Using action research to improve health and the work environment for 3500 municipal bus drivers more |
30 views |
Educational Action Research Vol. 15, No. 1, March 2007, pp. 75–106
Using action research to improve health and the work environment for 3500 municipal bus drivers
Kjeld B. Poulsena*, Søren H. Jensena, Elsa Bacha and John F. Schostakb
aThe
National Institute of Occupational Health, Denmark; University, UK
bManchester
Metropolitan
kbpoulsen@hotmail.com KjeldPoulsen 0 100000March 2007 15 2007 & Francis Original Article 0965-0792 Action Research EducationalFrancis Ltd 10.1080/09650790601151228 REAC_A_215050.sgm Taylor and (print)/1747-5074 (online)
During the past five decades occupational researchers have documented that bus drivers’ health is worse than in almost any other profession. The authors suggest that the reason there has not been any successful attempt to change this situation is because the focus until now on removing statistically associated external risk factors has been too narrow. The article describes a project whose purpose was to improve the health and well-being of 3500 Copenhagen bus drivers. At the end, more than 200 interventions were implemented. The authors adopted a new approach of combining epidemiological results and qualitative methodologies, creating a broader explanatory foundation for action, linked by repetitive processes of critical reflection, which was central to defining problems, explaining causes, developing sufficiently effective interventions and measuring effects. The project revealed the importance of several new and potentially preventable factors involving such issues as lifestyle, private stressors and inappropriate management. During the project period an evaluative framework was developed to explore and measure the complex effects of multiple interventions. Three years after the interventions were launched, follow-ups revealed remarkable improvements such as reductions in stress and body pains, an increase in satisfaction, and improvements in management and the drivers’ cabin. This article is the story of a methodological journey, from classical epidemiology to an approach combining the strengths of survey (broad coverage), qualitative methods (in-depth focus) followed by critical reflections and ending with action research.
Keywords: Bus driver; Critical reflection; Combined effect measure model; Large-scale intervention
*Corresponding author. Head of Quality Department, Specialist in Occupational Medicine, Prins Buris Vej 6, DK-4000 Roskilde, Denmark. Email: kbpoulsen@hotmail.com ISSN 0965-0792 (print)/ISSN 1747-5074 (online)/07/010075–32 © 2007 Educational Action Research DOI: 10.1080/09650790601151228
76 K. B. Poulsen et al. Introduction The health problems—issues of causal explanation and prevention Over half a century ago, Morris and his employees discovered that London bus drivers had an increased risk of cardiovascular diseases when they were compared with conductors (Morris et al., 1953a, b, 1966; Heady et al., 1961). Subsequent studies have not just confirmed the ubiquity of cardiac morbidity among bus drivers (Rosengren et al., 1991; Netterstrøm & Suadicani, 1993; Gustavsson et al., 1996), but they also found increased cardiac mortality and hospitalisation rates (Netterstrøm & Juel, 1988; Michaels & Zoloth, 1991; Alfredsson et al., 1993; Tüchsen & Endahl, 1999). In addition, bus drivers have higher incidences of lung diseases, including lung cancer, hypertension, prolapsed vertebral disks, dyspepsia, other stomach problems, and muscle and back pain (Netterstrøm, 1988; Rafnsson & Gunnarsdóttir, 1991; Jensen et al., 1996; Magnusson et al., 1996; Krause et al., 1998b; Ragland et al., 1998; Soll-Johanning et al., 1998; Tüchsen & Hannerz, 2000; Soll-Johanning, 2003). Does this mean, therefore, that bus driving is the cause of health problems? Research shows that drivers are exposed to increased amounts of exhaust gas and diesel particles, which can cause both pulmonary cancer and ishaemic heart disease (Hansen et al., 1999; Praml & Schierl, 2000). Back pain could as well be explained by ergonomic factors, such as inappropriate design of the driver’s cabin and road bumps (Magnusson et al., 1996; Krause et al., 1998a; Lydolf, 1999). But occupational researchers have especially focused on the multitude of common stressors in the work environment, such as a tight time-schedule, constant alertness and being exposed to threats or even assaults. By elevating adrenal hormone levels in the blood, chronic stress can also cause cardiovascular diseases (Greiner et al., 1998; Meijman & Kompier, 1998). Despite abundant evidence of potentially being able to prevent the factors causing illness, only a few intervention surveys on bus drivers have been published. For example, a survey from Stockholm found a significant association between minimising hassles and a reduction in systolic blood pressure (Rydstedt et al., 1998). Stress could thus be explained as the consequence of incremental accumulation of hassles, such as delays caused by boarding of walking-impaired passengers, heavy traffic and controlling tickets. Earlier, a more comprehensive project experimented with the implementation of self-rule. In the beginning it was a huge success, but after one year resource constraints and internal problems started to occur. Finally, the project was unable to survive the introduction of outsourcing (Netterstrøm, 1999). There are only a few more intervention surveys published on bus drivers (Johanning et al., 1996; Aust et al., 1997) but these do not include general health or health promotion programmes. Recently, occupational researchers argue that a participative stepwise approach is, in general, a more effective method to implement and evaluate epidemiological interventions at company level (Israel et al., 1996, 1998; Moir & Buchholz, 1996; Toulmin & Gustavsen, 1996; Kompier et al., 1998; Levin, 2002). In epidemiology, the clue is to find statistically associated causal explanations for disease, which could then be removed to improve health. However, this only works if
Improving health and the work environment
77
there are few significant exposures, from which the driver could be effectively protected. In this perspective the driver is supposed to stay passive to the intervention, because somebody from the outside is acting on behalf of the driver. Second, even though many of the solutions, theoretically, might be highly effective in protecting the drivers from disease, they are unrealistic in the real world: for example, substantially longer breaks and elongated time-tables to unwind from stress or having co-drivers/ conductors on each bus. Even using a stepwise approach, the epidemiological approach will still be a long way from being an action research approach. What is missing is an understanding of the necessary political processes through which change can be encouraged and sustained. There is a need for reflection to be periodically interpolated between the combined quantitative–qualitative knowledge base and actions for change. With action research we find a method to implement social improvements in a democratic multidisciplinary and theoretical–practical dialectic way (Greenwood & Levin, 1998), having a deliberate and planned intent to solve a particular problem (McMahon, 1999). The purpose of the project was to improve the health of bus drivers, not merely describe the extent of the problem. This meant getting them involved and listening to their experiences in order to make change with them and not on them. Thus the project had to have a methodology that would ensure that all voices, not just the loudest and the most powerful, would be taken into account. Hence there was a concern for social justice. The approach, as it developed, was influenced by democratic methodologies (MacDonald, 1987; Schostak, 2002, 2006). In the beginning, the project team were not aware of all the methodological and political implications of this approach. It involved including people from different professions, different organisations and different positions in organisational hierarchies and bringing them into dialogue. This article provides some insights into the learning journey that was involved in integrating traditional epidemiological approaches with action research This approach was felt to be a radical one within the professional, social and political context of the researchers, the Bus companies and the trade unions. It was a major step methodologically and politically. Over time, as it became apparent that the improvements taking place were greater than in previous interventions, views were changed. Indeed, previous interventions had proven unsuccessful in reversing what was known from decades of research—that the health of bus drivers was seriously impaired by job conditions. Indeed, the project was supported because everybody seemed to agree that being a city bus driver was extremely stressful. The following sections present the design and three-year-follow-up results from what we believe to be the first successful large-scale intervention project to improve health, well-being and work environment among 3500 Copenhagen bus drivers. The research combined action-oriented interventions on a highly local, contextualised basis, with assessments of the magnitude of the effects of these interventions across local sites. Knowing that there is a consistent statistical association between being a bus driver and the increased risk of disease, together with plausible and preventable physiological and toxicological explanations, to support such a causal relationship is not enough
78 K. B. Poulsen et al. to bring about change. Thus we carried out a series of small exploratory stakeholder interviews with representatives from the unions, employers’ organisation, drivers, managers, Copenhagen Traffic and other researchers. This was also informed by personal experience of being a former bus driver (K.B.P.), and more generally as users of public transportation (McLaughlin, 2003). We considered that there must be relevant alternative causal explanations to these very common diseases and complaints (Rothman & Greenland, 1998). Some stakeholders believed it was due to there being more bus drivers than people in other occupations having a negative lifestyle. As an alternative to such explanations we can also ask: what are the consequences for the health profile of bus drivers that a substantial number were employed after long-term unemployment or that they were already harmed in earlier jobs? And what would be the importance of non-occupational causes of stress? For such reasons we argue interventions should also include the need to reduce major risk factors, which arise from outside the work environment. If not, bus drivers will not succeed in reducing their rates of morbidity and mortality compared with that of other Danish workers (Burr et al., 2002). A methodological journey At the very beginning, the idea was to make a classical epidemiological study: a largescale controlled trial involving randomly selecting companies to be exposed to one or more interventions, which were then to be compared with control companies without interventions. Statistically key outcome variables could be measured to see whether, for example, self-assessed health or stress improved more in the intervention companies compared with the controls (Clayton & Hills, 1993; Rothman & Greenland, 1998). This, for epidemiologists, would be the preferred method in this type of research setting. However, epidemiology has a problem handling very complex situations, where action research and complexity theory would have something new to offer (Phelps & Hase, 2002). The team believed that it would prove difficult to evaluate the exact exposure to the intervention, and that there would be a need to differentiate between many subcategories, such as gender, age groups, location, ethnicity, social background or lifestyle. Epidemiology has its origin and its strengths in finding causal associations between a simple exposure and a given disease in clinically controlled settings, but epidemiology faces difficulties when the focus is on how to evaluate the effects of multiple change processes in an everyday complex and messy work environment. To explore complexity, qualitative research processes are more appropriate (Bhaskar, 1975; Sayer, 1993; Schostak, 2002, 2006; Denzin & Lincoln, 2003a, b, c). An important difference between quantitative and qualitative approaches, methodologically, is how they relate to the issue of context. In epidemiology, statistical analysis requires the identification of variables and removing those that are not relevant to get a ‘pure’ measure of a specific factor influencing a given outcome. Qualitative approaches instead try to analyse how the interaction between the context and the problem could either facilitate or hamper the change process (Koelen et al., 2001).
Improving health and the work environment
79
However, from the team’s initial point of view, one major problem with many of the qualitative approaches is that they seemed extremely time-consuming. Secondly, as we wanted to improve health for the majority of bus drivers, we thought the weakness of qualitative methods was in the inability to measure the compound efficiency of several interventions on a broader scale, across different company cultures. However, the team learnt that the power of qualitative research derives from being able to map complex social realities from the accounts made by participants, gaining in-depth understandings of processes, identifying opportunities for change and development, empowering actors and engaging in dialogue. For example, some exploratory stakeholder interviews revealed that, among other things, previous research had given a rather homogeneous picture of the problems, leading to simplistic solutions, such as better seats, more frequent and longer breaks and less strained time schedules. Some stakeholders found this to be misleading, and that the situation was both more heterogeneous and dependent on specific contexts. Nevertheless, almost all stakeholders’ explanation models were at a very general and political level. For example, the two unions said that drivers’ health problems were caused by tendering out the bus service, insufficient financial resources and traffic problems due to the many private cars in the city. Their solution was therefore more money, removing the cars from the city and putting the companies back into public hands. However desirable, they could not be applied in current political realities. Unsurprisingly, the employers’ organisation and Copenhagen Traffic took the opposite view. They believed that because efficiency increases were much lower among bus drivers compared with the rest of the labour market during the past two decades, tendering had increased the service to the public while at the same time reducing the extraordinarily high costs. Since the result, from their point of view, has been very efficient, nobody now would expect a return to ‘the good old days’ (Jensen, 2003). The stakeholders also expressed the view that several important issues had never been investigated, such as lifestyle and health status at job entry. However, all stakeholders agreed that they would only participate if, this time, the project could provide sufficient and general improvements. The project ‘… must not end up as just another report’ (union spokesman). They would not accept a controlled design, because they didn’t believe in the existence of a so-called magic bullet solving all problems, and because all drivers needed improvements now—not just those from the intervention companies. Hence, it was important to include stakeholder experiences, and make the project with them, not on them. For these reasons, it seemed to us we needed a qualitative methodology to go into depth, find new explanations and explore contextual differences. At the same time, we considered we needed an epidemiological dimension to provide evidence of the prevalence of both these and earlier findings, and thus update the problem baseline. The epidemiological results combined with qualitative surveys might then be able to pinpoint special groups requiring further study. Thus both methodologies were to be part of the overall analytical framework through which the effects of the project could be analysed. Finally, a prerequisite for success, we believed, was a partnership, where all sides in the industry would take full ownership and responsibility for action to take
80 K. B. Poulsen et al. place. This is different from the partner being a critical friend in a one-to-one relationship (Robertson, 2000). Our partnerships were much more than one-to-one, with a local level including drivers, shop/safety stewards and managers, and a central level with companies, unions/employers and Copenhagen Traffic. And the researchers would have an obligation to work in partnerships with them all. Project design Ensuring coverage over four years of all Copenhagen bus drivers by the HealthyBus project meant integrating six different haulage contractors running a total of 20 garages. Large-scale changes in the industry in the recent past made it unlikely we could achieve sufficient stability to establish a randomised controlled design over such a long period. And time confirmed this view. Every year one-eighth of the total services were put out to tender, creating insecurity about where the lines would be garaged afterwards. There was an annual job turnover of 15%, and in the first three years we recorded both general and local strikes, bankruptcies, multinational takeovers, merging of companies and garages, implementation of driver certification and a huge number of project independent local changes. In short, all garages were influenced by major changes during the time of the project. It was therefore decided to construct a multiple case-based cohort design (Yin, 1994), where both results and processes are followed in all companies by means of a combination of quantitative and qualitative methods. Instead of comparing with a non-intervention (control) garage, a multiple case design collects evidence from mutual sources to document whether there is progress in each case. The more this total sum of evidence would support that intentional action could explain that improvements have taken place, the more confidence we could have in a causal effect of our combined and complex actions. Instead of looking for objective proof, we would be collecting circumstantial evidence. In the project, we drew on the following main sources of information before, during and after the interventions:
● ● ● ● ● ●
Epidemiological data from questionnaire studies (see below). In-depth qualitative interviews, focus groups and observations. Systematic activity plans for implementation of and follow-up on interventions. Logbooks on relevant activities at the workplaces. The project employees’ diaries of personal observations and informal information. Published material from newspapers, trade journals, garage journals, and so on.
To triangulate between the different sources of information, researchers from the following disciplines were employed or engaged in the project: occupational medicine, epidemiology, management, ethnography, engineering, communication, philosophy, applied educational research and health promotion experts. Validity could then be looked at as ‘… an integrated evaluative judgement of the degree to which empirical evidence and theoretical rationales support the adequacy and appropriateness of the inferences and actions based on test scores or other modes of assessment’ (Gray, 1997).
Improving health and the work environment
81
Politically, there was an increasingly supportive climate for change and the key stakeholders, defined as the stakeholders with political power to close the project (the highest level in the participating organisations), had a business interest in improving their reputation in relation to health and working conditions through an action research methodology (Toulmin & Gustavsen, 1996; Greenwood & Levin, 1998). Two drivers, two managers and one employee from Copenhagen Traffic were appointed and paid by the stakeholders to work half-time in the project. This was to ensure the trade’s commitment to and ownership of the project by enabling the representatives to act as the daily link between the research project and the trade. Together with the researchers they comprised the project group (PG). Both formal and informal training were given to all so that they would understand the basic principles of the different scientific and practical and quality management (Rothman & Thomas, 1994; Meredith & Mantel, 1995) approaches applied in the project. Implementation of the project The key steps in the project, shown in Figure 1, will be commented on in this section. Based on the exploratory stakeholder interviews, we asked all the drivers’ about their own judgement of health, work and living conditions, to see whether there were conflicts of perception between officially and individually experienced realities. Qualitative approaches were employed to provide the team with in-depth insights that went behind the stakeholder attitudes and statements. Epidemiological techniques were used to analyse the quantitative results from earlier surveys and develop a questionnaire based on the qualitative findings. In this way there was coverage of the individual views of all the drivers. Based on this evidence, the stakeholders would decide whether there was a need for action. Hence, in this way the qualitative and epidemiological approaches combined to inform decision-makers at key points during the course of the project, providing a mechanism for critical reflection. However, reflection mean thinking about the conditions for what one is doing, investigating the way in which the theoretical, cultural and political context of individual and intellectual involvement affects interaction with whatever is being researched (Alvesson & Skjöldberg, 2000, p. 245).
Figure 1. Flow chart showing the type and timing of major events in the project. Right-hand side, evaluation of how much epidemiology and qualitative research played a role as the conceptual scientific framework. The actual work was made in combination with researchers, drivers and managers employed and trained together. *, Recurrent activities; #, repetitive visits by foreign senior researcher; x, major influence; (x), minor influence; –, no important influence
Critical reflection Processes of critical reflection (CR) were central to the project design. First, we gathered some of the most competent and knowledgeable trade and research representatives to share their views at a meeting. The meeting was organised democratically in order to facilitate the communication of all views. The themes were always structured around key stages of the project to define the most important problems. Either prior to or at the beginning of the meeting, the participants were updated with excerpts from the relevant developing understandings and knowledge gained through the project. The reflexive process was intended to swiftly draw upon the participants’ immediate experience, insights, judgements and understandings in as efficient a
82 K. B. Poulsen et al.
Qualitative
Epidemiology
Quest ionnaire
Quest ionnaire
Quest ionnaire
Figure 1. Flow chart showing the type and timing of major events in the project. Right-hand side, evaluation of how much epidemiology and qualitative research played a role as the conceptual scientific framework. The actual work was made in combination with researchers, drivers and managers employed and trained together. *, Recurrent activities; #, repetitive visits by foreign senior researcher; x, major influence; (x), minor influence; –, no important influence
Improving health and the work environment
83
manner as possible, as when, for example, a correct diagnosis is made at a glance by a senior doctor. Similarly, drawing on the interpretations of experienced insiders provides a rich source of how they see and interpret their everyday world(s). In complex social circumstances, it provides the basis for what is seen in common and what is seen differently and disputed. Thus, the points of conflict and struggle are made visible. The possibilities for change depend on such disagreements being discussed openly in relation to the available evidence. That in turn depends upon there being a democratic underpinning to the project’s organisation enabling the discussion. CR aided by research-based evidence focusing on the conflicts and agreements enabled the identification of new approaches and points of agreement for action. These are the fundamental ideas behind the CR process. This approach was taken in seminars to discuss visions for future developments, and for plenary meetings, group works and 24-hour meetings: the vision seminars were extensive, the 24-hour meetings were very political, while plenary and group meetings could be short and narrow in focus. Sometimes meetings were audiotaped or videotaped and sometimes they were transcribed. Other times, only the important decisions were noted. Most of the times CR functioned as a catalyst, to initiate a longer process, ending with the presentation of, for example, an action plan. But it was always a process where practice meets theory, where the findings from research could be discussed in relation to experience in daily life in order to explore why things are related as they are, and what might realistically be the most effective preventive activity. Crucial phases in the CR process were those leading to a ‘stop or go’ decision. These were often in the context of serious problems where a decision regarding action was needed. In such circumstances the discussion focused on whether the group could agree on the circumstantial evidence behind these problems and whether the suggested interventions would be likely to prevent the problems. The CR process was, at times, very time consuming, as was the case in developing the questionnaire. Others were short, like the follow-ups for unproblematic action plans. Introducing critical reflection was thus a crucial step for a linkage between explanation and action. Gradually the team realised that they had reinvented (Rogers, 1995) action research as the most efficient organisational framework for complex change processes in daily practice. The CR process was clearly similar to the cycles of reflection typical in the action research literature. To us, it was thought-provoking that communication between scientific traditions is often so lacking that we do not learn from each other. From this point on, we found action research methodology to be a very inspiring paradigm to nourish the constant interplay between quantitative and qualitative methodologies and between theory and practice. Achieving a decisive foundation for interventions The first assignment of the PG was to explore how to develop an updated questionnaire that was sufficiently detailed, action-oriented and tailored to Copenhagen bus drivers (Figure 1, point 2). At the beginning of our journey, we did not use the
84 K. B. Poulsen et al. terminology reflection, but simply felt it imperative to use all relevant knowledge in the trade and not only what was documented by research. However, it turned out to be an advantage to structure such periodically repetitive processes of mutual exchange of thought and experience. By including an increasing number of research areas, such as ethnology, management, communication and action research, we realised that there was a new land of theories to be used and that reflection was also critical for the exchange of scientific experience and being a potential for developing new theory (Mellor, 2001). And by creating continued dialectical theory building (Robertson, 2000), where theory interacts with practice creating new theory (Greenwood & Levin, 1998; Watts & Jones, 2000), we would make a change project into action research project. The PG started one of the most central CR processes with a series of future/ brainstorming seminars (French & Bell, 1995), resulting in hundreds of useful possibilities for survey questions. The analysis of interviews was used to challenge different views and, after a series of working group sessions, a questionnaire was pilot-tested, revised and feasibility-checked (an English version of the final baseline questionnaire is available online: www.ami.dk/upload/SUNDBUS-spskema1-UK.pdf). Figure 1, point 3. When the questionnaire was ready, representatives from the trade, the haulage contractors and all the garages were invited to listen to a presentation of the project. At this meeting, a set of governing rules was presented for the project: business is to run as usual during the project period, the project is to be founded upon developing real partnership, multiple interventions will take place, there should be a focus on all stakeholder needs, it is the trade who is expert on management and, therefore, the trade takes full responsibility for action to take place. In addition, no extra money would be provided from project sources for project-based interventions, since all changes, if they are to be sustainable beyond the life of the project, must be effectively integrated into everyday practice. Figure 1, point 4. At the end of 1999, the baseline questionnaire survey was launched. A tremendous amount of work was done beforehand, especially by the trade representatives in the PG, to establish credibility for the project, and to secure a high response rate—as it turned out, an amazing 76%. The epidemiological results were presented mostly as easily interpretable graphs. Although there may have been many surprising results to each stakeholder, all the stakeholders from the trade said that they were able to recognise the core results as true. This, then, was a basis for agreement. Figure 1, point 5. To ensure feedback to all drivers, an eight-page newsletter was mailed directly home to every driver who had participated. Further newsletters were published regularly during the project. Figure 1, point 6. In April 2000 we arranged a 24-hour meeting, where the key stakeholders went through the results and agreed on a policy for a mutual project. It
Improving health and the work environment
85
had by then taken almost two years to create the potential for action (Poulsen, 2004), and it was only then that the key stakeholders really committed themselves to do the intervention part of the project. The key stakeholders agreed on a political coronation charter, stating that no single issue should be able to dominate the project. Therefore a four-leaf clover model was created, where each leaf illustrated the main intervention categories: workplace organisation, training, lifestyle and physical/psychological work environment. The stakeholder meeting agreed that if a garage had at least one intervention within each of the four leaves, it could call itself a spearhead garage. However, it further decided that lifestyle was not allowed to take up more than one-quarter of the intervention activities and must never be the only intervention. This meeting proved to be the most important CR, since it was here the intervention part of the project could have been stopped before it started. Compared with the initial CR around creating the conceptual framework for the development of the questionnaire, the latter CR was a very delicate political process, where the PG had much less influence on the outcome. However, the recognition of the high quality of the combined baseline process facilitated the key stakeholders’ decision of a GO!
Figure 1, point 7. In May 2000, shop and safety stewards from most of the garages met with their local and company managements at the National Institute of Occupational Health, to prioritise needs for action and prepare how to carry out that action in practice. In the majority of workplaces this triumvirate constituted the local action groups. They were provided with graphs of the results of the baseline survey—a booklet of around 200 pages. Here, results were presented at three levels: the Copenhagen area, each haulage contractor, and each garage. In group sessions representatives from garages sat together company-wise. First, they went through the results using a simple specifically developed two-page form where they were guided through prioritising their problems. Based upon the desire to create the conditions for quality management, a second form was used to help them to formulate causal explanations and actions to be taken, name the person to be responsible, and detail the milestones, budgets, processes and result measures. After this one-day facilitation seminar, they went home and continued the process. In addition, the project had developed a one-page action plan registration form to be used for the description of the specific interventions and for follow-up purposes (Figure 2).
Figure 2. Based on quality management principles, an action plan registration form was developed to help describe and follow the process of intervention
Figure 1, point 8. At a ministerial conference, with the participation of the ministers of work and traffic, in September 2000, the companies and Copenhagen Traffic presented a comprehensive intervention catalogue that was the result of the previous CR and homework processes. Thus, this was the day the key stakeholders committed themselves publicly. Figure 3 shows the titles of the 116 interventions presented as the four-leaf-clover intervention catalogue.
86 K. B. Poulsen et al.
Figure 2.
Based on quality management principles, an action plan registration form was developed to help describe and follow the process of intervention
Figure 3.
The initial four-leaf-clover intervention catalogue (September 2000)
Improving health and the work environment
87
88 K. B. Poulsen et al. Figure 1, point 9. Several new ideas for action emerged during the project period and, at the end, 208 specified interventions were either commenced or were still running. They were distributed between 75 interventions to improve organisation at work, 36 lifestyle interventions, 69 interventions to improve the competences of drivers and management, and 28 focused on the physical/psychological work environment. Of course, many of the interventions were alike, but to succeed they had to be interpreted differently in order to integrate into local settings. Also many of the interventions were overlapping in scope, such as management education, which for example might have an influence on the improvement of communication, garage management, duty rosters and timetables at the same time.
Figure 3. The initial four-leaf-clover intervention catalogue (September 2000)
Figure 1, point 10. As mentioned above, observations, interviews, focus groups and reflections were undertaken throughout the whole project period whenever feasible and reasonable. Feed-back and follow-up surveys Figure 1, point 11. All action plans were updated quarterly, either by a PG member paying a visit to the local action groups or having a telephone meeting with them. The project kept a track record of all changes in the plans, and each intervention got a progress score between 0 (turned down), 1 (active but delayed), 2 (just in time) and 3 (activity completed). All participants got a short comprehensive quarterly report stating purpose, specific action, score, contact person and comments for each intervention. The reports were followed by a half-day plenary meeting where all stakeholders, action groups and researchers could exchange information and experiences between companies (Robertson, 2000). However, this last part turned out to be a very slow process because of competing agendas, except at four major spearhead garages from two contractors, where a remarkable collaboration was established quite early. This is an example of a not dramatic, but sometimes rather effective repetitive CR process. Figure 1, point 12. Precisely two years after the baseline survey, a questionnaire follow-up survey was carried out in exactly the same way. This time the response rate was 72%. Although many of the interventions had run only for a maximum of one year, several improvements were already observable, such as bus maintenance, training, general work environment and retention rates. Importantly, both practitioners and researchers learned from it. Practitioners got a more complex perception of the situation, and the researchers also learned that they needed simultaneously to incorporate theories from other scientific fields and thus create the necessary holistic insight required for action research. Figure 1, point 13. On their own initiative the key stakeholders therefore made a joint press release stating that the project was already a success. This was the first time for
Improving health and the work environment
89
decades that the unions, employers and Copenhagen Traffic combined to agree such a mutually positive statement. They stated that a new collaborative culture had been founded:
The HealthyBus project has already established persuasive documentation for this development [and] … these garages are not just producing improvements in the work environment, but at the same time they exhibit improved economic results alongside figures that reveal that passengers give high scores to reputation and customer satisfaction which translate into high quality ratings overall. [The trade] … has adopted this conclusion … [and] … are committed to disseminate the experiences from the four [spearhead] garages to the remaining garages in the Greater Copenhagen area and later to the rest of the country.
However, the researchers were still sceptical and although it looked like a turning point there was still a long way to go before success could be claimed. Nevertheless, this was a significant practical and political outcome following the CR cycle of the Healthy Bus project. A clear signal had been given to the trade as a whole by a key stakeholder that, in future, the tendering process will include an increased emphasis on improvements in health and well-being. Figure 1, point 14. An external senior researcher with experience in project monitoring was hired to evaluate whether the epidemiological results were supported by qualitative investigations. He interviewed the PG, drivers, managements, shop and safety stewards and representatives from the unions, employers and Copenhagen Traffic. Observations were made at garages, during the project work and during seminars. The material was transcribed and followed by dilemma (Winter, 1982) and discourse analyses (Torfing, 1999) primarily based on action research theory and grounded theory (Glaser & Strauss, 1980). The method of using democratic participation and establishing ownership was studied to see whether it was successful and whether the method had an influence on the initiation of health and the work environment promoting initiatives. The evaluation also contributed insight into the type of qualities an adviser needed in order to be able to offer intervention projects to companies. The evaluation, in short, played the role of monitoring the research and implementation activities, describing where the project was a success, where it had its problems, and proposing ideas for future challenges with advice and guidelines for further work. Figure 1, point 15. The second follow-up survey was made four years after the baseline, just a few months before the closedown of the project in March 2004. The amazing response rate of 76% confirms a significant engagement and supports the conclusion that the project was held to be highly credible by the drivers (see the results section). A much more detailed description of the organisation, reflections and evaluation of the project, especially concentrating on the first years, is freely available in English on the Internet (Poulsen et al., 2005).
90 K. B. Poulsen et al. Results Explanations for bad health What was the bus drivers’ baseline exposure to risk factors and disease patterns? A total of 2677 bus drivers responded to the baseline questionnaire. The mean age was 45 years, and mean seniority in the job 10 years. Eighty-six per cent were male and 30% belonged to an ethnic minority group. Ten per cent assessed their general health as bad, which could have an impact in the future, since it is known to be a very reliable measure of upcoming morbidity and mortality (Heistaro et al., 2001). Muscular complaints within the previous two weeks were common, especially from the neck/ shoulder (54%) and from the back/loin (52%). Fatigue (55%) and headache (42%) were also common. Such results were typical and thus expected. However, the prevalence of heart disease was low (1%), probably explained by subsequent selection out of the job. Figure 4 provides baseline frequencies of exposure to potentially harmful factors. It shows that formerly documented factors, such as bad drivers’ cabin, tight time schedules, threats and road bumps (solid bars), were still very common. It also shows that it seems very relevant to get exposure to thermal risk factors back into focus (Figure 4, hatched bars). In Danish buses, the drivers’ cabin is not closed and there are no air-conditioning or antiviolence protective screens. It is also suggested that there may be new explanatory candidates for heart disease, such as bad management and disagreements/conflicts between colleagues. Nuisance factors in private life are relevant for the working life, since troubles do not respect any demarcation line between job and private life. More generally, lifestyle is very important to look at when discussing causes of heart disease (Johansson & Sundquist, 1999). However, in occupational research on bus drivers, the issue of lifestyle had not previously been in focus. A crude comparison of bus drivers with a representative sample of the Danish population (Kjøller & Rasmussen, 2002) shows in excess of 95% are heavy smokers, twice the amount are severely overweight and more than the double have a sedentary lifestyle. Do the working conditions encourage this negative lifestyle? We asked whether changing their eating or smoking habits was more difficult because of the job. Respectively, 61% and 57% confirmed this point of view. Talking about problems is essential to resolving them, thus in the research strategy both survey and dialogue were combined. The benefits of the reflective dialogue between the quantitative and qualitative methods can be further illustrated. Occupational epidemiologists draw upon particular statistics to estimate the amount of, for example, heart disease that could be explained from exposure at work. To measure the risk attributable to driving it is typically presupposed that the new drivers’ health and social conditions at the point of being hired is similar to that of the average working population. But maybe it is not? Interviews with the manger of the drivers training school revealed that today around 25% of all new drivers are referred to be retrained as a consequence of long-term unemployment. Indeed, ‘All have a work history. Usually it is people’s second or third
Figurevariable; *, measure wasbaseline survey withproject; BMI, body mass index odds ratios are adjusted for age and gender. Solid bars, exposures well established from previous epidemiological surveys; hatched bars, exposures known to be common years back, but that have not been in focus in the most recent research; Open bars, frequencies for potential alternative explanatory exposures that have not been evaluated for bus drivers before. (A/O), Always or often exposed to the 4. Examples from the changed during the frequencies of exposures. The
Improving health and the work environment
91
Figure 4. Examples from the baseline survey with frequencies of exposures. Solid bars, exposures well established from previous epidemiological surveys; hatched bars, exposures known to be common years back, but that have not been in focus in the most recent research; Open bars, frequencies for potential alternative explanatory exposures that have not been evaluated for bus drivers before. (A/O), Always or often exposed to the variable; *, measure was changed during the project; BMI, body mass index
92 K. B. Poulsen et al. job’. From the baseline survey we also found that eight per cent had taken the job because of health problems in a prior job. As a reason for turning to driving as a job, a manager stated that ‘It is incredibly simple to drive a bus, actually like driving a bumper car, no gear to worry about and a bus swings easily so if the person’s body is worn down it is easy’. There might thus be an unrecognised relationship between the job and the physical–psychological fitness of the new drivers. The combined methodology has thus pointed out the possible significance of the drivers’ state of health at job entry, as well as also indicating potentially important indicators of heart disease from causes, such as lifestyle, social stress or problems due to former jobs. However, especially small qualitative surveys might have a problem with validity (Tricoglus, 2001), and we found that the impact of in-depth analysis on major changes would improve if it also had a high epidemiological validity. Secondly, do bus drivers have more illnesses and symptoms than other workers? We compared the bus drivers statistically with 5377 workers participating in the representative Danish Work Environment Cohort Study (Burr et al., 2002). We were able to make comparisons for selected exposures and a few resulting health parameters. Figures 4 and 5 are examples where epidemiology usefully provides descriptions of prevalence and prevalence ratios across large populations that help management and policy-makers to assess the severity and the extent to which a problem is widespread. Thirdly, did sufficient improvements happen during the project period? Again, we used a multiple logistic regression analysis to calculate the change between baseline and follow-up as an odds ratio, which is this time adjusted for age, gender and ethnicity. Analysis not shown here made it obvious that there were no important differences if only drivers employed for all four years are included or if the total population was used in the analysis. Thus, to increase the statistical strength we compare the 2677 respondents from the baseline survey with the 2729 respondents from the second follow-up survey four years later. Table 1 shows that, at baseline, 28% of the bus drivers assessed themselves as always or often stressed. This was reduced to 19% four years later. The odds ratio indicates a 37% reduction in stress. The measure is quite reliable because the confidence interval is narrow around the odds ratio, and as the interval does not overlap 1, it is also statistically significant. To obtain a more reliable and differentiated picture of stress we used a somatic and cognitive stress index. Both indices consists of four questions and were calculated using a Danish reference method where each response was evenly categorised between 0 and 100 points, and the stress score was then calculated as the mean of the four questions (Kristensen et al., 2002). If the driver scored more than 33 points, they were categorised as being stressed. Vitality was calculated in the same way, where a score below 45% was selected to indicate a low vitality, which is negative. Incredibly, on average, it appears all bus drivers in Copenhagen had a remarkable reduction in stress during the project period. A crude comparison with the Danish reference database on stress showed that bus drivers were now much below the average figures of the reference database. Most of the physiological symptoms traditionally associated with bus drivers were also reduced. At the same time we recorded an increase from 41% to 48% of drivers who
Figure 5. Comparison of bus drivers with a representative cohort of Danish workers. The graph shows am=n excess prevalence of bus drivers’ exposure and health complaints with 95% confidence limits. The odds ratios are adjusted for age, gender and ethnicity
Improving health and the work environment
93
Figure 5. Comparison of bus drivers with a representative cohort of Danish workers. The graph shows excess prevalence of bus drivers’ exposure and health complaints with 95% confidence limits. The odds ratios are adjusted for age, gender and ethnicity
felt that their health had improved (data not shown). However, it is remarkable that at the same time we found an increase in drivers who had negative expectations of future health status. Looking at the results, there were large improvements for what were among the key candidates for action when we started. From triangulation with the qualitative part of the project, we find it very likely that the improvements are a result of the mutual efforts by all parties organised by the HealthyBus project. We are not able to follow up on exposure to heat, noise and vibration. They were excluded in the second follow-up questionnaire because we needed new questions to reflect other issues that emerged during the project period, such as the importance of driver certification and the introduction of floating time tables during rush hours. Is action research a facilitating vehicle for the process? It was the combination of a multiple case-study design, followed by periodically critical reflections resulting in an action research methodology that made the difference
94 K. B. Poulsen et al.
Table 1. Changes in selected exposures, diseases and symptoms from the baseline survey in 1999/2000 to the second follow-up survey in 2003/04 Baseline (%) Symptoms and health Psychological symptoms Self-/assessed stress Somatic stress index positive Cognitive stress index positive Vitality reduced Physical and general health Self-/assessed health scored bad Pain in shoulder/neck Pain in the back/loin Headache Fatigue Stomach pains Exposure Psychosocial Rush hours time table too tight Cannot take full length of break Think of violence and threats Poor well-being at job Only negative feedback Management creates insecurity Managers do not treat bus drivers well Considered to change employer Physical Drivers’ cabin is bad Lifestyle Daily smokers Body mass index ≥ 30 Exercise > 3.5 hours per week Eating habits regarded as healthy Follow-up (%) Odds ratio 95% confidence limit
28 18 17 23 10 54 52 35 55 19
19 8 7 11 12 41 43 32 40 12
0.63 0.69 0.65 0.70 1.17 0.81 0.87 0.96 0.75 0.77
0.53–0.74 0.62–0.76 0.58–0.72 0.64–0.77 1.06–1.29 0.76–0.87 0.81–0.93 0.90–1.04 0.70–0.81 0.70–0.85
72 43 23 43 54 36 29 40 33 50 19 20 53
43 30 17 20 39 17 12 17 20 45 21 33 66
0.54 0.79 0.82 0.62 0.77 0.66 0.65 0.55 0.72 0.86 1.10 1.45 1.31
0.50–0.57 0.74–0.85 0.76–0.90 0.57–0.68 0.72–0.82 0.61–0.72 0.59–0.71 0.51–0.59 0.69–0.78 0.80–0.92 1.02–1.20 1.34–1.58 1.22–1.40
Note: Odds ratios and 95% confidence limits are shown adjusted for gender, age and ethnicity. Figures in bold are statistically significant.
between success and failure. It provided the necessary sensitivity to the politics involved in making changes. It is central to the research process as described above, where the keywords embodied in the action research process were: democratic dialogue, mutual respect, and equal partnership. It required enough time, openness to new possibilities, multi-paradigmatic and cross-professional approaches where practitioners work together with researchers in a critical and reflexive way, the right
Improving health and the work environment
95
people at the right place and time, flexibility—and a degree of luck! At least we can say that our luck was enhanced through an action research approach. It is also important to realise, drawing on existentialism (Feldman, 2002), that the project would have been totally different if it had been manned with other persons than us, even if they had had exactly the same scientific backgrounds. And we can confirm the project’s dependency on how well each individual contributed to the action research group (Dadds, 2003). We therefore consider it quite evident that we would not have succeeded with a traditional epidemiological design. This is because, right from the outset, the managers in the bus companies made it explicit that they would only participate if real improvement could be brought about through the research. Real improvement meant facing the politics. The trade unions were sceptical following a decade of conflicts due to issues stemming from outsourcing. There were sufficient agreements about the ‘causes’ of the problems for them to collaborate. However, where the trade unions saw the sources of the health problems to be in working environment, management saw it in lifestyle. After the first follow-up report, both sides realised that small but important changes in the areas they agreed upon (such as bad management, bad equipment, defective buses) were being made through the cooperation of employees, employers and the contractors, Copenhagen Traffic. From this point they became more positive about the methodology, gradually accepting that it was possible for the different stakeholders to work together. It paved the way for looking at the wider issues of lifestyle. The external evaluator for the innovation agreed that:
● ●
● ● ●
● ● ● ● ● ●
HealthyBus established a participative structure that worked well. We made the main parties interested in cooperation as a long-term and regular development process. We changed the participants’ perceptions and conduct. HealthyBus made the participants believe in their ability to create change. The project established better internal dialogues between the drivers and the management/shop stewards. Focus changed from blame to common understanding of the problems. An effective dialogue structure was established. Processes of change became institutionalised. The project opened new ways of communicating. The industry was well briefed even though it was very demanding. It is better to describe and analyse the dilemma of the processes and choices than having a success/failure mentality.
From the above experience of combining statistical with qualitative and action research approaches, a conceptual model will next be outlined. A model for the exploration of complex interventions and the measurement of their effects In this section, a conceptual model to be used for measurement of effects of complex interventions will be sketched. The model was developed ad hoc during the life of the
96 K. B. Poulsen et al. project and is thus an example of the action research theory–practice–theory pathway (Greenwood & Levin, 1998). It is important to recognise that the model gives equal credit to ideas or theories independent of who makes them: practitioners or researchers (Dadds, 2003). The model is an empirical description of how the researcher– practitioner partnership got enacted (Mol, 2002). The team call it the Multiple Intervention Complex model (MIC). The single most important use of the model is to assist deciding what the most serious problems are. Only when this has been established can measurements be developed that are able to reflect the extent to which interventions are properly developed and efficiently implemented (Pfeffer & Salansic, 1978). Note that it is also important to be aware that all parties, including the researchers, have their own opinions, personality and emotions (Dadds, 2003) and that such opinions, if unreflectively assumed and not submitted to critical reflection, may lead to severe bias (Hawe, 1996). Similarly, not including alternative causal explanations of heart disease because too politically sensitive will also undermine the credibility of the research. Even so-called objective research draws upon subjective emotional elements in thinking, viewing the familiar differently, making and suspending judgements, being creative, drawing conclusions, taking action and working with others (McLaughlin, 2003). The MIC model is based on the assumption that doing intervention research in real-life settings is heterogeneous by nature, because normally there are multiple causal factors behind each of the problems (Qx) one wishes to change. As an example, let us take ‘driving a defective bus’ as the problem we wish to affect (Qa) (Figure 6). Even if we know that 34% say they drive a defective bus at least once a week, we do not know how they interpret the word ‘defective’. Thus when the haulage contractors decided that something had to be done to reduce this figure, it became quite evident that there were several possible causal explanations depending on which contractor or garage we looked at. Therefore local CR processes were used to explore and develop a series of analyses. For illustrative reasons we have chosen a hypothetical example where three garages prioritised this problem. Imagine that garage A suggests four explanations as to why they are driving in defective buses: (1) defects are not reported, (2) mechanics are not sufficiently trained, (3) there is a lack of mechanics and (4) the service intervals are too long. Formally, we use the following notation: CaA(1–4). In garage B they have only one explanation CaB(1), and garage C has seven explanations CaC(1–7). The total contextualised complex of causes from the three garages can thus be written as CaA(1–5), CaB(1), CaC(1–7)—or in general terms as CaA(1 – n) … CaX(1 – n) if there are X intervention localities. Based on these specific causal complexes, a new CR process is used to formulate the most optimal interventions that the participants, with all their mutual professional experience, believe would be sufficient to alleviate Qa. In the example of garage A, the CR resulted in two suggestions for interventions, IaA(1–2). If there are respectively four and three possible solutions for the remaining two other garages, then the three intervention complexes can be written as IaA(1–2), IaB(1–4), IaC(1–3)—or in general terms as IaA(1 – n) … IaX(1 – n).
Improving health and the work environment
97
As the world is in constant change, it is necessary to adjust the intervention complex regularly, especially during long-lasting intervention processes. Therefore systematic follow-ups must document the extent to which the intervention complex is actually carried out and ensure appropriate adjustments in order to make it work as planned (Poulsen, 2004; Poulsen et al., 2005). This was the purpose of the project’s quarterly feedback reports and meetings. In the example, imagine if new buses are bought, then the previous interventions aimed at preventing driving in a defective bus would probably be unnecessary overnight. In case studies, where it has been decided to do something about Qx, it is recommended that analyses are made for each level or case—such as the level of each garage (Yin, 1994)—it is possible to see whether the cases that used the complex intervention had better outcomes than those that did not initiate a specific intervention. This can be expressed: Qx(A … X) = f(IxA(1 – n) … IxX(1 – n)/Ix0), where Ix0 is the baseline against which the effect of the intervention can be measured. For each garage (unit of intervention), the formula could be used to find out whether they have improved by comparing themselves with the baseline (Ix0). The importance of this is that, by making the intervention complex the unit of analysis, one can use the entire epidemiological analysis system even on a mixed quality/quantity basis, because we now have a relevant interpretation of an intervention complex. Table 2 presents an actual example from the project of a milestone evaluation. Here the factors to be considered were the processes to be implemented and whether they would have the expected effect(s) at the time of the evaluation. Seven garages identified driving a defective bus as an important problem. The contextualised complex of causes contains two to five explanations, of which several were repetitions. The statistical analysis shows that the group of seven garages had reduced the problem from 36.7% to 29.8%, which is a statistically significant improvement. But the seven do still have a higher rate of problems compared with the remaining garages—these did not define this as a priority. So the analysis provided the information that, in total, the intervention complexes seem to be effective, but also that more could be done. As shown in Figure 6, the results could be accumulated into a common Grand Database as was done the HealthyBus project (Drewes & Poulsen, 2003).
Figure 6. The MIC model. See the text for a detailed explanation
But our task was to improve health Just as there were many different causes leading to a defective bus being driven, the same can be expected for such issues as bad health, fostering well-being or improving the work environment. These ‘causes’ may also be expected to be affected by differences in contexts. Figure 7 illustrate how the MIC model could be a help to formulate and optimise our efforts in surveys of large and/or complex interventions. In trade Z we see that the most important cause of bad health is smoking. But also time pressure, fear of getting fired, lack of exercise, nightshifts and exposures to chemicals and dusts are regarded as harmful. However, in trade W, time pressure and heavy lifts are the most important causes of health problems. Fear of getting
98 K. B. Poulsen et al.
Table 2. Changes in problem prevalence among drivers who answered both the baseline and the follow-up questionnaire in the HealthyBus intervention survey Baseline 1999 Intervention Defective bus > 1 time/week (% who had the problem) Intervention complexes from each of the seven different intervention garages 36.7 ‘Control’ 24.4 Follow-up 2001 Intervention 29.8 ‘Control’ 23.7 p value* 0.02
Round-the-clock work at the garage + faster renovation of old buses + new buses + priority of common problems Feedback from garage to drivers + new buses Improved communication between driver and garage + feedback from garage to drivers Improved fault reporting + communication Know your bus course + garage visits + more resource at the garage + feedback Meetings with focus on collaboration between driver and garage + bulletin board with technical information New buses + new drivers’ seats
*Wilcoxon rank-sum test of difference between baseline and follow-up.
fired, lack of exercise and smoking are also important, but much less so than in trade Z. Hence, different strategies are needed to improve health. There are no magic bullets, or one-size-fits-all strategies. For example, choosing ‘time pressure’ as a focus for intervention would lead to suboptimal results across the range of cases.
Figure 7. A generalised conceptual disease MIC model. See the text for a detailed explanation
Closing remarks With HealthyBus we wanted to have an effective methodology for research on how to make changes in the work environment. This involved a combination of qualitative (to gain in-depth insights into processes and structures), epidemiological approaches (to explore issues of coverage across large populations) and practical intervention through the research cycles of action research, in order to be as close to the real-life situation faced every day by the companies as possible and practical. One of the most positive personal experiences in the bus project was that of applying a cross-disciplinary focus on facilitating the participants’ ability to break boundaries (Stacey, 2003; Smith, 2004): for example, when local shop stewards hold positions in radical opposition to their own union. The project, simultaneously, had to deal with a complex and ambiguous set of problems at the organisational, intra-personal and interpersonal levels. Intervention research has been described as a ‘very complex hodgepodge’ (Campbell, 1986). In HealthyBus the major challenge was transforming the hodgepodge into a practical synthesis involving both practitioners and academics. In practical intervention research:
Improving health and the work environment
99
Figure 6.
The MIC model. See the text for a detailed explanation
100 K. B. Poulsen et al.
Figure 7.
A generalised conceptual disease MIC model. See the text for a detailed explanation
Improving health and the work environment
the socially responsible researcher engages in a process of best fitting the available methods to the various questions and purposes under examination, always being mindful of the unintended consequences of actions, but refusing to be paralysed by indecision and inaction, since knowledge and justice arise not from contemplation in the abstract but from simultaneous critical engagement with both theory and human life. (Burton & Kagan, 1998)
101
This is why it is so important to intersect critical reflections between the needs assessments and action potential assessments (Poulsen, 2004) and the change processes— and thus turn it into an action research project. The concluding remarks are in three parts. Firstly, the research showed there are still substantial health problems among bus drivers. Bus drivers have an increased risk of both negative health symptoms and exposure to harmful factors compared with those of other workers. Perhaps more interestingly, the research showed the existence of important additional factors from both work and private life relevant to most of the diseases and health complaints among bus drivers, because many of these are potentially preventable. This calls for renewed optimism, because it gives a much larger palette from which to choose interventions and thus a much higher chance of practical and relevant effects. Indeed, after the three years where HealthyBus interventions have been put into effect, there is evidence of extensive improvements in health-related parameters and widespread reductions in exposure to harmful factors. In addition, 45% of the bus drivers said that the work environment had improved during the final two years of the project. In the context of occupational intervention research, this is almost a revolutionary breakthrough. The combined action research method was able to reduce stress by more than one-third for 3500 bus drivers. Furthermore, to a great extent this has only happened because of the focus on changing factors such as management style and communication. And it was unpredictable what would work in each context. Comprehensive organisational change cannot be either forecasted or replicated in detail (Phelps & Hase, 2002), but based on the experience from HealthyBus we would argue that action research is a great help to contextualise deliberate change processes in a complex and chaotic world. Secondly, the team’s experience and evidence suggests action research has facilitated change. Indeed, besides our statistical analysis, the key stakeholders, the external evaluator and the Danish Working Environment Authorities have also claimed that the project has been effective. By making a combined quantitative and qualitative approach, we learned that there are many stories that have to be told before changes can be made in a new way. Because actions are related to the participants’ personal experience followed by reflection, a new way is needed that can generate positive transformations from former negative experience (McMahon, 1999). A central feature of the project is the process of critical reflection that should systematically be done at all crucial stages of a project: the initial problem formulation, the causal explanatory phase, the intervention formulation as well as during the process and outcome evaluations. Importantly, our methodological journey not only taught the practitioners how to get more
102 K. B. Poulsen et al. influence on their own lives, it also changed the researchers way of working, learning about researching our own research (Mellor, 2001) and thus validating ourselves (Tricoglus, 2001). Thirdly, by suggesting a conceptual model for framing complex interventions in a real-life setting, the research contributes to the expanding field of mixed methods evaluation (Chen, 1990, 2005; Dahler-Larsen, 1999; Pawson & Tilley, 2004; Stame, 2004). Thus the MIC model may increase the relevance of practical epidemiological and qualitative intervention research because it seems to comply well with the demands of an approach that includes both depth and coverage in evaluation at the same time (Schostak & Phillips, 1997; Kaufman & Cooper, 1999). Epidemiologists needed to go more into depth, but it also became very clear that qualitative researchers would benefit from epidemiology. We can only agree that it is time for qualitative inquirers to consider a rapprochement with experimental, measurement and verificational types of science and scientists (Smith, 2004). Finally, it is important that there is the right mix of very experienced researchers and practitioners who participate in the CR processes, because it is here that the important decisions are taken (Edmondsom, 1966; Yin, 1994; Schulte et al., 1999; Worren et al., 2002). Such applied researchers need to be able to manage the politics of change in the working contexts of everyday life. In these circumstances, what counts as knowledge is continually challenged by the interests and multiple agendas of participants, as between, say, the employers and the trade unions, or between epidemiologists and qualitative researchers. However, embedding real changes requires the maintenance of critical dialogue between all participants through which the practical bases of their disagreements can be explored and challenged by evidence. The degree of success that the HealthyBus project achieved was in large measure due to the maintenance of such dialogue. Acknowledgements The HealthyBus Project was funded by the National Board of Health (grant number 9801326) and the Danish Ministry of Work (grant number 204–0003). The authors wish to acknowledge the substantial contribution made by Terry Phillips (deceased), Centre for Applied Research in Education, University of East Anglia, Norwich, who was the external evaluator. References
Alfredsson, L., Hammer N. & Hogstedt, C. (1993) Incidence of myocardial infarction and morbidity from specific causes among bus drivers in Sweden, International Journal of Epidemiology, 22(1), 57–61. Alvesson, M. & Skjöldberg, K. (2000) Reflexive methodology. New visitas for qualitative research (London, SAGE Publications). Aust, B., Peter, R. & Siegrist, J. (1997) Stress management in bus drivers: a pilot study based on the model of effort-reward imbalance, International Journal of Stress Management, 4(4), 297–305. Bhaskar, R. (1975) A realist theory of science (Hassocks, Sussex, Harvester Press).
Improving health and the work environment
103
Burr, H., Bach, E., Borg, V. & Villadsen, E. (2002) Arbejdsmiljø i Danmark 2000 [The Danish Work Environment 2002] (1st edn) (Copenhagen, AMI). Burton, M. & Kagan, C. (1998) Complemantarism versus incommensurability in psychological reasearch methodology. Available online at: http://homepages.poptel.org.uk/mark.burton/ QUALPAR3.PDF (accessed 7 May 2006). Campbell, D.T. (1986) Relabeling internal and external validity for applied social scientists, in: W. M. K. Trochim (Ed.) Advances in quasi-experimental design and analysis (San Fransisco, JosseyBass Inc. Publishers). Chen, H.-T. (1990) Theory-driven evaluations (Newbury Park, Sage Publications). Chen, H.-T. (2005) Practical program evaluation. Assessing and improving planning, implementation and effectiveness (London, Sage Publications). Clayton, D. & Hills, M. (1993) Statistical models in epidemiology (New York, Oxford University Press). Dadds, M. (2003) Dissidence, difference and diversity in action research, Educational Action Research, 11(2), 265–282. Dahler-Larsen, P. (1999) Den rituelle reflektion [The ritual reflection] (Odense, Odense Universitetsforlag). Denzin, N. K. & Lincoln, Y. S. (2003a) Collecting and interpreting qualitative materials (London, Sage Publications). Denzin, N. K. & Lincoln, Y. S. (2003b) Strategies of qualitative inquiry (London, Sage Publications). Denzin, N. K. & Lincoln, Y. S. (2003c) The landscape of qualitative research: theories and issues (London, Sage Publications). Drewes, K. & Poulsen, K. B. (2003) Gør noget ved dit arbejdsmiljø. 64 eksempler som kan inspirere til at forbedre chaufførers sundhed og arbejdsmiljø [Improve your work environment. 64 examples to inspire to, an improvement of drivers health and work environment] (København, Arbejdsmiljøinstituttet). Edmondsom, A. C. (1966) Three factors of Eden: the persistence of competing theories and multiple diagnoses in organizational intervention research, Human Relations, 49, 571–595. Feldman, A. (2002) Existential approaches to action research, Educational Action Research, 10(2), 233–251. French, W. L. & Bell, C. H., Jr. (1995) Comprehensive interventions, in: W. L. French & C. H. Bell, Jr. (Eds) Organization development (Englewood Cliffs, NJ, Prentice Hall International). Glaser, B. G. & Strauss, A. L. (1980) The discovery of grounded theory: strategy for qualitative research (New York, Aldline Publishing). Gray, B. T. (1997) Controversies regarding the nature of score validity: still crazy after all these years. Available online at: http://eric.ed.gov/ERICDocs/data/ericdocs2/content_storage_01/ 0000000b/80/26/47/cc.pdf (accessed 7 May 2006). Greenwood, D. J. & Levin, M. (1998) Introduction to action research. Social research for social change (London, SAGE Publications). Greiner, B. A., Ragland, D. R., Krause, N., Syme, S. L. & Fisher, J. M. (1998) Objective measurement of occupational stress factors-an example with San Francisco urban transit operators, Journal of Occupational Health and Psychology, 2, 325–342. Gustavsson, P., Alfredsson, L., Brunnberg, H., Hammer, N., Jacobsson, R., Reuterwall, C. & Östlin, P. (1996) Myocardial infarction among male bus, taxi, and lorry drivers in middle Sweden, Occupational and Environmental Medicine, 53, 235–240. Hansen, Å. M., Wallin, H., Willhardt, P. & Knudsen, L. E. (1999) Monitoring of urban air exposure in Danish bus drivers and mail carriers (Copenhagen, Arbejdsmiljøinstituttet). Hawe, P. (1996) Needs assessment must become more change-focused, Australian and New Zeeland Journal of Public Health, 20, 473–478. Heady, J. A., Morris, J. N., Kagan, A. & Raffle, P. A. B. (1961) Coronary heart disease in London busmen: a progress report with particular reference to physique, British Journal of Preventive Social Medicine, 15, 143–153.
104 K. B. Poulsen et al.
Heistaro, S., Jousilahti, P., Lahelma, E., Vartiainen, E. & Puska, P. (2001) Self rated health and mortality: a long term prospective study in eastern Finland, Journal of Epidemiology and Community Health, 55, 227–232. Israel, B. A., Baker, E. A., Goldenhar, L. M., Heaney, C. A. & Schurman, S. J. (1996) Occupational stress, safety, and health: conceptual framework and principles for effective prevention interventions, Journal of Occupational Health Psychology, 1(3), 261–286. Israel, B. A., Schulz, A. J., Parker, E. A. & Becker, A. B. (1998) Review of community-based research: assessing partnership approaches to improve public health, Annual Review of Public Health, 19, 173–202. Jensen, M. V., Tüchsen, F. & Ørhede, E. (1996) Prolapsed cervical intervertebral disc in male professional drivers in Denmark 1991–1990. A longitudinal study of hospitalizations, Spine, 21, 2352–2355. Jensen, S. H. (2003) Arbejdsmiljø som organiserende koncept -en sund strategi for busbranchen? [The work environment as the organising concept—a healthy strategy for the bus trade?] (Copenhagen, Copenhagen Business School). Johanning, E., Lansbergis, P., Geissler, H. & Karazmann, R. (1996) Cardiovascular risk and backdisorder intervention study of mass transit operators, International Journal of Occupational Medicine and Environmental Health, 2, 79–87. Johansson, S.-E. & Sundquist, J. (1999) Change of lifestyle factors and their influence on health status and all-cause mortality, International Journal of Epidemiology, 28, 1973–1980. Kaufman, J. S. & Cooper, R. S. (1999) Seeking causal explanations in social epidemiology, American Journal of Epidemiology, 150(2), 113–120. Kjøller, M. & Rasmussen, N. K. (2002) Sundhed og sygelighed i Danmark 2000 … & udviklingen siden 1987 [Danish Health and Morbidity Survey … & trends since 1987] (Copenhagen, Statens Institut for Folkesundhed). Koelen, M. A., Vaandrager, L. & Colomer, C. (2001) Health promotion research: dilemmas and challenges, Journal of Epidemiology and Community Health, 55, 257–262. Kompier, M. A., Geurts, S. A. E., Gründemann, R. W. M., Vink, P. & Smulders, P. G. W. (1998) Cases in stress prevention: the success of participative and stepwise approach, Stress Medicine, 14, 155–168. Krause, N., Ragland, D. R., Fisher, J. M. & Syme, S. L. (1998a) 1998 Volvo award winner in clinical studies, Spine, 23, 2507–2516. Krause, N., Ragland, D. R., Greiner, B. A., Syme, S. L. & Fisher, J. M. (1998b) Psychosocial job factors associated with back and neck pain in public transit operators, Scandinavian Journal of Work Environment and Health, 23, 179–186. Kristensen, T. S., Borg, W. & Hannerz, H. (2002) Socioeconomic status and psychosocial work environment: results from a Danish national study, Scandinavian Journal of Public Health, 30, 41–48. Levin, M. (2002) Researching enterprise development. Action research on the cooperation between management and labour in Norway (Amsterdam, John Benjamin Publishing Company). Lydolf, M. (1999) Undersøgelse af vejbumps påvirkning på erhvervschauffører (Copenhagen, SID). MacDonald, B. (1987) Evaluation and the control of education, in: R. Murphy & H. Torrance (Eds) Evaluating education: issues and methods (London, Harper and Row). Magnusson, M. L., Pope, M. H., Wilder, D. G. & Areskog, B. (1996) Are occupational drivers at an increased risk for developing musculoskeletal disorders?, Spine, 21, 710–717. McLaughlin, C. (2003) The feeling of finding out: the role of emotions in research, Educational Action Research, 11(1), 65–77. McMahon, T. (1999) Is reflective practice synonymous with action research?, Educational Action Research, 7(1), 163–169. Meijman, T. F. and Kompier, M. A. (1998) Busy business: how urban bus drivers cope with time pressure, passengers, and traffic safety, Journal of Occupational Health and Psychology, 3, 109– 121.
Improving health and the work environment
105
Mellor, N. (2001) Messy methods: the unfolding story, Educational Action Research, 9(3), 465–484. Meredith, J. R. & Mantel, S. J. (1995) Project management. A managerial approach (New York, John Wiley & Sons Inc.). Michaels, D. & Zoloth, S. R. (1991) Mortality among urban bus drivers, International Journal of Epidemiology, 20, 399–404. Moir, S. & Buchholz, B. (1996) Emerging participatory approaches to ergonomic interventions in the construction industry, American Journal of Industrial Medicine, 29, 425–430. Mol, A. (2002) The body multiple: ontology in medical practice, (Durham, Duke University Press). Morris, J. N., Heady, J. A., Raffle, P. A. B., Roberts, C. G. & Parks, J. W. (1953a) Coronary heart disease and physical activity of work, Lancet, 2, 1111–1120. Morris, J. N., Heady, J. A., Raffle, P. A. B., Roberts, C. G. & Parks, J. W. (1953b) Coronary heart-disease and physical activity of work, Lancet, i, 1053–1057. Morris, J. N., Kegan, A., Pattison, D. C., Gardner, M. J. & Raffle, P. A. B. (1966) Incidence and prediction of ischaemic heart-disease in London busmen, Lancet, i, 553–559. Netterstrøm, B. (1999) Denmark: self-rule on route 166. An intervention study among bus drivers, in: M. Kompier & C. Cooper (Eds) Preventing stress, improving productivity: European case studies in the workplace (London, Routhledge). Netterstrøm, B. (1988) Cancer incidence among urban bus drivers in Denmark, International Archives of Occupational and Environmental Health, 61, 217–221. Netterstrøm, B. & Juel, K. (1988) Impact of work-related and psychosocial factors on the development of ischemic heart disease among urban bus drivers in Denmark, Scandinavian Journal of Work Environment and Health, 14, 231–238. Netterstrøm, B. & Suadicani, P. (1993) Self-assessed job satisfaction and ischaemic heart disease mortality: a 10-year follow-up of urban bus drivers, International Journal of Epidemiology, 22(1), 51–56. Pawson, R. & Tilley, N. (2004) Realistic evaluation (London, SAGE Publications Ltd). Phelps, R. & Hase, S. (2002) Complexity and action research: exploring the theoretical and methodological connections, Educational Action Research, 10(3), 507–523. Pfeffer, J. & Salansic, G. R. (1978) The external control of organizations: a resource dependence perspective (New York, Harper & Row). Poulsen, K. B. (2004) The Healthy Bus project in Denmark: need for an action potential assessment, Health Promotion International, 19(2), 205–213. Poulsen, K. B., Drewes, K., Grøn, S., Petersen, P. T. & Bach E. (2005) Reflections on interventions (Copenhagen, Arbejdsmiljøinstituttet). Available online at: www.ami.dk/upload/reflections.pdf (accessed 7 May 2006). Praml, G. & Schierl, R. (2000) Dust exposure in Munic public transportation: a comprehensive 4year survey in busses and trams, International Archives of Occupational and Environmental Health, 73, 209–214. Rafnsson, V. & Gunnarsdóttir, H. (1991) Mortality among professional drivers, Scandinavian Journal of Work Environment and Health, 17, 312–317. Ragland, D. R., Greiner, B. A., Holman, B. L. & Fisher, J. M. (1998) Hypertension and years of driving in transit vehicle operators, Scandinavian Journal of Social Medicine, 25, 271–279. Robertson, J. (2000) The three Rs of action research methodology: reciprocity, reflexivity and reflection-on-reality, Educational Action Research, 8(2), 307–326. Rogers, E. M. (1995) Diffusion of innovations (New York, The Free Press). Rosengren, A., Anderson, K. & Wilhelmsen, L. (1991) Risk of coronary heart disease in middleaged male bus and tram drivers compared to men in other occupations: a prospective study, International Journal of Epidemiology, 20(1), 82–87. Rothman, J. & Thomas, E. J. (1994) Intervention research. Design and development for human service (New York, The Haworth Press). Rothman, K. J. & Greenland, S. (1998) Modern epidemiology (Philadelphia, Lippincott-Raven).
106 K. B. Poulsen et al.
Rydstedt, L. W., Johansson, G. & Evans, G. W. (1998) The human side of the road: improving the working conditions of urban bus drivers, Journal of Occupational Health Psychology, 3, 161–171. Sayer, A. (1993) Method in social science. A realist approach (London & New York, Routledge). Schulte, P. A., Goldenhar, L. M. & Connally, L. B. (1999) Intervention research: science, skills, and strategies, American Journal of Industrial Medicine, 29, 285–288. Schostak, J. F. (2002) Understanding, designing and conducting qualitative research in education. Framing the project (Buckingham, Open University Press). Schostak, J. F. (2006) Interviewing and representation in qualitative research projects (Buckingham, Open University Press). Schostak, J. F. & Phillips, T. (1997) Evaluation of the assessment of practice at diploma, degree and postgraduate level in pre and post registration nursing and midwifery education. Proposal submitted to the English National Board for Nursing, Midwifery and Health Visiting, CARE, Norwich. Smith, L. M. (2004) Yesterday, today and tomorrow: reflections on action research and qualitative inquiry, Educational Action Research, 12(2), 175–195. Soll-Johanning, H. (2003) Luftforureningens betydning for kræft og andre luftvejssygdomme i to erhvervsgrupper, der på grund af deres arbejde er udsat for luftforurening [The importance of air pollution causing cancer and other airway diseases to happen in two occupations exposed to air pollution] (København, Arbejdsmiljøinstituttet). Soll-Johanning, H., Bach, E., Olsen, J. H. & Tüchsen, F. (1998) Cancer incidence in urban bus drivers and tramway employees: a retrospective cohort study, Occupational and Environmental Medicine, 55, 594–598. Stacey, R. D. (2003) Strategic management and organisational dynamics. The challenge of complexity (Harlow, FT Prentice Hall). Stame, N. (2004) Theory-based evaluation and types of complexity, Evaluation, 10(1), 58–76. Torfing, J. (1999) New theories of discourse. Laclau, Mouffe & Zizek (Oxford, Blackwell Publishers). Toulmin, S. & Gustavsen, B. (1996) Beyond theory: changing organizations through participation (Amsterdam, John Benjamins Publishing Company). Tricoglus, G. (2001) Living the theoretical principles of critical ethnography’ in educational research, Educational Action Research, 9(1), 135–148. Tüchsen, F. & Endahl, L. (1999) Increasing inequality in ischaemic heart disease morbidity among employed men in Denmark 1981–1993: the need for a new preventive policy, International Journal of Epidemiology, 28, 640–644. Tüchsen, F. & Hannerz, H. (2000) Social and occupational differences in chronic obstructive lung diseases in Denmark 1981–1993, American Journal of Industrial Medicine, 37, 300–306. Wang, P. D. & Lin, R. S. (2001) Coronary heart disease risk factors in urban bus drivers, Public Health, 115, 261–264. Watts, B. & Shirley, J. (2000) Inter-professional practice and action research: commonalities and parallels, Educational Action Research, 8(2), 377–382. Winter, R. (1982) Dilemma analysis: a contribution to methodology for action-research, Cambridge Journal of Education, 12(3), 161–174. Worren, N., Moore, K. & Elliot, R. (2002) When theories become tools: toward a framework for pragmatic validity, Human Relations, 55(10), 1227–1250. Yin, R. K. (1994) Case study research: design and methods (Thousands Oaks, CA, Sage).