Effects of a Dutch Work-Site Wellness-Health Program: The Brabantia Project Stan Maes, PhD, Chris Verhoeven, PhD, France Kittel, PhD, and Hetty Scholten, PhD Introduction The "Healthier Work at Brabantia" pro- ni~~'teject, named after a Dutch manufacturer of household goods, was inspired by American work-site wellness health projects."' The phi- losophy underlying the project, however, dif- fers from that of the typical American project. American programs focus almost exclu- sively on changes in lifestyle.' In the Braban- tia project, it was assumed that, above all, an improved working environment will enhance wellness and health. This basic philosophy is expressed in the Dutch working conditions law, which defines wellness at the work site in terms of explicit wellness conditions.3 The demand-control-social support model4 was used as a comerstone for the formulation of these conditions. As a means of operational- izing the text of the law, 7 operational well- ness conditions were defined5: completeness of the function, challenge, involvement in organizational tasks, autonomy, social con- nX. s tacts, cycle length, and information. Each of these conditions gives rise to a fundamental issue. For example, in tenns of V;i<.;!5completeness of the function, what are the proportions of time spent on the performance, preparatory, and supportive tasks entailed in a specific job? In this context, performance tasks are tasks that involve the mere execu- tion of actions (e.g., the actions necessary to assemble a specific product), preparatory tasks are tasks that precede performance tasks (e.g., obtaining the necessary raw material from the stock or collecting and preparing the necessary tools), and supportive tasks are tasks that support performance tasks (e.g., maintenance of tools and machines or quality control of the product). If a job consists of more than 95% performance tasks, this is ee;i'. considered to constitute a wellness risk. >^<-' Other issues are as follows: (1) Is the function/task suited to the individual's capaci- ties, or is it too difficult or too easy for the worker? (2) To what degree is the employee participating in the organization of his or her work? (3) Does the employee have enough autonomy concerning work rhythm, method, and sequence? (4) Can the employee establish and maintain a sufficient number of social contacts at the work site? (5) Does the job involve a sufficient number of tasks with cycles that are longer than 90 seconds? and (6) Does the employee have enough informa- tion concerning the objectives and outcome of the work? As additional wellness risks, lack of decision latitude (i.e., when problems occur) and ergonomic problems were considered. Inspired by the Dutch working condi- tions law, the project is concemed with the following question: Do combined interven- tions, directed at both lifestyle and the con- tent and organization of work, lead to (1) improved health behavior, (2) a reduction in health risks, (3) a reduction in general stress reactions, (4) improved quality of work, and (5) a reduction in absenteeism? Methods A quasi-experimental pretest/posttest control group design with repeated measures was used to evaluate the effects of the inter- ventions. Workers from 3 Dutch Brabantia sites participated in the study. Employees at 1 site constituted the experimental group, and employees at the other 2 sites formed the con- trol group. The experimental site and the larger control site were chosen at random; it was determined that the third site would be The authors are with the Section of Clinical and Health Psychology, Leiden University, Leiden, the Netherlands. Requests for reprints should be sent to Stan Maes, PhD, Section of Clinical and Health Psychology, Leiden University, PO Box 9555, 2300 RB Leiden, the Netherlands (email: poppel@rulf- sw.fsw.leidenuniv.nl). This paper was accepted January 15, 1998. American Journal of Public Health 1037 VI: 2 .l4
Transcript
Effects of a Dutch Work-SiteWellness-Health Program:The Brabantia Project
Stan Maes, PhD, Chris Verhoeven, PhD, France Kittel, PhD, and Hetty Scholten, PhD
Introduction
The "Healthier Work at Brabantia" pro-ni~~'teject, named after a Dutch manufacturer of
household goods, was inspired by Americanwork-site wellness health projects."' The phi-losophy underlying the project, however, dif-fers from that of the typical American project.
American programs focus almost exclu-sively on changes in lifestyle.' In the Braban-tia project, it was assumed that, above all, animproved working environment will enhancewellness and health. This basic philosophy isexpressed in the Dutch working conditionslaw, which defines wellness at the work sitein terms of explicit wellness conditions.3 Thedemand-control-social support model4 wasused as a comerstone for the formulation ofthese conditions. As a means of operational-izing the text of the law, 7 operational well-ness conditions were defined5: completenessof the function, challenge, involvement inorganizational tasks, autonomy, social con-
nX.s tacts, cycle length, and information.Each of these conditions gives rise to a
fundamental issue. For example, in tenns ofV;i<.;!5completeness of the function, what are the
proportions of time spent on the performance,preparatory, and supportive tasks entailed in aspecific job? In this context, performancetasks are tasks that involve the mere execu-tion of actions (e.g., the actions necessary toassemble a specific product), preparatorytasks are tasks that precede performance tasks(e.g., obtaining the necessary raw materialfrom the stock or collecting and preparing thenecessary tools), and supportive tasks aretasks that support performance tasks (e.g.,maintenance of tools and machines or qualitycontrol of the product). If a job consists ofmore than 95% performance tasks, this is
ee;i'. considered to constitute a wellness risk.>^<-' Other issues are as follows: (1) Is the
function/task suited to the individual's capaci-ties, or is it too difficult or too easy for theworker? (2) To what degree is the employee
participating in the organization of his or herwork? (3) Does the employee have enoughautonomy concerning work rhythm, method,and sequence? (4) Can the employee establishand maintain a sufficient number of socialcontacts at the work site? (5) Does the jobinvolve a sufficient number of tasks withcycles that are longer than 90 seconds? and(6) Does the employee have enough informa-tion concerning the objectives and outcome ofthe work? As additional wellness risks, lack ofdecision latitude (i.e., when problems occur)and ergonomic problems were considered.
Inspired by the Dutch working condi-tions law, the project is concemed with thefollowing question: Do combined interven-tions, directed at both lifestyle and the con-tent and organization of work, lead to (1)improved health behavior, (2) a reduction inhealth risks, (3) a reduction in general stressreactions, (4) improved quality of work, and(5) a reduction in absenteeism?
Methods
A quasi-experimental pretest/posttestcontrol group design with repeated measureswas used to evaluate the effects of the inter-ventions. Workers from 3 Dutch Brabantiasites participated in the study. Employees at 1site constituted the experimental group, andemployees at the other 2 sites formed the con-trol group. The experimental site and thelarger control site were chosen at random; itwas determined that the third site would be
The authors are with the Section of Clinical andHealth Psychology, Leiden University, Leiden, theNetherlands.
Requests for reprints should be sent to StanMaes, PhD, Section of Clinical and HealthPsychology, Leiden University, PO Box 9555, 2300RB Leiden, the Netherlands (email: [email protected]).
This paper was accepted January 15, 1998.
American Journal of Public Health 1037
VI: 2 .l4
Maes et al.
part of the control group because of its muchsmaller size. The sites were highly compara-ble, producing similar or identical householdgoods. The pretest was administered to bothgroups in 1990 and the posttests 1, 2, and 3years later. The control group received nointerventions. Participation in the study wason a voluntary basis. Data from each individ-ual were known to the researchers but werenot available to the company.
Subjects
The total eligible population consistedof 552 individuals. The research population,defined as the individuals for whom usabledata were available from all measurements,consisted of 264 respondents (134 in theexperimental group and 130 in the controlgroup).
Measurement Instruments
The measurement protocol was identi-cal for all measurement points andincluded (1) a structured interview com-pleted during work time (30 to 45 minutesin duration), (2) self-report questionnairesto be completed at home (approximately 45minutes), and (3) biomedical measure-ments (approximately 15 minutes) takenduring working hours. At the last point ofmeasurement, no biomedical measureswere taken because of the financial coststhat would have been incurred.
Interview. A structured interview wasused to collect sociodemographic data, dataon health behavior, and data on quality ofwork. The interview on health behaviors,based on the protocol for the World HealthOrganization Monica Gent Charleroi study,6consisted of questions on smoking, alcoholconsumption, use of medicines, quality ofsleep, physical exercise, and nutrition.
The data on quality of work were col-lected by means of the structured Wellnessat Work interview.7 8 By means of thisstructured interview, several wellness con-ditions were measured: completeness(score range: 0 to 2), challenge (0 to 3),involvement in organizational tasks (0 to4), autonomy (0 to 8), social contacts (0 to4), cycle length (100 to 500), and informa-tion (0 to 2). In addition, questions wereasked on decision latitude (score range: 0to 26) and ergonomic job aspects (0 to 13).
Biomedical measurements. After theinterview, biomedical measurements weretaken: heart rate, diastolic and systolic bloodpressure (measured twice by means of anErkameter-type sphygmomanometer), bodyweight and height (to calculate body massindex), and a blood sample to determine
serum cholesterol level. No fasting instruc-tions were given.
All employees were individuallyinformed about the results of these measure-ments. If the results exceeded risk levels, theperson was referred to the general practi-tioner. The following criteria for "high risk"were used: (1) serum cholesterol level of 6.5mmol/L or 250 mg/L or higher, (2) diastolicblood pressure level of 105 mm Hg orhigher and/or systolic blood pressure levelof 160 mm Hg or higher, (3) body massindex of 30 or higher, and (4) smoking habit(1 cigarette or more per day).
Questionnaires. A set of self-reportquestionnaires was used to assess stress reac-tions and social support. Employees com-pleted 5 subscales of the validated Dutch ver-sion of the Symptom Checklist-90 anxiety:(score range: 10 to 50), depression (16 to 80),somatic complaints (12 to 60), hostility (6 to30), and sleep problems (3 to 15). They alsocompleted 2 subscales measuring social sup-port (from colleagues and from supervisors)from a validated, shortened Dutch versionl1of the Workstress Questionnaire. 1
Registration. Employee absenteeismwas continuously monitored at all companysites according to a standardized registrationsystem (EMPLOS).
Construction ofEffect Variables
To evaluate the effects of the inter-ventions, we constructed several outcomevariables.
Lifestyles. A total lifestyle score(range: 0 [no healthy lifestyles] to 6) wascalculated on the basis of 6 variables:smoking, amount of physical exercise,hours of sleep, body mass index, use ofalcohol, and fat consumption.
Health risk. A probability score (0 [norisk] to 1) was calculated on the basis of thefollowing variables: age, serum cholesterol,systolic blood pressure, and smoking habit.The coefficients from the FraminghamStudy'2 were used in calculating this score.The probability value represents a subject'srisk of developing coronary heart diseaseduring the subsequent 8 years.
General stress reactions. A total scorewas calculated on the basis of scores on the 5Symptom Checklist-90 subscales describedearlier. High scores indicated that many gen-eral stress reactions were reported.
Working conditions. By means of aprincipal components analysis, the condi-tions measured by the structured Wellness atWork interview were reduced to 3 dimen-sions or scales. These dimensions, whichcould be interpreted as "control," "ergono-mic conditions," and "psychological
demands," explained, respectively, 20%,10%, and 7% of the variance. The psycho-logical demands scale included 5 items (e.g.,"Does your work require a lot ofmental con-centration?" and "Is your work too easy [ortoo complex]?'), the control scale consistedof 10 items (e.g., "Can you correct your ownmistakes?" and "Can you determine the wayin which you perform a given task?"), andthe ergonomic conditions scale contained 10items (e.g., "Do you have the right means[tools, machines] to perform your work?"and "Do you have sufficient light to do yourwork properly?"). A social support factorwas constructed on the basis of the 2 sub-scales from the Dutch version of the Work-stress Questionnaire (lack of social supportfrom colleagues and from supervisor). Thismade it possible to use the dimensions of thedemand-control-social support model4 toevaluate the effects on working conditions.
Absenteeism. The percentage of absen-teeism was calculated every half year on thebasis of the following formula: [number ofdays absent in half-year period/hr(meannumber of personnel x possible workdaysduring the period)] X 100.
Statistical Analyses
Chi-square analyses and t tests wereused to assess differences at pretest betweenthe experimental and control groups. As ameans of assessing the effects of the inter-ventions, repeated measures analyses ofcovariance (ANCOVAs) were carried out; inthese analyses, pretest scores, educationallevel, gender (for all variables), and age (forhealth behaviors and health risks) werecovariates. Additional ANCOVAs were usedto determine effects at specific measurementpoints. Furthermore, when significant effectsover time were found on global scores, addi-tional ANCOVAs were performed on thecomponents ofthe effect variables.
The Interventions
The Brabantia project was based on inter-vention principles that have been describedelsewhere.8"3 The project began in 1990. Dur-ing the first year of intervention, activitieswere targeted at lifestyles. The interventionsdirected at quality of work were implementedduring the second and third years because theyrequired a longer period ofpreparation.
At the individual level, all employeeshad the opportunity to participate in a half-hour intervention session (held duringlunchtime at the work site) 3 times a week.Half of the time spent in each session wasconsidered (paid) work time, while the other
July 1998, Vol. 88, No. 71038 American Journal of Public Health
Effects of the Brabantia Project
half was considered to be part of the employ-ees' free (lunch) time. The sessions consistedprimarily of physical exercise; during the firstyear of the intervention, however, 1 session ofeach 6 was devoted to health education. Thetopics of these sessions were nutrition (7 ses-sions), alcohol and drug consumption (2 ses-sions), working conditions and stress, smok-ing behavior, headaches, and back pain (1session each). In addition to the relevant infor-mation, participants in these health educationsessions were given the opportunity to joingroups oriented around smoking, head aches,or back pain.
The content of the physical exercisesessions was chosen by the participants onthe basis of video recordings of varioustypes of physical exercise. The health edu-cation sessions were discontinued after thefirst year of intervention because of low par-ticipation (100/o-20% of the eligible popula-tion). In the second year, the initiative forinterventions directed at lifestyles was trans-ferred to a special lifestyle committee (agroup of workers elected by employees).This resulted in fewer but more comprehen-sive activities, such as a health fair in thesecond year and a health exhibition in thethird year, with high levels of participation(600/o-70% of the eligible population).
A second type of intervention at thislevel consisted of about 40 hours of trainingin social skills and leadership, along withtraining on how to lead work consultationmeetings for upper and middle management.
At the organizational and environmen-tal level, an intervention was introduced thatgave support to the interventions at the indi-vidual level. The measures included, forexample, creation of on-site exercise facili-ties; a smoking policy in the cafeteria;advertising of the program by means of aninformation corner in the cafeteria, alongwith posters, videos, internal radio mes-sages, and newsletter articles; and providingof healthy food (and information aboutnutrition) in the cafeteria. In addition, incen-tives to promote participation in the pro-gram were used (e.g., T-shirts, sweatshirts,sport bags, and the chance to win a weekendstay at a health and leisure resort).
The second type of intervention at theorganizational and environmental level wasbased on screening for wellness risks at workby means of the structured Wellness at Workinterviews with each employee. The result-ing information was used to construct well-ness risk profiles for each function categoryand each of the 11 production units. Theseprofiles were examined by a "wellness com-mittee" consisting of the management teamand members of the project team. On thebasis of this information, the committee
developed proposals for modifying specificfunctions and/or aspects of the work organi-zation and environment. After extensive con-sultation with the participating workers, thewellness committee guided the implementa-tion and evaluation ofthe proposed changes.
For example, production of the Braban-tia potato-chipper had previously beendivided into short-cycled tasks. The work wassimple, and each worker had always per-formed the same repetitive task. Decision lati-tude was almost nonexistent. Opportunitiesfor the workers to influence the rate andsequence of work and/or the planning of pro-duction were few. Workers had been wellinforned about immediate but not about moredistant outcomes of their work. In addition,opportunities for social contacts between theworkers were limited, partly because of theway in which work areas were designed.
A "task group" of workers was estab-lished and given authority over the entireproduction process, from collection of rawmaterials to delivery of the product to thesales department. This implied additionaltasks such as initiating work orders, arrang-ing supply and transport ofraw materials andfinished products, calculating hours spent ontasks, and performing quality checks. Inaddition to a greater variety of tasks, rotationof tasks became possible. Workers couldtake tums in the transport and wrapping offinished products or any other constituentactivity. For such changes in working condi-tions, additional training was necessary forboth leaders and the other employees. Inaddition, a reorganization of the productionline was necessary to support these changesand to improve ergonomic conditions.
Thus, the entire organizational structureat the experimental site was changed from aproduct-oriented structure to a more func-tional one. For example, there were 26 fore-men and 6 transporters before the interven-tion, while at the end there were only 13foremen and no transporters, since many ofthe tasks were now carried out by the pro-duction workers themselves. Another conse-quence was that the evaluation and gratifica-tion system changed from one based onindividual performance to a combined indi-vidual and group evaluation system.
Results
Participation and Dropout
At pretest, 346 people participated in themeasurements. Respondents differed fromnonrespondents on gender but not age. Menparticipated significantly more than women(56.5% vs 46.3%; x2=4.99, P=.026).
The number of experimental grouprespondents decreased from 175 at pretest to167, 157, and 134 at posttests 1, 2, and 3,respectively; the number of control grouprespondents decreased from 171 to 169, 157,and 130, respectively. At all points of mea-surement, there were no significant differ-ences in demographic characteristicsbetween the 2 groups other than the differ-ences already existing at pretest.
The experimental group did not differfrom the dropouts on any of the demo-graphic characteristics, work status variables,or dependent variables. The control groupdiffered from the dropouts on gender as wellas on control/decision latitude. More womendropped out, and dropouts had lower levelsofcontrol than those who remained.
Pretest Differences
Chi-square analyses and t tests wereused in comparing the experimental groupand the control group on sociodemographiccharacteristics. No significant differenceswere found for age (t262 = 1.74, P=.08) ormarital status (X22= 1.22, P=.54). The meanage for the experimental group was 38.6years (SD= 10.48), and 81.3% were mar-ried; in the control group, the mean age was40.9 years (SD= 10.44), and 82.3% weremarried. There were, however, significantdifferences between the experimental andcontrol group, in terms of gender and educa-tional level. In the experimental group,26.1% of the population was female, ascompared with only 12.2% in the controlgroup (X21 =824 P<.01). The percentage ofemployees with only an elementary schooleducation was also higher in the experimen-tal group than in the control group (61.2%vs49.2%; X22= 10.96, P<. 01). For this reason,gender and educational level were enteredas covariates in all effect analyses.
At the pretest, the experimental groupdiffered from the control group on 3 of thedependent variables: psychologicaldemands, control, and ergonomic condi-tions. Employees in the experimental groupreported fewer psychological demands (t262=-3.45, P<. 01) and a lower level of control(262=-2.55, P<.01) than employees in thecontrol group. In addition, the ergonomicconditions of the experimental group werenot as good as those of the control group(t262=-3.14, P<.01). There were no othersignificant differences between the experi-mental and control groups at pretest.
Program Effects
Effects on lifestyles. No statistically sig-nificant effects over time were found on
American Joumal of Public Health 1039July 1998, Vol. 88, No. 7
Maes et al.
lifestyle variables (see Tables 1 and 2) inanalyses using pretest scores, age, gender,and educational level as covariates. Addi-tional ANCOVAs did not reveal significantdifferences on any of the posttests.
Effects on cardiovascular health risk.When pretest scores, age, gender, and edu-cational level were used as covariates on
posttest scores, a statistically significant dif-ference was found on health risk to theadvantage of the experimental group. Addi-tional ANCOVAs showed that this effectwas mainly caused by a significant differ-ence at the first posttest measurement in
favor of the experimental group, a differ-ence that could be attributed to a change in
serum cholesterol levels in men (men: F 1194=5.61, P=.02; women: F1,42=2.97, P=.09).
Effects on general stress reactions.
When pretest scores, gender, and educa-tional level were used as covariates, no sig-nificant differences were found between the
experimental and control groups. Addi-
tional ANCOVAs revealed no significantdifferences on any of the posttests.
Effects on working conditions. There
was a significant difference between the
experimental and control groups in per-ceived psychological demands over time.
Further ANCOVAs indicated that this differ-
ence was mainly attributable to a significantchange in perceived psychological demands
between the second and third posttests(F1,260= 10.15, Pr.s01).
The change over time in perceivedcontrol was also significantly different for
the 2 groups. Further ANCOVAs showed a
large increase in control in the experimen-tal group (relative to the control group)between the first and second posttests(F1 260=15.74, Psr.0O1).
There was also a trend for ergonomicconditions in the expected direction. Further
ANCOVAs revealed a significant increase in
ergonomic conditions in favor of the experi-mental group between the first and second
posttests (F1,60=3.94, P<.05).The interventions did not lead to signifi-
cant changes over time in terms of social
support from supervisors and colleagues.
Additional ANCOVAs on the separateposttests did not reveal significant effects.
Effects on absenteeism. A standardized
registration system for absenteeism was
introduced in the experimental and control
groups in mid-1990, at the time of the
pretest. Standardized data were obtained
until the end of 1993. Before the interven-
tion was initiated, the percentage of absen-
teeism in the experimental group was
15.8%, as compared with 14.3% in the con-
trol group. At the end of the program, absen-
teeism in the experimental group had
decreased to 7.7% (vs 9.5% in the control
group). Thus, there were decreases in absen-
teeism of 8.1% in the experimental groupand 4.8% in the control group. In compari-son with the national Dutch absenteeism
percentages for workers in the light metal
industries, the experimental group showed a
clear reduction over this time period. In fact,
the percentage of absenteeism in the experi-
mental group was below the mean national
percentage. In contrast, the trend in the con-
trol group followed the national trend and
July 1998, Vol. 88, No. 71040 American Journal of Public Health
TABLE 1-Means and Adjusted Means of Outcome Measures for Experimental and Control Groups: The Brabantia Project,the Netherlands, 19901993
Experimental Group Control GroupOutcome Variable Adjusted Adjustedand Measurement Mean (SD) Mean n Mean (SD) Mean n
remained, on average, 2% to 3% higher thanthe national percentage.
Discussion
Interventions at the individual leveldirected at lifestyle changes, which wereintroduced during the first year of interven-tion, brought about a favorable change inhealth risk at the first posttest. This effectwas mainly due to the significant decreasein cholesterol levels in the experimentalgroup, which can be attributed to the factthat half of the health education sessionswere devoted to nutrition education and toan accompanying cafeteria project in whicha variety of healthy foods were added tothe traditional foods offered in the cafete-ria. The fact that the initial effect on healthrisk disappeared at posttest 2 illustrates thatcontinuous and more extensive interven-
tion is required to produce permanent andbroader effects. Because the emphasis onlifestyles was much weaker in this studythan in comparable American projects,2 itmay not come as a surprise that Americanprograms generally have produced superioreffects in terms of lifestyle and health-related variables.
General stress reactions (anxiety,depression, hostility, somatic complaints,and sleep problems) did not change overtime in either the experimental or the controlgroup and were thus unaffected by the inter-ventions. One could argue, however, thatchanges in work-related stress reactions aremore realistic targets of wellness-health pro-motion programs. The program did havefavorable effects on work-related variables,including psychological demands, control,and ergonomic conditions. The significantimprovement in perceived control over timein the experimental group can be considereda particularly important intervention effect,since a change in psychological demandsdoes not guarantee beneficial wellness/health effects, while an increase in control ordecision latitude usually does, in fact, resultin such benefits.4 Effects on ergonomic con-ditions may be spillover effects of the reor-ganization ofwork.
The results also show a reduction inabsenteeism in the experimental group. Thisreduction was so substantial that, since thesecond year of intervention, the companyhas had a positive financial return on itsinvestment in the project. While thisundoubtedly represents a success, the absen-teeism rate in the company was high tobegin with. Unless large populations areinvolved, it will be difficult to obtain cost-effectiveness in companies with lower initialabsenteeism rates.
In conclusion, this project emphasizedthe organization and content of work ratherthan lifestyles. Therefore, it is not surprisingthat the project had its strongest effects onperceived work conditions and absenteeism.American programs tend to score better onhealth-related variables and thus seem toinfluence outcomes such as absenteeism viaa different pathway. The different philoso-phies at the base of these approaches maynot facilitate reconciliation.'4 A combinationof the 2 approaches may produce more per-manent effects on a variety of wellness-health targets, especially in a population ofblue-collar workers such as those employedat the Brabantia factory. L]
AcknowledgmentsThis study was made possible by grants from theDutch government.
Effects of the Brabantia Project
We thank 2 anonymous reviewers and 0.Gebhardt, PhD, for their valuable comments.
References1. Fielding JE, Piserchia PV. Frequency of work-
site health promotion activities. Am J PublicHealth. 1989;79: 16-20.
2. Kasl SV, Serxner S. Health promotion at theworksite. In: Maes S, Leventhal H, JohnstonM, eds. International Review of Health Psy-chology. Chichester, England: John Wiley &Sons Inc; 1992:111-142.
3. De Arbeidsomstandighedenwet Compleet[Dutch Labor Act]. Amsterdam, the Nether-lands: Nederlands Instituut voor Arbeidsom-standigheden; 1990.
4. Karasek R, Theorell T. Healthy Work: Stress,Productivity, and the Reconstruction of Work-ing Life. New York, NY: Basic Books; 1990.
5. Projectgroep WEBA. Welzijn bij de Arbeid[Wellness at Work]. The Hague, the Nether-lands: Ministerie van Sociale Zaken, Direc-toraat Generaal van de Arbeid; 1989.
6. World Health Organization. Monica project(monitoring trends and determinants in cardio-vascular diseases): a major international study.J Clin Epidemiol. 1988;41:105-114.
7. Maes S, Kittel F, Scholten H, Verhoeven C.Interviewprotocol Inhoud en Organisatie vande Arbeid [Interview Protocol: Content andOrganization of Work]. Tilburg, the Nether-lands: Gezondheidspsychologie, KatholiekeUniversiteit Brabant; 1989.
8. Maes S, Kittel F, Scholten H, Verhoeven C.Gezonder Werken bij Brabantia, Effecten vaneen Totaal Bedrijfsgezondheidsprogramma[Healthier Work at Brabantia, Effects ofa TotalHealth Promotion Program at the Work Site].Den Haag, the Netherlands: Staatsdrukkerij enUitgererij Plantijnstraat; 1994.
9. Arrindel WA, Ettema JHM. SCL-90 Handleidingbyj de Symptom Checklist-90, een Multidimen-sionele Psychopathologie-Indicator [Manual ofthe Symptom Checklist-90, a MultidimensionalPsychopathology Indicator]. Lisse, the Nether-lands: Swets Test Services; 1986.
10. Bergers G, Marcelissen F, de Wolff Ch. VOS-DVragenlyst Organisatiestress-Doetichem, Han-dleiding [Questionnaire on Organizational Stress:Manual]. Nijmegen, the Netherlands: Stressgroep,Psychologie van Arbeid & Organisatie; 1986.
11. Caplan RD, Cobb S, French JRP, van Harrison R,Pinneau SR. Job Demands and Worker Health:Main Effects and Occupational Differences.Washington, DC: National Institute for Occupa-tional Safety and Health; 1975:75-160.
12. Kannel WB, McGee D, Gordon T. A generalcardiovascular profile: the Framingham Study.Am JCardiol. 1976;38:46-51.
13. Maes S. Theories and principles of healthbehaviour change. In: Drenth PJD, SergeantJA, Takens RJ, eds. European Perspectives inPsychology. Volume 2: Clinical Health, Stressand Anxiety, Neuropsychology. Psychophysiol-ogy. Chichester, England: John Wiley & SonsInc; 1990:193-207.
14. Dugdill L, Springett J. Evaluation of workplacehealth promotion: a review. Health Educ J.11994;53:337-347.
July 1998, Vol. 88, No. 7 American Journal of Public Health 1041
