Drivers of HGVs drivers have been identified as a high-risk occupational group who have traditionally been underserved in terms of health promotion initiatives.4,25 This trial aimed to evaluate the effectiveness and cost-effectiveness of the multicomponent SHIFT intervention in a sample of long-distance HGV drivers. Participants were recruited across 25 transport sites across the Midlands region of the UK, with sites operating within the transport, retail, hospitality, health-care, pharmaceutical, construction, oil and gas, and automotive industries. The average age of our sample at baseline [48 (SD 9) years] and our sex split (99% male) matches the average age of HGV drivers and the sex proportions seen nationally.26
A high prevalence of overweight and obesity were observed in our sample at baseline, which exceeds the prevalence of overweight and obesity seen in males aged 45–54 years across the general population (89% vs. 79%).126 Five per cent of our sample had severe obesity (i.e. a BMI ≥ 40 kg/m2) at baseline, which is more than double the prevalence of severe obesity seen in a national sample of aged-matched males (2%).126 Furthermore, over half the sample had pre-hypertension (51%) or hypertension (28%), 84% had clinically elevated circulating LDL-C concentrations (i.e. > 2 mmol/l), and 67% had high total cholesterol levels (i.e. > 4 mmol/l). Participants accumulated high volumes of sitting, particularly on workdays, and high levels of physical inactivity. The characteristics of our recruited sample support previous observations of the high-risk health profile of UK-based HGV drivers,36 and highlight the need for health promotion initiatives to be prioritised in this workforce.
Main findings from the randomised controlled trial
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
activPAL variables on workdays and non-workdays
Complete-case analyses for these secondary outcomes revealed statistically significant differences, in favour of the SHIFT group, in time spent sitting, standing and stepping, and time in MVPA, at 6 months’ follow-up, across all monitored days. Further analyses revealed that the positive changes in overall activity and sitting seen at 6 months were driven by differences in these behaviours occurring between groups on non-workdays. No statistically significant differences were observed in any variables assessed using the activPAL between groups at 6 months (or at 16–18 months) on workdays.
A common theme, which emerged as part of the process evaluation, was the irregularities of shifts and the long duration of shift patterns, which many drivers reported as a barrier to being able to engage in beneficial health behaviours. Owing to the constraints of their job, it appears, therefore, that participants in the SHIFT arm were more likely to adopt positive behaviours in terms of physical activity and reduced sitting on non-workdays than on workdays. Relative to the control group, at 6 months, participants in the SHIFT group accumulated 2012 more steps per day on non-workdays. This was accompanied by an extra 21 minutes per day spent stepping, which was broken down into an extra 10 minutes per day spent in light physical activity and 11 minutes per day spent in MVPA. Similarly, participants in the SHIFT arm accumulated 40 minutes per day less sitting, relative to control participants, at 6 months. The mean differences in MVPA and sitting observed between groups on non-workdays are greater in the present study than those observed in Ringeval et al.’s133 meta-analysis of Fitbit interventions, where mean differences in MVPA of 6 minutes more per day and 11 minutes less per day of sedentary time were seen in intervention groups, relative to control groups.
As with the primary outcome, further interrogation of the data revealed that the favourable changes in behaviours observed on non-workdays at 6 months between intervention and control participants were largely driven by the reductions in physical activity and increases in sitting seen in control group participants, alongside small positive behaviour changes seen in the SHIFT group. It appears, therefore, that the SHIFT intervention was successful in mitigating the unhealthy behaviour changes seen at 6 months in the control group. Furthermore, as highlighted in the recent World Health Organization physical activity guidelines update,134 doing some physical activity is better than none, and even modest increases in activity seen in the SHIFT arm on non-workdays could be beneficial to health.
At baseline, all participants accumulated high volumes of sitting on workdays (≈ 12 hours/day) and non-workdays (≈ 9 hours, 40 minutes), which, unsurprisingly, owing to the nature of their work, demonstrates that HGV drivers accumulate greater sitting times than most occupational groups.135 At baseline, there was no evidence that participants compensated for their highly sedentary occupation by being more active on non-workdays, with participants actually accumulating less physical activity on non-workdays. Sedentary behaviour, defined as ‘any waking behaviour characterised by an energy expenditure ≤ 1.5 metabolic equivalents while in a sitting, reclining or lying posture’,136 has been identified as a risk factor for a number of chronic conditions, including CVD, type 2 diabetes and all-cause mortality.120,137–140 Although recent studies suggest that the detrimental effects of sedentary behaviour can be mitigated by engagement in regular MVPA, with at least 150 minutes of moderate intensity activity accumulated per week required,141 the relatively low volumes of MVPA seen in the present sample is unlikely to reduce the risk of the detrimental health effects associated with sedentary behaviour. Furthermore, recent studies have reported potential thresholds, ranging from 6–8 hours per day140 to 9.5 hours per day,120 spent sedentary where all-cause mortality risk is substantially increased, independent of physical activity. Our sample exceed both of these thresholds when looking at their overall daily sitting times.
Periods of prolonged sitting have been associated with negative health outcomes, and regularly breaking up sitting (every 20–30 minutes) has been associated with favourable changes in blood glucose control, particularly in individuals who are overweight or have obesity and/or individuals who are at high risk of type 2 diabetes.142 Accumulating prolonged periods of sitting is unavoidable in long-distance HGV drivers on workdays; however, non-workdays provide an opportunity where prolonged bouts of sitting can be minimised. A noticeable observation from the descriptive analyses of the activPAL data revealed that, at 6 months, control participants exhibited an increase in the time spent sitting (and the proportion of sitting) in prolonged bouts. No such changes were observed in the intervention group, again suggesting that the SHIFT intervention likely mitigated increases in time spent in prolonged sitting bouts at 6 months.
Despite the favourable differences seen in the SHIFT arm, relative to the control arm, at 6 months, particularly on non-workdays, limited differences between groups were seen in the majority of activPAL variables assessed at 16–18 months’ follow-up. Although not statistically significant (p = 0.10), at 16–18 months, daily step counts on non-workdays were 1391 steps per day more in the SHIFT group, relative to the control group, suggesting some evidence of sustainability. The COVID-19 pandemic, however, is a major confounding factor that occurred for the majority of participants between the 6- and 16- to 18-month follow-up assessments. Furthermore, a disproportionately larger number of control participants (58%) were furloughed at some point between the 6- and 16- to 18-month follow-up assessments, relative to participants in the SHIFT arm (24%). Questionnaire-based data collected from a subsample of participants during the first national lockdown, along with qualitative responses provided on the 16- to 18-month follow-up CRFs, indicated that participants who were furloughed were more likely to engage in new forms of physical activity while away from work. In contrast, it is likely that drivers who continued to work throughout the national lockdowns had extended driving hours and, therefore, even less time to engage in healthy lifestyle behaviours because of the relaxation in drivers’ hours that came into force.101
Markers of cardiometabolic health and functional fitness
The changes in weight and BMI observed at 6 months demonstrated favourable trends in the direction of the SHIFT group. At 6 months, participants in the SHIFT arm recorded an average weight loss of 1.4 kg (i.e. a change of –1.2 kg relative to control participants; p = 0.08) and a reduction in BMI of 0.4 kg/m2 (i.e. a change of –0.4 kg/m2 relative to control participants; p = 0.09). Fifty-eight per cent of participants in the SHIFT arm experienced a reduction in weight at 6 months, compared with 48% of control participants. Although these findings look promising, it should be cautioned that this level of change in weight (≈ 1.4%) would not be considered clinically meaningful and could be an artefact of natural variations in hydration status occurring between measurement sessions. Interventions predominantly focusing on physical activity have been shown to have small to no effects on weight loss.143 To have a bigger impact on weight, the SHIFT intervention could be revised to include a greater emphasis on diet. In a weight-loss intervention conducted in US truck drivers, Thiese et al.30 reported a median weight loss of 3.2 kg in participants following the completion of their 12-week intervention. However, this was a single-arm trial involving only 12 participants.30
There were no other beneficial changes in markers of cardiometabolic health (e.g. blood pressure, waist circumference, waist–hip ratio, biochemical measures) seen in the SHIFT arm relative to the control group at 6 months. Given the strong links between adiposity and a number of these cardiometabolic markers, and the small change in weight, it is perhaps not surprising that no changes in markers of cardiometabolic health were observed. Albeit in a smaller sample, similar findings were observed in the weight-loss intervention in US truck drivers reported by Thiese et al.30 Similarly, no noticeable differences were observed between groups in the present study in their psychophysiological reactivity to stress at 6 months.
Descriptive analyses suggested that the SHIFT group demonstrated favourable increases in average grip strength at 6 months, whereas no changes were detected in the control group. Lower hand grip strength, indicative of lower muscle function, has been shown to be strongly associated with a wide range of adverse health outcomes, including all-cause mortality and incidence of, and mortality from, CVD, respiratory diseases and cancer.144,145 The potential improvements in grip strength observed in the SHIFT group are promising, and are likely to be linked to the inclusion of a hand gripper as part of the cab workout equipment. Although the process evaluation revealed that the cab workout was the least-favoured part of the intervention, participants did highlight that they enjoyed using the hand gripper. Therefore, this simple piece of equipment, which could help maintain and/or improve upper-limb muscle function, holds promise as an effective tool for drivers to use during breaks.
Dietary quality and fruit and vegetable intake
There were no statistically significant differences observed between groups in reported fruit and vegetable intake or overall dietary quality at both 6 months and 16–18 months. These findings contrast with the numerous comments made as part of the process evaluation from drivers, where favourable changes to their diets were reported. This contrast in findings may be attributable to the sensitivity of the FFQ used to assess diet, as previous studies146 have demonstrated questionable validity of FFQs when compared with 4-day weighed food records. However, the feasibility of assessing dietary intake using weighed records in the present study population was uncertain at the planning stages of this trial. The overall dietary quality score derived from the FFQ for our sample (11/15 at baseline) is comparable to that observed from a large randomly selected general population sample from Northern England (11.4/15).60 In comparison to this population sample, overall intake of fruit and vegetables appears to be lower in our driver sample (≈ 240 g/day), with intake decreasing further at 16–18 months (≈ 200 g/day), indicating that participants are falling short of the government’s recommendations of at least 400 g/day of a variety of fruit and vegetables.147 This finding suggests that more needs to be done to support drivers in making healthier dietary choices, with improved access to fresh fruit and vegetables.
Sleep
A notable observation within this trial was the short sleep duration observed across the sample at baseline and at 6 months’ follow-up. Although the SHIFT intervention did not specifically target sleep in detail, sleep duration and efficiency were assessed in the present study as secondary outcomes using a wrist-worn device, and processed using a validated algorithm.89 Of the participants providing valid GENEActiv data at baseline (n = 349), the average sleep duration across all monitored days for the whole sample was 6 hours and 10 minutes (SD 54 minutes), and this reduced slightly to 6 hours (SD 60 minutes) on workdays. At baseline, 41% of participants exhibited an average sleep duration across all monitored days of < 6 hours per 24-hour period, and 82% of participants exhibited an average sleep duration of < 7 hours per 24-hour period. These proportions increased further on workdays to 45% and 85%, respectively.
Of concern, a consistent finding observed across both the SHIFT and control groups at 6 months’ follow-up was a further reduction in sleep duration. The average sleep duration for the sample at 6 months across all monitored days was 5 hours and 56 minutes (SD 57 minutes), and this fell to an average of just 5 hours and 21 minutes (SD 69 minutes) on workdays. At 6 months, just over half of the sample (51%) providing valid GENEActiv data (n = 221) exhibited an average sleep duration across all monitored days of < 6 hours per 24-hour period, and 87% exhibited an average sleep duration of < 7 hours per 24-hour period. These proportions increased further on workdays to 71% and 91%, respectively. The reductions in sleep duration appear to be solely driven by reductions in the duration of the sleep window (i.e. at follow-up, drivers were allowing themselves less time in bed to sleep, as opposed to reductions in overall sleep quality). At the 6-month follow-up, although there was a consistent trend across groups for sleep duration (and sleep window duration) to increase on non-workdays [mean increase across the sample: 41 (SD 99) minutes/24-hour period], participants were still only accumulating 6 hours and 52 minutes (SD 85 minutes) of sleep on these days, which falls short of the recommended minimum of 7 hours per 24-hour period required for optimum health.148 Similar to that discussed above in relation to the activPAL data (see Primary outcome), as both baseline and 6-month follow-up measures were distributed evenly over a 6-month data collection period for both groups, it is unlikely that seasonal changes can fully explain the net reduction in sleep observed.
Systematic review-level evidence has demonstrated that people habitually sleeping less than 6–7 hours per night have a significantly increased prevalence of type 2 diabetes, obesity and CVD, higher cortisol and cholesterol levels, reduced cognitive functioning, depression and other psychiatric conditions, and premature all-cause mortality.149,150 Some of these associations may be mediated by sleep-related changes in glucose metabolism and appetite regulation. Sleep restriction impairs glucose tolerance,151 reduces circulating leptin, and increases hunger and the consumption of carbohydrate-rich foods.152 Short sleep can also lead to daytime fatigue and suppresses the volume and intensity of physical activity undertaken.153 Indeed, a common theme that emerged from the process evaluation when discussing the cab workout component of the intervention was that a high proportion of participants reported prioritising trying to catch-up with their sleep when at a rest stop, as opposed to using the cab workout equipment. As a result, the cab workout was a less favourable intervention component.
In addition to the individual-level cardiometabolic risks associated with short sleep duration,150 and of particular relevance and concern within the present sample, is the association between short sleep duration and reduced driving performance and increased accident risk,154,155 as this has wider public health and safety implications for all road users. For example, a US Department for Transportation study observed that both severe sleep apnoea (a condition common in commercial drivers, which drivers are required to inform the Driver and Vehicle Licensing Agency about156) and sleeping < 6 hours per night were equally, and independently, associated with impaired driver performance.157
A limitation of the measurement of sleep used in the present study is the fact that naps were not assessed, and it appears from the process evaluation that a number of participants did attempt to nap during their breaks. Therefore, it is possible that total sleep durations are underestimated in this study. Nevertheless, sleep duration was a recurrent theme highlighted within the process evaluation, and this, in combination with the sleep data collected from the GENEActiv, suggests that the drivers in this sample are chronically sleep deprived. These findings have important implications, suggesting that participants are at an increased risk of excessive daytime sleepiness, road traffic accidents and chronic disease.158 Indeed, a UK Department for Transport review concluded that insufficient sleep, leading to daytime sleepiness, impaired vigilance and poor concentration, is responsible for the ‘disproportionately high number of fatigue related accidents’ involving drivers of large goods vehicles (contains public sector information licensed under the Open Government Licence v3.0).159
The findings from this secondary outcome measure, along with the concerning observation of a further reduction in sleep duration and sleep window duration at 6 months in this sample of drivers, suggests that the SHIFT intervention should be expanded to include a much greater focus on sleep. Increasing sleep quantity through interventions targeting improved sleep management in drivers will potentially offer dual public health benefits of reducing accident risk (through reduced fatigue and improved vigilance performance) and reducing cardiometabolic risk within the individual (through improved glucose tolerance and appetite regulation, and increased engagement in physical activity). This recommendation is particularly pertinent at the present time, given the increased number of HGV driver shortages within the UK29 and the relaxation of drivers’ hours rules as a result of COVID-19 and Brexit.160 There is a risk that the current sleep profile of HGV drivers may be even worse than that observed in this study, given our 6-month follow-up assessments were completed just prior to the COVID-19 outbreak and Brexit, and the associated relaxation in drivers’ hours rules and substantial increase in driver shortages. The long hours worked by our participants suggests that drivers may also not completely recover from work-related fatigue between shifts. High levels of ‘need for recovery’ have been associated with sleep complaints in coach drivers,161 and with longer-term sickness absence in HGV drivers.162 Further work examining interventions to improve drivers’ sleep should also take into account, therefore, working hours and the potential impact of the need for recovery between shifts.
In the present study, within both groups, no changes in device-measured sleep quality (i.e. sleep efficiency) or chronotype score were observed between baseline and follow-up. Despite the reduction in device-measured sleep duration observed in both groups at 6 months, there were no changes in ratings of situational sleepiness observed across groups, although this measure should be treated with caution because of the variability in the exact timing within the day/night that this questionnaire was completed across follow-up periods. Furthermore, other studies have demonstrated no associations between self-reported sleepiness and reduced cognitive performance across a range of tasks, including driving, resulting from sleep deprivation.157,163
Mental well-being, cognitive function, musculoskeletal symptoms and work-related psychosocial variables
In contrast to previous observations of relatively high levels of poor mental health within drivers,4 reported symptoms of anxiety and depression were low in the present sample at baseline, with limited changes in symptoms occurring across the follow-up assessments in either group. At baseline, 13% of participants reported borderline symptoms of depression and 17% of participants reported borderline symptoms of anxiety, whereas 2% and 5% of participants reported abnormal scores for depression and anxiety, respectively. Similarly, low levels of social isolation were reported across all assessment points throughout this study. No noticeable differences in changes in cognitive function were observed between groups at 6 months’ follow-up. Likewise, there were no observable differences between groups in changes in musculoskeletal symptoms or any work-related psychosocial variables (i.e. work engagement, occupational fatigue, job satisfaction and performance, sickness absence and presenteeism, work ability and perceived job demands) occurring at either follow-up. In addition, no differences were observed between groups in terms of reported driving-related safety behaviours.
Lifestyle-related behaviours and cardiovascular disease risk
At baseline, 25% of participants reported drinking more than 14 units per week of alcohol, and this is a lower proportion than that reported126 in a nationally representative sample of aged-matched males, where 35% of the sample reported drinking more than 14 units per week of alcohol. No noticeable differences were observed between groups at any assessment point in terms of alcohol intake, and alcohol intakes observed in the present sample appear lower overall than what has been reported elsewhere in HGV drivers from other countries.4 However, this observation should be treated with caution, as the tools used to assess alcohol intake in HGV drivers have varied extensively across studies, making it difficult to draw comparisons.4
The prevalence of smoking within the sample at baseline (19.4%) was similar to that seen in males aged 45–54 years living in England (20%).126 When split by study group, 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. For participants completing the baseline and 6-month follow-up assessments, smoking prevalence changed from 17% to 19% in the control group, and from 13% to 11% in the SHIFT group. In the smaller sample of participants who completed the baseline and 16- to 18-month follow-up assessments, smoking prevalence decreased by 1% in both groups at 16–18 months. The impact of the SHIFT intervention on smoking is, therefore, uncertain, and limited effects on smoking (and alcohol intake) are perhaps to be anticipated, as these topics were covered only briefly in the structured education session, with the focus of this session being predominantly on physical activity, diet and sitting.
When examining the proportion of participants with an estimated CVD risk of ≥ 10% over the next 10 years, 23.6% of control participants and 24.3% of SHIFT participants fell into this category at baseline, and this increased to 26.4% in the control group and reduced to 23.4% in the SHIFT group at 6 months. These findings suggest that participants in the SHIFT group experienced a modest reduction in risk of a cardiovascular event over the next 10 years, relative to control participants. Reducing the risk of a CVD-related event in HGV drivers has important implications, not only for the individual, but also for the wider public, given the serious consequences should a driver have a CVD event while driving. Although not specifically related to CVD events, Ronna et al.23 reported that, based on 10-year CVD risk calculated using the Framingham Risk Scale, the odds of having an accident doubled in US truck drivers with a Framingham Risk Scale score > 13. Ronna et al.23 also observed a statistically significant association between prevalence of accidents and increased risk scores, further highlighting the public health importance of improving the overall health of this occupational group.
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.
Main findings from the cost-effectiveness analysis
The within-trial analysis showed that the SHIFT intervention reduced QALYs and increased costs. The small improvements in physical activity seen as a result of the intervention generated potential for slight improvements in QALYs in the longer term. Despite this, under a range of alternative scenarios and assumptions, the SHIFT intervention in its current delivery format is unlikely to be considered cost-effective when compared with usual practice at commonly used threshold values of a QALY.
Main findings from the process evaluation
The process evaluation indicated that the SHIFT intervention had a positive impact on the intervention participants, as reported in both the questionnaire and interview responses. Participants reported an increase in knowledge, awareness and motivation regarding the importance of increased physical activity and a healthy diet. The Fitbit was the most favoured component of the intervention, whereas the cab workout appeared the least favoured and too cumbersome for the majority of participants. The most common suggested improvement to the intervention was to increase the frequency of communication with participants. The barriers to health were still very apparent throughout, with the irregularity and long duration of their shift patterns highlighted by many drivers. These barriers required a high level of extrinsic motivation to overcome within this at-risk occupational group to enable them to change health-related behaviours, and, therefore, regular contact from those administering any future interventions would likely be needed to help motivate participants to maintain improved behaviours.
Using the MRC process evaluation framework,94 the discussion of findings from the process evaluation will focus on the implementation process and the mechanisms of impact that influenced the findings, followed by the contextual factors that may have affected the RCT outcomes.
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.
Trial strengths and limitations
A major strength of this study was the implementation of a lifestyle health behaviour intervention within the workplace environment of a very underserved and at-risk occupational group. The characteristics of our sample at baseline highlight the poor health profile of HGV drivers within the UK, and emphasise the urgent need to improve the health of this shrinking, yet essential, workforce.29 The study involved 25 different transport sites spread throughout the Midlands region, operating within subcontracts across eight different industries. The range of industries represented by these sites, together with the demographic characteristics of our sample (mean age at baseline 48 years and 99% male, which matches exactly the characteristics of UK HGV drivers26), suggests that the included sample likely represents the 278,700 HGV drivers currently in employment.1
Our multicomponent lifestyle health behaviour intervention (i.e. the SHIFT intervention) was evaluated through a fully powered cluster RCT, where randomisation occurred at the site level (reducing the risk of contamination) after baseline assessments had been undertaken (reducing bias). The trial incorporated immediate (6-month) and longer-term (16- to 18-month) follow-up periods to enable the examination of the effectiveness and potential sustainability of the SHIFT intervention. The trial also included a mixed-methods process evaluation and a full economic analysis. To the best of our knowledge, this is the first cluster RCT to examine the effectiveness and cost-effectiveness of a lifestyle health behaviour intervention within HGV drivers, with the few earlier intervention studies25,30,31 reported in this workforce limited by small sample sizes, no control groups and limited follow-up durations. The SHIFT intervention has a strong theoretical underpinning.42 The SHIFT intervention was created and refined based on our earlier work,7,36–38 and the planning of this study, and subsequent conducting of it, has been informed by extensive PPI.
The use of the activPAL accelerometer as the primary outcome measure is a further strength, with this device being shown to provide a highly accurate measure of steps and posture.53–55 Furthermore, we were able to confirm the validity of this device in our particular sample by demonstrating that, within the HGV cab, the activPAL is not affected by vehicle vibrations. Compliance to the activPAL wear protocol was relatively high in the present study, and this was facilitated by checking the activPAL data on return of the devices and requesting re-wears where possible. At baseline, 90% of participants provided at least 1 day of activPAL data. Of the sample of participants returning the device at 6 months, 89% provided at least 1 day of activPAL data, of whom 84% provided valid activPAL data at both baseline and 6 months. On average, participants wore the activPAL for 6.8 days at baseline and 7.2 days at 6 months. These compliance rates are similar to those seen recently in a large sample of office-based workers.172 Although a minimum number of days of device wear are usually specified to allow for day-to-day variation in behaviours,173 to maximise our sample, owing to the high loss to follow-up experienced (discussed below), in our main analysis we included all participants who provided at least 1 day of activPAL data, as applied elsewhere.86 However, to test the robustness of our findings, we performed a sensitivity analysis including only participants who provided more valid days of activPAL data, and our findings remained unchanged. Although the activPAL provides a device-based measure of physical activity (and participants were blinded to the data recorded), reducing bias associated with self-report measures, participants were still aware of the purpose of the activPAL. Therefore, reactivity to this measure may have occurred, although any potential reactivity is likely to have affected the SHIFT and control groups equally. The trial included a range of validated secondary outcomes, enabling a comprehensive evaluation of the SHIFT intervention on markers of adiposity and cardiometabolic risk, mental well-being, a range of lifestyle health-related behaviours and measures of work-related psychosocial factors.
A major limitation of the present study was the high loss to follow-up experienced, which was beyond that initially predicted. We experienced a 31.4% loss to follow-up at the 6-month assessments, with the sample included in the primary outcome analysis reduced further (55% of the initial randomised sample) after taking into account activPAL compliance across the two assessment points. Further losses to follow-up were experienced at the final follow-up, with 54% of the original sample attending this assessment. We also lost two sites/clusters during the trial due to the collapse of their contracting companies. It was emphasised by managers as part of our process evaluation that HGV drivers are notoriously transient workers, with a high staff turnover rate. A large proportion of drivers not completing this study had left their role before the cessation of the programme. Sick leave and missed assessment sessions were also common reasons for non-completion. Future trials with this, or similar, occupational groups will need to take into account potentially high loss to follow-up rates within sample size calculations, along with consideration of compliance rates to device-based measures, if appropriate. Within the present study, we overrecruited at baseline, which is perhaps further evidence of the need for such health improvement interventions in HGV drivers. Nevertheless, the initial larger sample recruited meant that the larger than expected loss to follow-up rates were mitigated to a certain extent within our primary analysis, where sufficient statistical power remained to detect a significant difference between trial arms in our primary outcome.
The overall day-to-day running of the trial was extremely complex, and it was very challenging to schedule the measurement sessions in some sites because of the demand on the workforce, which led to overall delays with data collection. Owing to the 24/7 working nature of the logistics sector, a number of site visits took place during the night/very early hours of the morning, which led to further challenges for the research team in terms of scheduling and undertaking these visits. It was also extremely challenging to schedule the 6-hour education sessions within intervention sites, as, owing to the pressures faced by the industry, a number of managers found it difficult to facilitate the time for their drivers to be away from their driving duties. The overall challenges associated with the scheduling of measurement visits and education sessions, along with the challenges faced by the drivers to incorporate healthy lifestyle behaviours on workdays, emphasise and confirm the hard-to-reach nature of this male-dominated occupational group.
Owing to the multicomponent nature of our intervention, it should be highlighted that the SHIFT structured education session did not focus on one specific element of lifestyle health behaviours. It was not focused on physical activity, diet or sitting alone; all three elements were included. With the education session linked to the feedback participants had received from their baseline health measurements, intervention participants could choose to work on and improve any single behaviour or a combination of behaviours. Therefore, for some participants, step count targets may have increased if this is what they chose to focus on; for other participants, it could have been dietary choices and/or weight. Therefore, in some respects, owing to the multiple health behaviours covered, our overall results for each individual behaviour (i.e. steps, diet, weight, sitting) could have been ‘watered down’.
Conclusions and recommendations
The SHIFT intervention may have had a degree of success in positively impacting physical activity levels and reducing sitting time in HGV drivers at 6 months’ follow-up. Owing to the nature and demands of the occupation, the statistically significant differences observed between groups in these behaviours were largely driven by changes occurring on non-workdays, and are also largely attributable to the maintenance of physical activity levels in the SHIFT arm and a decline in physical activity levels in the control arm. The process evaluation revealed favourable attitudes towards the SHIFT intervention from both drivers and managers, with drivers highlighting that the education session, Fitbit and step count challenges were particularly effective for facilitating behavioural changes. Managers and participants reported enthusiasm and a sense of necessity for the SHIFT intervention to be included in future CPC training for professional drivers in the UK.
Although most intervention participants reported positive improvements to both knowledge and behaviour around their dietary intake within the process evaluation, the dietary outcome measures did not substantiate these findings within the RCT. Owing to the modest differences in physical activity seen between groups, and there being no differences between dietary variables, no statistically significant differences were observed between groups in terms of markers of adiposity or cardiometabolic outcomes. No differences in any outcome measure were seen between groups during the final follow-up assessments, suggesting that the positive impacts of the SHIFT intervention were not sustained beyond the duration of the 6-month intervention. However, the pandemic presents a major confounding factor that limits our ability to draw firm conclusions regarding the sustainability of the SHIFT intervention, particularly in light of the imbalance in participants on furlough between the two trial arms. The economic evaluation revealed that the SHIFT intervention is not likely to be cost-effective in its current delivery format.
The high prevalence of drivers with obesity, along with the poor cardiometabolic health profile and sleep deprivation seen in our sample, accompanied by the challenges experienced in scheduling data collection and the education sessions, highlight substantial health inequalities in this at-risk and hard-to-reach occupational group. Given the current, and increasing, shortfall of HGV drivers in the UK, which has risen from 60,00028 to an estimated 100,000 in 2021,29 the government and sector urgently need to address working conditions and the poor health profile of this ageing workforce to attract employees to the role. The already challenging working conditions are likely to be only exacerbated currently, as the small number of drivers have to compensate for driver shortages by expanding their own working hours, as relaxations in drivers’ hours rules have been re-introduced as a result of driver shortages, COVID-19 and Brexit.160 Driver recruitment and a prioritisation of driver health is essential to combat the current challenges seen in maintaining critical supply chains, and to support the UK’s economic recovery from the COVID-19 pandemic. In addition, improving drivers’ health has significant implications, not only for the individual or their employer (through reductions in sickness absence and staff turnover), but also for the wider public through improving road safety for all users. Although the longer-term impact of the SHIFT intervention is unclear, the intervention (with ongoing development and refinement) offers potential to be incorporated into driver training courses to promote activity in this at-risk, underserved and hard-to-reach essential occupational group.
Based on the findings of the present study, we recommend the following:
- To support the development and implementation of the SHIFT intervention as a CPC training module for HGV drivers, further work involving stakeholder engagement is needed to refine the content of the intervention, based on findings of the present study, and to examine an appropriate delivery mode that is cost-effective with maximal reach. On the translation of the SHIFT intervention into a CPC module, further work should be conducted to evaluate the scaling-up of this intervention over the longer term, in a real-world setting.
- Effective strategies targeting improvements in dietary behaviours that, in turn, promote weight loss in HGV drivers need to be researched and incorporated into the SHIFT intervention to further impact the high prevalence of drivers with obesity.
- Effective interventions targeting improvements in drivers’ sleep duration need to be created and evaluated and, subsequently, incorporated into the SHIFT intervention to combat the high levels of sleep deprivation observed in this study. Increasing sleep quantity through interventions targeting improved sleep management in drivers will potentially offer dual public health benefits of reducing accident risk (through reduced fatigue and improved vigilance performance) and reducing cardiometabolic risk within the individual.
Further research
Based on the findings of the present study relating to the high levels of sleep deprivation seen in our sample, members of the research team, along with colleagues with expertise in sleep science, have been awarded a MRC Public Health Intervention Development grant (reference MR/W004070/1; principal investigator Dr Iuliana Hartescu; start date 1 November 2021) to co-develop (with target users and stakeholders) an app-based intervention to improve sleep quality and quantity in commercial drivers within the road freight sector.