The (cost-)effectiveness of an individually tailored long-term worksite health promotion programme on physical activity and nutrition: design of a pragmatic cluster randomised controlled trial

Cardiovascular disease (CVD) is the leading cause of disability and mortality in most Western countries. CVD causes nearly half of all deaths in Europe (49%) [1]. Major modifiable risk factors for CVD include smoking, alcohol use, low physical activity (PA), and poor nutrition. The prevalence of several risk factors is very high, most notably low PA and poor nutrition (low fruit and vegetable consumption and high saturated fat intake).

According to a survey in European Union countries in 2002, 56% of the Dutch population over 15 years was insufficiently physically active for health [2]. The Dutch recommendation on PA stipulates that an adult should engage in PA of at least moderate intensity for at least 30 minutes a day on five days a week, and preferably every day in order to obtain health benefits [3]. In 2006, about half of the Dutch adults (25–55 year) met this recommendation [4]. In order to improve physical fitness it is recommended to engage in PA of vigorous intensity for at least 20 minutes on at least three days a week [3]. Exercise capacity has been found to be a powerful predictor of mortality [5]. It has been estimated that the life expectancy for people with low PA levels at age over 50 is 1.4 years less than for people with moderate PA levels and even 3.8 years less than for people with high PA levels [6].

The results of a recent meta-analysis on cohort studies indicate that fruit and vegetable consumption is inversely associated with the occurrence of coronary heart disease. The risk of coronary heart disease decreased by 4% for each additional portion of fruit and vegetables per day [7]. In the last representative Dutch food intake survey in 1997/1998 less than a fourth of the Dutch population met the recommendation for vegetable (200 grams a day) and fruit intake (200 grams a day) [8]. Regarding saturated fat intake, only 9% of the Dutch adult population met the recommendation (a maximum of 10 percent of energy intake as saturated fat) in 1997/1998 [9]. A high intake of saturated fat increases the risk of coronary heart disease [10].

The imbalance between PA and nutrition is an important cause of overweight and obesity, which in turn are important risk factors for CVD [11]. In the Netherlands self-reported overweight (body mass index ≥ 25) in adult men increased from 37% in 1981 to 51% in 2004, and in adult women from 30% in 1981 to 42% in 2004 [12].

In the prevention of cardiovascular disease, lifestyle behaviour changes are of great importance. Worksites have specific features that make them a promising place for health promotion. Worksites offer an efficient structure to reach large groups, enable the introduction of social support, and make use of a natural social network for peer support [13, 14].

Literature shows contradictory results of randomised controlled trials (RCTs) on worksite health promotion programmes (WHPPs). A recent systematic review concluded that there is strong evidence for effectiveness of WHPP, based on two RCTs with a small effect on exercise behaviour and on energy expenditure [15]. However, another review on worksite PA programmes reported a small average effect size of 0.04 (95% CI -0.04–0.12) based on RCTs on self-reported level of PA 1–144 months after the intervention ceased [13]. A third review on environmental and policy interventions presented preliminary evidence that combined health education, screening, counselling, peer support, and access to (on-site) exercise equipment had positive effects on fitness levels, frequency of exercise, cholesterol levels, and systolic blood pressure. Several randomised studies on point-of-purchase nutrition interventions, some in worksites, showed positive effects on fruit and vegetable consumption, self-reported fat intake, cholesterol, and body weight but other studies have failed to corroborate these findings [16].

The overall picture emerges that WHPP may increase PA and improve nutritional intake among targeted groups, depending on the critical features of the interventions. Amongst others, as success factors of WHPP have been identified: (1) interventions tailored to the individuals' readiness for exercise adoption, (2) programmes that integrate specific components (nutrition, smoking, PA) into a combined approach, and (3) linking individual approaches to environment and policy conditions [17]. Marcus and colleagues showed that workers receiving self-help exercise promotion material tailored to the individual's readiness were significantly more likely to have increased exercise [18]. An individualized approach of high risk employees within the framework of a comprehensive program proved to be a critical feature of worksite interventions [19]. Recent studies have shown that web-based education tailored to personal characteristics may increase fruit and vegetable consumption and PA level, and decrease fat intake. In these interventions people received personalized feedback and advice that directly matched their individual behaviour, motivation, perceived (dis)advantages, and self-efficacy beliefs [20]. Based on results of their study on email messages to promote health behaviours, Franklin et al. suggest that emails may contribute tot the effective deliverance of health promotion programmes [21].

In contrast, three factors have been identified as greatest risks for ineffective WHPP: (1) a low, selective participation, (2) lack of adherence to the WHPP, and (3) an intervention effort too short for sustainable change in behaviour [13, 15, 16]. In several worksite studies intervention and evaluation periods were too short to determine the sustainable impact of environmental conditions [16].

In conclusion, previous WHPPs have shown contradictory results. Studies are needed on a WHPP that counteracts the three main factors for ineffectiveness.

In the study protocol described in this article, a long-term WHPP will be evaluated that adds the following four critical features to a traditional WHPP: (1) a computer-tailored advice on PA and diet (to increase awareness and adherence to the WHPP) (2) insight in progress over time on health-related behaviours (to increase adherence to the WHPP, compliance with the lifestyle recommendations and sustainability of a healthy lifestyle), (3) continuous feedback and support through monthly e-mails (personal coach) for 12 months (to increase adherence to the WHPP and compliance with and sustainability to the lifestyle recommendations), and (4) opportunities to seek personal advice from a variety of professionals (to increase adherence to the WHPP).

The aim of this pragmatic cluster randomised controlled trial is to evaluate the cost-effectiveness of a new investigator-driven WHPP with individually tailored advice in PA and nutrition and individual counselling to increase compliance with the lifestyle recommendations and sustainability of a healthy lifestyle.

The study is a single blind pragmatic cluster randomised controlled trial with the worksite as the unit of randomisation. The (evaluation of the) intervention is targeted at the individual level. The study population consists of workers in companies that offer a standard WHPP to their employees, as provided by an organisation specialized in health management (LifeGuard Inc., Utrecht). Eligibility criteria for individual workers in the study are: 1) paid employment, 2) working at least 12 hours a week, and 3) being literate enough to read and understand simple e-mail and Internet-based messages in the Dutch language. All participants are blinded to the type of intervention. Data collection starts in September 2007 and will continue until August 2010. Participants will be requested to fill out a questionnaire at baseline and after 12 and 24 months. A physical examination (the 'health check') will take place at baseline and after 24 months. The study design and participant flow are shown in Figure 1. The Medical Ethics Committee of Erasmus MC, University Medical Center Rotterdam, the Netherlands, has approved the study protocol.

Within each company, units will be randomised by a researcher not involved in the study, based on a table of random numbers (SAS command Ranuni). Within each company, worksites with comparable work activities and a comparable number of workers will be randomly allocated to the intervention or the control group. Subsequently, workers within each unit will be asked to participate in the study, presented as an evaluation study of different types of WHPP. All participants from one worksite will be randomised together rather than individually because individual randomisation may lead to contamination of the control group. Written informed consent at individual level is collected after agreement of the employer and randomisation at cluster level. Since it is deemed not possible within companies to withheld participation in a WHPP, workers within the control group will receive a standard WHPP.

The standard WHPP consists of:

On top of the standard WHPP the intervention group will have full access to the personalized Health Portal on Internet. The Portal contains four critical features: a computer-tailored advice, a monitoring function, a personal coach, and opportunities to contact professionals at request.

Possible confounders include demographics, smoking behaviour, and general health. The demographic variables of importance are age, ethnicity, educational level, sex, and marital status. Smoking is defined as current smoking status. General health will be assessed using the Short Form-12 questionnaire [35]. In addition, some questions on occupation will be asked: job title, years in current job, days and hours of work (including overtime work), main job requirements (physical or mental), and working conditions [36]. During a worksite visit, environmental determinants will be assessed, especially available resources in the company to provide and sustain healthy behaviour (e.g. fitness room, financial compensation for membership of sport/fitness club, stairs, fruit and vegetables in canteen). Further, participants will be asked if they have Internet availability at home.

The cost-effectiveness analysis will be performed from a societal perspective as well as a company perspective. The following direct costs will be determined: cost price of the standard WHPP, costs for the Health Portal and direct costs of medical consumption. Direct costs of medical consumption will be based on frequency of contacts with a variety of health professionals and average remuneration fee, assessed by an adapted version of the Dutch Trimbos and iMTA Questionnaire on Costs Associated with Psychiatric Illness (TiC-P) [37]. The direct costs will be measured over the complete follow-up period of 24 months with annual questionnaires with 12-month recall.

The indirect costs will be based on assessment of days with loss of productivity at work due to health problems and productivity loss due to sickness absence, using the Dutch productivity and disease questionnaire (PRODISQ) [38]. The estimated days of productivity loss will be multiplied by the average wage per day for each worksite.

In a second step cost-effectiveness ratios will be calculated on two measures of health: general health and the SCORE risk profile for cardiovascular events.

The sensitivity analysis will start with the analysis of the effectiveness of the intervention in specific subgroups, most notably those workers with a low physical activity level, with a low intake of vegetables and fruit, and with a high body mass index. A sensitivity analysis will also be performed on the individual cost-effectiveness ratios by means of bootstrapping. This part of the sensitivity analysis will be used to determine the minimum level of effectiveness required to make the Health Portal more cost-effective than the standard WHPP.

The assumptions for the power calculations were: an intra-cluster correlation of 0.05 (as observed in a previous cluster RCT in companies [39]), an average of 20 workers per cluster, a power of 80%, and a level of significance of 5% (one-sided). Under these assumptions, we anticipate to be able to detect a difference of at least 12% in prevalence between intervention and control group (e.g. primary outcome measure 30% compliance with the recommendation for PA up to 42%) with 350 workers with completed questionnaires assigned to the intervention. Without a noticeable intra-cluster variance the detectable difference will increase by 9%. With an initial participation of 70% and loss-to-follow-up of 30%, the cluster RCT should invite 2*700 workers.

An intention-to-treat analysis will be used with last available information carried forward to missing data in subsequent measurements. A multilevel linear regression model with repeated measurements will be used (SAS proc Mixed) for continuous outcomes and a hierarchical logistic regression (SAS proc Genmod) for dichotomous outcomes. The effect of self-efficacy, environmental determinants, Portal use, and drop-out during follow-up on primary and secondary outcome measures will be evaluated for their potentially differential effects.

Although the discriminatory power will be limited, an analysis will be carried out as to which subgroups participate in the WHPP and the Health Portal and which subgroups have the best adherence to the lifestyle recommendations (age, sex, education, ethnicity).

In addition, preliminary analyses will be performed for workers with cardiovascular complaints and workers with obesity in order to evaluate whether these subgroups are more or less amendable to changing their lifestyle.