Group-based healthy lifestyle workplace interventions for shift workers: a systematic review

Objective Shift work is a risk factor for many chronic diseases and has been associated with unhealthy lifestyle behaviors. Workplaces have great potential for promoting and supporting behavior change. We conducted a systematic review of group-based lifestyle workplace interventions for shift workers to (i) identify adaptations and intervention components that accommodate shift working and (ii) assess their impact on weight, physical activity, sedentary behavior and healthy eating. Methods A systematic search was conducted in Scopus, Web of Knowledge, EBSCO and Ovid databases. Using pre-established criteria, independent pairs of researchers conducted the study selection, quality appraisal and data extraction. Results In total, 22 studies on group-based workplace interventions for shift workers were included. Many demonstrated organizational level adaptations, such as flexible delivery times and paying employees’ time for their involvement. Delivery locations near the workplace and management support were other key features. Common intervention components included competitive group activities, individualized goal setting, self-monitoring and feedback, staff involvement in intervention delivery, and incentives. There was moderate evidence for effectiveness on weight and physical activity outcomes, but insufficient evidence for healthy eating outcomes. No interventions focusing on sedentary behavior among shift workers were found. Conclusion Current evidence demonstrates that group-based workplace interventions can be effective for supporting shift workers to lose weight and increase physical activity, while further research is needed to change healthy eating and sedentary behaviors. Our findings offer decision support on organizational-level adaptations and intervention components that are important to make interventions that promote healthy lifestyles more accessible to shift workers.

support on organizational-level adaptations and intervention components that are important to make interventions that promote healthy lifestyles more accessible to shift workers. key terms group-based; healthy eating; healthy lifestyle; intervention; physical activity; review; sedentary behavior; shift work; shift worker; systematic review; weight; workplace intervention Healthcare expenditure has risen continually over recent decades (1)(2)(3)(4)(5), and a significant proportion of these costs can be attributed to chronic health conditions such as diabetes, cardiovascular disease and asthma (6). Chronic diseases are the leading causes of death and disability worldwide and affect all socioeconomic groups. Poor lifestyles, including an unhealthy diet and physical inactivity, are associated with being overweight or obese as well as many chronic health conditions, reduced physical functioning, functional impairment and early exit from the labor market onto disability pensions (6)(7)(8). These factors can hinder individuals' opportunities for extending their healthy working lives, and present employers with challenges such as higher sickness absence rates, reduced productivity, and premature loss of valued employees (9).
Unhealthy lifestyles are amenable to change, but achieving sustained behavioral change is difficult. Workplaces, as physical and social environments, have great potential for facilitating more positive lifestyle choices. Over 75% of working age people are in paid work and spend much of their waking time working (10). Employers have legislative responsibilities for health and safety, and there is a strong business case for investing in appropriately-scaled initiatives to promote healthy lifestyles in the workplace, reduce employee turnover, increase productivity and employee engagement, and contribute to corporate social responsibility for the UK's ageing workforce (10,11). Workplace lifestyle interventions have been shown to improve employee health, increase productivity and be cost effective (6), and many are tailored to suit the specific operational and organizational requirements of different workforces (12-21). Group-based workplace interventions offer the advantage of peer support and cost-effectiveness, and are often the preferred option for healthy lifestyle initiatives in the workplace (22-24). There is evidence of the positive impact of workplace programs on health behaviors (25): increases in selfreported physical activity have been demonstrated, particularly in workplace interventions targeting physical activity (including walking) as opposed to general lifestyle change (26,27). Workplace dietary interventions have also been shown to improve eating behaviors, for example increased fruit and vegetable intake and decreased fat intake (28). Other positive outcomes include: improvements in psychosocial health, quality of life and emotional wellbeing (29); and reduced presenteeism (29-31), absenteeism (31) and sickness absence (32). been associated with physical inactivity and poor diet (39). A recent study of the dietary characteristics of shift workers found that that while diet quality does not differ between shift workers and those working regular hours, doing night shifts was associated with higher energy intake (36). Fujishiro et al (34) examined the cumulative exposure to rotating night shifts among more than 50 000 women from the Nurses' Health Study II, and demonstrated an independent contribution of night shift work to weight gain. The development of workplace lifestyle interventions specifically for shift workers is an emerging field (33). Such interventions require an approach that considers and accommodates not only the worker and operational characteristics, but also the organizational determinants that can act as barriers or enablers to successful implementation (33,40). This study aims to identify the existing evidence for group-based interventions delivered within workplace settings to help shift workers lose weight, increase physical activity, improve healthy eating, or reduce sedentary time.

Inclusion criteria
We included any studies and study protocols that described group-based interventions delivered in work-place settings, specifically targeting shift workers in the public or private sectors. Group-based interventions were defined as any interventions that delivered the intervention or specific components of the intervention to groups of employees rather than on an individual basis. For instance, this could include group educational seminars, teambased physical activity challenges and group physical activity training sessions to name a few. The intervention had to target at least one of the following: weight loss; physical activity; dietary improvement; and/or reduced sedentary behavior. The target population had to be aged 18-70 years. Outcomes of interest were weight, physical activity, diet, and/or sedentary time.

Exclusion criteria
Interventions delivered to the self-employed or employees working in small-to mediumsized enterprises were not included. Studies were excluded if there was no mention of shift work and/or interventions were delivered in a work setting that the research team agreed was not likely to involve shift working. Additionally, studies detailing interventions that were purely web-based were excluded from the review, as were studies with participants who had undergone weight loss surgery.

Search strategy
To devise our search strategy, the four main concepts relevant to the review were identified: (i) lifestyle; (ii) interventions; (iii) shift workers; and (iv) setting. A search term list was then developed for each concept, which included free-text terms and comprehensive controlled vocabulary items. We included both UK and US spellings and used truncation to capture plural and singular forms of words. We also used search techniques such as Boolean and proximity operators and phrase searching. We applied limits to year of publication from Our final search strategy was: [(Weight W/2 (Manage* OR Los* OR Reduc*)) OR ((Physical OR Exercis*) AND (Capacity OR Perform* OR Train* OR Effort* OR Exert*)) OR (Healthy W/2 (Diet* OR Nutrit* OR Eat* OR Food)) OR (Lifestyle)] AND [(Group AND (Program* OR Interven*))] AND [((Employee* OR Staff* OR Worker* OR Personnel* OR Workforce)) OR (Shift* W/2 Work*) AND ((Job OR Work) W/2 (Place* OR Site*))]. The search strategy was adapted to the specific requirement of each different database used.
To be as inclusive as possible, we consulted databases across the medical, public health, and social science disciplines: Scopus, Web of Knowledge, EBSCO (CINAHL, PsychInfo, Francis, and SocIndex), Ovid (EMBASE and Medline). All authors participated in screening the titles, abstracts and full papers using the inclusion criteria defined in the previous paragraphs. After excluding duplicates, a total of 5626 citations were obtained from the electronic search (figure 1). A further 298 citations were identified through hand-searching reference lists and the grey literature, leading to 5924 citations for screening. Each unique title was independently assessed by two reviewers, and where they did not agree, a third reviewer was consulted in order to reach a consensus.
After title screening, two reviewers independently reviewed 696 abstracts for eligibility. Of these, 281 publications were identified for full-text review. Two reviewers assessed all full texts, and a third reviewer independently read 10% for quality control. Any discrepancies were resolved by discussion. After full-text screening, 242 papers were excluded, and reasons for exclusion were noted. A further 17 papers were subsequently excluded as shift work was not specifically evidenced.

Data extraction
In order to develop the final data extraction table, members of the research team used a pilot version to independently extract data from a sub-sample (16%) of the included papers. The final version contained seven fields: first author and publication year; country; study design; workplace setting; study participants/sample size, intervention aim; and intervention description (including components).

Quality assessment
Two reviewers independently assessed the methodological quality and sources of potential bias of all included studies using the Consolidated Standards of Reporting Trials (CONSORT) Statement (41) for randomized studies (25 items) and the Transparent Reporting of Evaluations with Nonrandomized Designs (TREND) Statement (42) (22 items) for non-randomized studies. Both quality assessment tools were used to systematically examine and appraise: title and abstract; scientific background and introduction; methods; results; and discussion. All items were rated as 1 when the condition was satisfied, 0.5 when it was partially satisfied and 0 if the condition was not met. The included studies were then classified as high, moderate or low quality if their final assessment score was >80%, 60-79%, and <60% of the maximum possible score (42), respectively. The two reviewers discussed any disagreements and, if necessary, resolved with reference to a third reviewer. Detail results of the quality assessment of all included studies can be found in the supplementary material (tables S1 and S2, www.sjweh.fi/show_abstract.php? abstract_id=3715).
Following quality assessment, three papers were rated as low-quality (43-45); these were included in the descriptive analyses, but excluded from the evidence synthesis (see below). Reasons for low quality included: lack of detail on the method used to generate the random allocation sequence or on the type of randomization; no information on the mechanism used to implement the random allocation sequence; and the absence of eligibility criteria for participants. Five study protocols were also excluded from the evidence synthesis. Seventeen studies of high or moderate methodological quality were included in the evidence synthesis (figure 1 and supplementary tables S1 and S2).

Analysis
To identify themes emerging from the review and key intervention components, a narrative synthesis approach was undertaken using three distinct steps: collating, summarizing, and reporting the results (46). The information was collated in tables, and the main findings summarized and reported by outcome of interest. This involved an iterative process, examining the evidence for intervention components that may have influenced the outcomes. Study protocols were included in this stage.
In order to assess the effectiveness of group-based workplace interventions, we performed an evidence synthesis based on the quality assessment rating and the significance or nonsignificance in relation to the outcomes of interest (weight, healthy eating, physical activity, sedentary behavior) and other relevant outcomes (objective and self-reported health, sickness absence and other work-related outcomes). The criteria used for the evidence synthesis were: "strong evidence" -consistent results (in terms of statistical significance between ≥2 high quality studies; "moderate evidence" -consistent results between ≥1 high quality and ≥1 intermediate quality study, or between ≥2 intermediate quality studies; "insufficient evidence" -identification of a single study or inconsistent results across studies; and "evidence of no association" -consistent results of a non-association in ≥2 studies (47,48). Based on the definitions of Stennstra et al (48), a significant effect in one study and a nonsignificant effect in another were considered consistent findings, while a negative effect in one study and a positive effect in another were considered inconsistent findings.

Characteristics of included studies
In total 22 studies were included in our review, and 17 of these in the evidence synthesis (figure 1). The main reasons for exclusion were: not a group-based intervention; not adapted for shift work; email, postal, purely web-based, environmental or individual (ie, one-to-one delivery) intervention; not a workplace setting; and target health behaviors and outcomes (eg, smoking, alcohol consumption) beyond the scope of this review.
The included studies were conducted over four continents (

Intervention delivery
We found a range of delivery formats for workplace lifestyle group-based interventions for shift workers; from programs based around face-to-face information sessions (50,64) (53) or paid for employee time both to deliver (51) and participate (52) in the intervention. Another common measure was to ensure that physical activity sessions were held as near as possible to the workplace (52,66). Management support (45, 50-53, 57) and encouragement (12, 50-52) for employees to join and continue to take part in the intervention were also used to support program delivery, and in two cases both management and shift working employees took part in co-production activities during intervention development (44, 45).

Intervention components
The interventions all included multiple components (

Components of interventions targeting weight
Interventions focusing on weight loss (43, 51, 53, 58, 64, 65) included group education sessions, sometimes combined with one-to-one information or counselling sessions or individualized support and feedback. A range of resources was used, including dedicated websites to complement workplace delivery, handbooks, pedometers, diet logbooks and healthy eating resources or supplies (eg, provision of free fruit). A strong emphasis was placed on the importance of group activities and peer support, and team competition was often included. Incentives in the form of financial prizes for teams (64,65) or individuals (53) were also used. Environmental components included healthy options and portion sizes in the cafeteria (51).

Components of interventions targeting physical activity
The types of physical activity exercises offered were wide ranging (eg, aerobics, walking sessions, weight training, dancing, step challenges).

Components of interventions targeting healthy eating
The components of interventions with a major focus on dietary improvement (45, 50-52, 62, 63, 65, 66) were often highly similar to the weight loss and physical activity interventions described above. These included free access to health clubs, personal training, food logs, cookbooks and healthy eating supplies (51,52,62,63,65,66). Environmental changes included free coupons for healthy meals at the workplace cafeteria, changes in the price for healthy foods or the establishment of "healthy eating chat tables" at the cafeteria (52,62,63), and the provision of healthy options and smaller portion sizes (45, 51, 62, 63). One of the interventions provided weekly educational classes delivered by a registered dietician (50), and incorporated an analysis of participants' health beliefs, nutrition knowledge and dietary behaviors prior to commencing the intervention (50), while another intervention combined both nutrition education sessions with environmental changes at the workplace cafeteria (62,63).

Components of interventions targeting sedentary behavior
There were no studies focusing on changing sedentary behavior. However, one study targeting physical activity in truck drivers reported a change in the number of truck drivers sitting for >9 hours each day at work (self-reported) after implementing different interventions or combinations of interventions, including displaying healthy eating posters, supplying free fruit, promoting online resource, group educational sessions or step challenge (45).

Intervention effectiveness
Evidence synthesis on the effectiveness of the interventions on the reported primary outcomes (as well as other outcomes of interest, eg, health, sickness absence, work ability) was performed on the 17 studies rated as moderate or high quality following the quality assessment. Table 2 indicates that there is moderate evidence for improvements in weight and physical activity, and insufficient evidence for improvements in healthy eating (see also supplementary table S3, www.sjweh.fi/show_abstract.php?abstract_id=3715). Five of the nine studies reporting weight loss outcomes showed positive and significant impacts (52-54, 64, 67); three studies showed no significant difference between intervention and control groups (12, 49, 61, 62), one showed inconsistent results between the intervention and the control arms (62), and one moderate quality study demonstrated a modest negative impact (57). Physical activity had four high or moderate quality studies reporting significant positive impacts (59,64,65,67), three studies reporting a non-significant positive change or inconsistent results (49,57,68) and one study reporting positive self-reported change in physical activity levels without indicating if the change was significant or not (66). There were seven studies targeting healthy eating (49, 50, 62, 64-66, 68, 69), only two studies-one high and one moderate quality-reported significant positive impacts (66,68), whereas five studies reported non-significant or inconsistent results (49,50,(62)(63)(64)(65).
All studies examined a range of health and wellbeing indicators, both objectively-measured (eg, blood pressure, resting heart rate, body fat, fasting lipids, VO 2 max) and self-reported (eg, perceived health status, self-reported mental health, work ability . Heterogeneity meant that it was not possible to assess strength of effect, conduct a meta-analysis, or assess the effectiveness of specific intervention components on our target behaviors.

Discussion
Group-based workplace interventions to promote weight loss, physical activity and healthy eating behaviors in shift workers require a number of adaptations at the organizational level, including flexible delivery, proximity of intervention sites to the workplace, and management support and encouragement. The flexibility in delivery that was demonstrated reflected the complexity of intervening in workplaces via group-based interventions to improve shift workers' health behaviors and especially adapting to the specific challenges associated with differing work patterns. The interventions included in this review often targeted more than one of the outcomes of interest (sometimes with other outcomes) and had many components. Competitive group activities, behavioral modification strategies, such as individualized goal setting and feedback, and incentives were key components that featured widely. The results demonstrate moderate evidence of the effectiveness of group-based workplace interventions on weight and physical activity, but insufficient evidence for healthy Demou  eating. Moderate evidence was also demonstrated for health and work-related outcomes, but no significant impact on sickness absence was observed.

Research findings in context with previous studies
Previous research has shown that workplace interventions at the organizational level alone have modest effects on lifestyle behavior (72)(73)(74) and that the best evidence for effectiveness is from multi-component interventions that work across different levels (23, [74][75][76][77]. This was also evident in the studies included in this review: many were multicomponent and included a number of adaptations to reflect shift working patterns and constraints in order to support, promote and implement the interventions. These adaptations and components ranged from changes to the cafeteria environment ( Significantly positive impacts on diet have previously been observed in interventions delivered in the workplace during work-time, and those that involved staff in delivery and were multicomponent (78,79). Similarly, the interventions tailored for shift workers included in this review were all either delivered at the workplace, or very close by. Staff involvement was often demonstrated by having peer champions (49), staff or management role models (44), and group leaders (51). Examples of changes at an organizational level were evident in all interventions, including management involvement and support (12, 44, 45, 50-53, 57, 62, 63).
Another review on the general working population suggests that workplace health promotion can improve health outcomes and productivity (80). Goetzel et al's (80) evidence synthesis on the impact of workplace health promotion interventions on health outcomes showed insufficient evidence overall but did include a number of individual studies demonstrating positive impacts. The studies included in our review demonstrate similar results for physical and mental health outcomes, with a number of studies demonstrating significant positive impacts (12, 49, 52, 54, 60). While we did not assess productivity per se, our review addressed productivity-related outcomes, including sickness absence and work ability (56,59,60,(69)(70)(71). Brox et al (60) demonstrated an increase in sickness absence in the intervention group, but a non-significant difference in self-certified sickness absence. While Jakobsen et al (56,(69)(70)(71)) demonstrated a significant increase in self-reported sickness absence in the last year as measured by one item of the work ability index. Pohjonen et al (59) showed no change in work ability. Jakobsen et al (56,(69)(70)(71) reported a small to moderate significant effect in work ability, but no changes in other work ability measures (eg, work disability, influence at work).

Study strengths
A strength of this systematic review is that it included a comprehensive scrutiny of databases covering the medical, public health, and social science literatures. It also covers a large population (total N=9725), from ten countries and four continents. Additionally, it encompasses a breadth of workplaces, including hospitals and care/nursing homes, manufacturing, fire and prison services, hospitality, casinos, transport, and other public and private sectors, each with unique opportunities and often significant challenges for intervention development and delivery, and included sectors (eg, hospital, fire service, care home) which are recognized to have high levels of work-related health problems and sickness absence (81).

Study limitations
While the range of targeted behaviors and other reported outcomes, workplaces covered and intervention components included in this review is a strength, at the same time the heterogeneity of the included studies and intervention components do not allow for a metaanalysis, or assessment of the effectiveness of specific intervention components on our target behaviors. While the majority of identified studies were RCT, which are regarded as being methodologically robust, in the final evidence synthesis only seven RCT were included.
Most of the studies were classified as moderate quality; only three (all RCT) were high quality. The pre/post design of a number of the studies makes it difficult to draw conclusions about causal relationships. Other limitations include the variability, validity and reliability of the reporting of shift work. As only papers that explicitly made some mention of shift work, however tangentially, were included in this review, it is possible that we could have inadvertently excluded other high quality studies, for example where interventions were aimed at an entire workforce, and not only shift workers.

Implications for policy and practice
Research into the development and implementation of interventions tailored specifically for shift workers is a new and evolving field with many evidence gaps. Papantoniou et al (33), highlight the increasing evidence that shift work increases the risk of major chronic diseases, draw attention to the large proportion of the current workforce exposed to shift work, and call for more workplace interventions addressing health-related outcomes among shift workers. A recent study on the barriers and facilitators to a healthier lifestyle and the impact the working environment can have on shift workers found that the workplace environment was key in assisting shift workers to adopt and lead healthier lifestyles (82). Discussions are ongoing about whether to classify shift work as a workplace hazard qualifying for compensation (33). Denmark already considers breast cancer an occupational disease in shift workers, and compensates women with >20 years of night work who develop breast cancer (33).
Interventions for chronic disease risk reduction and prevention in shift workers require novel approaches to reflect the constraints of shift working. This review suggests a number of adaptations, including flexibility in timing of delivery (49-51, 53, 54, 56, 62, 63), allowing time off work for participation (53) and paying for employee time for intervention delivery and participation (51), that should be considered in developing future workplace healthy lifestyle interventions for shift workers. However, although there is some evidence in relation to weight loss and physical activity, more research is needed in order to maximize impact on lifestyle (including sedentary behavior), health and work-related outcomes.

Concluding remarks
Workplaces, as physical and social settings, have great potential for promoting health and wellbeing (6,(12)(13)(14)(15)(16)(17)(18)(19)(20)(21)82). Shift work has been associated with unhealthy lifestyle behaviors, which can contribute to increased risk for disease. Our findings suggest that workplace healthy lifestyle interventions with a group-based element can be implemented for shift workers by ensuring flexible delivery modes and organizational level adaptations, and can be effective in promoting weight loss and physical activity. This review can inform the development and implementation of future workplace interventions for shift workers to ensure that this specific workforce population can benefit from their workplace environments by promoting behaviors that protect against chronic diseases.