activity during the commute

To explore associations with levels of physical activity during the commute, logistic univariable analyses and likelihood ratio tests were conducted. The following explanatory variables for analysis were selected a priori: gender, age group, annual household income, education, weight status, occupational activity and commute distance. A multivariable logistic regression model was developed using ‘inactive during commute’ as the reference group. In the order of the strength of association, variables were selected for inclusion and retained in the model if there was an associated improvement in fit (p < 0.05). The final model adjusted for weight status, occupational activity and commute distance.

Associations with objectively measured mode of travel

The objective of the next stage of the analysis was to identify associations between different modes of travel to work and individual, interpersonal and workplace variables. Analyses were restricted to participants who were classified as ‘walkers’ (n = 74), ‘public transport users’ (n = 76) or ‘car users’ (n = 422). Participants classified as cyclists (n = 68) or whose mode of transport was unknown (n = 14) were excluded. Initially, associations were examined using logistic univariable analyses and likelihood ratio tests. Multicollinearity between variables was tested for through correlations. Using the same methodology as described previously, two separate multivariable logistic regression models were developed for ‘walkers’ and ‘public transport users’, both using ‘car users’ as the reference group. Individual, interpersonal and workplace characteristics and perception-of-commute variables were eligible for inclusion if they were associated with an improvement in fit of model (p < 0.05). The final ‘walkers’ model adjusted for limited access to a car, commute distance and availability of workplace car parking. The final ‘public transport users’ model adjusted for age group, limited access to a car, combines commute with caring responsibilities, availability of workplace car parking and perception of commute environment. Finally, a description of reasons for car use by car users was presented as counts and percentages.

Potential clustering by workplace was adjusted for using a robust standard errors approach allowing for workplace-level random effects in the final model. For each model, results were presented as odds ratios (ORs), adjusted odds ratios (aORs) and corresponding 95% CIs and p-values. Through sensitivity analyses, separate logistic models that were restricted to males only and females only were developed, with no major effect sizes by variables observed. Interactions were not fitted owing to small sample sizes. Analyses were undertaken to explore whether or not associations varied by gender. All analyses were conducted using Stata^® version 14 (StataCorp LP, College Station, TX, USA).

Associations with undertaking physical activity during the commute

Valid accGPS data from at least 1 day were provided by 597 participants. After adjustment for weight status, occupational activity and commute distance, there was strong evidence that participants were more physically active during their commute if they had sedentary jobs (aOR 1.96, 95% CI 1.26 to 3.04) or had a commute distance of < 2 km (aOR 2.73, 95% CI 1.69 to 4.41) or between 2 km and 4 km (aOR 2.74, 95% CI 1.58 to 4.73). There was weaker evidence that participants in the underweight or normal weight category (aOR 1.48, 95% CI 1.04 to 2.12) were more physically active during the commute (Table 9).

Of 542 participants (82.4%) who provided 3 days of valid accelerometer data, a minority (n = 60, 11.1%) met current UK public health physical activity guidelines. 1 A substantially higher proportion of walkers (n = 24, 38.7%) and public transport users (n = 10, 16.1%) met public health physical activity guidelines than car users (n = 17, 4.7%) (p < 0.001). There were marked differences in time spent in MVPA by main mode of travel. Overall, both walkers (mean 71.3 minutes, SD 21.3 minutes) and public transport users (mean 59.5 minutes, SD 26.6 minutes) accumulated more MVPA throughout the day than car users did (mean 46.3 minutes, SD 20.6 minutes). Walkers (mean 34.3 minutes, SD 18.6 minutes) and public transport users (mean 25.7 minutes, SD 4.0 minutes) were also, on average, more active during the commute than car users (mean 7.3 minutes, SD 7.6 minutes). There was no strong evidence for differences in time spent in sedentary behaviours (p = 0.12) or accelerometer wear time (p = 0.43) by main mode of travel (Table 10).

Individual characteristics, interpersonal responsibilities and workplace and environmental characteristics associated with mode of travel

After adjustment for having a current driving licence, household access to a car, commute distance to workplace and free work car parking, there was strong evidence that not having a driving licence (aOR 8.74, 95% CI 2.45 to 31.2) and not having access to a car (aOR 23.1, 95% CI 5.0 to 106.5) were positively associated with walking to work. The workplace characteristics ‘commute distance of less than 2 km’ (aOR 51.4, 95% CI 19.7 to 134.3), ‘commute distance between 2 km and 4 km’ (aOR 15.2, 95% CI 5.84 to 39.4) and lack of free parking (aOR 2.97, 95% CI 1.38 to 6.39) were also positively associated with walking (Table 11).

After adjustment for age group, having a driving licence, household access to a car, combining the commute with caring responsibilities and free work car parking, there was strong evidence that being < 35 years old (aOR 1.98, 95% CI 1.02 to 3.84), not having a driving licence (aOR 20.2, 95% CI 7.25 to 56.2), not having access to a car (aOR 5.00, 95% CI 1.13 to 22.2), not combining the commute with the school run or caring responsibilities (aOR 4.77, 95% CI 1.13 to 20.2) and the workplace characteristic ‘absence of free parking’ (aOR 3.32, 95% CI 1.56 to 7.09) were positively associated with using public transport (see Table 11).

Sensitivity analyses indicated no marked differences by gender across all of the models.

Reasons for car use

The reasons that were frequently provided for car use (Table 12) included that it was quicker than alternative modes of travel (n = 329, 78.0%), reliability (n = 275, 65.2%), comfort (n = 275, 65.2%), having to visit more than one place (n = 179, 42.4%), that it was cheaper than alternative modes of travel (n = 179, 42.2%) and lack of alternatives (n = 174, 41.2%). This order changed slightly when participants were asked to choose their single most important reason (see Table 5). Being quicker than alternatives (n = 100, 28.7%), lack of alternatives (n = 51, 14.7%) and reliability (n = 39, 11.2%) continued as main reasons, but dropping off and collecting children (n = 34, 9.8%) and a car being essential for their job (n = 29, 8.3%) appeared to take priority over comfort, cost or having to visit more than one place.

Lessons from the feasibility study: behaviour change techniques

During the Walk to Work feasibility study, the intervention focused on nine BCTs to guide participants to consider a more active travel mode through to maintaining the new behaviour: intention formation, instruction, barrier identification, goal-setting, general encouragement, self-monitoring, social support, review of goals and relapse prevention. Evidence from the process evaluation suggested that different participants found different BCTs more useful than others. 82 This was supported by other research: for example, ‘intention formation’ was identified by Bird et al. 47 as a key BCT for use in walking and cycling interventions, but not by Malik et al. 28 in their review of workplace physical activity. A qualitative evidence synthesis of workplace smoking interventions concluded that workplace interventions should employ a range of different elements because different employees have different requirements. 83 Overall, the nine BCTs that formed the basis of the Walk to Work intervention during the feasibility study appeared acceptable to participants and were retained for the main trial. However, the term ‘relapse prevention’ had been challenged by some participants as being inappropriate for an intervention that promotes walking. This is supported by Michie et al. ,44 who argue that different BCTs from those used to stop negative behaviour, such as smoking, should be used to support positive behaviours, such as physical activity and healthy eating. Therefore, for the main trial, the techniques used to avoid ‘relapse’ from increased walking back to less active modes of travel were couched in terms of ongoing support and encouragement for walking.

Lessons from the feasibility study: Walk to Work promoter training and support

During the feasibility study, it was originally proposed that the Walk to Work promoters from different workplaces would be invited to an external training event organised by the researchers. However, not all workplaces were able to release an employee for the time and date proposed, and so two smaller external training sessions were organised and some individual workplace-based training was provided. Having received their training and booklets to guide them through their role, the Walk to Work promoters were not provided with any additional support or guidance during the 10-week intervention. However, during post-intervention interviews, the Walk to Work promoters indicated that they would have welcomed additional prompts and encouragement during the intervention period. 82 In the light of these findings, for the main trial an individual training package for Walk to Work promoters was developed for delivery in the workplace and ongoing support for Walk to Work promoters was incorporated into the 10-week intervention.

Lessons from the feasibility study: workplace support for the intervention

Advocates of the socioecological model argue that public health interventions need to consider influences at the policy, community, organisational, interpersonal and intrapersonal levels. 84 During the feasibility study, the importance of support at the workplace (organisational) level was acknowledged. However, the majority of employers who took part in the feasibility study were unclear about how to give practical support for employees who walk to work. 85 Additional materials were, therefore, developed for the main trial to give employers information and ideas about how to support the Walk to Work intervention.

Walk to Work promoter recruitment

Following randomisation, workplaces in the intervention group were asked to identify a Walk to Work promoter. This could be a volunteer with an interest in the study, someone nominated by their employer as being suitable for the role (e.g. an employee already tasked with developing and implementing a travel plan) or an employer willing to take on the role themselves. Potential Walk to Work promoters were provided with an information leaflet explaining the role (see Appendix 5) and a consent form. Written consent was provided before potential Walk to Work promoters received the training and undertook the role.

Walk to Work promoter training and resources

The research team delivered a training session, lasting for approximately 1 hour, to the Walk to Work promoters at their workplace and at a time and place to suit their needs. The training was summarised in a DVD (digital versatile disc), developed by the research team, which promoters retained after the session. The training included information about the health, social, economic and environmental benefits of walking; using BCTs to promote increased walking, either the whole route or as part of a mixed-mode journey; providing support and encouragement to participating employees; and accessing relevant websites and resources for additional information and resources. Walk to Work promoters were given booklets, also developed by the research team, to assist them in the role. The content of these booklets (see Appendix 6) was discussed in detail as part of the training session.

Walk to Work promoters were given the names of participating employees in their workplace and were asked to provide the employees with Walk to Work booklets and an optional pedometer, highlight the benefits of increased walking, discuss barriers and solutions to walking during the commute, help participants to identify routes and methods of incorporating walking in their journeys and provide ongoing support through four contacts over the following 10 weeks. These contacts could be face to face, via e-mail or via telephone, in groups or on an individual basis, depending on the preferences of the workplace and participants. The Walk to Work promoter’s booklet contained guidance about providing support and encouragement to participating employees over the 10-week intervention period and optional diary pages to record their activities.

Walk to Work promoters were prompted and encouraged in their role through three short newsletters from the research team over the 10-week intervention period (see Appendix 7). These were supplemented by newsletters to pass on to participants (see Appendix 8) focusing on specific BCTs and providing additional information about the benefits of walking. The newsletters were provided by e-mail and/or in paper form to suit the workplace.

Participant booklets

The Walk to Work promoters were given the names of all employees in their workplaces who were participating in the study and were asked to provide each of them with a participant booklet. These booklets provided information and guidance relating to the BCTs that comprised the behavioural intervention (see Appendix 9). To encourage intention formation, the booklets began by considering the benefits of increasing walking during the commute. The focus then moved to identifying personal benefits, barriers and solutions, and goal-setting. Self-monitoring was encouraged through the use of diary pages at the back of the participant’s booklet or the use of optional pedometers, which had been provided free of charge to the study by the charity Living Streets (www.livingstreets.org.uk; accessed 6 June 2018), to record daily steps. Participants were prompted to seek encouragement and social support during their attempts to increase walking during the commute.

Participant newsletters

Three newsletters were sent to the Walk to Work promoters for circulation to participating employees during the following 10 weeks (see Appendix 8). These provided additional information, to stimulate continued interest in the benefits of walking, as well as continuing the focus on key BCTs: highlighting social support at around week 3, reviewing goals at week 5 and following up participants to support the maintenance of behavioural change (relapse prevention) at around week 7.

Information and ideas for employers

An employer pack was provided to all workplaces in the intervention group of the study (see Appendix 10). This comprised a letter outlining the intervention, the booklets being used by the Walk to Work promoters and participants, additional booklets specifically designed for employers and free Walk4Life poster templates for display in the workplace (www.walk4life.info; accessed 6 June 2018). The employer’s booklet contained ideas for promoting walking to work, including providing information about walking distances to train and bus stops, providing lockers or improved cloakroom facilities, giving financial assistance for public transport season tickets or for walking clothes/shoes, offering free incentive items for those who switch to walking (e.g. umbrellas, rucksacks or breakfast vouchers) and supporting competitions and challenges for those who enjoy taking part in such activities. Employers were encouraged to record any support they had provided for the intervention, together with associated costs, in a section at the back of their booklet.

Adverse events

Details of the procedure for reporting adverse events are provided in Chapter 2. No accidents or adverse events relating to the intervention were reported during the study.

Logic model

A logic model was developed, indicating the theoretical underpinning of the Walk to Work intervention, inputs, activities, outputs and outcomes as well as contextual factors likely to have an impact on the implementation of, and response to, the intervention (Figure 2).

FIGURE 2.

Walk to Work intervention logic model.

Data collection

The planned process evaluation methods are summarised in Table 14.

Data relating to the context, delivery and response to the intervention were collected from employers, Walk to Work promoters and employees through survey questionnaires at baseline, post intervention and at the 12-month follow-up, and through interviews with a purposive sample of employers, Walk to Work promoters and employees immediately post intervention. Observations and notes were also recorded during contacts with workplaces.

Questionnaires

For the process evaluation, the post-intervention questionnaires included questions about whether or not participants were aware of efforts within their workplace to increase or support walking during the commute and whether or not the research activities (particularly wearing activity monitors) had influenced their physical activity levels. All questionnaire data were entered into a secure database. Descriptive statistics are displayed as counts and percentages. Some questions allowed participants to add short free-text responses. These were entered in full into the database and grouped in accordance with the issues raised, including identifying similarities and differences in relation to those issues. The quotations included in this chapter were chosen to illustrate the key points that were made.

Interviews

Semistructured interviews were conducted with a purposive sample of employers, Walk to Work promoters and employees in the intervention group. The aim was to include views and experiences from a wide range of workplaces, Walk to Work promoters and employees who self-reported that they had, or had not, increased walking during the commute to work. A gender balance was also sought. Interviews were conducted in the workplace at a time to suit the interviewee. The topic guides were tailored to suit the participant’s role (employer, Walk to Work promoter or employee) and whether they were in the intervention or the control group of the trial (see Appendix 11). Potential interviewees were sent a letter of invitation and written consent was obtained prior to the interview. Interviewees were given a £10 gift voucher to thank them for their contribution to the study.

All interviews were digitally recorded, fully transcribed and anonymised. For this report, preliminary analyses focused on a key outcome for the process evaluation: barriers to, and facilitators of, walking during the commute to and from work. Thematic analysis was undertaken using the Framework method of data management. 87 The Framework method entails examining the transcripts and creating charts of sections of text relevant to key research questions. Streamlined versions of the charts are produced as the process of coding and analysing the data progresses: key terms and phrases are retained, repetition and extraneous text are removed and similarities and differences are examined within emerging themes.

Workplace descriptors

The classification of workplace activities is based on the UK Standard Industrial Classification (SIC) Categories 2007. 68 ‘Micro’ usually refers to workplaces with between one and nine employees,88 but for this study at least five employees were needed to meet the eligibility criteria and so micro refers to workplaces employing between five and nine people. Small workplaces are defined as employing 10–49 people, medium-sized workplaces are defined as employing 50–249 people and large workplaces are defined as employing ≥ 250 people. 88

Participant identification numbers

All participants in the study were given a unique identification number: the first three digits refer to the workplace and the last two digits refer to the participant. IDs (identifiers) preceded by ‘9’ indicate that these participants, either Walk to Work promoters or Walk to Work employers, contributed to the study but did not take part in the outcome data-collection activities.

Workplace and participant recruitment

The process of recruiting workplaces and participants is considered in detail in Chapter 3. Following baseline data collection, 44 workplaces were randomised to receive the intervention. However, two workplaces withdrew from the intervention group: one because of unanticipated relocation to another country and one because the key contact felt that the study activities were too onerous.

Walk to Work promoter recruitment

In the remaining intervention workplaces, the employers were asked to identify a Walk to Work promoter to implement the 10-week intervention. In 19 workplaces, the employer took on this role themselves: of these, six were micro-sized, seven were small, five were medium-sized and one was large (Table 15). One large workplace had two Walk to Work promoters, but in the majority of workplaces one person was identified to undertake the role. Fourteen of the Walk to Work promoters were male and 24 were female.

During the first year of recruitment, five workplaces in the intervention group were unable to identify someone to take on the role of Walk to Work promoter. Rather than lose these workplaces from the study altogether, it was decided that the relevant researcher would take on the role of ensuring that the intervention materials were distributed to participating employees. However, it was not felt appropriate for the researcher to go into the workplace to give face-to-face support for employees to encourage them to change their travel behaviour. These workplaces were therefore considered to have received a ‘light-touch’ intervention. In the second year of recruitment, researchers stressed the importance of being able to identify someone in the workplace who might be suitable to be a Walk to Work promoter should the workplace be randomised to the intervention group, and no further workplaces were recorded as not recruiting a Walk to Work promoter.

Walk to Work promoter training and intervention materials

The methods and resources used for the Walk to Work intervention are described in detail in Chapter 5. The training session, DVD and booklets offering guidance about their role were delivered to all designated Walk to Work promoters at the start of the intervention. Three newsletters offering additional information and encouragement were also sent to all Walk to Work promoters during the 10-week intervention. All Walk to Work promoters were provided with booklets and optional pedometers to distribute to participating employees in their workplaces, plus three newsletters to distribute to participants during the 10-week intervention period. The pedometers, which were provided free of charge for participants who wished to use them, were not of high quality. Participants who wanted to self-monitor preferred to use their own wrist-worn monitoring devices or an app (application) on their smartphone, details of which were supplied in newsletter 2 of the intervention.

All employers were provided with the employer pack containing a letter reminding them of the aims of the intervention and the role of Walk to Work promoter, booklets with ideas about how to support the intervention and templates for posters to display as appropriate in their workplaces.

Activities of the Walk to Work promoters

In relation to fidelity (the extent to which the intervention was delivered as planned), we were able to examine the recruitment, training and ongoing support of the Walk to Work promoters but it was not possible to directly observe their activities in their respective workplaces. However, the post-intervention questionnaires included questions to shed light on the Walk to Work promoters’ activities. Questionnaires were administered to all study participants, including those in the control group. Descriptive statistics show that participants in the intervention group were more likely to indicate that someone had tried to encourage them to change their travel behaviour in the last 2–3 months (the intervention period): 19.9% (n = 66) in the intervention group compared with 10.8% (n = 35) in the control group (Table 16).

The non-response rate for this question was noticeably higher in the intervention group (40.5%) than in the control group (18.6%), and there were noticeably higher non-response rates in the intervention group for other questions relating to intervention activities. It is not clear why this was the case; it may relate to the additional participant burden in the intervention group, as participants had just come to the end of a 10-week intervention, or there may have been some reluctance among participants in the intervention group to confirm that they, or the Walk to Work promoter, had not engaged with the intervention.

In the intervention group, 64 participants from 28 of the 42 intervention workplaces made use of the free-text box to give some additional information about the encouragement they were given. In some cases, the Walk to Work promoter or the Travel to Work study were specifically mentioned; in others, it was less clear if the participant was referring to other active travel initiatives. Half (n = 32) of comments in the intervention group suggested that sustainable transport, active travel and cycling were promoted, rather than specifically focusing on walking. For example:

E-mails about cycling to work, conversations about parking further away from office, e-mails about bike pods.

Employee, ID 106-05

Work colleagues supported my decision to start biking, but also tried suggesting alternatives but unfortunately nothing that would work for me.

Employee, ID 224-08

I have had a workplace induction and the health and safety talk included protecting the environment. They mentioned the benefits (personal and environmental) that can be achieved by alternative travel.

Employee, ID 330-13

The remainder of the comments in the intervention group focused more clearly on encouragement given for walking:

Colleague involved in project gave me a pedometer and suggested I park in the industrial estate and walk the rest of the way to work. Articles on benefits of walking also sent round.

Employee, ID 106-21

Walk to Work champion discussed some strategies to increase number of days I walk.

Employee, ID 141-02

There were posters around the office. Some people who took part in the study live close by so we walked together and encouraged each other to walk. In the past we were sharing car lifts more.

Employee, ID 155-04

Although there was evidence of some interest and enthusiasm, other study participants offered explanations as to why they could not increase walking during the commute:

Talked to [name of Walk to Work promoter] in work, the liaison for Travel to Work. But with weather changes + nights getting darker. I didn’t feel happy to walk more.

Employee, ID 303-02

Tried to get me to park further away, but due to the hours I work, it’s safer for me to park closer to work.

Employee, ID 307-08

I cannot walk to work as I live 6 miles away. I walk most lunchtimes – 1.75 miles but this programme hasn’t adjusted that behaviour.

Employee, ID 335-03

In the control group, there were fewer comments from study participants (n = 35 from 18 of the 43 workplaces). About half (n = 17) who said that they had been encouraged to change their travel behaviour focused on cycling:

Travel to work schemes advertised, car sharing – cycling – new cycle centre opened – discussed with boss whether to cycle to work and he suggested borrowing the electric loan bike to try out for the 10 miles.

Employee, ID 162-14

Posters at work showing examples of employees cycling to work, their routes, how long it takes and the benefits.

Employee, ID 112-12

We have some very good advocates for cycling to work here.

Employee, ID 126-10

Others mentioned more general sustainable transport initiatives:

Conversation with the sustainability leader. Consideration how to try and expand the carshare from my home town.

Employee, ID 104-06

We took part in [name of city] commuter challenge which encourages employees to travel more sustainably/actively for 1 month.

Employee, ID 1151-01

Six respondents indicated that they had been encouraged to walk:

Colleague who lives near me persuaded me to walk home with her one evening rather than get the bus – it was a good experience.

Employee, ID 125-29

Manager encourages us to walk more.

Employee, ID 150-05

The Civil Service walking challenge – an annual campaign to raise money for the Civil Service Benevolent Fund and get us to walk at least 10,000 steps each of 50 days in the summer.

Employee, ID 153-02

Other related activities and campaigns

Descriptive statistics suggest that very few participants were aware of, or taking part in, activities outside the workplace that aimed to change the way they travelled to work (Table 17).

The free-text box to provide more information about other schemes was completed by 28 participants, of whom five appeared to have misunderstood the question and referred to the Travel to Work study. Ten participants mentioned cycling schemes, for example ‘Cycling UK “cycle-to-work” day’ (Employee, ID 128-09), ‘bike to work campaign’ (Employee, ID 136-05) and ‘cycling schemes to buy bikes’ (Employee, ID 224-07). Only three participants mentioned walking, although not specifically walking to work: ‘I went on a 3-month holiday – trekking’ (Employee, ID 125-07), ‘Living Streets’ (Employee ID 150-04) and ‘I chatted with colleagues about it and walked more after work on a regular basis. I took walks between 6–8 p.m. for around 10–30 mins + took part in an exercise diary’ (Employee, ID 131-01). Other participants referred to more general sustainable transport initiatives: ‘commuter challenge’ (Employee, ID 121-09), ‘Local campaign about car sharing/using the car less’ (Employee, ID 121-10) and ‘Travel Smart in my organisation. Big commuting challenge. Green Impact Award’ (Employee, ID 162-16).

Hawthorne effects

Data from the post-intervention questionnaires provide some evidence of a potential Hawthorne effect in relation to the research activities. Table 18 indicates that around 18–28% of all participants ‘often’ or ‘sometimes’ changed their behaviour (either walking more on journeys to/from work or being more active in general) while wearing the monitors. However, this affect was reported in both the intervention and the control arms and therefore is unlikely to have changed the results of the study.

Free-text comments suggest that taking part in the research study, rather than receiving the intervention, appeared to have influenced some participants:

The Travel to Work survey got me to cycle a few days a week in the summer when my wife, a teacher, was on school holiday.

Employee, ID 112-05

This one [Travel to Work questionnaire] made me think. When I was first approached I was tied as I had to pick mum up – although if she could of [sic] walked, she would have – she was known for walking and was on no medication when she died, so I have decided to do more.

Employee, ID 112-21

This study enabled me to think about my activity levels. If I could travel to work by bicycle that would be great, but it is impracticable, and unsafe, too time consuming.

Employee, ID 137-02

Discussed ideas in staff room with other people taking part in Travel to Work survey.

Employee, ID 160-08

To encourage participants to wear the monitors, the research team printed and distributed personal graphs of activity levels and maps showing the routes that were taken. Just over 60% of participants recalled receiving the graphs or maps (Table 19).

Of those who recalled receiving a graph or map, around 18–28% indicated that they ‘often’ or ‘sometimes’ changed their behaviour (either walking more on journeys to/from work or being more active in general) after seeing the graph or map of their activity (Table 20). This influence appeared to be balanced between the intervention and the control arms in relation to walking to work, but was somewhat higher in the control group in relation to undertaking more activity in general.

Workplace support for the Walk to Work intervention

Questions were included in the post-intervention questionnaires to explore the extent to which employees were aware of support for the intervention in the workplace. Again, rates of ‘no response’ were noticeably higher in the intervention group than in the control group. Overall, around 10% of respondents were aware of an increase in their workplace of information about walking. Table 21 suggests that those in the intervention group were more aware than those in the control group of posters (13.6% vs. 7.1%) and information about walking or walking routes (14.0% vs. 4.0%).

Changes requiring more input from employers (such as improved facilities, changes to parking arrangements, incentives to walk during the commute or subsidised sustainable transport initiatives) were less evident (Tables 22 and 23).

Interviews were conducted with 70 participants: 11 employers (Table 24), and a further 12 employers who also took on the role of Walk to Work promoters (Table 25); 11 Walk to Work promoters (Table 26); and 36 employees, of whom 19 self-reported no increase in walking to work (Table 27) and 17 self-reported an increase (Table 28). The age range of interviewees was 24–63 years and there were 33 men and 37 women.

For the process evaluation, a key outcome of interest was the identification of barriers to, and facilitators of, walking during the daily commute. Preliminary analyses are presented here, examining key issues from the perspectives of employers, Walk to Work promoters and employees. Several key issues were identified: distance, alternatives to car use (public transport and park and walk facilities), caring responsibilities and the stress of the commute.

Distance

The distance between home and work was frequently highlighted as an important factor in whether or not walking was a feasible mode of travel to work:

I suppose the most important is distance the person lives away from their workplace and that’s how, and I think is the main consideration of how they decide how they’re going to get to work.

Employer promoter, ID 302-03

Distance was expressed in terms of both mileage and time:

1 to 2 miles, I think, you know, most people would walk, which they do, um, there are a lot of people who just live within probably yeah, probably three or four live in a mile, or mile and a half, radius so they walk to work.

Employee, ID 161-03

For staff that live more than a mile or so away it’s not a convenient route to be able to walk into work.

Employer promoter, ID 228-01

I live literally 15 minutes approximately from here. I always have tried to walk to work with most of my jobs, um, it’s only once I had to commute ‘cos I worked at a hospital which was a bit too far away. So I always try to walk as much as possible, um, anywhere really, but yeah I think living close to my workplace is what gives me the motivation to do it really.

Promoter, ID 141-03

I walk to work, but only because of the distance, like it’s only a 20-minute walk.

Employer, ID 9-232-15

Interviewees also linked distance with location:

I would walk to work if I worked in town, like in the centre, I would walk to work definitely but it’s just too far here.

Employee, ID 102-06

Distance is a big one I mean where we live, um, we’re in [name of city] but I know a lot of people here especially at work live quite rurally, live a long way away from [name of city], um so distance is a big factor.

Employee, ID 335-06

If you are in a business park, it’s, it’s too long a hike for most people.

Employer, ID 9-128-20

Some workplaces suggested that the majority of employees taking part in the study lived too far away from the workplace to walk the entire commute:

There was something about this project that attracted people from a long way away. I’m not quite sure what there was in our pre-marketing recruitment and I don’t know whether that’s been the same with other workplaces, but we’ve had people from [English city] and [English town] and [English city] and all over that um commute hours each way . . . I think yeah they would have had that barrier from the beginning.

Promoter, ID 9-226-18

Everybody was like ‘Look, we just live too far, too far away to get here’ so whether that was because of the group make-up, um, if it was a larger group maybe we would have had people that were closer.

Promoter, ID 9-335-15

The location of workplaces was also linked to perceptions of available walking routes:

We don’t park on site, we park a little further away and then it takes about sort of ten minutes to walk from the car park to the door and it’s, it sounds really daft it’s probably not that much quicker than walking from my house but it’s safer. Where I live I have to walk along to the dual carriageway, down a lane which is incredibly creepy, um and sort of through another car park in order to get here um and especially with the light, with the nights being darker. In summer months it wouldn’t be so much of an issue but where it’s later, it’s darker, it’s really creepy so where, where we park now is lovely and light, lovely bright um and just generally feels a little less like somethings going to jump out in the bushes . . . there’s probably nothing to be concerned about um I don’t think this area’s got a particularly high crime rate but it’s um, it’s more of a comfort thing.

Promoter, ID 111-02

It’s some distance for me but the routes, I wouldn’t even know the safest route, and the route that I drive in, half of that has no pavements that you could walk anyway. So I wouldn’t even know which route to take um to get to work, safely and on my own, I mean especially in the nights or in the um in the mornings when it’s dark.

Employee, ID 317-02

At one workplace, concerns about the safety of a walking route had led to calls for improvements:

Since they’ve build the new houses up by um junction – up by the [Welsh village] roundabout . . . They’ve actually now put in the new footpath for those houses on that side of the road, so they can get, so that the children can go from that site straight into the primary school. Um and one of the girls who works in [workplace], she lives there and she’s saying it takes her 20 minutes to get by car from there to here, so she’s been looking at walking but of course the other pavement that we need to come in, which is the opposite side of the road, is the one where there’s no lighting and you have to cross the dual carriageway, so it’s the safety aspect of that and it’s the same with me if I parked um on [Welsh village] and walked back that way, it’s crossing that dual carriageway is the problem. Now further up they’ve put, now put the um traffic lights with the pelican crossing facilities on it, and we were just wondering whether they would have considered doing it further down so that we can use, so people can actually walk across there.

Promoter, ID 305-01

At the time of the interview, as a result of this concern, the local authority had set up monitoring devices to check vehicle speeds at the site of a requested new crossing.

Analyses of baseline characteristics (see Table 8) confirmed that a large majority of participants lived > 2 km from their workplace. This meant that increased walking was more likely to be part of a mixed-mode journey involving the use of public transport or combining a car journey with walking.

Public transport

There were some positive comments about the availability of public transport:

We’re very lucky we’ve got a train station you know not even a 3-minute walk up the road so um you know location to public transport helps um I’m not familiar with the bus but I’ve used the train a few times so I know that, I know the train station’s very close.

Employee, ID 106-23

Access to this workplace is good because we’re in the centre of [name of city] so you’ve got buses, the trains, a bit of a walk, it’s a 15-minute walk, some people are OK with that, some people are not OK with that.

Employer promoter, ID 165-01

There is good public transport links so I don’t think that will be a major problem for people to change from using a car to public transport really.

Employer promoter, ID 302-03

However, the majority of interviewees expressed dissatisfaction with public transport as a means of travelling to and from work. Although the study had focused on workplaces in urban areas, this did not mean that their employees lived in urban areas served by regular bus or train services:

It is either ‘Do I get in the car?’ or ‘Do I get the bus?’. You know it is that simple for me. That may not be the case for somebody who lives a little bit closer who could walk all the way, but the four of us that participated, none of us are in that category. In fact, the other three, I don’t know how they’d get to work with public transport. They would be hours if they had to do that. Because one lives down at [English village] I think it is, and the other at [English village] so they are miles outside of [English city] so I think they’re limited to what changes they could make really.

Promoter, ID 161-02

Because we’re quite rural the public transport isn’t great, um you couldn’t get a train from where I live to here, you couldn’t get one bus from, from, where I live to here, you’d have to catch several.

Employee, ID 335-06

There were also concerns about whether or not public transport would enable employees to get to work on time:

A bus from my house [laughs] would actually get me to [name of town] town centre about 15 minutes after I was supposed to start work so, you know, that, the timing of the bus probably, you know, is probably the biggest influence.

Employee, ID 226-08

I have to be in work for 7 o’clock in the morning, there’s no buses, there’s no trains, I am forced to use a car. If there were public transport I would use it, so that’s a big factor.

Employee, ID 310-02

Because of concerns about getting to work on time, the idea of getting off the bus a few stops early to include additional walking in the journey could be problematic:

It’s whether you are running late or not, and that can be down to the public transport being late. You know, you might, you might intend to get off the bus earlier and then the bus might not turn up till 10 minutes later so that cuts your getting to work time . . . you might go out thinking ‘Right, I am going to get off the bus two stops earlier’ but then the bus might be 10 minutes late and you think ‘Actually I can’t because I am not going to get into work on time’.

Promoter, ID 161-02

In some areas, using public transport might result in quite complex journeys compared with a single-mode car journey:

I live in the [name of place] which is about 30 miles away from here where I work, um, it’s quite a remote village, um, there’s very little public transport, unless it’s a Tuesday or a Thursday [laugh] so travel by public transport would mean a cab to a rail station and then three changes of train, or a cab to a local town (name of place) and then two buses at least, so the only practical way to travel to work is really by car.

Employee, ID 122-05

More-complex journeys may also be perceived as extending the working day:

The bus station is only a very short walk from the office so anyone travelling in by bus wouldn’t have a problem, the same as the train station that’s not far away. Whether or not there would be a direct bus or a direct train, I suspect a direct train is more likely, um but if they were to come in by train they may have to get a bus to the train station and vice versa. I think for a lot of people one of the deterrents there would be the fact that it makes the working day that much longer.

Employer promoter, ID 9-319-01

In some areas, public transport was also perceived as being expensive:

Financially it would be far more expensive. I can’t, I can’t remember how much it is something like £4.50 one way to get on the bus to come to work, and I think I can probably drive that for a lot less.

Employee, ID 226-08

I think public transport is relatively expensive in this area, on a few occasions I’ve used it um I‘ve been surprised how expensive it is, I’m sure if someone was using it on a regular basis and they’re buying season tickets that type of thing it’s probably cheaper, but I think um sort of on an ad hoc basis given that the price of public transport would probably put people off as well.

Employer promoter, ID 9-319-01

Park and walk

Where public transport was not perceived to be a viable alternative to car use, increased walking might be achieved by parking further away from the workplace and walking the remainder of the journey:

What I do now is I park 10 minutes away from the office, so I drive, it’s probably, yeah it’s probably just 2 and a bit miles and so I park 10 minutes away and then I have that 10-minute walk into the office, and I do that most of the time . . . I’ve done it in the rain and I’ve done it when I’ve not been feeling great, and I’ve done it when I’ve had a lot to carry, so all those things that previously I would have said ‘I can’t do it today’ because I’ve, because I think that was one of your questions in one of your booklets wasn’t it . . . I guess that made me think about it and I thought well it’s, you know, it’s not really a valid excuse because you, I can carry that or I might feel better – feeling a bit grotty I might feel a bit better after a walk and sure enough you were right [laugh] and I did.

Employee, ID 106-12

Other interviewees encountered difficulties in finding a suitable place to park their car:

I haven’t found a way of making the walk part of the commute and making the journey times the same for me. The, the, my main route to work there isn’t actually anywhere I can leave the car to walk that last bit, so for me to, to park further away I am actually increasing the journey in terms of the driving part of it as well as the walking part.

Promoter, ID 106-13

One of the options was to park in a local park and then walk um and then they changed the parking restrictions so you can only park there for 2 hours before you move on, rather than all day . . . I guess having somewhere to park for free which then allows me a short walk to work would be useful but the, yeah, with the parking restrictions going crazy all over the place that’s becoming less and less feasible.

Employee, ID 207-01

There isn’t anywhere around here where you could actually park your car to be able to walk in, sort of thing. I think it’s more down to the council and the government then, who are actually stopping anybody from doing it because there aren’t facilities around here to do be able to leave your car off the premises where you can walk that distance then into work.

Employee, ID 305-08

One participant decided to pay for the park-and-ride facility but then walk to work rather than take the bus. This led to some confusion when an attendant at the park-and-ride said it was not permitted and attempted to issue a fine:

There’s been quite a few changes at the park and ride, one of which was introducing civilian um security guards who seem to be of the opinion they could issue fines for people that left their car and walked, um that was misdirection it now appears, um but it did cause 2 weeks where um we couldn’t walk at all because of the threat of being fined . . . through various emails and correspondence we managed to clarify the position and find out that it wasn’t the case you couldn’t walk – it was a misunderstanding and they were happy for you to walk.

Employee, ID 310-02

Where workplaces provided free or relatively stress-free parking, this could undermine attempts to encourage study participants to park and walk:

The only thing that could seriously affect it [current park-and-walk journey] is if I reached the holy grail and get a parking space in the basement car park . . . it’s not only financial but obviously the ease. I can come straight in to work and literally jump in the lift and I’m at work so it’s a big time saving as well. I mean with the best will in the world I spend at least half an hour getting to and from work where if I could park in work I could drive straight in and drive straight out again.

Employee, ID 310-02

If there’s somewhere I know I can park, and I can walk in, like a 25- to 30-minute walk, and I used to do that quite a lot, but then we had more parking at work so, I actually feel bad [laugh], so they can buy you more parking and so it’s so much easier now to come and drive in and park whereas before it wasn’t and so I had to use this park and walk thing um but now we’ve got more parking.

Employee, ID 223-02

They often live a fair distance out, there’s convenient parking at the place of work, so if somebody was to try and drive part-way and park somewhere it would actually make life a lot more difficult for them. Um most of the streets around the office are either residents’ parking or you’ve got to pay to park and people aren’t going to do that if they can just drive straight to the office and make life easier for themselves.

Employer promoter, ID 9-319-01

One possible solution would be for workplaces to provide parking facilities away from the workplace itself to encourage employees to park and walk:

I suppose one thing that might help if there was spaces that the company, I just don’t think it’s really viable so I wouldn’t even, I don’t [laugh], but if there were spaces that were a 10-minute walk way. If the company kind of had, were going to buy more parking spaces, it almost would be a good thing in a way but to be a 10-minute walk.

Employer promoter, ID 223-01

They’re putting a charge on it [parking] so it’s £20 a month if you park in one of the car parks close by but if you then park on site it’s £40 a month . . . for £20 a month you can’t really beat that so um it’s about £1 a day so um yeah the car park’s probably about half a mile away so then I’d walk about half a mile . . . if I park in the car parks which are a little bit, little bit further away then just walking that half mile in is now probable.

Employee, ID 226-03

Car drivers appeared to discount the cost of purchasing and maintaining their vehicle, and focused on the immediate costs of running and parking their vehicles. This resulted in perceptions that car driving was cheaper and more convenient than other modes of transport, especially where workplaces provided free or subsidised parking:

One person said she used to get the bus and walk the rest of the way but um, it didn’t save any money and the weather’s always rubbish so she couldn’t really be bothered with that. Um, most of the employees feel it’s too far to walk and as the building has free parking, and we’re in the city centre there’s nowhere really to park further away um, without paying or getting parking tickets.

Promoter, ID 317-03

We get free parking at work but if you park further away there’s a lot of residential parking zones so there aren’t many areas where you can actually park [laugh] and leave your car all day without paying . . . I think this project has prompted people higher up than myself such as [name] to um thinking about actually our policy on car parking isn’t an incentive at all to make people walk or cycle to work so, and we knew that anyway, and we did have a try a few years ago to bring in car park charging, but it got thrown out. But now perhaps is the time in the days of austerity to try again [laugh] . . . there’s lack of incentives really, in terms of just providing people free and ample parking, to try and change behaviour.

Promoter, ID 9-226-18

If people had to pay for parking and we didn’t have um, parking here, then I think it would change how people would travel to work. It would definitely be more of an encouragement for them to, to look at alternatives, possibly to walk.

Employer, ID 317-01

Caring responsibilities

Interviewees with caring responsibilities, especially for young children, appeared to feel that they had no alternative but to combine the journey to work with transporting their children to school or child-care facilities:

My children go to school near here and I bring them here to work and then they walk to school from here. Um, they could all, we could all walk but actually, you know, they have, during their spring and summer terms they have cricket bags. They are at different ages so the days that they bring their stuff to school is different and, you know, yes they could walk with their cricket bags and their school bags and their sports bags but actually it’s a lot!

Employee, ID 158-03

My personal family life is definitely one of the barriers. I have a young son who is at a childminder and my wife works part-time hours so it’s a combination of making sure we were picking him up at the childminder and my wife also works evenings and making sure I was back in time for that.

Employee, ID 228-04

My journeys include dropping the kids off to childminders/school and then coming in to work and finishing early to get back for the school run and so on, so I’m restricted with times.

Employee, ID 317-02

Concerns about timing the journey were particularly evident in relation to public transport:

I have to drive for my daughter because there’s no other option um well there is a bus, but it doesn’t, the timing isn’t really, doesn’t work for everyone taking their daughters to school so that’s, so that’s not an option.

Employer promoter, ID 223-01

I’ve for example got two children and I often have to drop them off and pick them up at the same time to, as I go to and from work, so um that makes public transport impossible for me.

Employee, ID 337-02

One participant who was keen to change his travel arrangements was able to negotiate a solution with his employer:

Getting the children, well picking them up for me from school um was something to consider so, how much walking or public transport could I do, you know, with that in mind um and actually some things I would change, I have changed my hours since I filled in the questionnaire and I should mention that, so it makes it easier for me to catch the train um three times a week to [Welsh city] and then I get home in time to pick my daughter up then from the childminder . . . I always find the train times don’t quite correlate, so I tend to have about a half an hour wait in the station. Um so I think obviously more regular transport would help but I have found work have been a bit more flexible I think, with thinking about how people do get to and from work now and they have been flexible by giving me 15 minutes at the end of 4 days a week to catch the train.

Employee, ID 337-05

The stress of the commute

For some participants, it was clear that the journey to and from work was a stressful part of the day. Some interviewees felt that changing their mode of travel would be complicated, less convenient and would increase the level of stress:

For maybe half of the team they possibly could get a bus, but then they would have to walk from the bus stop, but a lot of that, they would probably choose not to do it because of the reliability of the bus. Um, that being the biggest factor, it would be the reliability. They would probably end up having to leave a lot earlier to get to work, but yeah, I would say that probably, definitely, half of them are on a bus route that comes this direction. I wouldn’t say it’s necessarily round the corner but it’s this direction. They’d have to then maybe do a walk. ‘Cos even myself, I could get a bus to the top of [name of road] and I could walk down and have done that, but then it would take me, I would have to get, make sure that that bus is 100 per cent guaranteed to leave at that time, so, cos I’d be panicking.

Employer promoter, ID 127-05

If I had the time, I’d have quite a good walk in to work and back. I would definitely prefer to do it. Um but yeah just, just the last few months has been mental . . . I just have to get kids to school, get to work, not you know as quickly as I possibly can. Um I don’t have time to walk, um I don’t have time for lunch breaks. I work on reduced, I’d be working, I have a break because I have to have a break but I have the shortest break that I, you know, because I just have, I have to leave the office early to get my daughter so to work my hours I don’t have, I can’t have, an hour lunch break um and so really I supposed I felt like why, why set myself targets that I can’t, you know, I haven’t, I just thought, I haven’t got the space in my life right now to set myself goals . . . it would just go down the source of stress.

Employer promoter, ID 223-01

I thought I would try to increase my levels of activity across the board um by walking a little bit on my way in to work, etc., and then the things such as the park closing and then having to take a detour or um having to get on to the main road from my parking space and it adding another 15 minutes to my day, these – what would consider to be relatively minor incidents in the grand scheme of things, just to have such a negative knock-on effect on my, onto my, the most stressful part of my day which is getting to work and getting, with the kids, and getting home from work with the kids, um that it left me very little wiggle room and I just found it stressful and unhelpful so I stopped doing it.

Employee, ID 337-02

Others felt able to adjust their journey and found some enjoyment in doing so:

When we started doing it, it became quite enjoyable and we got the maps back and we saw [laugh] where we had walked. We’re all there then going ‘How much did you do?’. It’s all about like gamification you almost want to be ‘Oh I’ve done the most steps’ or ‘I’ve walked the furthest’ that’s, creates a bit of team building really’.

Employer promoter, ID 165-01

I have walked more, um, my commute’s about 4 miles but over the summer I was trying to do it once a week and I’ll, I’m a little bit embarrassed to admit that it was Pokémon GO that started me doing that. So I found that the commute was a great time to catch Pokémon on the way to work.

Promoter, ID 9-226-18

I’ve found a really pretty route. I’ve also started listening to serialised podcasts on my walk to and from the bus, so like it’s time that I can spend on my own listening to something . . . I go from working with loads of people to being with my partner all the time so having like half an hour or so on my own listening to something and chilling out is really nice.

Employee, ID 335-06

When you first contacted me I was working 20 miles, or 18 miles away, um, and I looked into the train, couldn’t use the train, um, because of times to get to work and certainly when I’m on call at the weekends, nights and so on it just doesn’t happen so I didn’t have any option other than to drive . . . now it’s completely different because it’s 12 minutes’ walk for me so I’m not going to drive . . . now I live and work in the same town, I do not drive, and even though I know the parking is going to be OK I elect not to drive.

Employee, ID 337-07

Data collection

The physical activity outcomes, data-collection methods and timings are summarised in Table 29.

Outcome measures

Physical activity was objectively measured using accelerometers (ActiGraph GT3X+). Validated accelerometer thresholds were used to compute daily time spent in MVPA (all 10-second epochs with ≥ 1952 c.p.m.) and being sedentary (all 10-second epochs with < 100 c.p.m.);72 these are listed in Table 30. The primary outcome was the daily number of minutes of MVPA. To be included in the analyses, participants were required to provide at least 3 valid days of accelerometer data: any 3 days were permissible, as our primary aim was to see if increasing walking during the commute led to an increase in overall physical activity, rather than on work days only. The secondary outcomes were overall levels of physical activity (c.p.m.), daily number of minutes of sedentary time, daily number of minutes of MVPA during the commute and modal shift (number of journeys, over the previous 5 working days, when walking was the major commuting mode). Primary and secondary outcomes were measured at baseline and at the 12-month follow-up.

To identify physical activity during the commute, participants wore a GPS receiver (QStarz BT-1000X) during their journeys, set to record positional data every 10 seconds; GPS settings are listed in Table 31. As described in Chapter 4, participants’ workplaces and homes were geocoded using the full postcode and imported into a GIS (ArcMap v10.2.2). The merged accGPS files were imported into ArcMap and participants’ journeys to and from work were visually identified and segmented from other accGPS data using the ‘identify’ tool to provide a measure of duration of the journey and associated MVPA.

Randomisation procedures

Randomisation took place at the workplace level after consent to participate was secured and baseline data were collected, thus ensuring allocation concealment. Employers in participating workplaces completed a short questionnaire to optimise matching pairs (or triples) of workplaces with similar characteristics, including size (micro-sized, 5–9 employees; small, 10–49 employees; medium-sized, 50–249 employees; large, ≥ 250 employees), location (south Wales, Bath, Swindon, Bristol and South Gloucestershire) and type of business [using UK Standard Industrial Classification (SIC) Categories]. 68 The assignment of workplaces was carried out at the BRTC by a statistician not involved in workplace recruitment. Allocation was based on random numbers generated by Stata version 14, such that one workplace from a matched set was randomised to the control group and one (or two in a triple) to the intervention group. Given the nature of the intervention, it was not possible to blind participants following randomisation.

Statistical methods

The primary statistical analysis was prespecified in a statistical and health economics analysis plan, which was made publicly available prior to release of outcome data to the statistical team. 90 Individuals providing a measurement of the primary outcome were included in the primary analysis, which compared workplaces allocated to the intervention and workplaces allocated to usual practice. The treatment effect was estimated as the mean difference between the intervention and the control groups using multivariable linear regression (adjusted for baseline MVPA), accelerometer wear time at follow-up, workplace size, location and type of business as covariates. A normally distributed random effect accommodated any variation between workplaces in the mean outcome of their employees. This model was adapted for the continuous secondary outcome measures (overall levels of physical activity, daily number of minutes of sedentary time and daily number of minutes of MVPA during the commute). A zero-inflated negative binomial regression model, with robust standard errors, estimated the treatment effect on the modal shift measure (number of journeys when walking was the major mode of travel). This model was adjusted for baseline modal shift, workplace size and location, with variation across participants in the total number of journeys recorded being accommodated (as an offset). In this model, variation between workplaces in the average outcome of their employees was accommodated by the robust standard errors. 91

Sensitivity analyses on the primary outcome only were pre-planned to assess the impact on the primary analysis of any imbalance in baseline covariates, any non-normality in the distribution of the primary outcome missing data (through the use of multiple imputation) and different quality-assurance thresholds for accelerometer data. This last analysis included a greater number of participants in the primary analysis and so explored the influence of missing values. Prespecified subgroup analyses of the primary outcome measure explored whether or not age at baseline (above/below the median), gender (male/female) or household income (above/below £30,000) modified the intervention effect; these analyses proceeded by adding interaction terms to the regression models used in the primary analysis. The statistician was blinded while the code for analysis was developed and unblinded to run the final analysis.

Recruitment and retention

Recruitment is discussed in detail in Chapter 3. Recruitment took place in two phases: May to July 2015 and March to May 2016. The initial intention was to recruit across three urban areas in south-west England and south Wales. However, because fewer workplaces were recruited during the first year, we included four additional urban areas in the same regions during the second year. We received 271 expressions of interest and, after screening for eligibility and giving further information about the study, we recruited 87 workplaces (Figure 3): 10 micro-sized, 35 small, 22 medium-sized and 20 large.

FIGURE 3.

Flow of workplaces and participants through the study. This figure is reproduced from Audrey et al. ,69 published by BMC Public Health journal. This is an Open Access article distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. The figure includes minor additions and formatting changes to the original.

This figure is reproduced from Audrey et al. ,69 published by BMC Public Health journal. This is an Open Access article distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. The figure includes minor additions and formatting changes to the original.

Following baseline data collection, 44 workplaces (331 participants) were randomised to receive the intervention and 43 workplaces (323 participants) were randomised to the control group (see Figure 3). Two workplaces (21 participants) withdrew during the 10-week intervention period: one relocated overseas, and in the other workplace the main contact indicated that they were too busy to continue with the study. A further intervention workplace (with only one participant) withdrew at the follow-up data-collection point because of a lack of interest among staff. At the 12-month follow-up, 84 workplaces (41 intervention and 43 control) and 477 employees (73% of those originally recruited to the study) took part in data-collection activities. Response rates in the control group were higher than those in the intervention group (Tables 32 and 33; see Figure 3).

Outcomes and estimation

At baseline, study participants were achieving, on average, > 50 minutes of MVPA per day (Table 34). At the 12-month assessment point, no increase in MVPA was observed in either the intervention group or the control group, and hence there was no evidence of an effect of the Walk to Work intervention on the primary outcome measure (p = 0.92) (see Table 34). The ICC for the primary outcome was estimated at 0.020 (95% CI 0.001 to 0.292).

Including participants with at least 1 day of valid accelerometer data, at least 2 weekdays and 1 weekend day, using data from either working days or non-working days only, or log-transforming the primary outcome data (at baseline and follow-up, to model the non-normality), in sensitivity analyses supported the same conclusion (Table 35). We did not carry out the multiple imputation of the primary outcome as specified in the statistical and health economics analysis plan88 as it was clear that using at least 1 day of valid data did not change our conclusions and, therefore, it was agreed with the Trial Steering Committee that this was acceptable. We also did not further adjust our primary outcome analysis by any baseline characteristics, as there was no imbalance in these between the intervention and the control groups at baseline.

The unadjusted means of the primary outcome of daily number of minutes of MVPA and the secondary outcome of overall physical activity decreased in both the intervention group and the control group over the 12-month follow-up (see Table 35). There was no evidence of an effect of the intervention on these measures for any of the adjusted analyses. The unadjusted means of sedentary time increased over the 12-month follow-up in the control group but reduced in the intervention group, again with no evidence of an intervention effect. However, daily number of minutes of MVPA during the commute increased in the control group but reduced in the intervention group, with evidence of a differential effect after adjustment for covariates (p = 0.036). For both the intervention group and the control group, at all assessment points, the median number of journeys walked to work was 0 (interquartile range 0–0), with no evidence of an effect of the intervention on this measure (p = 0.395).

There was no evidence that the effect of the intervention differed between different age groups, men and women or participants with different household incomes (Table 36).

Assessment of costs

Promoter training was conducted in the workplace. Trainer and promoter time and trainer travel costs were recorded on time sheets during the 10-week intervention period. Trainer time was costed using University of Bristol pay scales and included basic salary, National Insurance and superannuation (Table 37). Trainer travel costs were self-reported. If trainers travelled by car, they reported the number of miles travelled and the cost was estimated using a cost per mile of 59.9 pence. 92 The numbers of promoters and employers and the numbers of individuals participating in the intervention at each workplace were recorded. The unit cost of promoter time was calculated by dividing the upper quartile weekly earnings by the median number of hours worked per week, as reported in the annual survey of hours and earnings. 93 The upper quartile was applied as some promoters were also the employer and it was believed that promoters were likely to be on a higher wage than the average worker.

The intervention resources required at each workplace included employee, employer and promoter booklets and folders, employer letters, newsletters, incident forms, poster sets, postage, optional pedometers and DVDs. Materials were printed by the University of Bristol print services, with 60% of the costs covered by Living Streets (www.livingstreets.org.uk; accessed 6 June 2018), a UK charity that promotes everyday walking. Pedometers were donated by Living Streets but were costed at £3.00 per unit in the analysis to reflect the true cost if the intervention was to be rolled out.

Health service use in the previous 4 weeks was self-reported by study participants at baseline and at the 12-month follow-up. Unit costs that were used to value health service use are presented in Table 38.

Primary care, including general practitioner (GP) visits, nurse visits, GP calls and nurse calls, was costed using national unit cost estimates. 94,95 Hospital-based care was costed using NHS Reference Costs 2015–16. 96 Osteopathic and chiropractic care were costed using the NHS Choices website. 98,99 Out-of-hours care and NHS 111/NHS Direct use were costed using published estimates. 100,101 Admitted care and all other care reported by participants, including use of NHS 111/NHS Direct, out-of-hours care, osteopathic care and chiropractic care were grouped together under ‘other care’ for the analysis. All health service use was valued for the year 2016; when 2016 unit costs were not available, costs were inflated to 2016 prices using the Hospital and Community Health Services Index. 94 For five appointments in which participants were asked to provide details of other care they received, some participants reported only the type of care and not the number of appointments; for these, we assumed that the participant had one appointment. Routine dental and optician appointments and pharmacist contacts were excluded from the analysis and a few health-care items were excluded because of a lack of information on what service the participant had received. Medications were costed using the Prescription Cost Analysis – England. 97 When medications could not be costed because of missing information on the number of days taken, it was assumed that medications for chronic conditions and contraception were taken daily. For medications with missing information on the number of doses per day, an assumption of one dose was made. Any medications that could not be costed owing to a lack of sufficient information were excluded.

Self-assessed productivity was measured on a 10-point scale based on the extent to which health problems had affected productivity in the previous 7 days at baseline and in the previous 3 months at the post-intervention and the 12-month follow-ups: 1 indicated that health problems had no effect on an individual’s work and 10 indicated that health problems had completely prevented work. 102 Self-assessed lost productivity was valued using median weekly earnings,93 assuming that a 1-point decrement on the scale equated to a 10% loss in productivity. Absence from work was also self-reported, whereby participants were asked to report the number of hours of work they had missed because of health problems in the previous 7 days at baseline and the number of days they had missed because of health problems in the previous 3 months at the post-intervention and the 12-month follow-ups. Lost productivity due to hours and days of missed work was also valued using median weekly earnings. 93

Workplaces were contacted after the 12-month follow-up and asked to report the total number of sickness absence days and the number of full-time-equivalent employees between baseline and the 12-month follow-up. The overall workplace absentee rates were estimated assuming that full-time employees have 28 days of paid annual leave per year.

Information regarding the commute was recorded for 1 week at baseline and the 12-month follow-up in a travel diary. Time spent commuting by mode of transport, daily expenses (such as bus fares) associated with the commute and occasional commuting expenses (including bus and train passes and parking permits) were all recorded. When reported commute modes and times were found to be inaccurate when compared with GPS data, diary data were replaced with GPS data. For some participants, it was observed that their commute time included time spent doing other activities (such as shopping); for these journeys, when possible, the commute time was replaced with the participant’s average commute time for the given mode of transport. A cost per mile of 59.9 pence92 and an average speed of 23.43 miles per hour19 were used to calculate a cost per minute of driving; this was multiplied by the number of minutes of driving in order to estimate the cost of commuting by car. Costs and durations of parking permits, bus passes and train passes were used to estimate daily parking permit, bus and train costs. Car sharing was incorporated into the costs of car travel by deducting the contribution from passengers from the cost of commuting by car.

Quality of life

Scores on the ICECAP-A, a measure of well-being, were collected at baseline, post intervention and at the 12-month follow-up to capture participant well-being. 52 ICECAP-A is a five-attribute questionnaire, with each attribute containing four levels, which range from no capability to full capability. The ICECAP-A tariff was applied to questionnaire scores to estimate participant quality of life. 103 ICECAP-A scores are anchored on 0 indicating no capability and 1 indicating full capability.

Costs of intervention materials

The costs of intervention materials to deliver the intervention are provided in Table 39.

Cost per employer and participating employee

The costs of the intervention per workplace and employee, by workplace size, are presented in Table 40. Costs varied because of different numbers of promoters in each workplace and depending on the number of employees participating in the intervention at each workplace. The number of trainers delivering the promoter training and travel distance and cost also affected the overall cost. The costs per workplace and per participating employee by workplace size demonstrate that there is no clear association between employer size and cost per participating employee; this may be in part because the number of participants recruited was not consistently larger in larger workplaces.

Health service use

Table 41 provides the mean number of units and the mean cost of each service per participant, by group and by time point. There was a total of 331 participants in the intervention group and 323 participants in the control group. At baseline, the response rate varied between 88% and 91% in the intervention group and between 81% and 88% in the control group, depending on the question asked. At the 12-month follow-up, response rates, including only participants who also recorded their baseline health-care use, varied between 52% and 56% in the intervention group and between 63% and 67% in the control group. On average, the response rate was 36% lower than baseline in the intervention group and 21% lower than baseline in the control group at follow-up.

At baseline, the mean total cost of health services, including medications, in the intervention group was lower than that in the control group, at £50.43 (SD £174.64) compared with £60.61 (SD £236.81). At the 12-month follow-up, the mean cost of health service use was similar in both arms, at £139.75 (SD £372.56) in the intervention group and £140.51 (SD £339.55) in the control group. The adjusted incremental difference of £23.19 (95% CI –£56.09 to £102.47) at the 12-month follow-up was not statistically significantly different from zero.

Commute

A summary of the commute data is presented in Table 42. At baseline, 90% of the intervention participants and 90% of the control participants provided information on their weekly commute to and from work. This information was provided at both baseline and the 12-month follow-up for 56% of the intervention participants and 65% of the control participants.

At baseline, the average daily time of commuting to work was similar between groups, at approximately 71 and 70 minutes for the intervention and the control groups, respectively. At the 12-month follow-up, there was a trend for a lower total daily mean commute time in the intervention group [60.90 minutes (SD 41.48 minutes)] than in the control group [73.56 minutes (SD 57.40 minutes)], with an adjusted incremental difference of –9.17 minutes (95% CI –18.05 to –0.28 minutes).

The percentage of participants with complete information available for costing was 78% of the intervention group and 72% of the control group at baseline. This number dropped at the 12-month follow-up, with 42% of the intervention participants and 41% of the control participants with complete information available for costing. As shown in Table 43, at baseline the mean daily commuting cost was £11.03 (SD £9.74) in the intervention group and £11.37 (SD £10.21) in the control group. At the 12-month follow-up, the mean daily commuting cost was £9.32 (SD £7.67) in the intervention group and £10.99 (SD £12.19) in the control group. The adjusted incremental difference at follow-up was £1.15 (95% CI –£3.10 to £0.79).

Productivity

Response rates to the questions on self-rated productivity and absenteeism declined over the study period (Table 44). At baseline, approximately 89% of participants in the intervention group and 88% of participants in the control group answered these questions. At the 12-month follow-up, response rates had fallen to approximately 59% in the intervention group and 72% in the control group.

At baseline, the mean productivity score in both the intervention group and the control group was low [1.77 (SD 1.47) and 2.02 (SD 1.83), respectively], suggesting a low self-perceived impact of health on productivity. Productivity scores were slightly lower (better) in the intervention group than in the control group at the 12-month follow-up [2.26 (SD 1.83) and 2.85 (SD 2.45), respectively]. When converted to lost wages, on average, the control participants lost more wages than the intervention participants did [£1056.44 (SD £1397.15) and £719.30 (SD £1041.80), respectively], with a difference in wages of –£231 (95% CI –£424.77 to –£37.92) based on a repeated-measures analysis.

The average number of hours missed from work because of health problems was slightly higher in the control group at baseline, at 0.60 hours (SD 3.48 hours) compared with 0.45 hours (SD 2.36 hours) in the intervention group. Immediately post intervention, the mean number of days missed was higher in the control group, at 1.44 days (SD 5.13 days) than 1.01 days (SD 2.79 days) in the intervention group. At the 12-month follow-up, the mean number of days was higher in the intervention group [2.01 days (SD 6.11 days)] than in the control group [1.71 days (SD 4.58 days)]. Conversion into wages suggested that the intervention participants lost, on average, more wages than the control participants [£176.57 (SD £536.05) and £149.92 (SD £401.70), respectively]; however, this result was not statistically significant (mean –£12.50, 95% CI –£84.34 to £59.33).

Data required to estimate overall workplace absentee rates were reported by 15 intervention and 13 control workplaces. The average overall workplace absentee rate between baseline and the 12-month follow-up was 1.68% in intervention workplaces and 2.91% in control workplaces.

Quality of life

At baseline, 90% of the intervention participants and 88% of the control participants completed the ICECAP-A questions. For both trial arms, the percentage reduced over the follow-up period; however, the reduction was larger in the intervention group, with 59% of the intervention participants and 82% of the control participants answering the ICECAP-A questions at the 12-month follow-up.

The ICECAP-A scores, presented in Table 45, were similar between trial arms at baseline, with the mean intervention score being 0.839 (SD 0.133) and the mean control score being 0.831 (SD 0.136). A repeated-measures analysis suggests that the mean ICECAP-A score in the intervention group was higher than that in the control group over the follow-up period (mean 0.018, 95% CI 0.000 to 0.036; scores anchored at 0 indicated ‘no capability’ and scores anchored at 1 indicated ‘full capability’).

Recruitment and retention

Recruitment took place in two phases, during May to July 2015 and March to May 2016, across seven urban areas in south-west England and south Wales. Information sheets with invitations for expressions of interest were sent to approximately 9800 workplace addresses. We received 271 expressions of interest and 87 workplaces were recruited (10 micro-sized, 35 small, 22 medium-sized and 20 large workplaces) involving 654 participants. Following baseline data collection, 44 workplaces were randomised to receive the intervention and 43 workplaces were randomised to the control group. At the 12-month follow-up, 84 workplaces (41 intervention and 43 control) and 477 employees (73% of those originally recruited to the study) took part in data-collection activities. Response rates in the control group were higher than those in the intervention group at the 12-month follow-up.

Baseline characteristics and physical activity

In line with other studies reporting objective measures of physical activity,104–106 a low proportion of participants in our study (≈11%) met current physical activity recommendations of ≥ 150 minutes of MVPA per week in bouts of ≥ 10 minutes. 1 Compared with car users, walkers and public transport users accrued substantially higher levels of daily MVPA during the commute. The absence of free workplace car parking was independently related to walking to work and using public transport. Shorter commuting distances were also related to walking to work, and public transport users were less likely to combine their commute with caring responsibilities.

Intervention delivery

Of the 44 workplaces that were randomised to receive the intervention, two workplaces withdrew during the intervention. A total of 37 workplaces identified a Walk to Work promoter who received training and materials to undertake the role; however, five workplaces were unable to identify a Walk to Work promoter and the local researcher took on the role of distributing intervention materials to participating employees. All employers in the intervention group were provided with booklets and posters encouraging them to support the intervention.

Process evaluation

Descriptive statistics from the process evaluation suggested that a majority of participants in the intervention group were aware of the Walk to Work promoters in their workplace. Employers’ support for the intervention tended to focus on the provision of information rather than initiatives with a higher cost or requiring organisational change, such as improved washing and changing facilities, changes to parking arrangements, incentives to walk during the commute or subsidised sustainable transport initiatives. Key factors influencing whether or not participants increased walking during the commute were identified through the behavioural questionnaires and qualitative interviews, and included commuting distance and workplace location; the availability, reliability and cost of alternatives to private car use; and caring responsibilities, especially child care.

Outcome evaluation

At the 12-month follow-up, we found no effect on the participants’ MVPA levels, overall physical activity or travel mode. There was no evidence that the effect of the intervention differed between different age groups, males and females or participants with different household incomes.

Economic evaluation

On average, the Walk to Work intervention cost £186.20 per workplace and £24.19 per participating employee, representing a relatively inexpensive intervention for employers to adopt. There was weak evidence that self-rated productivity and well-being scores were better in the intervention group than in the control group over the 12-month follow-up period. However, the small differences, the lack of improvement in MVPA and active commuting and the higher loss to follow-up in the intervention group caution against overinterpreting these findings.

Recruitment

Although the target for workplace recruitment was achieved, this was after a large mailout to workplaces across seven urban areas. It was not possible to check if all workplaces on the lists were extant, if addresses and contact details were accurate and if the information reached someone with the authority to make a decision on study participation. However, with only 271 responses from > 9000 letters, we can conclude that the study, which was clearly related to workplace travel behaviour with an emphasis on changing travel mode to increase walking, was not of interest to the majority of employers. Furthermore, there was a relatively low cluster size, even within larger workplaces.

Baseline characteristics and physical activity

There are some notable strengths of this study. To our knowledge, this is the largest study that has combined robust, objectively measured data derived from accelerometer and GPS devices to identify characteristics associated with physical activity and walking as the main mode of travel to work. Data were collected over a relatively short period of time (May to July 2015 and March to May 2016) to minimise the effect of seasonality on physical activity and travel behaviours. In this study, 65% and 11% of the study participants were categorised as car users and walkers, respectively. Similarly, findings from the National Travel Survey: England 2016 showed that 64% and 11% of commute journeys were made by car and walking, respectively. 19 This suggests that the findings we report are similar to national patterns of commuting behaviour and could be more widely generalisable. However, as analyses were undertaken on cross-sectional data, we are unable to establish causal relationships between the variables of interest. We used perceptions of the commute environment, rather than objectively quantified characteristics of the environment through a GIS. Participants of the study were relatively young, predominantly well-educated and employed in sedentary occupations. Therefore, the findings may not be applicable to a population with different characteristics.

Process evaluation

The promoters all received the Walk to Work training session and DVD, booklets and resources relevant to their role and four newsletters during the intervention period. However, they were encouraged to deliver the intervention to colleagues in a way that suited their workplace routines. This model was relatively cheap to deliver, as shown by the economic evaluation, but leaves room for variation and uncertainty about fidelity and reach. On the other hand, the results may be more relevant to a ‘real-world’ situation than an intervention that is highly monitored throughout and is therefore more likely to accrue Hawthorne effects.

The post-intervention questionnaires sent to all study participants, which included questions about interventions relating to travel to work, had noticeably higher non-response rates in the intervention group than in the control group. It is not clear why this is the case; it may relate to the additional participant burden in the intervention group, as these participants had just come to the end of a 10-week intervention, or it may be that there was a reluctance among participants in the intervention group to confirm that they, or the Walk to Work promoter, had not engaged with the intervention in their workplace. A strength of the process evaluation is the number of interviews (n = 70) with employers, Walk to Work promoters and employees. This has provided a substantial data set that has provided additional understanding of some of the statistical findings and will be subject to detailed future analyses.

Outcome evaluation

The use of objective measures, and a 12-month follow-up period, contribute valuable evidence for those who have called for greater rigour in assessing the effectiveness of physical activity interventions,26,28 and interventions aiming to change the travel mode of commuters. 25 It would have been interesting to collect these objective measures immediately post intervention to assess whether or not there was an immediate impact of the intervention on physical activity and commuting behaviour, and the response rate may have been higher. However, this would have increased the burden on research participants and would not have addressed the important concern that long-term follow-up has been identified as a weakness in other studies.

The study included workplaces in geographically distinct areas and of different sizes and industrial classifications, which might add to its generalisability. However, it should be noted that participants were predominantly educated to degree level and with a household income above the national average. It might be thought that study participants were fairly active with a baseline mean daily number of minutes of MVPA of > 50 minutes; however, this was largely not achieved in the recommended bouts of ≥ 10 minutes. Furthermore, there are concerns that feedback from activity monitors is not easily reconciled with current physical activity recommendations because the guidelines refer to the amount of activity required on top of normal activities. 107

Missing data

Three workplaces, all randomised to the intervention group, did not continue with the study. The loss of these workplaces contributed to a higher response rate in the control group at the 12-month follow-up than in the intervention group, but does not completely explain the difference observed. In the intervention group, 142 out of 331 participants (43%) provided a measure of the primary outcome; in the control group, 180 out of 323 participants (56%) provided that measure. The number of missing data is disappointing, and the reasons are unclear. It may be that the interest in the research, and especially the novelty of wearing the monitors, decreased over the 12-month period. At the follow-up data collection point, 477 participants (73%) were still involved with the study and returned some data (questionnaire, travel diary, accelerometer or GPS data). However, it may be that those who had not changed their travel behaviour were less inclined to have this confirmed by wearing the monitors. Moreover, the waist-worn accelerometer involves a research burden for the participants: it is usually worn on an elastic belt (as was the case with our study) and requires a degree of commitment to remember to put it on at the beginning of the day and to wear it for ≥ 10 hours, especially if a change of clothes is required when arriving home.

Although this low response rate is clearly a limitation to the strength of conclusions that can be drawn, we do not believe that the missing measurements cause the study results to be misleading. If the lower completion rate in the intervention group was due to some individuals who did not benefit from the intervention not wanting to have their physical activity assessed, had these participants provided outcome data it would have been in line with the same conclusion of no intervention effect. Furthermore, measuring the primary outcome for all participants who provided 1 day or more of accelerometer data (rather than 3 days or more) provided an outcome measure for 189 out of 331 participants (57%) in the intervention group and 217 out of 323 participants (67%) in the control group; repeating the analysis with these data led to the same conclusion of no effect of the intervention.

Measuring physical activity

The primary outcome of the Travel to Work study was compromised by missing data at the 12-month follow-up and this may have related to the burden of wearing waist-worn accelerometers in free-living conditions. Wrist-worn monitors are easier to wear and may increase compliance. 108 At the outset of this study, wrist-worn accelerometry was evolving and we wished to use an established technology so that we could relate the findings to other studies. Research suggests that the results from wrist-worn accelerometers differ from waist-worn accelerometers, with the wrist attachment producing a higher average step count than the waist attachment in free-living conditions. 109 A study comparing the wrist-worn Fitbit Flex (Fitbit, San Francisco, CA, USA) and waist-worn ActiGraph wGT3X-BT (ActiGraph, Pensacola, FL, USA) in free-living adults found that the Fitbit Flex and ActiGraph respectively classified 51.5% and 37.5% of the days as active. 110 Clearly, caution is needed in the choice of instrument used to measure physical activity in free-living conditions. The potential of wrist-worn devices to collect more data over a longer period of time and in larger samples is obviously attractive to researchers and health professionals. However, more work needs to be done to examine the accuracy of wrist-worn devices in measuring walking-related physical activity.

Economic evaluation

There was a high percentage of missing health economic data, in particular for commuting costs: at the 12-month follow-up, costs could only be estimated for 42% of the intervention participants and 41% of the control participants. To estimate daily commuting costs, participants were required to record complete times and costs for each work day at baseline and follow-up. It was evident that some participants reported details of bus and train costs in multiple places in the travel diary; when possible, this information was included only once. When possible, fuel and parking permit costs were excluded from daily costs as these costs were also recorded elsewhere in the diaries. Some participants reported that they were reimbursed for travel costs by their employer, but these costs were still included in the analyses. Some missing information was observed in the medication section of the questionnaire; as a result, some medications could not be costed and were not included. Consequently, the medication cost for these participants is an underestimate. Although we could have used multiple imputation to impute missing data and obtain estimates for all participants, given that no evidence of an intervention effect was observed, imputation was not conducted as it would not have altered the conclusion.

Recruitment

Overall, it can be concluded that a large proportion of workplaces that were sent information about the study were unwilling or unable to participate. This would suggest that active travel interventions may need to be carefully targeted at workplaces. Issues such as the location of the workplace, the type of business activities, working hours and the availability and reliability of public transport are important factors to be taken into consideration.

Two decisions taken by the research team may have restricted workplace recruitment and employee participation in the intervention. The first relates to the size of workplaces. The research team decided that workplaces with fewer than five employees would be ineligible. This was because of the increased cost of delivering the intervention to only one or two people. However, in the UK in 2014, micro-sized businesses (0–9 employees) accounted for 96% of all businesses and 33% of all employees,87 and so a large pool of workplaces were ineligible for the intervention.

The second issue relates to the distance between where the study participants lived and their workplaces. The original feasibility study,49 which was not powered to provide evidence of effectiveness, had nevertheless shown some promise that the intervention might have an effect. However, the feasibility study had initially focused on recruiting employees who lived within 2 miles of their workplace. After discussion within the team, it was felt that this might have been an unnecessary restriction, making recruitment difficult and failing to reach people who might be willing to increase walking as part of a mixed-mode journey. For the current study, the baseline statistical analyses and qualitative analyses of interview transcripts suggest that commuting distance is a critical factor in whether or not people walk during the commute. Focusing on those who live within walking distance of their workplace but who do not currently walk to work may narrow the field of participants but may increase the likelihood of success in changing travel mode.

Changing travel mode

Our findings suggest that workplace-based interventions focusing on travel mode are more likely to be of interest to ‘motivated subgroups’, as Ogilvie et al. 24 also suggest, rather than a broad spectrum of workplaces and employees. It is worth noting that 89% of participants in our study lived > 2 km from their place of work and, to increase walking during the commute, mixed-mode travel would be more feasible than walking the whole route. Our findings indicate that such a change is likely to be influenced by the availability, cost, convenience and reliability of public transport. The analyses of the baseline data in Chapter 4 indicate that other contextual factors, such as the location of the workplace and the availability of car parking, influence travel mode. 76,111 In addition, interventions may be more effective when they coincide with naturally occurring disruption in travel habits,112 such as moving house, changing employment or reductions in workplace parking, suggesting that tailoring the timing or target group of travel mode interventions in accordance with the wider context could also be an important factor in their impact.

Implications for policy and practice

We have shown that walking to work, either the whole route or combined with public transport, is an important contributor to objectively measured physical activity levels in a large sample of adult employees recruited from diverse workplaces and settings in the UK. We believe that the picture is sufficiently clear to assert that supporting walking during the daily commute (either as the main mode or as part of a mixed-mode journey) should be a priority for both transport and public health disciplines. However, interventions to increase walking to work should take into account the wider determinants of commuting behaviour. Our research suggests that, at organisational and policy levels, consideration needs to focus on commuting distances, availability of car parking and the availability of convenient and reliable public transport. Our findings support the argument that attention should be directed towards a systems approach that focuses on interactions between the correlates of physical inactivity, rather than BCTs focusing on individuals. 113

Previous qualitative research has suggested that child-care commitments may restrict choice of travel mode for the commute. 114,115 Within our sample of employees, we found that child-care commitments were perceived as an important reason why participants were unable to change travel mode from private car use. This was particularly evident in relation to the reliability, frequency and timing of public transport, suggesting that employees with caring responsibilities may require greater flexibility in their working hours if they are to be encouraged to use public transport.

We found that walking and public transport use were both positively associated with a lack of free car parking at work. The removal of this ‘perk’ is unlikely to be popular with employees who value driving to work. Previous qualitative research has suggested that, where removing parking might be perceived as punitive, employers would prefer this to be imposed from outside the workplace. 49 This might, for example, be a directive from a more distant ‘head office’ or because of policies imposed by the local or national government.

In other UK-based studies, a short distance to the workplace75,76,116 and a lack of on-site car parking84,112,117 have been shown to act as facilitators of walking travel modes. In the current study, the majority of participants had a commute distance of > 2 km (n = 555, 84.8%). Therefore, switching to walking as the main mode of travel to work may not have been feasible for many of our study population. However, a mixed-mode commute, combining walking with public transport, may be possible. A case study of 20 UK workplaces showed that limiting parking, by either introducing parking charges or reducing the number of spaces available, and providing payments for public transport users were critical factors contributing to decreased car use. 118 The implementation of workplace policies to limit or charge for workplace car parking, while offering subsidies for public transport, may make modes of travel other than a car more appealing. Providing, or subsidising, car parking away from city-centre workplaces, to enable and encourage employees to park and walk, may be a more palatable option for employees than the removal of parking ‘perks’ altogether, and may be attractive to employers if it proves to be cheaper than the provision of city-centre parking.

In our study, although univariable analyses suggested that walkers and public transport users had more positive perceptions of their commute environment that car users, there was no evidence of a difference after adjustment for other variables. As we did not objectively quantify characteristics of the environment through GIS-based measures, we cannot eliminate physical differences in participants’ commute environments. Our qualitative data suggest that concerns about the walking environment may be a deterrent to walking. However, Guell et al. 119 suggest that some participants will walk despite adverse environmental conditions, having overcome the issue through experience or weighing up the perceived benefits and costs. In the UK, nationwide construction of walking and cycling routes has been shown to be associated with increased physical activity levels and walking and cycling as modes of travel. 120,121 However, distance to the infrastructure was identified as a mediating factor, suggesting that fragmented improvements to infrastructure may not be sufficient to bring about behaviour changes. 122

Recommendations for future research

We believe that there is sufficient reliable evidence to show that walking to work, or incorporating walking into a mixed-mode commute, can enable working adults to build physical activity into their daily routine and meet the physical activity recommendations. We would also argue that there is sufficient evidence to suggest that focusing on individual behaviour change has limited capacity to change travel mode at the population level.

We would recommend that future research considers targeted interventions for particular subgroups (e.g. employees with good public transport links between their workplace and home and employees wanting to increase physical activity for health reasons), interventions that address specific barriers that have been identified in this and other research (e.g. employees with caring responsibilities) and interventions that operate at organisational, environmental and policy levels (e.g. reductions in car parking and improvements to pedestrian routes). Some of the environmental and policy research may be more suited to evaluation as high-quality natural experiments rather than in RCTs.

Some areas for future research to encourage active travel include the acceptability, impact and cost-effectiveness of interventions relating to the reduction or relocation of workplace parking, examining the links between the school run and the commute to work, the effectiveness and cost-effectiveness of targeting travel-to-work interventions at micro-sized workplaces and the impact of infrastructure changes to the walking environment on travel mode.

Future work

The current study has a comprehensive dissemination plan, including publications and presentations to academics, practitioners and policy-makers. This focuses on analysing both the statistical and the qualitative data to (1) further interpret and explain the results relating to physical activity and travel mode, (2) examine contextual issues and (3) explore the wider determinants of workplace travel behaviour.

Travel to Work: randomised controlled trial

Audrey S, Cooper A, Hollingworth W, Metcalfe C, Procter S, Davis A, et al. Study protocol: the effectiveness and cost effectiveness of an employer-led intervention to increase walking during the daily commute: the Travel to Work randomised controlled trial. BMC Public Health 2015;15:154.

Batista Ferrer H, Cooper A, Audrey S. Associations of mode of travel to work with physical activity, and individual, interpersonal, organisational, and environmental characteristics. J Transp Health 2018;9:45–55.

Audrey S, Fisher H, Cooper A, Gaunt D, Garfield K, Metcalfe C, et al. Evaluation of an intervention to promote walking during the commute to work: a cluster randomised controlled trial. BMC Public Health 2019;19:427.

Walk to Work: feasibility study

Audrey S, Procter S, Cooper AR. The contribution of walking to work to adult physical activity levels: a cross sectional study. Int J Behav Nutr Phys Act 2014;11:37.

Procter S, Mutrie N, Davis A, Audrey S. Participants’ views of using behaviour change techniques to encourage walking to work: a qualitative study. BMC Public Health 2014;14:868.

Audrey S, Procter S, Cooper A, Mutrie N, Hollingworth W, Davis A, et al. Employer schemes to encourage walking to work: feasibility study incorporating an exploratory randomised controlled trial. Public Health Res 2015;3(4).

Audrey S, Procter S. Employers’ views of promoting walking to work: a qualitative study. Int J Behav Nutr Phys Act 2015;12:12.