Introduction

Workers in the construction industry are often exposed to physically demanding work tasks. These include, amongst others, the lifting of heavy loads and working in awkward postures. High physical work demands increase the risk of the development of musculoskeletal symptoms (da Costa and Vieira 2010; Holtermann et al. 2010). In blue collar construction workers, musculoskeletal disorders (MSD) are the most prevalent work-related health problem (Boschman et al. 2012; Peterson and Zwerling 1998). In addition, in the Netherlands, the workforce in physically demanding work is ageing and the risk of MDS also increases with age (de Zwart et al. 1997; Hildebrandt 1995). As such, MSD are a major cause for sickness absence, work disability, and early exit from work, are related to lower work performance, and consequently constitute an extensive social, medical, as well as economic problem (van den Heuvel et al. 2010; Arndt et al. 2005).

The prevalence of overweight and obesity in construction workers is higher than in the general adult population (Proper and Hildebrandt 2010; Santos and Barros 2003; Arndt et al. 1996). Both MSD and a high BMI are negatively associated with several work-related outcomes, but are also associated with each other (Claessen et al. 2009; Viester et al. 2013; Roos et al. 2013; Robroek et al. 2013; Alavinia et al. 2009). Since both factors are highly prevalent in blue collar construction workers, these might contribute to the high risk of developing health disorders and associated adverse work-related outcomes compared with workers in other industries and the general population (Snashall 1990; Taimela et al. 2007). This emphasizes the importance to reduce the burden of overweight and obesity in this particular group of workers.

Both diet and physical activity are considered of importance in achieving and maintaining a healthy body weight (Swinburn et al. 2003, WHO 2004). Worksite health promotion programmes aimed at physical activity and diet were found to be effective on weight-related outcomes (Verweij et al. 2011; Groeneveld et al. 2010; Anderson et al. 2009). Moreover, workplace health promotion programmes that improve physical activity levels have been shown to also reduce the risk on MSD (Proper et al. 2003). A lifestyle intervention amongst those with jobs involving moderately heavy or heavy work also showed a reduction in prevalence of low back pain (Mattila et al. 2007). Although intervention studies with MSD as primary outcome have not often been targeted at lifestyle factors, there is evidence from observational studies, suggesting that health promotion should be considered in the prevention of MSD (Hildebrandt et al. 2000; Holth et al. 2008; Morken et al. 2007; Burton et al. 2006). Beneficial effects on work-related outcomes, including sickness absence, productivity, and work ability, have been reported resulting from preventative measures targeted at healthy lifestyle (Cancelliere et al. 2011; Proper et al. 2004; Kuoppala et al. 2008; von Thiele Schwarz and Hasson 2012). Consequently, implementation of worksite programmes targeted at lifestyle factors may be a promising strategy to improve worker health and other outcomes relevant to employers.

In the vitality in practice (VIP) in construction study, it was hypothesized that a worksite health promotion intervention, aiming at improving physical activity and diet, could positively change body weight-related outcomes, musculoskeletal symptoms, and work-related measures (Viester et al. 2012). The aim of the present study was to evaluate whether the intervention programme for blue collar construction workers reduced musculoskeletal symptoms, limitations in physical functioning and sickness absence, and increased work-related vitality, work performance and work ability.

Methods

Study design and population

The effectiveness of the programme was assessed in a randomized controlled trial (RCT). The research population consisted of consenting blue collar employees of a construction company. All employees who attended a non-compulsory periodic health screening (PHS) and who were not on sick leave for more than 4 weeks prior to the PHS were eligible for inclusion. In total, 314 participants were recruited over a 15-month period (March 2010–June 2011) and randomized to an intervention (n = 162) or control group (n = 152). Participants completed questionnaires at baseline (T0), at 6 months (T1), and at 12 months (T2). Written informed consent was obtained from participants before enrolment in the study.

The study design and procedures were approved by the Medical Ethics Committee of the VU University Medical Center, and the trial has been registered in the Netherlands Trial Register (NTR, www.trialregister.nl): NTR2095.

Randomization, blinding, and sample size

Following baseline measurements, participants were randomly assigned to either the intervention or the control group by a computer-generated list using SPSS 15 (SPSS Inc. Chicago, Illinois, USA). The randomization was prepared and performed by an independent researcher. Whereas participants could have been aware of the allocated arm, data collectors and analyst were kept blinded to the allocation. The sample size was calculated to identify an effect on body weight (Viester et al. 2012). Based on that calculation, in each study group (intervention and control), 130 participants were needed for follow-up.

Intervention

The intervention programme aimed at the prevention and reduction of overweight and musculoskeletal symptoms and was developed and implemented via the Intervention Mapping protocol (Bartholomew et al. 2006; Viester et al. 2012). The full programme has been described previously (Viester et al. 2012). In short, the intervention consisted of an on-site lifestyle coaching programme tailored to the participant’s weight status (BMI and waist circumference), physical activity level, and stage of change. The intervention programme focused on improving (vigorous) physical activity levels and healthy dietary behaviour. The programme consisted of tailored lifestyle information, lifestyle coaching sessions, exercise instructions, and the ‘VIP in construction toolbox’. This toolbox consisted of an overview of the company’s health promoting facilities, a waist circumference measuring tape, a pedometer, a BMI card, a calorie guide, healthy recipes, and a lifestyle knowledge test.

The intervention was delivered face-to-face and via telephone by personal health coaches (PHC) who were trained specifically for the study. Face-to-face coaching sessions took place at the worksite during working hours. An overview of the timing and duration of the contacts is presented in Table 1. Participants additionally received ‘Personal Energy Plan’ (PEP) forms to record their goals and action plans, and to be used during the follow-up health coaching sessions. Intervention providers were not involved in the outcome assessment.

Table 1 Coaching contact schedule
Full size table

The control group received care as usual and was only contacted for the baseline and follow-up measurements.

Outcome measures

The present study investigated the effectiveness of the intervention on musculoskeletal symptoms, physical functioning, and work-related outcomes (work ability, work performance, work-related vitality, and sickness absence). Sickness absence data were obtained from the company’s registration system after follow-up measurements were completed. All other data were obtained using questionnaires.

Health-related measures

Musculoskeletal symptoms

The prevalence of musculoskeletal symptoms during the past 3 months was assessed using the Dutch musculoskeletal questionnaire (DMQ), which has been validated for different body regions (Hildebrandt et al. 2001). The occurrence of pain or discomfort was rated on a four-point scale (never, sometimes, frequently, and prolonged). For the current analysis, the measure was dichotomized; answer categories ‘frequently’ and ‘prolonged’ were classified as having musculoskeletal symptoms, whereas categories ‘never’ and ‘sometimes’ were classified as having no musculoskeletal symptoms. Body regions were grouped into back (upper and lower back), neck/shoulders, upper extremities (elbows and wrist/hands), and lower extremities (hips/thighs, knees, and ankle/feet).

Physical functioning

Physical functioning was measured using a subscale of the RAND-36, evaluating functional status (Hays et al. 1993; VanderZee et al. 1996). The RAND-36 cluster on role limitations caused by physical problems consists of 4 items and ranges from 0 to 100 points (higher scores indicating less limitations), with a score of 79.4 considered average (VanderZee et al. 1996). The RAND-36 health survey is a widely adopted, and reliable and valid measurement of health-related quality of life (Brazier et al. 1992). In the present study, the validated Dutch version was used.

Work-related measures

Work ability was assessed with the work ability index (WAI) (Ilmarinen and Tuomi 1993; de Zwart et al. 2002; Nygard et al. 1991). The WAI covers 7 dimensions; current work ability, work ability in relation to job demands, number of current diseases, work impairment due to diseases, sickness absence days during past 12 months, own prognosis of work ability in next 2 years, and mental resources. Total scores over all dimensions range from 7 to 49, with 4 categories: poor (7–27 points), moderate (28–36 points), good (37–43 points), and excellent (44–49 points).

Work-related vitality, defined as vigour, was assessed through a subscale of the Utrecht Engagement Scale (UWES) that refers to high levels of energy and resilience, the willingness to invest effort, not being easily fatigued, and persistence in the face of difficulties (Schaufeli and Bakker 2003). The answers were rated on a seven-point scale from never (0) to daily (6). The mean score of the items resulted in the work-related vitality score, with a higher score indicating a better work-related vitality.

Work performance was measured using a single item from the Health Work Performance Questionnaire (WHO-HPQ)(Kessler et al. 2003, 2004) asking workers to report their overall work performance on a 10-point scale over the past four weeks.

Sickness absence data were collected directly from company records. For the analysis, cumulative sickness absence data over 6-month periods were used (pre-, during-, and post-intervention). Sickness absence has a skewed distribution with a substantial fraction clustered at the value zero. Therefore, sickness absence was dichotomized into no or short-term sickness absence (≤7 days), and long-term sickness absence (>7 days).

Statistical analysis

The analysis was conducted with all available subjects at 6 and 12 months of follow-up. All available data of the participants, regardless of whether or not they actually (fully) received the intervention, were used for analysis. Data on potential confounders and effect modifiers were assessed through the baseline questionnaire and included age, smoking status, education level, and marital status. For all variables, potential baseline differences were checked between intervention and control group.

Linear and logistic regression analyses were performed for the different outcome measures, with both 6- and 12-month follow-up as the dependent variables. Analyses were adjusted for the baseline levels. Analyses were performed using SPSS 20.0 (SPSS Inc. Chicago, Illinois, USA). For all analyses, a two-tailed p value of <0.05 was considered statistically significant.

Results

In total, 314 workers responded to the baseline questionnaire. At 12 months of follow-up, 83 % of the participants completed all measurements; 22 workers of the control group (14 %) and 31 workers of the intervention group (19 %) did not complete all follow-up measurements. Figure 1 presents the flow chart of the participants throughout the trial. Baseline characteristics are presented in Table 2. No differences between groups were found for key variables.

Fig. 1
figure 1

Flow chart of the study participants

Full size image
Table 2 Baseline characteristics
Full size table

Table 3 shows complete case intervention effects on work-related vitality, work performance, work ability, and physical functioning. For all outcome measures, a positive value for B, which represents the estimate (unstandardized coefficient) resulting from the regression analyses, can be interpreted as a positive intervention effect. No statistically significant differences were found for any of the outcome variables after 6 and 12 months of follow-up.

Table 3 Intervention effects for work-related vitality, work performance, and work ability after 6 and 12 months of follow-up
Full size table

Musculoskeletal symptoms

The intervention did not result in statistically significant effects on musculoskeletal symptoms (Table 4). Although for back symptoms at 6 and 12 months of follow-up (OR 0.69, 95 % CI: 0.36–1.36, and 0.76, 95 % CI: 0.38–1.52, respectively) and lower extremity symptoms at 12 months (OR 0.61, 95 % CI: 0.32–1.16), the odds ratios were in favour of the intervention group, differences reached no statistical significance.

Table 4 Intervention effects (OR (95 % CI) or B (95 % CI)) for musculoskeletal symptoms and sickness absence after 6 and 12 months of follow-up
Full size table

Sickness absence

Table 5 shows mean days of sickness absence in the past 6 months, and Table 3 presents the course of sickness absence for the study group, dichotomized into no or short-term, and long-term sickness absence. Directly following the intervention, the 6-month prevalence of long-term sickness absence was lower in the intervention group than in the control group. At 12 months, sickness absence was slightly higher in the intervention group compared with the control group. However, at both 6 and 12 months, the between-group differences were not statistically significant.

Table 5 Average number of sickness absence days for the intervention and the control group during 6-month periods before the baseline and follow-up measurements
Full size table

Discussion

The aim of this study was to evaluate the effectiveness on secondary outcomes of a health promotion intervention aiming at increasing physical activity and improving dietary behaviour in construction workers. No significant short- or long-term intervention effects were found on musculoskeletal symptoms, physical functioning, work-related vitality, work performance, work ability, or sickness absence. These findings will be discussed for the different outcome measures.

Musculoskeletal symptoms

The lack of observed statistically significant intervention effects on musculoskeletal symptoms is in line with other intervention studies in the construction sector (Gram et al. 2012a, 2012b; Oude Hengel et al. 2013). Overall in the present study, the prevalence of workers reporting musculoskeletal symptoms declined. For back and lower extremity symptoms, odds ratios were in favour of the intervention group, although not statistically significant. Since sample size calculations were performed to determine effects on the study’s primary outcome measure (body weight), for other outcome measures, the study could have been underpowered.

In the current study, it was hypothesized that an improvement in physical capacity through increased physical activity, and a decrease in workload through a reduction of overweight, would be effective in preventing or reducing musculoskeletal symptoms.

Although it is still not clear what type of exercise should be recommended, several reviews support the use of exercise as an effective strategy for the prevention or treatment of musculoskeletal conditions, including a wide range of interventions, such as increasing general physical activity levels, general exercise, and specific body-region exercises for strength and flexibility (Roddy et al. 2005; Hayden et al. 2005). The current intervention consisted of a combination of exercise prescription and coaching on improving physical activity levels, which implied that participants self-selected their physical activity goals. Although an increase in vigorous physical activity in the intervention group was found, this may not have been exercise or physical activity selected for the purpose to prevent or reduce musculoskeletal symptoms, and might as a result not have been the most appropriate type of activity or exercise to reduce or prevent specific symptoms. Additionally, the increase in physical activity levels may not have led to sufficient physical capacity improvements to be effective on musculoskeletal symptoms.

Presumably, the effects on outcomes related to body weight, as found in this study, were not substantial enough to have a direct effect on MSD. Another explanation could be that the intervention period was not long enough for effects on MSD to occur. However, prevention of body weight gain or reducing excess body weight could have future effects by lowering both systemic and metabolic risk factors. Systemic risk factors include a combination of mechanical load on weight-bearing joints and work postures. Obesity is one of the components of the metabolic syndrome, and metabolic risk factors are increasingly being recognized as a possible cause of MSD (Berenbaum 2013; Sellam and Berenbaum 2013).

To reduce or prevent musculoskeletal symptoms, it has been suggested that multi-component interventions are potentially more effective (Kennedy et al. 2010). In these programmes, exercise or training interventions are combined with components addressing environmental and/or organizational issues. For example, the physical and psycho-social work environment has been recognized as risk factors for MSD in the construction sector. This is supported by findings from interviews with employees during the development of the present study as well as in the study of Oude Hengel et al. (2010), Viester et al. (2012). Combining health and lifestyle promotion with efforts to decrease workload and/or change working conditions is probably necessary for programmes to be effective.

Work-related vitality, physical functioning, work performance, and work ability

In addition to the explanation of the lack of effect as described in the section on musculoskeletal symptoms, the initially high scores for work-related vitality, physical functioning and WAI could explain the lack of further detectable increase in these outcomes, i.e. a ceiling effect. For work-related vitality, this was also found in previous studies (Strijk et al. 2013). The lack of effect on the WAI in the current study is in accordance with previous studies on work ability (Oude Hengel et al. 2013; Nurminen et al. 2002; Pohjonen and Ranta 2001). The average baseline WAI score of 40.7 was only slightly higher compared with the average score of Finnish men in the same age group and engaged in physical work (Ilmarinen et al. 1997), and scores ranging from 37 to 43 are regarded as good work ability. For the physical functioning dimension of the SF-36, baseline values of the study population largely exceeded norm values of a reference population.

Sickness absence

With regard to sickness absence, the lack of effects is in line with other studies amongst blue collar worker (Oude Hengel et al. 2013; Jorgensen et al. 2011). During the trial period, several factors in addition to illness, which are related to sickness absence, may have influenced the results. Not all absence can be attributed to sickness; sickness absence has been associated with, for example, socio-economic factors, organizational features, job content, and attitudes to work (Briner 1996). This is especially of concern when using total sickness absence data, compared with the absence related to a specific condition, such as MSD. The current economic recession that strongly affected the construction sector during the trial period may have distorted effects on total sickness absence or patterns of sickness absence. Stress, increased (perceived) workload, and fear of job loss are factors that might have played a larger role under these circumstances during the study period.

For all outcome measures, the lack of intervention effects can in part be attributed to the level of implementation of the programme. In a process evaluation of the programme, it was concluded that the extent to which the programme was implemented as intended was modest (Viester et al. 2014). Although participants’ satisfaction with the programme and dose delivered by the health coaches was high, exposure and fidelity were not optimal. The compliance to the coaching sessions was acceptable, but the implementation of the exercise component was not successful. Although approximately two thirds of the participants indicated to have done the exercises, only a small percentage exercised regularly as prescribed by the programme.

The trial findings could be applicable to a larger population of manual labour workers. The intervention was implemented in a diverse group of blue collar workers with comparable participation rates for the subunits of the construction company. However, when generalizing the results from the specific setting of the RCT to a larger worker population, it should be taken into account that compared to the original population, older workers were slightly overrepresented in the study population (Viester et al. 2014).

Strengths and limitations

Strengths of the study include the randomized controlled trial design, and obtaining sickness absence data from company records. The use of sickness absence data from company records is preferred since it is more accurate than data gathered via self-report (Ferrie et al. 2005).

Some limitations have to be addressed as well. First, power calculation was performed on the primary outcome measure of the study, i.e. body weight. As a result, group sizes might have been below the required number to establish inter-group differences for other study outcomes. Further, missing data on items of the work ability index resulted in a reduced number of complete cases. For participants who did not complete all 7 items, the index could not be determined. With exception of sickness absence, all outcome measures were obtained using self-report which may lead to overestimation or underestimation of the outcomes. Finally, although contamination of the control group participants was expected to be minimal, since only intervention participants had access to coaching and the toolbox, it could not be completely ruled out. Behaviour change in colleagues working at the same worksites could have influenced control participants.

Implications for practice and future research

Maintaining a healthy and productive workforce depends on a wide variety of factors. It is recommended that future interventions aiming to improve work-related outcomes also include organizational and/or environmental components to more effectively target factors related to work ability and performance.

Theoretically, improving physical capacity (i.e. improving muscle function or increasing oxidative capacity) by increasing physical activity and exercise might prevent or reduce musculoskeletal symptoms. In the present study, we did not include measures to monitor possible effects of increased physical activity levels on physical capacity. To increase knowledge on the relevance of increasing physical capacity in this group of workers and to contribute to insight into optimal type, duration, and intensity of exercise, future studies should include such measures related to physical capacity.

Conclusions

The results of this RCT did not show effects of the programme on musculoskeletal symptoms, physical functioning, work-related vitality, work performance, work ability, or sickness absence. Although the intervention programme improved physical activity levels, dietary outcomes, and weight-related outcomes at 6 months, it was not successful in improving other health-related and work-related outcomes. In conclusion, for all outcome measures in the present paper, it could be argued that they are affected by additional factors to those included in the current conceptual model of the study (Viester et al. 2012). Based on the results of the present study, organizations attempting to improve worker health- and work-related outcomes should provide additional programme components. Although a non-significant decline in musculoskeletal symptoms was observed, without co-intervening on (psycho-social) organizational aspects in a more multifaceted intervention, the potential of improving these outcomes by health promotion is probably limited.