Around 16–20% of the workforce in Europe and Australia is represented by shift-workers who have work hours outside of the standard daytime work hours (7:00 a.m. to 6:00 p.m.), including both evening and late-night work [1
]. They play a vital role in the 24/7, 365-day operations of many industries, yet this comes at the cost of an increased risk of all-cause mortality, cardiovascular disease (CVD), type 2 diabetes, metabolic syndrome, and poor mental health for shift-workers [3
]. The risk of any CVD event is 17% higher among shift-workers than non-shift-workers, and the risk of CVD and coronary heart disease (CHD) mortality is 20% higher [7
The poorer health status of shift-workers is in part driven by external factors (e.g., work hours) that alter their circadian rhythm, which in turn influences their metabolism and other physiological processes, including increased stress and allostatic load [5
]. Shift-workers’ poorer health is also influenced by more modifiable factors such as engaging in high-risk lifestyle behaviors [9
]. The unique demands of shift-work include the need to resist the homeostatic drive to sleep at night and live on a schedule that is not in sync with that of their family or society at large, and that is frequently changing (e.g., rotating shifts, on/off roster). Consequently, the ability to prioritize, schedule, and allocate enough time for health-promoting lifestyle behaviors such as sufficient physical activity, eating a good-quality diet, and getting enough good-quality sleep [10
Poor sleep health is the most commonly reported high-risk behavior in shift-workers [11
] and significantly increases the risk of chronic disease [13
]. Circadian misalignment occurs when our behavioral cycles like sleep/wake and fasting/feeding are mismatched with the endogenous circadian rhythms. This can result in dysregulation of feeding behaviors, as well as changes in appetite-stimulating hormones, glucose metabolism, blood pressure, temperature, and heart rate [15
]. Differences in dietary patterns between shift- and non-shift-workers have been reported [16
]. Specifically, greater frequency in eating events, high-fat snacks, and caloric intake peak during the night shift was demonstrated in shift-workers, as well as a shorter maximum fasting period, and this could generate unfavorable metabolic responses such as hyperlipidemia and hyperglycemia, both increasing chronic disease risk [18
]. Reports on differences in physical activity between shift- and non-shift-workers are mixed [23
], and may have a stronger association with occupational tasks rather than shift pattern [24
]. However, a recent study of female nurses found that both rotating shift work and unhealthy lifestyle, as assessed by low physical activity level, poor diet quality, body mass index over 25 kg/m2
, and current smoking, were associated with a higher risk of type 2 diabetes, and the joint effect of these four risk factors was higher than the addition of the risk associated with each individual factor [6
]. This indicates that shift-workers may benefit from a multiple health-behavior change over and above that of non-shift-workers [6
Although many lifestyle behavior interventions exist, relatively few specifically target shift-workers, and those that do utilize resource-intensive methods such as one-on-one consultations and financial incentives which make scaling the program to reach a large number of shift-workers unfeasible [25
]. A useful approach may be to modify existing efficacious lifestyle interventions, but little evidence exists on how to modify “typical” lifestyle behavior interventions in order to suit the unique requirements of shift-workers [28
]. Utilizing mobile health (m-health) may be especially suitable in this population due to the ease of access any time of the day or night, the low cost involved with scaling the intervention, and the lack of geographical barriers. As few multiple behavior interventions exist for shift-workers, we conducted a feasibility study of an existing m-health intervention with process evaluation to inform the co-design of m-health intervention tailored to the unique needs of shift-workers. The current study aims were as follows:
To evaluate the feasibility of an intervention aimed at improving physical activity, diet, and sleep quality not tailored to shift-workers (Move, Eat, and Sleep [29
]) in a shift-worker population and receive feedback on improvements to increase acceptability.
To estimate the effect of the shift-worker Move, Eat, and Sleep intervention on improving physical activity, diet quality, and sleep quality in shift-workers.
This pilot RCT tested the feasibility of the shift-worker Move, Eat, and Sleep intervention in Australian shift-workers aged 18–65 years, who reported either insufficient moderate-to-vigorous physical activity, poor-quality sleep, or a poor-quality diet. The research procedures (recruitment, randomization, data collection and retention) were feasible, with 40 eligible shift-workers recruited in a week and low attrition with 85% completing the follow-up questionnaire. User engagement was acceptable compared to other app-based interventions [50
], with 70% accessing the app at least once and 55% still using the app at four-weeks follow-up. The proportion accessing the app at least once was comparable to that of other app-based interventions [50
], but as it is well known that engagement tapers off with time in technology-based interventions, studies with longer follow-up time often report lower engagement rates at the end of the study [51
]. When used in a non-shift-worker population, the app scored better on the system usability score, with 70.8 (19.7) points compared to the shift-worker population with 62.7 (12.7) points [33
]. The non-shift-workers also used the app for longer, with an average of 37.0 days of data logged compared to 11.6 days in the shift-worker population; however, as a proportion of study length, the logging events were similar (44% of 84 days vs. 41% of 28 days, respectively) [33
]. As it was shown that engagement with the intervention (i.e., app) significantly predicts behavior change, understanding why participants disengage is critical to improving the app for future use [52
The quantitative and qualitative feedback on implementation outcomes indicated a need for modifications to both the app and handbook to better suit the shift-worker population. Key modifications for future versions of interventions for shift-workers are suggested below.
Ability to track multiple sleep periods within one day. Date and a.m./p.m. or 24-h time when adding sleep periods could be included to lessen confusion when entering.
More detailed tracking in the app to be able to see progress better (i.e., different activity levels/types, number of serves of fruit and vegetables compared to just ticking, e.g., if they had five servings of vegetables or not, and a tally of total hours of sleep when having multiple sleeps). This relates back to building self-efficacy and goal-setting (Table S2, Supplementary Materials
), with participants indicating they would like to measure more incremental progress toward their goal.
A different format for the handbook to improve acceptability and accessibility, yet still allowing scalability. Delivering the information incrementally via regular emails, via a website with interactive action planning tools, as integrated chapters in the app, or as a combination of these may be feasible options. Making the information more accessible for those with lower levels of literacy may also be required by using simpler language, presenting information in audio-visual formats (i.e., short videos in the app), tailoring content to individual needs (i.e., based on goals), and other forms of interactivity [53
Introduce interactive features to encourage engagement and “checking in”, for example challenges or automated messages with update on progress. This relates back to increasing self-efficacy by using challenges, praise, and rewards as a type of relapse prevention.
These recommendations are based on the findings that, while 76.9% found the overall intervention slightly or moderately useful in motivating them to participate in more physical activity, sleep, and healthier eating habits, only half or less (41.7% to 50.0%) of the participants found the “move”, “eat”, and “sleep” sections in the app useful. Those who completed the qualitative interviews noted a lack of motivation to continue using the app because it did not allow them to log what they would like to log, i.e., it was not tailored to their needs. Furthermore, the information in the handbook tended to rank higher (61.5–69.2%) than the planning tools and action plans included (30.8–61.5%), perhaps because they were provided and emailed in a pdf document and not as print material. From the qualitative interviews, it was clear that the primary barrier for using the handbook was its length and, despite having intentions of returning to read more at a later time, not doing so.
These recommended changes broadly fit into the categories of “tailoring to individual needs” and “combining digital behavior change interventions (DBCIs) with human support”, which were issues identified in an expert consensus paper regarding promoting effective engagement with DBCIs [53
]. To examine the program in a format that would allow scalability, the current intervention did not include any human contact apart from emails conveying information on eligibility, group allocation, and automated progress reports. This may have limited uptake of the intervention and engagement with the app. Although few studies directly contrasted different levels of support [53
], some evidence suggested that multi-component as opposed to stand-alone app interventions may be more effective [52
]. Gamification by adding challenges within the app was suggested by one of the participants interviewed and may increase engagement. Social features within the app or via a third-party medium (Facebook/Snapchat/WhatsApp group, website with chat rooms, etc.) are also a possibility, but would require greater resources due to the need for a facilitator and modifier within those spaces. Further research is needed to determine which features of apps increase engagement [52
The current study also aimed to estimate the treatment effect of the shift-worker Move, Eat, and Sleep intervention on primary outcomes (minutes of moderate-to-vigorous physical activity, diet quality, and sleep quality) and secondary outcomes (percentage meeting physical activity guidelines, frequency of take-away purchases, consumption of discretionary foods, Sleep Hygiene Index). Findings indicated a positive effect on diet quality, while no significant effect was observed for other measures. At the start of the intervention, the entire sample (n
= 40) reported exceeding 300 min of MVPA a week, which is twice the minimum minutes of MVPA recommended [56
] and far higher than Australian population average [57
]. At follow-up, the minutes of MVPA in the intervention group was significantly lower in the intervention group compared to the wait-list group (−307 min/day; 95% CI = −638 to 24; p
= 0.069, d
= 0.67). This may be because the intervention group was more aware of the activity they were doing and recorded it more accurately, or the validity of the assessment method. Furthermore, the high levels of activity at baseline likely limited any potential for improvement. The use of activity monitors (pedometers or accelerometers) to objectively record baseline and follow-up activity may improve the accuracy of results. Both the wait-list and intervention groups reported a slight improvement in sleep quality, with a non-significant group difference, but the intervention time frame may not have been sufficient to demonstrate changes in sleep. While a meta-analysis of interventions in individuals without a sleep disorder with an average of five weeks duration (range: 2–10 weeks) was found to have a significantly positive effect on sleep quality [58
], in individuals with insomnia, longer treatment duration was associated with larger effect sizes (range: 4–48 weeks) [59
]. It is possible that the inherent changing schedule of shift-workers necessitates a longer intervention time to see the benefits of implementing improved sleep hygiene practices.
The use of Facebook to recruit may have introduced a recruitment bias, and the use of self-report introduces reporting bias. The fact that all participants reported 300+ min of MVPA likely highlights the latter point. The sample may not be representative of the shift-worker population as a whole, as only 10% (n
= 4) lived away from home while working, and the sample was also primarily Caucasian. The small sample size may have influenced the magnitude and variability of the effect sizes described. Also, the current intervention did not consider workplace level factors or environmental characteristics (i.e., access to fresh food, fruit, and vegetables) which may be important to consider in future interventions [60
]. The difficulties in completing the qualitative interviews at follow-up limited the feedback available for modifying the app. Finally, as this was a pilot study, it was not powered to detect changes in health behaviors.