A multicomponent structured health behaviour intervention to improve physical activity in long-distance HGV drivers: the SHIFT cluster RCT

Primary objective

• To investigate the impact of the 6-month SHIFT programme, compared with usual care, on device-measured physical activity (expressed as steps/day) at 6 months’ follow-up.

Secondary objectives

• To investigate the impact of the SHIFT programme, compared with usual care, at 6 months’ follow-up on:

  • time spent in light physical activity and moderate or vigorous physical activity (MVPA)

  • sitting time

  • measures of adiposity (i.e. BMI, per cent body fat, waist–hip ratio, neck circumference)

  • cardiometabolic risk markers [i.e. glycated haemoglobin (HbA_1c), total cholesterol, high-density lipoprotein cholesterol (HDL-C) and low-density lipoprotein cholesterol (LDL-C)]

  • fruit and vegetable intake and dietary quality

  • blood pressure

  • psychophysiological reactivity

  • sleep duration and quality

  • functional fitness (i.e. grip strength)

  • cognitive function

  • mental well-being (i.e. anxiety and depression symptoms, and social isolation)

  • work-related psychosocial variables (i.e. work engagement, job performance and satisfaction, occupational fatigue, presenteeism, sickness absence and driving-related safety behaviour)

  • health-related quality of life (HRQoL)

  • health-related resource use [i.e. general practitioner (GP) visits].

• To investigate the longer-term impact of the SHIFT programme, compared with usual care, at 16–18 months’ follow-up on:

  • steps per day

  • time spent in light physical activity and in MVPA

  • sitting time

  • fruit and vegetable intake and dietary quality

  • sleep

  • mental well-being (i.e. anxiety and depression symptoms, and social isolation)

  • work-related psychosocial variables (i.e. work engagement, job performance and satisfaction, occupational fatigue, presenteeism, sickness absence and driving-related safety behaviour)

  • HRQoL.

• To conduct a mixed-methods process evaluation throughout the implementation of the intervention (using qualitative and quantitative measures) with participating drivers and site managers.

• To undertake a full economic analysis of the SHIFT programme.

The SHIFT programme

The SHIFT programme is a multicomponent lifestyle–behaviour intervention that is designed to target behaviour changes in physical activity, diet and sitting in HGV drivers. The 6-month intervention, grounded within social cognitive theory (SCT) for behaviour change,42 consists of a group-based (4–6 participants) 6-hour structured education session, tailored for HGV drivers and delivered by two trained educators. The education session includes information about physical activity, diet and sitting, and details risk factors for type 2 diabetes and CVD. The educational component is founded on the approach used in the award-winning suite of DESMOND (Diabetes Education and Self-Management for Ongoing and Newly Diagnosed) programmes, including the PREPARE (Prediabetes Risk Education and Physical Activity Recommendation and Encouragement)43 and Let’s Prevent Diabetes programmes,44 created by researchers at the Leicester Diabetes Centre (Leicester, UK) and used throughout the NHS,45 while being tailored to meet the needs of HGV drivers. 7 During the education session, participants are not ‘taught’ in a formal way but are supported to work out knowledge through group discussions. Participants are also encouraged to develop individual goals and plans based on detailed individual feedback received during their health assessments (see Outcome measurements) to achieve over the 6-month intervention period. The education session is supported by specially developed resources and participant support materials for HGV drivers. The education session includes the discussion of feasible strategies for participants to increase their physical activity, improve their diet and reduce their sitting time (when not driving) during working and non-working hours. The content of the educational session is summarised in Table 2.

During the education session, participants were provided with a Fitbit Charge 2 (Fitbit, Inc., San Francisco, CA, USA) activity tracker. Participants were encouraged to use the Fitbit activity tracker to set goals (agreed at the session) and gradually increase their physical activity, predominantly through walking-based activity. The Fitbit activity tracker and associated smartphone app provided participants with information on their daily step counts and was used as a tool for self-monitoring and self-regulation. Physical activity tracking using step counters (traditionally pedometers) has been associated with significant reductions in BMI and blood pressure, with interventions incorporating goal-setting being the most effective. 49 Participants were provided with instructions on how to link their Fitbit account to an online monitoring system (Fitabase, Small Steps Labs LLC, San Diego, CA, USA). Participants were encouraged to link their account to Fitabase and to regularly upload their Fitbit data from their device to their mobile phone via Bluetooth. When participants sync their Fitbit through the Fitbit app, their step count data are automatically updated on the Fitabase website. Participants’ data on the Fitabase website were accessible to only two members of the research team, who used the step count data to provide participants with individually tailored step count challenges throughout the 6-month intervention period.

The education session adopted the promotion of the ‘small changes’ philosophy, using the specific, measurable, achievable, relevant, time bound (SMART) principle50 to encourage participants to gradually build-up their daily activity levels, within the confines of their occupation, to meet the UK physical activity guidelines. 51 For example, participants were encouraged to establish their own personalised action plan, which may have included making dietary improvements in addition to increases in physical activity, with SMART goals throughout the 6-month intervention. ‘Step count challenges’ were run every 6 weeks throughout the 6-month intervention and were facilitated by members of the research team via a text messaging service (TextMagic™, TextMagic Ltd, Cambridge, UK).

A ‘cab workout’ was introduced and practised at the education session, and participants were provided with resistance bands and balls, and grip strength dynamometers to take away. Participants were encouraged to undertake the cab workout during breaks when they were not permitted to leave their vehicle. Participants were able to keep the intervention tools beyond the 6-month intervention period; however, the step count challenges, as well as the supportive text messages sent by members of the research team, ended after the 6-month intervention period. A logic model detailing the underlying theory behind the intervention components is shown in Figure 1.

FIGURE 1.

A logic model for the SHIFT programme. CPC, Certificate of Professional Competence; PA, physical activity.

The structured education session was delivered by trained members of the research team in collaboration with trained personnel from DHL Supply Chain. Individuals from DHL Supply Chain co-delivering the education session were predominantly HGV drivers who also acted as driver trainers in each site as part of their role. The ‘driver trainers’ were trained by specialist educators from the Leicester Diabetes Centre and mentored by trained members of the research team. The education sessions took place within appropriate training rooms within the intervention depots. Personnel co-delivering the education sessions in each intervention depot were also trained to act as local champions, which has been shown to enhance the effectiveness of worksite physical activity interventions. 52 They provided ongoing health coach support, along with members of the research team (who provided support via the text messaging service), to intervention participants (during the 6-month intervention period).

The control arm

Sites assigned to the control arm (i.e. usual practice) were asked to continue with their usual-practice conditions. Participants in the control sites received an educational leaflet at the outset, detailing the importance of healthy lifestyle behaviours (i.e. undertaking regular physical activity, breaking up periods of prolonged sitting and consuming a healthy diet) for the promotion of health and well-being. Control participants completed the same study measurements as participants in the intervention worksites, at the same time points, and received the same health feedback immediately following their health assessments (i.e. outcome measurements).

Protocol for the outcome measurements assessed at baseline and at 6 months’ follow-up

Baseline measurements took place prior to randomisation of the sites into the two study arms. A second set of identical measurements occurred at 6 months’ follow-up. The two sets of measurements were undertaken in suitable rooms within participating DHL Supply Chain sites by trained researchers and lasted approximately 2 hours per participant. Participants were scheduled to attend these measurements, during their working time by their transport manager either before or following their driving shift.

Participants completed a range of self-report questionnaires and had a series of physiological health assessments taken (described below) at baseline and immediately following the completion of the 6-month intervention. Participants were also issued with two devices [an activPAL and a GENEActiv (Activinsights, Kimbolton, UK) accelerometer] to wear over a period of 8 days following the measurement sessions. Participants received detailed feedback on their physiological health assessment measures during these two measurement sessions. If a potential health issue was evident during the measurements, such as undiagnosed hypertension or high cholesterol levels, then participants were advised to visit their GP for further checks. A standard referral letter was provided for participants to give to their GP, which summarised the findings from our point-of-care (i.e. blood markers) and automated (i.e. blood pressure) measures.

Protocol for the 16- to 18-month follow-up assessments (undertaken during the COVID-19 pandemic)

A third set of outcome measures were originally planned to take place 6 months following completion of the intervention (i.e. a 12-month follow-up); however, owing to the COVID-19 pandemic, these measurements were unable to be completed within this time frame. The easing of government COVID-19 restrictions enabled a range of secondary outcome measures to be collected at 16–18 months (approximately 10–12 months following completion of the intervention). Owing to restrictions on external visitors to DHL Supply Chain sites throughout the pandemic, face-to-face physiological measurements were not able to be conducted at the final follow-up phase. Instead, the case report form (CRF), which contained a series of self-report questionnaires and recording sheets for the physiological measures used during data collection at baseline and immediately following the intervention (see Report Supplementary Material 1), was modified into a self-administered questionnaire booklet (see Report Supplementary Material 2).

Individual participant packs were prepared, which contained an instruction leaflet, a questionnaire booklet, a consent form, an activPAL and logbook, and a return envelope. On prior arrangement with transport managers, a member of the research team delivered the participant packs to each site. The transport managers distributed the packs to participating drivers, who completed the relevant paperwork in their own time and, on request, wore the activPAL for a period of 8 continuous days. After this 8-day period, participants returned their activPAL and their completed logbook, questionnaire booklet and consent form in a sealed envelope to a collection point within their site. Once all packs were returned, the packs were collected from the site by a member of the research team. This protocol was also followed for the one remaining intervention site for its 6-month follow-up (which had initially been due to take place at the beginning of the first national lockdown and, therefore, was unable to be completed as intended). Table 3 summarises all measurements collected at the three time points during the trial. All measurements are described in detail in the following sections.

Secondary activPAL variables

Sitting, standing, time in light intensity physical activity and time in MVPA were assessed using the activPAL micro accelerometer. The activPAL is regarded as the most accurate method of assessing sitting behaviour in free-living settings,55,57,58 and is recommended for use in interventions when sitting is an outcome measure. 54 From the data provided by the device, the following variables were derived by calculating the average across the number of valid days provided during each measurement period:

• average total daily sitting time (minutes/day)

• average total daily sitting time (minutes/day) accumulated in prolonged bouts lasting ≥ 30 minutes

• average total daily standing time (minutes/day)

• average total daily stepping time (minutes/day)

• average number of transitions from sitting to an upright posture

• average total daily time in MVPA (minutes/day), calculated as total stepping time at a step cadence threshold of 100 steps per minute (in bouts lasting ≥ 1 minute)

• average total daily time in light physical activity (minutes/day)

• number of valid days

• average waking wear time (minutes/day)

• average percentage of the day spent sitting

• average percentage of the day spent standing

• average percentage of the day spent stepping

• average percentage of total sitting time spent in prolonged sitting bouts (lasting ≥ 30 minutes).

The variables below were calculated and summarised for three different time periods within each measurement period: (1) daily (i.e. across all waking hours on all valid days), (2) during workdays only and (3) during non-workdays only.

Anthropometry and markers of adiposity

Height was measured at baseline only, without shoes and to the nearest millimetre, using a portable stadiometer (seca 206, seca Ltd, Birmingham, UK). Weight (kg) and body fat percentage were assessed via bio-impedance analysis using Tanita DC-360S body composition scales (Tanita Corporation, Tokyo, Japan). A clothing allowance of 1.5 kg was entered into the scales, along with participants’ age, sex and height. BMI (kg/m²) was calculated as weight (kg)/height (m²). Waist, hip and neck circumferences (cm) were measured using standard anthropometric measuring tape (seca Ltd, Birmingham, UK), and waist-to-hip ratio was calculated.

Biochemical assessments

Capillary blood samples were collected via finger-prick blood sampling. Participants were requested to place their hand in a bowl of warm water (provided) for 5 minutes prior to the sample being collected. Participants were also requested to fast for at least 4 hours prior to attending their health assessment. HbA_1c (mmol/mol) was measured using an A1CNow^®+ point-of-care analyser (PTS Diagnostics, Indianapolis, IN, USA). Triglycerides (mmol/l), HDL-C (mmol/l) and total cholesterol (mmol/l) levels were assessed using a Cardiocheck^® point-of-care analyser (PTS Diagnostics, Indianapolis, IN, USA). LDL-C (mmol/l) was calculated using Friedewald’s formula. 59

Dietary quality and fruit and vegetable intake

Dietary quality and fruit and vegetable intake (g/day) were assessed using a short-form Food Frequency Questionnaire (FFQ). 60 Using this measure, a dietary quality score was derived from reported fruit, vegetable, oily fish, non-milk extrinsic sugar and fat intakes. The dietary quality score calculated using this short-form FFQ has been shown to demonstrate a significant agreement (κ = 0.38) with dietary quality determined using a 217-item FFQ. 60

Sleep duration and quality, subjective sleepiness and chronotype

Sleep duration and quality were assessed using a GENEActiv tri-axial accelerometer (ActivInsights Ltd., Huntingdon, UK), which was worn (concurrently with the activPAL) on the non-dominant wrist continuously for 8 days. The GENEActiv has been shown to provide an accurate measure of sleep and activity behaviour patterns over a 24-hour period. 61 The device collected data at 100 Hz with a ± 8 g dynamic range. Participants were asked to note any time they removed this device on the same logbook used for the activPAL.

Situational sleepiness was assessed using the Karolinska Sleepiness Scale, which has been shown to be a valid measure of sleepiness when validated against electroencephalography and performance outcomes. 62,63 Participants’ chronotype was determined using the short version of the Morningness–Eveningness Questionnaire (MEQ). 64

Blood pressure

Blood pressure and heart rate were measured from the left arm of the driver after a 20-minute period of quiet sitting using an automated monitor (Omron HEM-907, Omron Corporation, Kyoto, Japan), in accordance with recommendations from the European Society of Hypertension. 65 Three separate measurements of blood pressure and heart rate were taken at 5-minute intervals. The mean systolic and diastolic blood pressures, and heart rate, recorded from the second and third assessments, were calculated and used in the analyses.

Cognitive function and psychophysiological reactivity

The Stroop test was administered over a 5-minute period using a validated software package (SuperLab 5, Cedrus Corporation, San Pedro, CA, USA) to provide a measure of reaction time, sensitivity to interference and the ability to suppress an automated response (i.e. reading colour names in favour of naming the font colour). 66 The Stroop test was utilised to provide a measure of cognitive function and as part of a battery of measures to induce acute stress to support the assessment of psychophysiological reactivity.

The mirror tracing task (Campden Instruments Auto Scoring Mirror Tracer 58024E, Campden Instruments LTD, Loughborough, UK) was used as the second stress task, which has been routinely used to induce stress in field- and laboratory-based studies. 67 The mirror tracing task immediately followed the Stroop test. The mirror tracing task involved tracing an adonised star pattern using a metal-tipped stylus with the right hand continuously for 5 minutes. Participants were, however, permitted to use only the reflection of the star in an adjacent mirror for reference. The machine beeped if the metal-tipped stylus left the star pattern, and each mistake was recorded on the machine. Participants were told to aim for at least five complete stars in the time frame. 68 Measurements of blood pressure and heart rate were repeated during the mirror tracing task at 2 minutes 15 seconds, and again at 4 minutes 35 seconds, into the task to measure psychophysiological reactivity to acute stress. The mean stress-induced blood pressure and heart rate readings were calculated from these two measurements. Blood pressure and heart rate psychophysiological reactivity were calculated by subtracting the average resting systolic and diastolic blood pressures, and resting heart rate, from the average systolic and diastolic blood pressures, and heart rate, taken during the stress task.

Functional fitness

Grip strength (kg) was assessed from both hands using the Takei Hand-Grip dynamometer (Takei Scientific Instruments Co., Ltd, Niigata, Japan).

Mental well-being

Depression and anxiety symptoms were self-reported using the validated Hospital Anxiety and Depression Scale (HADS). 69 The HADS consists of two subscales containing seven questions for anxiety symptoms and seven questions for depressive symptoms. The Cronbach’s alpha for HADS anxiety and HADS depression has been reported as 0.83 and 0.82, respectively. 70 Each answer is scored on a scale from 0 to 3. Therefore, total scores for each construct range from 0 to 21. For each construct, a score of ≤ 7 would be classified as ‘no symptoms’, whereas scores of 8–10, 11–14 and 15–21 are classified as the presence of mild, moderate and severe symptoms, respectively. 70 Social isolation was assessed using the 8-item Social Isolation Short Form from the Patient-Reported Outcomes Measurement Information System. 71,72

Musculoskeletal symptoms

Musculoskeletal symptoms were assessed using the standardised Nordic Musculoskeletal Questionnaire, which is a self-reported measure of musculoskeletal pain covering nine body regions. 73

Work-related psychosocial variables

A series of self-reported questionnaires assessed a range of work-related psychosocial variables. Work engagement (characterised by vigour, dedication and absorption) was measured using the Utrecht Work Engagement Scale. 74 Occupational fatigue was measured using the Occupational Fatigue Exhaustion Recovery (OFER) scale. 75 Perceived job performance76 and job satisfaction77 were measured using single-item 7-point Likert scales. Perceived work ability was assessed using the single-item Work Ability Index. 78 Sickness presenteeism and absenteeism were assessed using a single-item questionnaires. Participant’s perceptions of their work demands and support was assessed using four subscales from the Health and Safety Executive Management Standards Indicator Tool. 79 Reported driving-related safety behaviour was assessed using a six-item measure. 80

Health-related quality of life and resource use

Health-related quality of life was measured using the EuroQol-5 Dimensions, five-level version (EQ-5D-5L). 81,82 The EQ-5D-5L measure comprises a short descriptive questionnaire and a visual analogue scale. On the descriptive questionnaire, participants rate their current health state across five dimensions (i.e. mobility, self-care, usual activities, pain/discomfort, anxiety/depression) and across five levels of severity (ranging from ‘no problem’ to ‘unable to/extreme problems’). The visual analogue scale (which ranges from 0 to 100) records the participant’s overall current health, where the end points are labelled ‘the best health you can imagine’ (100) and ‘the worst health you can imagine’ (0).

Information on health-related resource use was collected using a questionnaire designed for this study. Using this tool, participants were asked to report information on the quantity and duration of GP and nurse practitioner visits, inpatient and outpatient appointments, and visits with other relevant health professionals. The information obtained from the EQ-5D-5L and health-related resource use questionnaire was used to inform the within-trial cost-effectiveness analysis (see Chapter 4).

Demographics and additional lifestyle health-related behaviour and risk measures

At baseline, participants completed a brief questionnaire collecting basic demographic information, including date of birth, sex, ethnicity, highest level of education, marital status, postcode (to determine Index of Multiple Deprivation as an indicator of neighbourhood socioeconomic status), working hours, years worked as a HGV driver, shift pattern and years worked at DHL Supply Chain. At each follow-up assessment, participants were asked if there have been any changes in these variables. During each assessment, information on smoking status and typical alcohol intake [using questions 1 and 2 from the Alcohol Use Disorders Identification Test (AUDIT)83] was gathered. Using information collected from the self-report questionnaires, and data collected within the health assessments (i.e. systolic blood pressure, cholesterol/HDL-C ratio, height and weight), participants’ 10-year risk of having a cardiovascular event was calculated using the Cardiovascular Risk Score (QRISK3) calculator [URL: https://qrisk.org/2017/ (accessed 16 July 2021)].

activPAL

activPALs were initialised and downloaded using manufacturer proprietary software (activPAL Professional v.7.2.38, PAL Technologies Ltd, Glasgow, UK). Event files were generated and processed using the freely available Processing PAL software [URL: https://github.com/UOL-COLS/ProcessingPAL (accessed 24 August 2022), version 1.3, University of Leicester, Leicester, UK]. The software provides information on valid waking wear time, sleep time, extended non-wear time and invalid data, according to a validated algorithm. 84 Once data were processed, heat maps were created, showing valid waking wear data and invalid data. The heat maps were visually checked independently by two researchers for any occasions where the algorithm had misclassified waking wear data, and vice versa. On any occasion where suspected misclassifications had occurred, the participant’s self-reported logbook wake and sleep times were compared with the processed data. If a misclassification was confirmed, then the data were corrected. The logbooks were also checked for scenarios where data should be removed, for example if participants reported removing the device for any reason. Once this process was completed, summary variables were calculated (see Secondary activPAL variables). A valid activPAL wear-day was defined as having ≥ 10 hours wear time per day, ≥ 1000 steps per day and < 95% of the day spent in any one behaviour (e.g. sitting, standing or stepping). Participants were included in the primary outcome analysis if they provided at least 1 valid wear-day at both baseline and 6 months’ follow-up (i.e. immediately following completion of the 6-month intervention). One valid day was chosen to maximise our sample and is in line with previous studies. 85,86

GENEActiv

GENEActiv devices were initialised and downloaded using manufacturer proprietary software (GENEActiv v.3.1, Activinsights Ltd, Huntingdon, UK). Accelerometer files were processed in the R package GGIR version 1.11-0 (The R Foundation for Statistical Computing, Vienna, Austria)87 to generate sleep outcome variables, with sleep duration (i.e. minutes/24-hour period) and sleep efficiency [i.e. sleep duration/sleep window duration × 100 (%)] the variables of interest for this report. ‘Sleep windows’ (i.e. the time between ‘lights out’ and out of bed time) were detected from the accelerometer data using a validated algorithm. 88 Sleep duration within the sleep window period was calculated using a validated sleep detection algorithm, which has been shown to demonstrate high sensitivity and specificity in detecting sleep periods. 89 A device wear time of ≥ 16 hours per 24-hour period was required to determine a valid night of sleep data. 89 Individual nights of data with a sleep window > 13 hours or < 2 hours or sleep duration > 12 hours or < 1 hour were identified as erroneous and removed. As with the activPAL data, participants were required to have provided at least 1 valid wear-day at both baseline and follow-up (i.e. immediately following completion of the 6-month intervention) to be included in the analyses within this report.

From the data provided by the GENEActiv, the following variables were derived by calculating the average across the number of valid days provided during each measurement period:

• sleep window duration [i.e. average duration between ‘lights out’ and ‘out of bed’ time (minutes)]

• sleep duration [i.e. average time spent asleep during the sleep window (minutes)]

• sleep efficiency [i.e. sleep duration/sleep window duration × 100 (%)]

• average number of valid days (days).

The variables below were calculated and summarised for three different time periods within each measurement period: (1) daily (i.e. across all 24-hour periods on all valid days), (2) on workdays only and (3) on non-workdays only.

Sensitivity analyses

Sensitivity analyses were conducted (see Full intention-to-treat analysis and Effects on the number of valid activPAL days), using similar methodology as the primary outcome analysis. There was no formal adjustment for multiple significance testing. The sensitivity analyses were conducted for the primary outcome (i.e. average daily step counts at 6 months’ follow-up). All tests and reported p-values were two sided. Estimates are presented with 95% CIs.

Per-protocol analysis

The effect size was also estimated using a per-protocol analysis. The per-protocol population were participants who did not exhibit any protocol deviations, and excluded participants who:

• did not provide valid activPAL data at baseline or at the 6 months’ follow-up (as applied in the primary outcome analysis)

• had time window deviations for their follow-up (> ± 2 months) assessment.

Full intention-to-treat analysis

Sensitivity analyses were performed to assess the impact of missing data on the primary results and to account for uncertainty associated with imputing data (full ITT analysis). To allow for analysis of the full data set, missing data from variables included in the primary analysis model (i.e. average daily steps at baseline and immediately following the intervention) were imputed using a multiple imputation procedure, which substituted predicted values from a regression equation. The following variables were used as predictors of the primary outcome in the regression equation: baseline BMI, sex, ethnicity, age, cluster size category, years worked as HGV driver and average waking wear time across baseline and 6 months. Missing values for these predictor variables were also imputed if needed. The imputation was carried out by the MI command in Stata^® (StataCorp LP, College Station, TX, USA). MI replaced missing values with multiple sets of simulated values to complete the data, performed standard analysis on each completed data set and adjusted the obtained parameter estimates for missing data uncertainty using Rubin’s rules to combine estimates. 103 Twenty imputations were estimated and a seed was set to allow reproducibility.

Additional worst- and best-case scenario ITT analyses using basic imputation methods were also carried out. A simple worst-case scenario ITT analysis was carried out, where missing covariate data in the final analysis model were replaced using cluster means. Where it was not possible to impute using the cluster mean, the mean for the respective arm was used instead. Missing outcome data in the final analysis model (i.e. at baseline and 6 months) were replaced using the mean for the standard care arm. Furthermore, a simple best-case scenario ITT analysis was also carried out using the same approach as above, but outcome data were replaced using the mean for the respective arm.

Effects on the number of valid activPAL days

We carried out further sensitivity analyses by assessing the effect of the number of valid activPAL days on the primary outcome analysis. This analysis was performed by including participants who provided valid activPAL data (including weekdays and weekend days) on:

• ≥ 2 valid days at both baseline and 6 months

• ≥ 3 valid days at both baseline and 6 months

• ≥ 4 valid days at both baseline and 6 months.

Recruitment, compliance and retention outcomes

Table 4 summarises the recruitment, compliance and retention outcomes observed from the internal pilot sites. Outcomes were reviewed by the TSC on the 11 December 2018; on the basis of the data reviewed, the TSC recommended continuation of the trial.

Participant recruitment

Participant recruitment across the remaining 19 sites commenced in January 2019, and baseline measurements for participants consented into the study were undertaken across these sites between February and July 2019.

As experienced in the internal pilot, a number of delays were encountered during the main trial phase. The delays were predominantly associated with challenges across sites in scheduling drivers for their measurement sessions, which required drivers to be released from their duties for 2 hours. The challenges were further exacerbated in the sites randomised to the SHIFT arm, which required drivers to attend the 6-hour structured education session, which was scheduled during work time. As a consequence, delays in the 6-month follow-up measures were experienced across the majority of sites.

Overall cluster and participant numbers

Figure 2 shows the flow of all participants through the study, combining data from the internal pilot and main trial phases. Overall, 386 participants across 25 clusters (i.e. sites) were recruited and consented into the study. The 25 sites were located across the Midlands region of the UK (1502 drivers were employed across these sites), and the sites operated within the transport, retail, hospitality, health-care, pharmaceutical, construction, oil and gas, and automotive industries. Of the 386 participants recruited, 382 participants were randomised into the two trial arms and four participants withdrew prior to randomisation. Thirteen sites (n = 199 participants) were randomised to the control arm and 12 sites (n = 183 participants) were randomised to the SHIFT arm. Between baseline and 6-month follow-up measures, two sites (i.e. one intervention site and one control site) dropped out of the trial. For both sites, this was because of site closures due to the collapse of the contracting companies.

FIGURE 2.

A CONSORT diagram of participant flow through the study.

Baseline characteristics

Characteristics of the clusters and baseline demographic characteristics of participants within each trial arm, and overall, are shown in Table 5. Table 6 displays the biometric measurements collected from the sample overall, and according to trial arm, at baseline. Table 7 displays the accelerometer-derived measures (i.e. physical activity, sitting time and sleep) collected at baseline. Descriptive comparisons between baseline characteristics of participants completing the trial and non-completers are shown in Appendix 1, Table 42. There were no noticeable differences between completers (i.e. participants who provided valid activPAL data at baseline and at 6 months) and non-completers in terms of cluster size, age, BMI, number of years as a HGV driver and number of steps per day at baseline.

activPAL-assessed secondary outcomes

Further activPAL variables

Across all monitored days and workdays, for the SHIFT group, there were no noticeable differences in time spent sitting in prolonged bouts (> 30 minutes), the number of transitions from sitting to standing and the proportions of time spent sitting, standing and stepping, and the proportion of sitting spent in prolonged bouts, between baseline and 6 months’ follow-up. Similar findings were observed for the control group, except for the time spent sitting in prolonged bouts, which tended to increase at 6 months’ follow-up (Table 15). On non-workdays, the control group exhibited increases in the time spent sitting (and the proportion of sitting) in prolonged bouts at 6 months’ follow-up, relative to baseline. The control group also exhibited an increase in the overall proportion of time spent sitting and a decrease in the proportion of time spent standing on non-workdays at 6 months’ follow-up. For the SHIFT group, no noticeable differences were observed for any variables on non-workdays between baseline and 6 months (see Table 15).

Across all monitored days, workdays and non-workdays, for both groups, there were no noticeable differences in time spent sitting in prolonged bouts (> 30 minutes), the number of transitions from sitting to standing, the proportions of time spent sitting, standing and stepping, and the proportion of sitting spent in prolonged bouts, between baseline and 16–18 months’ follow-up (Table 16).

Sleep duration and quality, subjective situational sleepiness and chronotype

Across all monitored days, between baseline and 6 months’ follow-up, both groups exhibited a decrease in their sleep window duration (defined as the time between ‘lights out’ and out of bed time) and a decrease in their overall sleep duration. These changes appeared to be driven by large reductions in sleep window duration and sleep duration on workdays at 6 months’ follow-up. In contrast, on non-workdays, increases in sleep window duration and sleep duration were observed for both groups at 6 months. There were no noticeable changes in sleep efficiency across any types of day between baseline and 6 months’ follow-up for either group (Table 17). There were no changes in ratings of situational sleepiness or chronotype score between baseline and either follow-up period for both groups (Table 18).

Blood pressure and psychophysiological reactivity

Small reductions in resting systolic and diastolic blood pressure were observed for both groups between the baseline and 6-month follow-up measures, but no noticeable changes were observed in resting heart rate for either group (Table 19). Mean blood pressure and heart rate measures increased during the mirror tracing task; however, the differences between resting values and values recorded during the task tended to be smaller for both groups during the 6-month follow-up measures (see Table 19). There were no differences in perceived stress ratings during this task between baseline and follow-up for either group. The control group tended to have fewer errors while undertaking this task at 6-months follow-up, whereas no evidence of a change in performance was observed for the SHIFT group (see Table 19).

Cognitive function

There were no noticeable differences in reaction times, measured using the Stroop test, between baseline and 6 months’ follow-up for both groups. No noticeable differences were observed between groups at baseline and 6 months (Table 20).

Functional fitness

No noticeable changes in grip strength were observed between baseline and 6 months’ follow-up in the control group, whereas modest improvements in grip strength for both hands were observed following completion of the intervention in the SHIFT group (Table 21).

Mental well-being

There were no noticeable differences in self-reported scores for symptoms of anxiety, depression or social isolation between baseline and 6 months’ follow-up for both groups, and similar findings were also observed for symptoms of anxiety and depression at 16–18 months’ follow-up. There was a tendency in both groups for perceived social isolation scores to increase marginally at 16–18 months’ follow-up (Table 22). No noticeable differences were observed between groups at any assessment point.

Musculoskeletal symptoms

Table 23 provides a summary of the prevalence of musculoskeletal discomfort reported in the past month for each body site, along with discomfort scores by body region, reported over the three time points. There was a tendency for the prevalence of musculoskeletal discomfort across the majority of body sites to decrease at the two follow-up assessments in both groups, with similar changes in prevalence occurring between groups. Similarly, there were no noticeable differences in discomfort scores (i.e. upper extremity, lower extremity and overall) between baseline and 6 months’ follow-up, and between baseline and 16–18 months’ follow-up, for both groups. No noticeable differences in discomfort scores were observed between groups at any assessment point.

Work-related psychosocial variables

Table 24 provides a descriptive summary of a range of work-related psychosocial variables assessed across the three time points. There were no noticeable differences in any of the outcome measures (i.e. work engagement, occupational fatigue, perceived job satisfaction and performance, sickness absence, presenteeism, perceived work ability and perceived job demands) between baseline and 6 months’ follow-up, and between baseline and 16–18 months’ follow-up, for both groups. No noticeable differences in any measure were observed between groups at any assessment point.

Driving-related safety behaviour

There were no noticeable differences in self-reported driving-related safety behaviour between baseline and 6 months’ follow-up, and between baseline and 16–18 months’ follow-up, for both groups (Table 25). No noticeable differences were observed between groups at any assessment point.

Health-related quality of life

There were no noticeable differences in perceived markers of HRQoL between baseline and 6 months’ follow-up, and between baseline and 16–18 months’ follow-up, for both groups (Table 26). No noticeable differences were observed between groups at any assessment point.

Lifestyle health-related behaviours and risk measures

There were no noticeable differences in reported alcohol intake between baseline and 6 months’ follow-up, and between baseline and the final follow-up, for both groups (Table 27). No noticeable differences were observed between groups at any assessment point.

There were no differences in QRISK3 scores between baseline and 6 months’ follow-up for both groups (see Table 27). Likewise, there were no noticeable differences in QRISK3 scores between the control group and the SHIFT group at any assessment point. When examining the proportion of participants with an estimated CVD risk of ≥ 10% over the next 10 years, 23.6% of control participants fell into this category at baseline, with this proportion increasing to 26.4% at 6 months. In contrast, 24.3% of participants in the SHIFT group exhibited a ≥ 10% risk of CVD over the next 10 years at baseline, but this fell slightly to 23.4% at 6 months.

Table 27 also provides a summary of the smoking prevalence reported by participants at each time point. There was a tendency for a higher smoking prevalence to be seen in the control group, than in the SHIFT group, across all assessment points. Between baseline and 6 months, two participants in the control group and one participant in the SHIFT group reported changing from a past smoker to a current smoker (i.e. re-starting smoking), and four participants in the SHIFT group reported stopping smoking. No control participants reported stopping smoking between baseline and 6 months. The average number of cigarettes smoked per day by current smokers in the SHIFT group (baseline, n = 16; 6 months, n = 13) was 14 (SD 5) cigarettes per day at both baseline and 6 months’ follow-up. The average number of cigarettes smoked per day by current smokers in the control group (baseline, n = 27; 6 months, n = 29) was 14 (SD 6) cigarettes per day at baseline and 13 (SD 7) cigarettes per day at 6 months.

Between baseline and 16–18 months’ follow-up, one participant in the SHIFT group reported re-starting smoking at and three participants (control group, n = 1; SHIFT group, n = 3) reported stopping smoking. The average number of cigarettes smoked per day by current smokers in the SHIFT group (baseline, n = 14; 16–18 months, n = 13) was 15 (SD 5) cigarettes per day at both baseline and 16–18 months. The average number of cigarettes smoked per day by current smokers in the control group (baseline, n = 21; 16–18 months, n = 20) was 14 (SD 6) cigarettes per day at both baseline and 16–18 months.

Participants

Of the 220 participants who were still enrolled in the study in May 2020, 91 (41.4%; control group, n = 48; SHIFT group, n = 43; 99% male) participants completed an additional online questionnaire that captured data on the effect of the pandemic on their working hours, mental well-being and health-related behaviours. The questionnaire was completed during the UK’s first national lockdown. At the time of completing the questionnaire, 20 (22%) participants [control group, n = 15 (31%); SHIFT group, n = 5 (12%)] were on furlough, and 44 (48%) participants [control group, n = 32 (67%); SHIFT group, n = 12 (28%)] reported being on furlough at some point during the pandemic.

There were no statistically significant differences in questionnaire responses between intervention or control participants, or between participants who were/had been furloughed and participants not furloughed. As a result, the responses received for the COVID-19 questionnaire are presented for the group as a whole. The only measure where a difference was reported was ‘sleep duration in the past 14 days’. Participants who had been or who were still on furlough reported a longer sleep duration [median 7.0 (IQR 5.5–8.5) hours/night] than participants who were not furloughed [median 6.5 (IQR 5.5–7.5) hours/night].

Baseline characteristics and measures of the 91 participants completing the COVID-19 questionnaire did not differ significantly from the remainder of the participants within the SHIFT study, and this suggests that the subsample of 91 participants are largely representative of the wider sample. Specifically, there were no differences at baseline between participants completing the COVID-19 questionnaire and the wider sample in terms of age, duration working as a HGV driver, duration working for DHL Supply Chain, hours worked per week, BMI, per cent body fat, waist circumference, self-reported symptoms of anxiety and depression, musculoskeletal complaints, physical activity levels, and sleep duration and efficiency.

Working hours and activity levels before and during the pandemic

Participants reported no changes to their working, driving, in-cab waiting or rest hours during the pandemic. Similarly, participants reported no changes in the time spent sitting, standing and walking/moving around on a workday during the pandemic (Table 28).

Anxiety, depression and fatigue before and during the pandemic

In comparison with baseline, there was a tendency for participants who completed the additional COVID-19 questionnaire to report lower levels of symptoms of anxiety during the pandemic (Table 29). There were no differences observed in symptoms of depression, levels of acute and chronic fatigue or intershift recovery during the pandemic, compared with baseline, for this subsample (see Table 29).

The impact of the pandemic on drivers’ lifestyle health-related behaviours

Twenty-three (25%) participants reported engaging in a new form of physical activity since the COVID-19 outbreak. New activities reported by participants included cycling (n = 10), walking (n = 5), gardening (n = 3), running (n = 2), weights at home (n = 2), boxing (punchbag training) (n = 1), exercises at home (n = 1), home workouts with a personal trainer (n = 1) and DIY (‘do it yourself’) (n = 1) (note that some participants reported more than one new activity). Seven of 47 (15%) participants who had not been furloughed reported engaging in a new activity, compared 16 of 44 (36%) participants who had been furloughed during the pandemic.

Twenty per cent of participants reported that their consumption of snacks (e.g. cakes, biscuits, crisps, chocolate and sweets) had decreased during the pandemic, whereas 19% reported an increase. Twenty-three per cent of participants reported that their consumption of fresh fruit and vegetables had increased during the pandemic, whereas 7% reported a decrease. Thirty-four per cent of participants reported not being able to access healthy food while at work during the pandemic, whereas 45% reported accessing healthy food, all stating that they brought their own from home.

Five per cent of participants reported a change to their smoking status during the pandemic. One participant reported starting smoking, two participants reported smoking less and two participants reported stopping smoking. Five per cent of participants reported a decrease in their alcohol intake during the pandemic, whereas 26% reported an increase.

Twenty-seven per cent of participants reported that their sleep duration had increased during the pandemic, whereas 13% reported a decrease. Of the participants reporting an increase in sleep duration, 23 of 25 participants had been furloughed. Participants currently furloughed at the time of completing the questionnaire (n = 20) reported a median sleep duration over the past 14 days of 8 (IQR 6–10) hours per day. Participants currently working (n = 70) reported a median sleep duration over the past 14 days of 7 (IQR 6–8) hours per day.

The impact of involvement in the SHIFT study on health behaviours during the pandemic, and lifestyle changes experienced

Participants were asked within the online questionnaire whether or not they felt that participating in the SHIFT study had given them the right knowledge to maintain a healthy lifestyle during the COVID-19 restrictions. A total of 63% of both intervention and control participants answered ‘yes’ to this question. Overall, 63% of both intervention and control participants answered ‘yes’. A range of qualitative quotes were provided by respondents on how the study had helped them maintain a healthy lifestyle during the pandemic. The responses received were similar between intervention and control participants, and largely centred around an increased understanding of the importance of activity and a better diet. The quotes are shown in Appendix 2, Box 1.

When asked within the questionnaire whether or not they had experienced any changes in lifestyle and/or work that had either a positive or negative impact on health, 40 (44%) participants answered ‘yes’. Of them, 20 (22%) participants reported that these changes had a positive impact and 20 (22%) reported that these changes had a negative impact. A range of qualitative quotes were provided from respondents regarding changes in lifestyle experienced (see Appendix 2, Box 2). Participants reporting positive changes tended to refer to having time off work as a result of being furloughed and, therefore, having more time to be physically active. Participants reporting negative changes tended to refer to a lack of access to facilities (e.g. gyms, swimming pools) to enable them to be active, increased snacking behaviours due to being at home and reductions in the overall quality of their diets due to limited food choices.

Resource use and costs

Health-related resource use was collected from participants using a service use questionnaire at baseline and at 6 months and 16–18 months. Responses at each time point related to participants’ resource use in the 6 months prior. Health-care costs were calculated for each trial participant by resource use category and at each follow-up period by applying unit costs to any resources used. It was assumed that the 16- to 18-month follow-up occurred at 18 months, and resource use between 6 and 12 months was equal to resource use between 12 and 18 months. Health-related resources and costs were categorised into primary care, secondary care, mental health care and occupational services. Unit costs were measured in GBP (2019–20) and were sourced from published UK sources (see Appendix 3, Table 43). 91,113 Unit costs were inflated to 2019–20 prices using inflation indices where necessary. 91 Costs of absenteeism were calculated based on firm-reported full-day driver-replacement costs.

The resource use and associated intervention costs for the SHIFT intervention comprised (1) exercise-related devices, equipment and materials [e.g. wearable device (i.e. a Fitbit Charge 2), THERABAND^® bands (Akron, OH, USA), intervention booklet] and (2) a 6-hour education session that required driver time, course materials and staff training. It was assumed that the education session would require a full worker day for each attendee and that the session was delivered by existing facilitators and, therefore, incurred no additional costs. The average SHIFT intervention cost per driver was treated as an up-front cost (i.e. with no follow-up costs) and calculated on an ITT basis. Usual practice was assumed to incur no intervention-related costs. Intervention unit costs were based directly on those incurred during the trial.

Outcomes

The primary outcome used in the cost-effectiveness analysis was QALYs, which is a generic measure of health that combines both length of life and HRQoL (1 QALY is equal to 1 year in perfect health). 108 Participants’ HRQoL was assessed using the EQ-5D-5L questionnaire, collected at baseline and at 6 months and 16–18 months. The EQ-5D-5L is a descriptive HRQoL instrument that comprises five levels of severity across the following five health dimensions: (1) mobility, (2) self-care, (3) usual activity, (4) pain/discomfort and (5) anxiety/depression. 81 In accordance with National Institute for Health and Care Excellence (NICE) recommendations, HRQoL weights were calculated from a published mapping of EQ-5D-5L responses onto HRQoL values calculated for the EuroQol-5 Dimensions, three-level version (EQ-5D-3L) instrument from a survey of the UK general public. 106,114 Trial QALYs were estimated through an area under the curve approach, with linear interpolation between observations. 108 It was assumed at the 16- to 18-month follow-up occurred at 18 months to allow for a common time horizon for estimation of QALYs. Longer-term cost-effectiveness analyses calculated QALYs using decision-analytic models (see Decision-analytic model and longer-term cost-effectiveness), which combined estimated within-trial QALY differences (i.e. within-trial analysis) with QALYs estimated over a longer time horizon, based on any observed impacts on levels of physical activity. It was assumed that the HRQoL of drivers beyond the trial were equal between arms (i.e. any within-trial differences did not persist beyond the trial period) and equivalent to age-specific values observed in the trial. The removal of estimated within-trial differences in outcomes and the use of HRQoL weights from the EQ-5D-5L value set were explored in scenario analyses.

Secondary outcomes included productivity, employee well-being and absenteeism. Productivity considered employee-assessed job performance and work ability on Likert scales that ranged from 0 to 7 and 0 to 10, respectively (with higher scores denoting more favourable outcomes). Participants’ work-related well-being considered employee-assessed job satisfaction on a Likert scale that ranged from 0 to 7 (with higher scores denoting greater satisfaction) and presenteeism according to the number of days drivers have worked despite feeling unwell.

Analysis

A health economics analysis plan (see Report Supplementary Material 4) was created before the health economics analysts had access to the data. From a public sector perspective, the cost-effectiveness of the SHIFT intervention (compared with usual practice) was assessed according to (1) the estimated differences in the QALYs gained by drivers and (2) the incremental costs to public services incurring over a 16- to 18-month time horizon (i.e. within trial) and over the longer term. Secondary analyses considered cost-effectiveness from a private sector perspective by assessing the differences in changes to measures of productivity and employee well-being at 6 months and 16–18 months from baseline, as well as the incremental costs. From a public perspective, intervention costs and all costs relating to health resource usage were considered. Only intervention and absenteeism costs were considered from the private sector perspective. Both public and private sector perspectives included SHIFT intervention costs, as these could feasibly be financed by either source.

For the public sector analysis, estimated costs and outcomes in each arm, and their differentials, are presented alongside ICERs and INHBs. ICERs represented the cost per additional unit of outcome for the SHIFT intervention compared with usual practice. The INHBs for the SHIFT intervention measured the intervention’s health gain less the health that would have otherwise been generated elsewhere had the additional resources (compared with usual practice) been allocated for alternative purposes (i.e. the opportunity cost estimated using a given cost-effectiveness threshold). Three cost-effectiveness thresholds (i.e. measures of health opportunity cost) were considered in the analysis: (1) £15,000 per QALY [i.e. the Department of Health and Social Care’s usual threshold (based on recent empirical estimates)],112,115 (2) £20,000 per QALY and (3) £30,000 per QALY. The thresholds (i.e. £15,000, £20,000 and £30,000) are used by NICE to assess cost-effectiveness of health-care interventions. 106 At a given threshold, the SHIFT intervention is considered cost-effective, compared with usual practice, when its ICER is below the chosen cost-effectiveness threshold and it has positive INHB. For the private sector perspective, ICERs are presented, showing the cost per unit change in the measure of productivity or well-being.

Treatment effect

Baseline physical activity (i.e. usual practice) was modelled at 15.36 minutes per day in MVPA and 670.31 minutes of sedentary time per day. In the first year, the SHIFT intervention was associated with an additional 5.84 minutes per day in MVPA and a reduction in sedentary time of 24.37 minutes per day (see Table 9). In the second year, the SHIFT intervention was associated with an additional 1.6 minutes per day in MVPA and a reduction in sedentary time of 12.16 minutes per day. In the subsequent years, the treatment effect was extrapolated from second year differentials (i.e. a 1.6-minute/day increase in MVPA and a 12.16-minute/day reduction in sedentary time).

Resource use and costs

Outcomes

Quality-adjusted life-years and HRQoL scores were similar between the arms, albeit with modest differences in baseline values and changes at 6 months’ follow-up (Table 32). In the SHIFT arm, EQ-5D-5L and mapped EQ-5D-3L scores fell over the trial period. In the usual-practice arm, scores rose between baseline and 6 months, before declining to levels below baseline (see Appendix 3, Table 54). Adjusted QALY estimates (i.e. QALY estimates estimated using imputed data while controlling for participant covariates) found –0.028 and –0.015 QALY decrements associated with the SHIFT intervention compared with usual practice over the trial horizon when using mapped EQ-5D-3L and EQ-5D-5L preference weights, respectively (see Appendix 3, Tables 61 and 62). Secondary outcomes are reported in Table 33. There were small and largely inconsistent changes within and between arms for each outcome considered.

Cost-effectiveness analysis

The within-trial and longer-term base-case mean cost, QALY and cost-effectiveness estimates for each arm and modelling approach are reported in Table 34.

The within-trial analysis found the SHIFT intervention to be more costly and less effective than usual practice, resulting in it being dominated. The probability of the SHIFT intervention being cost-effective in the within-trial period was low, with a probability of between 0.009 and 0.011 for the range of cost-effectiveness thresholds considered.

For the MVPA-based model, when using Ekelund et al. 120 all-cause mortality estimates with changes to MVPA minutes per day, the SHIFT intervention was found to be more costly and less effective than usual practice and, thereby, dominated. Incremental costs (£555) were the same as the within-trial analysis (the model did not extrapolate costs); however, QALY decrements were reduced to –0.022 per driver because of the increased physical activity in the SHIFT group reducing mortality. Similar results were found when using the sedentary-based model, with costs aligned to within-trial results and QALY decrements of –0.021.

The SHIFT intervention was also found to be more costly and less effective when using the MOVES tool. The inclusion of lifetime costs in the MOVES model increased overall cost estimates, but resulted in a small reduction in the incremental costs for the SHIFT group (£507). QALY decrements were reduced to –0.016 in the SHIFT group, relative to usual practice.

In all base-case analyses, the SHIFT intervention was found to be dominated by usual practice. The 95% credible intervals around incremental QALYs overlapped zero in the MOVES model, suggesting a significant level of uncertainty in the QALY differentials of the SHIFT intervention compared with usual practice. Appendix 3, Tables 57–63, present each regression analysis used to inform the cost-effectiveness analysis. Appendix 3, Figure 8, shows the cost-effectiveness acceptability curves for each modelling approach.

Scenario analyses

Table 35 presents the incremental cost, incremental QALY and associated ICER estimates for the scenario analyses considered for the long-term analyses, based on the three decision-analytic models. Base-case estimates are also provided for reference. A more detailed breakdown of each scenario can be found in Appendix 3, Tables 64–69.

In each modelling approach, removing estimated trial-specific differentials in non-intervention costs and QALYs between arms resulted in smaller incremental costs for the SHIFT intervention and positive incremental QALYs, compared with usual practice. For the MVPA- and sedentary-based models, the resulting ICERs were £65,072 and £51,174 per QALY, respectively, which are above the cost-effectiveness thresholds considered. For the MOVES model, SHIFT was associated with an ICER of £29,287 per QALY, thereby falling below the highest cost-effectiveness threshold considered. Under this scenario, for the MOVES model, the SHIFT intervention had a 12.4%, 26.1% and 46.6% probability of being cost-effective at the cost-effectiveness thresholds £15,000, £20,000 and £30,000 per QALY, respectively (see Appendix 3, Table 64).

Scenarios concerning the application of EQ-5D-5L preference weights, a 1.5% discount rate, generalised linear models to estimate costs, participant costs omitting inpatient-related resource utilisation and a complete-case analysis framework did not have marked impacts on results and, therefore, did not change base-case findings (i.e. the SHIFT intervention dominated by usual services).

Sensitivity and threshold analyses

Study findings were largely insensitive to changes in the decay rate of the treatment effect on physical activity. At base-case settings, the INHB of the SHIFT intervention remained negative for each threshold considered across the range of decay rates examined (i.e. 10–100%) in all models considered (Figure 3). When removing estimated trial-specific differentials in non-intervention costs and QALYs between arms, the INHB for the SHIFT intervention was positive at a decay rate of approximately 20% at a £30,000 per QALY threshold for the sedentary-based model. Likewise, for MVPA-based model, the SHIFT intervention was positive at a decay rate of approximately 15% at a £30,000 per QALY threshold (see Appendix 3, Figure 6). For the MOVES model, when within-trial differences were removed, the SHIFT intervention was cost-effective at all cost-effectiveness threshold values considered at a decay rate at, or below, approximately 20%.

FIGURE 3.

Incremental net health benefit of the SHIFT intervention relative to usual practice for alternative rates of treatment decay: MOVES model.

The base-case cost-effectiveness results were also largely insensitive to extensions in the duration of treatment effect on physical activity observed at 6 months. A comparison of SHIFT intervention ICERs (relative to usual practice) for alternative additional intervention costs and extension periods in treatment benefit are displayed in Table 36. With all other things remaining equal, the SHIFT intervention remained dominated for extensions up to and including 7 years for the MVPA- and sedentary-based models. For the MOVES model, ICERs only fell below £30,000 at a 6-year extension (see Table 36). Cost-effectiveness results were more sensitive to extensions in treatment effect when removing estimated trial-specific differentials in non-intervention costs and QALYs between arms. For the MVPA-based model, ICERs fell below £30,000 per QALY at 3 years, below £20,000 per QALY at 5 years and below £15,000 per QALY at 7 years. For the sedentary-based model, ICERs fell below £30,000 per QALY at 3 years, below £20,000 per QALY at 5 years and below £15,000 per QALY at 6 years. For the MOVES model, ICERs fell below £20,000 per QALY at a 1-year extension and below £15,000 per QALY for a 2-year extension (see Appendix 3, Table 72).

TABLE 36 - Two-way sensitivity analysis: continuation in treatment benefit and additional cost ICER matrix (Δcost)

Notes

Light purple shading indicates that the ICER is < £15,000 per QALY.

Blue shading indicates that the ICER is between £15,000 and £20,000 per QALY.

Dark purple shading indicates that the ICER is between £20,000 and £30,000 per QALY.

ΔCost (£)	Continuation of SHIFT treatment benefit and additional cost profiles: ICER (£)
	(1): Ekelund et al.120 MVPA						(2): Ekelund et al.120 sedentary						(3): MOVES model
	1 year	2 years	3 years	4 years	5 years	6 years	1 year	2 years	3 years	4 years	5 years	6 years	1 year	2 years	3 years	4 years	5 years	6 years
–370	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	69,399	Dominated	Dominated	28,422	8480	4742	2309
–200	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	133,256	Dominated	Dominated	85,269	29,189	19,027	12,303
–100	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	170,819	Dominated	Dominated	118,709	41,370	27,430	18,181
Base case	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	208,382	Dominated	Dominated	152,148	53,551	35,833	24,060
100	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	245,945	Dominated	Dominated	185,588	65,733	44,236	29,938
200	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	283,508	Dominated	Dominated	219,028	77,914	52,640	35,817
370	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	347,204	Dominated	Dominated	275,732	98,570	66,889	45,785
500	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	396,197	Dominated	Dominated	319,347	114,459	77,849	53,452
1000	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	Dominated	584,012	Dominated	Dominated	486,545	175,366	119,865	82,845

Secondary cost-effectiveness analysis from a private sector perspective

Absenteeism-related cost differentials between the SHIFT intervention and usual practice at baseline equated to a cost saving of £57.98 between the two arms (SHIFT intervention, £628.50; usual practice, £686.46). At the 6-month follow-up, larger reductions in absenteeism in the SHIFT arm (see Table 33) resulted in a cost saving equivalent to £169.43, compared with baseline differences. At the 16- to 18-month follow-up, an increase in absenteeism amounted to an additional £552.05, compared with baseline differences. Over the course of the trial, changes in absenteeism costs were £324.64 higher in the SHIFT group than in the usual-practice group, relative to baseline values. Incremental private costs per driver equated to £200.16 at 6 months and amounted to £694.23 at final follow-up (see Appendix 3, Table 72).

Employee-assessed job performance and presenteeism at 6 months were less favourable in the SHIFT arm, relative to the differences in changes from baseline. At 6 months, differences in employee-assessed work ability and satisfaction rose relative to baseline differences by approximately 0.148 and 0.036, respectively, in favour of the SHIFT arm, equating to a £1353 and £5560 cost per unit increase on each respective Likert scale. At the final follow-up, relative to baseline differences, differences in employee-assessed job performance, work ability, presenteeism and satisfaction changed in favour of the SHIFT intervention by approximately 0.081, 0.154, 1.19 and 0.151 less days worked while sick, respectively, equating to a £6816, £3585, £465 and £3656 cost per unit increase on each respective Likert scale. Average difference in results over the trial equated to the SHIFT intervention being dominated in presenteeism (compared with usual service), while having ICERs of £17,142, £4598 and £7425 with respect to one unit increases in employee-assessed job performance, work ability and job satisfaction, respectively (see Appendix 3, Table 72). Given the modest changes over time and largely inconsistent differences between arms, caution must be taken when interpreting the differences in costs, outcomes and associated ICERs.

Given the additional productivity costs (i.e. lost driver days), QALY decrements and higher public costs associated with the SHIFT intervention, relative to usual practice, a broader perspective that considers public costs and productivity would fail to alter base-case study findings.

Quantitative data collection

To inform attrition rates, records were gathered from all participants, including participation uptake per site (i.e. the number of participants who completed the baseline health assessment per site), dropout rate per site (i.e. the number of participants who failed to complete each follow-up), engagement with text messaging service (i.e. the number of responses) and compliance with activity monitor wear.

Qualitative data collection

Managers in intervention sites were asked to schedule three to six participants for an on-site focus group, during the drivers’ working hours. A semistructured focus group schedule was developed and piloted with participants from six pilot sites. Questions related to each SHIFT intervention component, barriers to achieving health goals and recommendations for improvements to the SHIFT intervention. A semistructured interview schedule was developed and piloted for the managers in the pilot sites. Managers were also asked to schedule time for an interview on the same day as participants. These interview schedules were then analysed and revised, as needed, before commencing data collection for the main trial sites (n = 19).

Data analysis

All focus groups and interviews were carried out by one or two researchers (AG and YLC). The focus groups were recorded on a digital audio device and transcribed verbatim (by AG) into Microsoft Word^® (Microsoft Corporation, Redmond, WA, USA), before inputting into NVivo 12 software (QSR International, Warrington, UK) for analysis. Transcripts were then re-read, coded and developed into themes using the deductive method of thematic analysis, where themes were already preconceived based on the interview schedule and existing knowledge. This was followed by an inductive method where themes were identified entirely based on the data (by AG). 125 Each stage of the analysis was critically analysed from an informed external perspective. (Dr Anna Chalkley, an expert in qualitative research methods and programme evaluation, was independent to the research team and acted as a ‘critical friend’ throughout the analysis and reporting stage of this process evaluation.) Quantitative data, including dose, attrition and questionnaire data, such as multiple-choice and Likert scale answers, were analysed descriptively in IBM SPSS v25 (IBM Corporation, Armonk, NY, USA).

Questionnaires

A total of 71.2% (n = 272) and 57.8% (n = 167) of the participants invited to complete a feedback questionnaire following the baseline and 6-month follow-up health assessments, respectively, responded.

Focus groups and interviews

During the internal pilot, on completion of the final follow-up measures, four focus groups and three interviews with 13 participants (i.e. an 83.3% site response rate) and eight individual interviews with managers (i.e. an 83.3% site response rate) were carried out to capture their views on and experiences of implementation. In the main trial phase, 15 telephone interviews were conducted with participants from the remaining 19 sites (i.e. an 82.4% site response rate), and 13 telephone interviews were conducted with managers (i.e. a 76.5% site response rate) after the 16- to 18-month follow-up assessments.

Through synthesising the data, we found there to be no substantial differences between pilot testing and main trial sites in both the questionnaire responses and the content of the interviews and focus groups. Therefore, the combined findings from both the pilot and main trial sites are presented below.

Dose

It was intended that participants in the SHIFT arm would receive a total of 12 hours of face-to-face contact with the research team over 12 months, broken down into three 2-hour health assessments and one 6-hour education session. Control participants would receive 6 hours of face-to-face contact, consisting of three 2-hour health assessments. Owing to the COVID-19 pandemic, a national lockdown prevented all main trial participants from receiving a face-to-face final follow-up health assessment, hence a reduction of 2 hours of face-to-face contact with participants from the main trial.

Fidelity

Adaptations

All baseline health assessments were conducted at the beginning of the drivers’ shift. Carrying out baseline health assessments for all drivers was logistically challenging for both the researchers and transport managers because of the 24 hours a day, 7 days a week (24/7) nature of the job (i.e. drivers started their shifts at different times of the day and night). As a result, post-intervention measures were carried out at a time of day that was more feasible and occasionally this was at the end of the drivers’ shifts. Drivers were still asked and reminded to fast for 4 hours prior to the assessment; however, this change in protocol may have affected body weight and blood pressure measurements.

The initial target was to include six participants per education session; however, it was logistically challenging for the transport managers to take six drivers off the road at any one time, particularly for the smaller sites, and, as a result, most education sessions consisted of fewer drivers per workshop (mean: 3.4 drivers; range: 1–7 drivers).

Owing to the COVID-19 pandemic, a national lockdown prevented all main trial participants from receiving a face-to-face final follow-up health assessment. Instead, in the 16- to 18-month follow-up health assessment, participants were asked to fill in the questionnaire, self-report their weight and wear the activPAL device (to monitor physical activity and sitting) while filling in the wear log continuously for 8 days. The pandemic also caused a 4- to 6-month delay in the final follow-up assessments for sites in the main trial phase, resulting in these taking place 16–18 months after randomisation.

Attrition

A total of 382 participants (SHIFT intervention, n = 183; control group, n = 199) received the baseline health assessment. Two hundred and sixty-two participants (68.6%) took part in the 6-month follow-up health assessments. The retention rate at 6 months was higher in the control group (73.9%) than in the intervention group (62.8%), and this may be because of intervention participants being ineligible to continue with the study if they failed to take part in the 6-hour structured education session. Of the 183 intervention participants, 145 (79.2%) took part in the education session. At 6 months, the retention rate for intervention participants who were eligible to continue after the education session (n = 145) was 79.3%. Participants who did not attend the session were removed from the trial. The dropout rate in the intervention group may have been a result of their expectations for greater health improvements among them:

I think that for them that dropped out they just . . . I think they’d realised that they’d not really made that much of a change, or they made them, but they were short-lived if you know what I mean?

Intervention manager, M11, FG1 – feedback from drivers (manager perspective) – reasons for dropping out

Table 40 shows reasons for discontinuation of the study.

Retention rate between the 6- and 16- to 18-month follow-ups was higher (78.2%). In this instance, retention was better in the intervention group (87.8%) than in the control group (70.7%). It may be that overall retention rate was higher after 6 months because participants who were disinterested in the programme had already dropped out. Participants in the intervention group were possibly more invested in the intervention, and yet little time spent engaged in the project was required of them during this time. From start to finish, 46.3% (n = 177) of participants dropped out of the RCT. Mangers were asked in interviews if they knew the reasons for participants dropping out. One of the primary reasons mentioned was that truck drivers are notoriously transient workers, with a high staff turnover rate. It was evident that a lot of drivers had left their company before the cessation of the programme:

Like I say, the drivers that had to drop out, they left, obviously that’s one of those things, they left and went on to different contracts.

Control manager, M12, interview – feedback from drivers (manager perspective) – reasons for dropping out

Other reasons for participants dropping out included long-term illness, and a few participants were isolating or on furlough:

Yeah, a couple on long term sick, and I think a couple that were shielding as well.

Intervention manager, M11, interview – feedback from drivers (manager perspective) – reasons for dropping out

Another commonly mentioned reason for participants dropping out was that some saw the timings of the health assessments as too inconvenient:

They just want to come in and go home. Yeah, they don’t want to do anything else that adds on to their day. Which I totally get.

Control manager, P05, interview – feedback from drivers (manager perspective) – reasons for dropping out

Barriers to behaviour change

In the 6-month questionnaires, all participants were open-endedly asked to identify their main barriers to a healthy lifestyle. Among the 145 responses, 46% suggested that the biggest barriers were work related, predominantly the length, irregularity and start times of their shifts. The second biggest barrier was family (e.g. child care) commitments (12%) and this was followed by self-motivation (10%), with drivers referring to themselves as being ‘lazy’ or ‘need[ing] more discipline’. Table 41 shows a summary of the mentioned health barriers.

Using an inductive approach in the thematic analysis, we discovered that participants discussed barriers to living a healthy lifestyle both at work and at home. We deemed it important to mention these barriers, as this may influence guidance for future health interventions in this demographic. Participants mentioned that there was little time in the day to fulfil a healthy lifestyle:

The trouble is, is when do you get the time, isn’t it? . . . You know, you finished at 10 o’clock in the morning, from 8 o’clock at night. All you want to do is go back and have a shower go to bed.

Intervention participant, P03, FG2 – barriers, time

You’ve got all the things to do, you know I don’t have any chance to do any exercise, I don’t.

Intervention participant, P03, FG2 – barriers, time

Linked to time pressure are the long, irregular shift patterns and early start times that almost all drivers from both control and intervention sites mentioned at least once in their interview:

. . . well, yeah, you know and the start times as well, I start between midnight and 4 a.m. in the morning and that’s no, no good for your body rhythm if you like. It’s not good for your body. By 2 o’clock in the afternoon you want to go to sleep.

Control participant, M09 – barriers, shift pattern

But I’ve done this for 35, 36 years now, umm, and it’s just, it’s just part of being a lorry driver, it is how it is, it’s the territory that you’re in. You know, the uncertainty of where you’re going to work, whether you’re going to go home that night, or it might be 5 days later.

Intervention participant, M14 – barriers, shift pattern

I think there should be, once you have worked a 12-hour shift, you should have a minimum of 12 hours off. Whereas you can actually work 15 hours and have 9 off. That’s the legal requirement, 9, 9 minimum . . . because it’s 15 and 9, that’s your 24 hours in a day isn’t it. Alright, so you could start at 6 in the morning, work till 9 at night and realistically you could start at 6 o’clock in the morning the next day.

Intervention participant, M14 – barriers, shift pattern

And, I mean, to say that the trouble with the HGV world is uh, it’s poor wages long hours, you know. You can make a good life out of it, but you’ve got to put a lot of hours in, and when you put the hours in, it is detrimental to your health at the end of the day. You know, unfortunately that’s the aim of the game, you know.

Control participant, M25 – barriers, shift pattern

Therefore, there’s often unrealistic, unrealistic expectations required of drivers, and long hours. And short breaks. And that doesn’t help at all with this side of things, which is looking at trying to keep healthy. Try to have a diet, a decent diet. When you’ve got guys going out for 13, 14 hours a day, when you get home . . . when I’m on 12 hours a day, I get home, I have a snack and go to bed practically. A snack, a shower and bed and up in the morning. And it’s a snack, and into work, when I’m on my 12-hour shift. So the guys that are getting a couple of 15 hour shifts on the bounce, they get home, go straight to bed. And they’re actually back the next day driving, munching as they go down the road because that’s their time for having something to eat. So, the fact that the pressures that are put on us in the workplace, and often or not the uhhh . . . the mismanagement of that is, is a factor towards the healthiness and well-being of the driver himself.

Intervention participant, P03 – barriers, shift pattern

A lack of managerial support and excessive expectations also affected drivers:

It’s operational requirements, like. It’s, it’s operational requirements, because you can’t always eat when you’re supposed to eat, or sleep when you’re supposed to sleep or anything like that. It all revolves around your working day. And that’s how it is.

Intervention participant, P03, FG2 – barriers, shift pattern

It’s not doing one 12-hour shift or one 15-hour shift, it’s one after another, after another, after another. You know, they think you’re machines like these trucks, you can drive them for 24 hours a day, 7 days a week, they’re just trucks, put some more diesel in, put some more oil in, they’ll go forever, we can’t do that. They think we can.

Intervention participant, M14 – barriers, shift pattern

I’ll be honest, well, me and my partner have now gone part time, because it had just got ridiculous. And they just expect so much of you. And like, especially . . . for some reason, I don’t know why. In our place, the night drivers are just the dirt on their feet . . . They’ve just, no respect for them . . . but you try working nights. They don’t seem to understand that, you know, your natural body clock, even no matter how many years you work nights, your natural body clock is, you go sleep at night.

Intervention participant, M18 – barriers, shift pattern

Another key barrier included unavailability of healthy foods while at work:

Well, umm, basically being lorry driver is . . . the diet is shocking, you can’t really get healthy food it’s all fast food. Basically working the shift that I do I finish my day at the end of the night, around 11 o’clock at night. And if I’m parked up somewhere, the only thing that’s available is McDonald’s [McDonald’s Corporation, San Bernardino, CA, USA], or Burger King [Jacksonville, FL, USA] or sandwiches from WHSmith [WHSmith, Swindon, UK].

Intervention participant, M14 – barriers, diet at work

. . . you’re feeling knackered, you’re feeling drowsy, you’re feeling, you know, you need to be doing something, you eat. And unfortunately that’s the nature of the beast. And it’s not necessarily good for you, but you can also have the mindset, well, it’s better that I eat than crash. Because I could kill somebody, or myself, you know. And it’s one of those, umm, it’s not ideal, and I do try and eat more healthily when I’m snacking, but sometimes you are thinking an apple ain’t doing it, a banana ain’t doing it, grape ain’t doing it, dried fruit’s not doing it. **** it, I’m having a choccy bar.

Control participant, M22 – barriers, diet at work

But yeah, the biggest problem is, when you go to a service station, what’s the first thing you see there? It’s fast food. They are trying to change it but it is fast food, burger bars, so . . . They are popping up aren’t they. You don’t see enough healthy food options, certainly in the UK, and whether that’s because of the climate we have I don’t know.

Intervention manager, P01 – barriers, diet at work

Less mentioned, and more debated, was physical activity, with some drivers suggesting that they do a lot of physical activity at work and others saying that they do not:

Well, the thing is my exercise regime did go out the window a little bit. Because I was starting an hour earlier . . . I was starting at . . . getting up at 2 and starting at 3 in the morning. And of course you know, that knocks you right up, you know, you come in and all you want to do is have a sit down an hour, half an hour, you have your food and then you’re back in the bed, you know.

Control participant, M25 – barriers, physical activity

It’s a shame, we haven’t sort of got a place, a room or something with gym equipment in it. Because I mean, you know, for example, like drivers, they sometimes sat down there for like 2 or 3 hours waiting for a job. I mean, if you have like gym facilities on site, rather than just sitting there, you could just come in and do a bit of something, you know what I mean?

Intervention participant, P03 – barriers, physical activity

Yeah, so, and the pumping weights, well I do enough of that at work, loading and unloading trailers.

Intervention participant, M11 – exercise

My job is quite physical anyway. Umm and you know you unload like 300, 400 tyres, umm, so yeah you know that’s quite a good workout.

Intervention participant, M19 – exercise

Self-motivation was also a key barrier that was mentioned, and this appeared to be at both home and work:

Just being lazy . . . That’s it. Me just telling myself ‘oi get your big fat ass out of there and go for a walk!’ or do this, or do that!

Intervention participant, M24 – barriers, self-motivation

I just think it’s that lazy mindset really, like I said I’ve been preparing, like when I get home now like, my fruit or anything that I’m going to bring with me today, umm, it was a lot easier before to maybe just get to shop to buy sandwich. It’s pretty packed. It’s just lazy that mindset. Isn’t it?

Intervention participant, P03, FG1 – barriers, self-motivation

Suggested improvements to the SHIFT intervention

By a significant margin, the most frequently suggested improvement from both drivers and managers was that the SHIFT team could have had more regular contact and engagement with the participants:

I suppose, how could I say it. I think if I was to say anything, and this is not being negative or nothing, but you just asked to think of something so . . . Umm, maybe just the odd phone call just to cheer you up and keep you going. You know what I mean . . . Barring that, I couldn’t imagine anything else. You know what I mean. But yeah, a bit of a phone call every now and again, how are you keeping, how are you doing? We’re keeping an eye on you like. If you know like, someone is watching you, it makes you, it pushes you along a bit more. You know what I mean.

Intervention participant, M15 – improvements to SHIFT

I haven’t really text backwards and forwards, but when you sent me the, the little challenges like I thought good, yeah, I enjoyed that. And that’s one thing I would say. There could have been a lot more of that. Yeah, a lot more challenges, like, you know, you know, let’s say, look, everybody’s gotta hit 15 . . . But if there was more goals in there . . . And sort of push us along because I would rise to the occasion with the goals. I like that sort of thing. If I’ve got a target, that’s my goal that, you know me. But when there’s no targets, you tend to take a backward step.

Intervention participant, P06 – improvements to SHIFT

I think for them they were given all the information and then there was quite a long gap, so that was the only negative thing I would say, that gap was too long, they were saying what’s happening and they were starting to lose interest.

Intervention manager, P01 – improvements to SHIFT

Lots of drivers and managers would have liked to have seen more feedback about the results:

Yeah, just better . . . more publication of results for, for me, personally, I would have got something out for that. And I would have probably been able to get more of the guys talking about it if they have that data as well.

Control participant, M22 – improvements to SHIFT

I’d have liked to have been able to have access to the results of the monitors that we were wearing to see over the period of time that we were wearing them how things fluctuated as well.

Control participant, M22 – improvements to SHIFT

From the perspective of the managers, more clarity at the beginning would have been helpful in organising and managing expectations:

Umm, I think the only thing for me would be, at the very, very beginning, having a more understanding of what was involved . . . [exhales] it was a little bit, um, it was a little bit, ‘it won’t affect your business, it won’t affect your everyday’ bla bla bla, bla bla bla. But it does affect it when it’s 2 or 3 hours and it, and you’ve got to umm, plan that around the drivers’ times, rest days, and all that kind of stuff. Umm you know that alone, the admin side from myself or one of my managers, to prep it, umm, probably, it’s a few hours in a day, each time, to try and prep that. And it’s not about not doing that part of the role, I don’t mind doing it, but it’s about understanding that’s what is involved. Because at first it was sort of addressed, that ‘yeah, there’s no problems, there’s no impact to yourselves’, etc., etc., etc., but that is 2 or 3 hours of the drivers days that then have to rearrange his job for the day so that he can complete his job within legal time.

Intervention manager, M17 – improvements to SHIFT

Future sustainability

All managers were asked if they could see the SHIFT intervention becoming sustainable in the future, and if they would support it. All managers said that they would support the SHIFT intervention:

Absolutely. 100%. I can say that without even speaking to my, my higher tiers . . . But I know they’ll be on board, they will, yes . . . Again, it’s all about the well-being of drivers isn’t it?

Control manager, M12 – support SHIFT in the future

I think to have it as part of a CPC [Certificate of Professional Competence] I think certainly, would be, be a much easier route and you’d obtain a lot more numbers to be involved.

Intervention manager, M17 – support SHIFT in the future

However, some managers caveated this support with comments about issues with practicalities:

Umm, I suppose within DHL to get everybody done for the assessments and stuff would take some major planning. You know what I mean, just on my contract alone you’ve got 650 drivers. Umm, within DHL you’ve got 7500 drivers. Umm, so you imagine trying to have consultations and assessments with their health and things on site with 7500 people . . . Umm, and that’s just like 7500 driving colleagues, staff at the moment. Within DHL . . . It’s quite big . . . If you want to start planning your days around, you can plan 7500 in if you want to haha . . . Keep you busy until 2029 I think! Haha.

Intervention manager, M18 – support SHIFT in the future

Yes, it could. Ummm, you would have to work around sort of different areas or different drivers, because not all drivers would want to participate. Not everybody would want to be part of it. Like I said, in the planning of the drivers and getting the drivers back, it’s sometimes not that easy to be able to manage small contracts and get drivers back on site, the time that you need them.

Control manager, M22 – support SHIFT in the future

Intervention participants were asked about their thoughts on turning the education session into a Certificate of Professional Competence (CPC) module, and all drivers said that it would be a good idea:

Now that would be interesting that would. That would be interesting . . . yeah, I think that would be a good thing actually. It would be well educational actually for a lot of the drivers.

Intervention participant, M24 – education session, CPC module

However, almost all participants said that the education session would need a medical professional rather than a driver trainer to lead the workshop:

Well I think I’d prefer it to be a health professional, uh, because it just seems more appropriate that it comes from a health professional. You’d think that you’d take it more seriously.

Intervention participant, M14 – education session, CPC module, driver trainer

But no with that thing I think a health professional could present it better than a driver trainer because driver trainers understand driving but they don’t understand . . . other things . . . it’s always better to have an expert talking about something they know, then somebody talking about something they’ve been told to talk about.

Intervention participant, M17 – education session, CPC module, driver trainer

No . . . because they haven’t got the depth of knowledge that you guys have, I don’t think it would work. I really don’t think it would work. All they’d do is just read it. And then ask you five questions at the end of it and say right, you sign that, then sign the form to say that you understood. And people will just be saying, god is it time to go home?

Intervention participant, P03, FG2 – education session, CPC module, driver trainer

One driver mentioned that cost would prevent DHL Supply Chain from doing the CPC module:

They have a programme that the whole of DHL use and that’s all you’re doing, not doing nothing else. Because we may have to register for something else, or we may have to pay out for something else. So we’re not going to do that. So . . . companies would then have to pay for it, they’d have less interest . . . Is it going to help them get a truck down the road?

Intervention participant, P03 FG1 – education session, CPC module

All drivers said that moving the education session to an app-based workshop would be detrimental:

I mean if you’re in the classroom and that, with other people and that, then you get involved more, I think if you’re on your own, doing an app, I think a lot of people probably won’t bother.

Intervention participant, M19 – education session, app based

I’d rather have face to face because you can ask questions and that can’t you really. And if you’re there with a group of people, which you can’t do at the moment, everybody would be coming up with different questions . . . I might be asking one question and the guy next to me asks another question that I never even thought about asking.

Intervention participant, M24 – education session, app based

Being honest. I wouldn’t have paid much attention to it at all. I’d have flicked it on and flicked it back off just to say I’d been on it! I’m being totally honest. I’ve always worked with face to face, and I work well that way. I’m more of your audio kinetic sort of learner and everything. Umm, on a web page or on an app, you’re going to have loads of writing and all that sort of thing. I’m not going to read it and that’s being honest.

Intervention participant, P03, FG1 – education session, app based

Control site participants

There was a lack of awareness from participants and managers in the control group with regard to whether they were in the control or intervention arm, suggesting that the SHIFT team could have provided more communication and explanation to both the managers and participants of their allocated arms and what this means for participants:

So were you aware that there was two different groups in the SHIFT study, one was called control and one was called intervention?

Yeah, yeah, yeah. That’s what we’ve been told before, yeah, before we started . . .

Do you know which one your site was put into?

The . . . I think that second one as far as I can remember . . .

Oh, the intervention?

Yes.

Were you aware that there were two different groups in the SHIFT study, one called control and one called intervention?

Umm I can’t remember, it possibly was mentioned to me but it’s that long ago that I can’t remember to be honest with you.

OK, and did you know that there were two different groups in the SHIFT study, one was called control and one was called intervention?

Yes, yes I did.

OK, and did you know which one your site was allocated to?

No.

Contamination

Randomisation occurred at the site level to ensure that there was a minimal risk of contamination between intervention and control participants. All participants and managers were asked in the interviews if they had spoken about the SHIFT intervention to anybody from any other depot, and all participants and managers confirmed that they had not.

The COVID-19 pandemic

A large, unforeseen confounding variable to the SHIFT intervention was the COVID-19 pandemic, which caused three major lockdowns in the UK from March 2020 to June 2021. Fortunately, almost all (254/262, 96.9%) participants who were still enrolled in the trial at the time had completed the 6-month follow-up health assessment before the national lockdown. However, the final follow-up was greatly affected by the pandemic and, as such, all main trial sites required remote data collection, where participants were asked to fill out the questionnaires, self-report their weight, wear the activPAL and complete the sleep/work diary by themselves. Although this method has clear shortcomings, this was deemed the most appropriate, pragmatic and safe form of data collection at this time to assess the key main outcomes. When the COVID-19 pandemic was discussed in the interviews, it appeared that all sites were affected by the pandemic in different ways. For example, participants who delivered essential items were busier and worked longer hours:

In fact, we went through a stage where we were actually really short of drivers.

Oh wow, was that due to sickness?

No it was due to operational demand because people were panic buying so our stores we deliver to, were ordering bigger and things like that . . . Oh lots of hours and double shifts and things like that, just to keep the business going and to keep stocks up and things like that.

Some drivers mentioned that it was business as usual:

It was the same run on the same start times, so not many changes if you like.

Control participant, M12 - Contamination, COVID-19, business as usual

The same, whatever happens. Our job is the same. It never ever changes. We go out, two runs a day. Fourteen pallets, it might be a bit less. But it all depends how much the shop’s selling. But yeah, that’s it, it never changes. It’s just the same every day.

Intervention participant, M15 - contamination, COVID-19, business as usual

Some drivers mentioned being on furlough for most of the year. However, there was ambiguity between the furlough status of all sites. Several sites furloughed staff on a rotational basis, some sites furloughed all staff, other sites made it optional for drivers to choose furlough and the remainder of sites, which provided essential goods, did not make furlough an option:

I was put onto furlough, I was put on that until August . . . then we were locked down again in November, so we were put on furlough then . . . Since April I have probably done less than . . . 10 days’ work?

Control participant, M21 – contamination, COVID-19, furlough

Some managers also mentioned that the pandemic halted the enthusiasm for the study, with priorities moving elsewhere:

It’s just after that the, the communication and the engagement because of the COVID-19, where that stopped, then obviously, this . . . the SHIFT stopped as well, well in the drivers’ minds. So they just said ‘it’s not continuing anymore obviously’. Nobody’s ringing us, and everybody else had obviously their personal interests in their minds during this pandemic as well. So they just lost focus on it.

Intervention manager, M10 – contamination, COVID-19, impacted study enthusiasm

Behavioural changes

The primary outcome measure was the difference in steps per day at 6 months between the SHIFT and control arms, measured using the activPAL3 accelerometer. Although a significant difference in daily steps was observed at 6 months, in favour of the SHIFT arm, drivers (from both trial arms) commented that the biggest modifications they made were to their diet, with 57 comments/references to dietary changes recorded, whereas only 27 comments/references were made with regard to changes in physical activity. There were no significant differences in fruit and vegetable intake or dietary quality observed between trial arms at either follow-up period:

Well, as I say I cut down my sugars and the, is it . . . saturated fats? Is it? . . . I try and do a little bit more exercise although that’s a little bit more difficult haha!

Mmm, yeah yeah, understandable. Right OK, and you mentioned there that exercise was a bit harder to control than the diet, what makes you think . . . what makes you say that?

Well, I mean it’s easy to change from butter to margarine you just buy a different one on the shop haha, whereas to get your mind to want to go out and do a bit of exercise is a bit harder, ha ha.

Control participant, M09 – behavioural changes, dietary

Well, the barriers are still there, but you’ve just got to, it just puts thoughts in your mind because of the training you’ve had, I consider this to be like a training exercise to me, you know, I’m thinking about things before you just go in. So I used to like, when I got meals at McDonalds, I used to get the large meal. Now, I never do that. Always just your standard meal. I know there might not be many calories difference between a large and a medium meal, but there is some difference. So, that’s, that’s what I do now, and I manage with that. You know, I used I used to think I’ve got to have a large meal to be sort of, satisfied but now I don’t. I just have a standard meal, I have that, and that’s it.

Intervention participant, M14 – behavioural changes, dietary

I don’t eat as much chocolate as I used to do now. I take an apple or something like that for work, whilst I’m travelling down the motorway.

Intervention participant, M17 – behavioural changes, dietary

The small snack thing, before I just, I don’t know. I’d eat . . . in the truck for example, I’d always have a . . . I don’t know, say humbugs or something. And I’m eating a packet of biscuits and you’d just happily munch on them. Whereas now, I deliberately won’t have them in the vehicle, because if they’re not in the vehicle, I can’t . . . I can’t . . . do you know what I mean? I can’t eat them . . . I’d rather take like a bag of like roasted monkey nuts or something like that instead.

Control participant, M20 – behavioural changes, dietary

I’ve started eating more fruit . . . Rather than when I get hungry going for a chocolate bar, I go for some grapes, a banana or an orange or something like that.

Control participant, M23 – behavioural changes, dietary

Well put it this way, I now eat vegetables around four times a week now, never touched them before.

Oh OK, you never ate them before?

Never. I ate them once a week on Sunday if I had a Sunday dinner . . . But like I say, I live on a marina and there’s a pub, when it was open we’d go across for a meal, but I’d get burger and chips or, steak and chips, and I never really touched the vegetables. But since the lockdown as well like, I just sort of looked at them and said ‘I’ll try them’ and I tried them and like I say I’m now on four times a week on vegetables. In summer every night I was having salad. So my way of thinking about food now, I’m trying as they said in the study, how to portion your food as well.

Intervention participant, M24 – behavioural changes, dietary

My diet changed very much straight away . . . from not eating any fruit because I wasn’t interested, because I was a biscuit sweet person, I take fruit with me at work, I got down from nearly 14 stone to 13 [stone], 3 [pounds].

Intervention participant, P01, FG1 – behavioural changes, dietary

At 6 months, significant differences in steps, sitting and standing time were evident on non-workdays between trial arms, with no differences in these variables seen between groups on workdays, and this is reflected by the drivers’ and managers’ opinions about the inherent characteristics of the job:

Were the shifts a barrier to actually live the healthy lifestyle, did you just do it around it?

Just had to do it around it.

It’s more just, not doing it at work, just everything at home.

Primary outcome

The primary outcome was device-measured physical activity, expressed as mean steps per day across all monitored days, assessed at 6 months. At baseline, the sample accumulated 8583 steps per day, which is comparable to daily step counts recorded previously in a sample of UK-based HGV drivers,36 and to daily step counts seen in office-based workers. 127 The complete-case analysis revealed a statistically significant difference in mean daily step counts at 6 months’ follow-up, in favour of the SHIFT group, with this group accumulating 1008 more steps per day than the control group. The findings showed a similar pattern in the sensitivity analyses (examining the effect of the number of valid activPAL days), although the results were mixed in the ITT and per-protocol analyses. Although the difference in the primary outcome measure between the SHIFT and control arms at 6 months (i.e. 1008 steps/day) was lower than 1500 steps per day, which formed the basis of our sample size calculation, it has recently been reported that 500 steps per day is the minimum clinically important difference for inactive individuals, applying equally to men and women. 128 Therefore, the difference observed in the intervention group relative to the control group is potentially clinically meaningful and potentially of a sufficient magnitude to impact longer-term health and mortality risk. 128

Closer inspection of the changes in mean daily step counts recorded between baseline and 6 months revealed that a decrease in daily steps occurred in the control group, whereas activity levels (i.e. steps/day) measured at baseline were maintained in the SHIFT group. Although large increases in overall daily steps were not observed in the intervention group, the SHIFT intervention appears to be effective in mitigating a reduction in overall activity over at least a 6-month period, observed in the control group. As baseline and 6-month follow-up measures were distributed evenly over a 6-month data collection period (i.e. baseline measures were undertaken between the months of January and July) for all groups, with the corresponding follow-up measures taking place 6–8 months later, it is unlikely that the reduction in steps seen in the control group could be explained by seasonal effects. As physical inactivity is widely associated with an increased risk of many adverse health conditions,129 the prevention of a decline in habitual activity in any population/individual is important when considering longer-term health outcomes. Therefore, the observed differences in steps between groups at 6 months remain potentially clinically important. 128

Despite the high-risk health profile of HGV drivers globally,4 limited health promotion interventions have been conducted in this occupational group. A systematic review25 of health promotion interventions in HGV drivers (which included only eight studies) observed that the interventions generally led to improvements in health and health behaviours; however, the review cautioned that the strength of the evidence was limited because of poor study designs, with no control groups, small samples and no or limited follow-up periods. 25 Of the available literature, only one other study31 of HGV drivers has examined the potential impact of a wrist-worn device to help monitor and self-regulate physical activity levels and healthy dietary choices. In a sample of 26 Australian HGV drivers, similar to the present findings, Gilson et al. 31 observed that participants’ daily step counts [measured using the Jawbone UP accelerometer (Jawbone, San Francisco, CA, USA)] remained constant across the 20-week intervention, with daily steps averaging 8743 steps per day across the first 4 weeks, and averaging 8944 steps per day across the last 4 weeks. Across the 20-week intervention, the logging of dietary choices using the associated Jawbone UP app declined steadily, and the authors concluded that step counts were more successfully monitored than dietary choices. 31

The process evaluation revealed that the Fitbit was a favoured component of the SHIFT intervention. Fitbits, along with similar commercially available wearable activity trackers, and their associated apps, contain a number of behaviour change techniques, including self-monitoring, feedback and goal-setting. 130 Recent systematic reviews and meta-analyses131–133 have revealed that commercially available wearables are associated with favourable increases in physical activity in controlled trials in adults over the short term (note that the duration of the interventions included in these reviews typically ranged between 3 and 6 months). In their meta-analysis, which included 12 controlled trials that incorporated the use of a commercial wearable as an intervention tool, Brickwood et al. 131 reported greater intervention effects when the wearable was part of a multicomponent intervention (as applied in the present study), as opposed to when the wearable was utilised as the primary intervention tool. Within both trial types, however, meta-analyses revealed significant increases in daily step counts in intervention groups relative to control groups (multicomponent interventions, +685 steps/day; wearable-only interventions, +475 steps/day).

Findings from a systematic review and meta-analysis,133 which specifically examined the use of Fitbits as an intervention tool, reported significant increases in daily steps across 16 studies, with a mean difference of +951 steps per day seen in intervention participants, relative to control participants. The majority of RCTs included in this meta-analysis examined the effectiveness of multicomponent interventions, and had a duration of < 5 months. Only five studies incorporated a 6-month follow-up (as applied in the present trial), with only two further studies including a 12-month follow-up. 133 Given the unique population targeted in the present study, and the limited scope to compare the present findings with other studies using HGV drivers,31 the difference in our primary outcome (i.e. +1008 steps/day) observed between intervention and control participants at 6 months appears promising, especially when compared with the findings reported in the recent meta-analyses of wearable interventions, highlighted above. 131,133

Secondary outcomes

COVID-19

The COVID-19 pandemic had a large impact on the overall running of this trial, with the first government national lockdown occurring at the time that the final follow-up measurements within the main trial phase were about to commence. In addition, 6-month follow-up measurements were scheduled to take place in one intervention site during the week commencing 23 March 2020 (i.e. the start of the first national lockdown), and this was the last site to undergo the 6-month follow-up measurements. As a result of the national lockdowns that followed, the 6-month follow-up assessments in this intervention site, as well as all final follow-up assessments, were severely delayed. A change to the original protocol was approved in June 2020, where it was confirmed that the primary outcome would be daily steps recorded at the 6-month follow-up assessment, as opposed to daily steps recorded at 12-month follow-up, which was not feasible given the suspension of data collection. This required change in protocol is a limitation of the trial, as the switch in timing of the primary outcome analysis (from 12 months to 6 months) means that we cannot completely rule out any seasonal changes in behaviour affecting our findings. However, it should also be acknowledged that this change in timing affects both the intervention and control arms.

Within the main trial phase, the easing of government COVID-19 restrictions enabled a range of secondary outcome measures to be collected approximately 16–18 months after randomisation in sites. Owing to restrictions on external visitors to DHL Supply Chain sites throughout the pandemic, face-to-face physiological measurements were not able to be conducted at the final follow-up phase. These follow-up assessments, therefore, did not contain the complete set of measures included at baseline and at 6 months. Furthermore, for the one intervention site due their 6-month measures at the start of the first national lockdown, the delayed 6-month assessments did not contain the physiological health measures included for all other sites, and this led to a reduction in the sample size within the intervention arm for some of these secondary outcomes. Although a strength of this study is the fact that we were able to follow-up participants at 16–18 months, the pandemic presents a major confounding factor that limits our ability to draw firm conclusions regarding the sustainability of the SHIFT intervention. In particular, a greater proportion (58%) of control participants than intervention participants (24%) reported being furloughed, which may have had a large impact on their lifestyle health behaviours and markers of well-being at the final follow-up assessments.

Despite the associated challenges, the pandemic also provided an opportunity to collect further information on its impact on our sample of HGV drivers, who were classed as a key worker group. A subsample of participants completed an additional questionnaire during the first national lockdown. The questionnaire was developed in partnership with colleagues at DHL Supply Chain in response to the relaxation of permitted maximum driving hours. 101 Despite the change in permitted driving hours, respondents to our COVID-19 questionnaire did not report any changes to their working, driving, in-cab waiting or rest hours. Similarly, participants reported no changes in the time spent sitting, standing and walking/moving around on a workday during the pandemic, and there were no negative impacts on symptoms of anxiety or depression, or markers of occupational fatigue. The responses to the COVID-19 questionnaire should be treated with caution, however, as the responses represent only 41% of the sample invited to complete the questionnaire, and non-responders may have been experiencing the pandemic very differently.

The questionnaire did enquire whether or not participating in the study had provided participants with the right knowledge to maintain a healthy lifestyle during the COVID-19 restrictions, and, interestingly, 63% of both intervention and control participants answered ‘yes’. Responses to this question were similar between intervention and control participants, and largely centred around an increased understanding of the importance of activity and diet. The responses received from control participants to this question support observations from the process evaluation that a number of control participants were not aware of the two trial arms, with some participants believing that they were experiencing an intervention as a result of the regular health assessments they were invited to (note that control participants received the same feedback on their physiological measures as the intervention participants).

The questionnaire also enquired whether or not participants had spent time in nature (which could include time in their garden/allotment, in parks, in woodland, at the coast and in open green spaces) during the pandemic, along with whether or not participants habitually spent time in nature prior to the pandemic. These questions were included following recent reports of a wide range of both physiological and psychological health and well-being benefits associated with exposure to nature. 164,165 In this subsample, we observed novel associations between reported time in nature and reductions in measures of occupational fatigue. Further analyses, reported elsewhere,102 revealed that after controlling for covariates, drivers who visited nature at least once a week exhibited 16% less chronic fatigue prior to the COVID-19 pandemic, and 23% less chronic fatigue and 20% less acute fatigue during the COVID-19 pandemic. These novel findings suggest that nature exposure may have the potential to provide a promising remedy for many of the negative health outcomes associated with HGV driving,102 and further research into the use of nature exposure as a potential low-cost intervention to promote physical and mental health in drivers is recommended.

Implementation process

This RCT was complex in terms of multiple components, environments and outcome measures. The intervention comprised the amalgamation of five different components (the 6-hour structured education session, the Fitbit, step count challenges, cab workout equipment and text messages) among 25 heterogeneous worksites (pilot sites, n = 6; main trial sites, n = 19) and aimed to influence the health behaviours of participants in numerous ways (e.g. daily steps, sitting and standing time, time spent in MVPA and nutritional intake).

The structured education session was regarded as valuable by all interviewed intervention participants, who reported that it increased their knowledge, particularly about healthy diets. However, only 145 of 183 (79.2%) intervention participants took part in the education session, mainly because of logistical challenges and operational requirements, which made scheduling the education sessions across sites and ensuring driver availability particularly challenging. This shows that although the education session was beneficial to those participants who attended, it was not wholly feasible in this occupational group, with key issues being the varying start times, operational demand and time-critical deliveries. However, in a ‘real-world’ context, it is estimated that currently only 15–30% of people in the UK newly diagnosed with diabetes attend structured education sessions organised through the NHS, despite high referral rates by GPs. 166 Based on this information, it could, therefore, be argued that, although challenging to organise, if such sessions can be embedded within the workplace of at-risk occupational groups, then their reach could be substantially improved. In the context of HGV drivers, if such health promotion programmes can be embedded within compulsory professional competency training that drivers are required to undertake to maintain their licenses, which take place within working hours, the potential reach and impact of such programmes could be considerable.

The Fitbit worked as an important tool for increasing understanding of current activity levels, providing participants with feedback on their activity and acting as a motivational tool to increase daily steps. There was high adherence to the Fitbit throughout the intervention, suggesting that it was an effective tool to encourage behaviour change, specifically physical activity, but less so regarding sleep (although the Fitbit provides feedback on sleep, this was not a primary focus of the SHIFT intervention). There was less agreement on the step count challenges. Some participants liked the competition of the step count challenges, but other participants did not like competing with ‘strangers’. The text messages were regarded as useful for logistical purposes (e.g. for reminding participants about their up-and-coming health assessments); however, overall, there were minimal replies to the messages, with an average response rate of 18.8%. Participants mentioned that more frequent, personalised messages would be required to stimulate motivation.

The cab workout was a less favourable intervention component, with participants stating that they had more important priorities than using this equipment in their breaks, particularly catching up on lost sleep and eating. However, some participants did use the cab workout equipment, with the most popular device provided being the hand gripper, and 20% of participants agreed the cab workout equipment increased their overall levels of activity. As the adherence to the cab workout equipment appeared low overall, however, the cab workout is regarded as a poor tool to encourage behaviour change within this occupational group.

Mechanisms of impact

The SHIFT intervention used Bandura’s SCT as the theory of behaviour change for intervention development. 42 Bandura’s SCT suggests that learning can occur through observing and imitating someone else’s behaviour, and is most effective when the observer witnesses a model with similarities (e.g. another HGV driver) carrying out the behaviour. Bandura’s SCT focuses on the triadic model, in which personal factors, environmental influences and behaviour continually interact. 167 Bandura argues that goal-setting and self-monitoring are relevant components in effective interventions. In addition, Bandura suggests that the key concepts that affect health behaviour change interventions include self-control, self-efficacy, observational learning and reinforcement. Based on the SHIFT logic model (see Figure 1), self-efficacy and self-monitoring were to be utilised with the Fitbit. The supportive social environment was to be facilitated via the education session and through health coach support from the text messaging service. The acquisition of the essential knowledge relating to behaviours came from the education session. However, the SCT has a shortcoming regarding this RCT, as truck drivers are inherently isolated from each other and, therefore, they rarely learn behaviours from each other’s doing. A further model applicable to the SHIFT intervention is the behaviour change wheel, which uses the capability, opportunity, motivation – behaviour framework, where participants require capability, opportunity and motivation to change their health behaviours. 168 The opportunities to foster motivation can be created through the health assessments, notifying the individual of their current health status and that they may be at risk of certain lifestyle-related diseases and conditions. Capabilities are highlighted through the education sessions, where individuals acquire essential knowledge relating to health behaviours and lifestyle choices. Opportunity is derived from receiving the Fitbit and cab workout equipment, and then turning these changes into habits through regular reminders and feedback from the Fitbit, step count challenges and health coach support from the text messages.

It is also important to recognise that there is no ‘one size fits all’ solution with regard to behaviour change, and this is explained by Resnicow and Vaughan’s169 chaos theory and complex dynamic systems. Theories such as the SCT view change as an interaction of self-efficacy, belief, knowledge, attitude and intention, which creates a linear mechanism for an individual to assess the positives and negatives in a consistent manner. However, Resnicow and Vaughan’s169 chaos theory and complex dynamic systems argue that it is impossible to make predictions on human behaviour, likening this to the impossibility of mathematically predicting the course of two identical balls rolling down a rocky mountain, with the balls ending up in two very different places because of an almost infinite number of variables. Behaviour change encompasses these infinite interacting variables that impact the outcome. 169 According to Resnicow and Vaughan,169 regarding human behaviour, there may be common patterns of behaviour change that occur across and within individuals that may follow complex non-linear patterns. Resnicow and Vaughan169 highlight that identifying these recurrent patterns of change will be useful to aid identification of target groups who could benefit from common intervention components.

Context

All participants were asked in the follow-up questionnaires during each measurement session about any major changes to their life over the past 6 months. The biggest changes reported were moving house, followed by family illness and relationship break-ups. There were no apparent biases between trial groups regarding external factors influencing study participation.

The COVID-19 pandemic caused three major lockdowns in the UK from March 2020 to July 2021, which had wide-ranging impacts on each site that was involved in the study. Although there appeared no systematic differences between intervention and control sites in terms of the impact of the pandemic, it was a rapidly changing, dynamic situation that was unable to be adequately reported. We cannot, therefore, say with certainty that there were no differences in the impact of the pandemic between intervention and control groups. Indeed, as highlighted above, a greater proportion of control participants reported being furloughed than intervention participants, which may have affected participants’ lifestyle health behaviours and markers of well-being, either positively or negatively, prior to the final follow-up assessments.

The outcomes of the study were measured using health assessments, which all intervention and control participants attended. The health assessments were followed by short feedback sessions where the results were explained to each participant. Although not part of the intervention, the health assessments did have an impact on awareness and knowledge about a healthy lifestyle in both intervention and control participants, and this was an unintended outcome of the study, which, although it did not in turn lead to observed behavioural changes in control participants, provided participants with a more holistic understanding of their own current health status.

Process evaluation strengths and limitations

The triangulation of data led to a more comprehensive understanding and rigorous analysis, as we were able to capture data using different dimensions of the same phenomenon. 170 Data were also collected at multiple levels, including driver-, manager- and site-level data, to provide a more complete understanding of the specific context of the RCT. Data for the process evaluation were collected from baseline to the completion of the study (i.e. 16–18 months later), and this enabled us to follow the participants’ reflections throughout their experience of the study. The length of follow-up at the end gives the participant and managers time to reflect and provide more holistic responses about their experiences. The representativeness of each depot was considered when stratified sampling of the drivers and managers for the interviews took place. This method gives the reader a more thorough comprehension of the study, as every site was heterogeneous. The process evaluation was undertaken primarily by a single integrated evaluator, which was beneficial for effective communication, avoids duplication of efforts and reduces participant burden. 94 Very much part of the intervention team, the evaluator used this first-hand experience to understand thoroughly every part of the intervention, and this, in turn, helped to minimise the Hawthorne effect while collecting observational data about the operational challenges for both for the implementation team and the sites.

Assessing the reach of the SHIFT intervention across the included 25 depots was not appropriate or feasible within the context of the programme, as the present trial aimed to recruit approximately 14 participants per site because of financial and time restrictions. It was apparent that in most sites there was a large interest in the study, highlighting the necessity of such health interventions in this at-risk population. Indeed, at baseline, the trial over-recruited, with 382 participants providing informed consent, which exceeded our recruitment target of 336 participants from our sample size calculation. However, despite the initial high interest in the study, the total loss to follow-up was high (46.3%) and this potentially may have resulted in attrition bias, whereby there may have been systematic differences between participants who left and participants who stayed.

All participants were asked to participate in the interviews and incentivised to do so, and this may have led to a sampling bias, although this was mitigated as best as possible by involving one participant from each depot. The limitation of having an integrated process evaluator may increase risk of potential biases in the process evaluation outcome. However, this was mitigated through having an external, and independent from the trial, ‘critical friend’ (Dr Anna Chalkley), and all findings were discussed with the principal investigator (SC). 171 As the process evaluation data were analysed without the knowledge of the main trial outcomes, bias was also minimised so as to reduce influenced interpretations.

Process evaluation conclusions and recommendations

The SHIFT intervention demonstrated effectiveness in the primary outcome (i.e. daily steps); however, future replication and extension of this study should consider more valid measures of nutritional intake to best capture dietary behavioural changes, as regularly reported in the interviews. More frequent contact with both control and intervention participants was suggested as a key improvement, which, in turn, would lessen attrition rates. Attrition rates were high throughout the study, which supports the existing understanding that HGV drivers are a hard-to-reach population,7 not least due to the transient nature of the workforce. The COVID-19 pandemic had a mixed impact on participating sites, which would make any conclusions about the final follow-up uncertain. Overall, participants were enthusiastic about the SHIFT intervention, with particular emphasis on the dietary lessons from the education session and the activity monitoring and motivation from the Fitbit.