Occup. Med. VoL 49, No. 8, pp. 540-548, 1999Copyright O 1999 Upptncott Williams & Wilklns for SOM

Printed in Great Britain. All rights reserved0962-7480/99

A systematic review of theeffectiveness of health promotioninterventions in the workplaceA. Harden, G. Peersman, S. Oliver, M. Mauthner and A. OakleyCentre for 4he Evaluation of Health Promotion and Social Interventions(EPI-Centre), Social Science Research Unit, Institute of Education,University of London, UK

The aim of this study was to identify and critically review evaluations of theeffectiveness of health promotion programmes in the workplace. In line with guidelinesfor 'good practice' within the literature on workplace health promotion, this studyaimed to assess the extent to which evaluated interventions considered employees'expressed needs or involved employee-employer partnerships. Overall, 110 outcomeevaluations were located. Only a quarter of these reported that interventions wereImplemented in response to the explicit needs and/or views of the employees andvery few involved partnerships. Most of the programmes targeted individual behaviourand supportive organizational change was limited. The majority of the outcomeevaluations were not sufficiently rigorous to make a strong case for the effectivenessof workplace health promotion. However, some pointers to success were identified. Itwas concluded that there seems to be a wide disparity between what counts as'good practice' within workplace health promotion and what Is reported in theevaluation of effectiveness literature. This is not to say that 'good practice' does notexist, but that either such programmes are not rigorously evaluated for theireffectiveness and/or that many of the evaluation findings remain outside the publicdomain.

Key words: Effectiveness; health promotion; needs assessment; outcome evaluation;partnerships; process evaluation; systematic review; workplace.

Occup. Mod. Vol. 49, 540-548, 1999

Rtawed 30 March 1999; acapud in final form 6 July 1999

This review was funded by the Health Education Authority (HEA), England. The Department of Health

(DH) fund» a programme of work at the EPI-Centre which support! a research team and the development of

a specialized database (EPIC) compiling the result! of systematically reviewing evaluations of health promotion

interventions. The views expressed in this publication are those of the authors and not necessarily those of die

HEA or DH.

INTRODUCTION

Health promotion has become a central feature of healthpolicy in many countries due to the epidemic in chronicdiseases, the ageing population and the widening socialclass gradient in health.1"5 With a high proportion ofthe adult population in employment and with work-place accidents and illness resulting in a considerableeconomic burden, the workplace has been recognizedboth as an important target6'7 and a determinant ofhealth.8 The workplace is considered to be an ideal set-ting for health promotion initiatives: it provides easy and

Correspondence to: A. Harden, The EPI-Centre, Social Science ResearchUnit, 18 Wobum Square, London WC1H ONS, UK. Tel: (+44) 171 612 6246;Fax: (+44) 171 612 6400; email: a.harden©loe.ac.uk •

regular access to a large number of people who make upa relatively stable population and it may encourage sus-tained peer support and positive peer pressure.8"10

In the USA, the number of workplace health promo-tion programmes has grown exponentially since 1980,with 81% of workplaces offering some kind of healthpromotion programme.7 In the UK, however, only 40%of workplaces in 1992 undertook at least one majorhealth-related activity.1J Such a comparison suggests thatwhile employers may endorse occupational health andhealth promotion in principle, few transfer this supportinto practice in the UK. Surveys of public and privatesector employers and trade unions in the UK have shownthat although health promotion programmes are receiv-ing increasing attention, they are by no means universal,especially in small- to medium-sized workplaces where

at Mem

orial Univ. of N

ewfoundland on A

ugust 1, 2014http://occm

ed.oxfordjournals.org/D

ownloaded from

A. Harden et a/.: Review: health promotion Interventions In the workplace 541

there are financial and human resource barriers.12'13

These findings have led to several recent policy initiativesin the UK which have targeted the workplace as animportant arena for developing health promotion inter-ventions.10'14*15 For example, the recent governmentgreen paper on health emphasizes support for healthyworkplaces14 and a Department of Health WorkplaceHealth Advisory Team has recently been launched.15

Workplace health promotion has been associated witha reduction in health risks and promotion of healthylifestyles and with improvements in economic and pro-ductivity factors including medical costs, compensationbenefits, employee absenteeism and job satisfaction.16"19

However, to serve disease prevention purposes effect-ively, workplace health promotion programmes need toattract a large proportion of employees. Although overallparticipation estimates range from 20-60%, employeeswho are at risk for adverse health outcomes such assmokers, employees with elevated blood pressure and/orcholesterol levels and those who lead sedentary lifestylesare less likely to participate.20 Participants are morelikely to be younger, well-educated, female, non-smokersand white collar workers16*21"27 and health promotioninitiatives are less likely to reach low earners and thosewho are intermittently employed.8

This variation in participation rates suggests thatworkplace health promotion programmes are not appeal-ing, acceptable and/or relevant to all employees. Recentguidelines for workplace health promotion have sug-gested several essential elements which could help toovercome some of these problems. The starting point forworkplace health promotion should include a thoroughneeds assessment amongst the target population, fol-lowed by a formal planning stage which actively involvesall the major stakeholders including managers, unionrepresentatives and employees in the workplace.6'9'28'29

The use of such partnerships has been advocated withincommunity empowerment approaches to health promo-tion.30"32 Major objectives of such partnerships are toensure that an intervention is both relevant and accept-able, and to encourage employee ownership of the pro-grammes which boosts sustained participation.

Workplace health promotion guidelines suggest thatprogrammes need to feed into different levels to facilitatesustained behaviour change. At the individual level, aprogramme should incorporate a range of educationstrategies; at the organizational level, support mechan-isms throughout the organization should reinforce andencourage positive health actions (e.g., providing healthyfood options in the canteen). Finally, at the communitylevel, workplace health promotion should have the poten-tial to be actively disseminated by employees to theirfamilies and social networks.7'8

The aim of the study reported in this paper was toidentify and critically review evaluations of the effective-ness of health promotion interventions in the workplace.In line with current guidelines for good practice in work-place health promotion, the study aimed to assess theextent to which these outcome evaluations evaluatedinterventions developed in response to employees' needs,in partnership with employees, and/or reported process

evaluations carried out to examine employee views onthe programme. Outcome evaluations which evaluatedinterventions displaying these characteristics were furtherassessed for their methodological quality. Those deemedto have sufficient methodological rigour, enabling reliableconclusions about the impact of health promotion in theworkplace to be made, were summarized.

MATERIALS AND METHODS

The study was conducted in four stages.

Identification of outcome evaluations

The aim was to locate available outcome evaluations ofhealth promotion interventions in the workplace (i.e.,studies assessing the impact of health promotion pro-grammes on a range of health-related outcomes). Thereview updated an earlier one by France-Dawson et al.which included studies up to early 1994.33 Hence, thesearches for the update were limited to finding morerecent studies (1994-97). These were sought by system-atic searching of key journals; electronic database search-ing on EMBASE, ERIC, Medline, PsycLIT and theSocial Science Citation Index; through personal contactsand by scanning the reference lists of already identifiedoutcome evaluation studies. The full search strategies arereported elsewhere.34

Classification of relevant studies

Full reports of all relevant outcome evaluations wereobtained and classified according to: the country wherethe study was carried out; the health focus; the type ofintervention and the extent of involvement of the targetpopulation in the planning and implementation of theintervention.

Studies with an exclusive focus on smoking prevention/cessation were excluded from further analysis as they arethe subject of systematic reviews by the Review Group onTobacco Control of the Cochrane Collaboration andare included in a recent cancer health promotion reviewcarried out for the Health Education Authority.35 Detailsof completed and ongoing systematic reviews of theeffectiveness of different smoking prevention/cessationinterventions are available on The Cochrane Library}6

Assessment of the methodological quality ofoutcome evaluations

Outcome evaluations reporting interventions which hadbeen developed in response to needs or views expressedby the employees, or in partnership with them; or thosethat also included at least one process measure (otherthan programme reach or monitoring of interventionimplementation), were further assessed for their meth-odological quality in order to identify those from whichreliable conclusions could be drawn about the impactof workplace health promotion programmes. Outcomeevaluations were classified as 'sound' if they employed a

at Mem

orial Univ. of N

ewfoundland on A

ugust 1, 2014http://occm

ed.oxfordjournals.org/D

ownloaded from

542 Occup. Med. Vol. 49, 1999

control or comparison group equivalent to the interven-tion group in terms of sociodemographic characteristicsand baseline outcome variables and reported both preand post-intervention data for each group on all out-comes targeted. This classification has been informed bycriteria used to appraise evidence within evidence-basedmedicine and the principles for reviewing health careinterventions established in the Cochrane Collabora-tion,37 and the work of other reviewers in the health, edu-cation and social welfare fields.38"46

A standardized data extraction framework was used,developed by the Centre for the Evaluation of Social andHealth Promotion Interventions (EPI-Centre) at theSocial Science Research Unit, Institute of Education.47

Two reviewers, chosen for their similar experiencein health promotion research, independently assessedeach study and any disagreements were discussed andresolved, if necessary, with a third reviewer. High levelsof concordance were achieved between the reviewers.Out of the 78 questions asked of each study, on averagethe reviewers' answers showed an agreement level of80.1% (range: 68-94%). In the main, disagreementsarose as a result of differences in reviewers' interpreta-tions of information provided in the study or as a resultof one reviewer missing some of the detail from thestudy, rather than from disagreements about methodo-logical quality. In these cases, differences in interpreta-tion were discussed and a common interpretation agreedupon.

Description of 'sound' outcome evaluations

Those interventions which had been soundly evaluatedwere described according to key features of the inter-vention and in terms of the impact on health-relatedoutcomes. Particular attention was given to describingthe features of any attempt to develop interventions inpartnership with employees and to the findings of anyprocess evaluation.

RESULTS

Search results

The searches identified 100 relevant outcome evalu-ations published between 1994 and 1997. Electronicsearching located 83% of the relevant studies; handsearching 41%; 5% were obtained through personal con-tacts and 1 % by scanning bibliographies (there was someoverlap in outcome evaluations identified by differentmethods). Of the 100 newly-identified outcome evalu-ations, 84 were obtained as full reports within the timeavailable for conducting this review. This included twolonger-term follow-up reports of studies included in theprevious review;28 hence, reports of a total of 82 separateoutcome evaluations were obtained. Combined with theoutcome evaluations of the earlier review,33 a total of139 separate outcome evaluations were assessed for thisreview.

Classification of relevant studies

Of the 139 outcome evaluations, one was excluded as aretrospective study and a further 28 were excluded onthe basis of their exclusive focus on smoking. Of theremaining 110 outcome evaluations, just under half(46%, n = 50) reported on at least some employeeinvolvement: 23 (21%) evaluated interventions developedin response to the expressed needs or views of the targetpopulation; 27 (25%) included at least some assessmentof programme acceptability; 15 (14%) described inter-ventions modified on the basis of pilot studies and only15 (14%) described the involvement of employee part-nerships in planning and implementation. (Note: someoutcome evaluations were included in more than onecategory.)

The majority of outcome evaluations (63%) evaluatedinterventions which aimed to change clinical risk factors(e.g., blood pressure, cholesterol level, weight) and/orbehavioural risk factors (e.g., diet, exercise) for cardio-vascular disease. Most interventions (84%) included aneducational component; half included a behavioural riskassessments and/or medical screening; 12% includedpersonalized advice and 10% included some form ofsocial support Few interventions included a supportivemeasure (22%) or a regulatory/legislative measure (6%)at the organizational level of the workplace to facilitateindividual behaviour change.

Quality assessment of outcome evaluations

The 50 outcome evaluations which reported at least someemployee involvement were further assessed for theirmethodological quality. Just under half of the 50 outcomeevaluations employed a control or comparison groupequivalent on baseline outcome measures and sociodemo-graphic variables (46%) or reported pre-interventiondata (44%); nearly two-thirds (60%) reported post-intervention data, and almost all the outcome evaluations(82%) reported on all outcomes as targeted in the aimsof the evaluation. In all, only 15 (30%) of the 50 out-come evaluations met all four of the methodological cri-teria needed for these to be classified as 'sound'.

Description of soundly evaluated programmes

Of the 15 outcome evaluations that were judged tobe methodologically sound, 13 were carried out in theUSA, one in Canada and one in the Nedierlands. Eightof these included a process evaluation examiningemployee acceptability of the programme, five reportedon an intervention which had been previously pilot-tested and modified according to employees' views, twowere based on the explicit needs or views of employeesand 11 had used some form of partnership in the devel-opment of the intervention. (Note: some studies wereincluded in more than one category.)

Table 1 illustrates the sound outcome evaluationsaccording to the type of intervention employed: healtheducation with incentives; education only; personalized

at Mem

orial Univ. of N

ewfoundland on A

ugust 1, 2014http://occm

ed.oxfordjournals.org/D

ownloaded from

A. Harden ef a/.: Revtew: health promotion interventions In the workplace 543

Table 1. Effectiveness" of interventions by intervention type and use of 'partnerships' for the methodologically 'sound'studies (n = 15)

Intervention type Number of outcome evaluations

Hearth education plus incentivesEducation onlyPersonalized advice/educationSkill developmentEnvironmentalComprehensiveUse of 'partnership' in the development of intervention15

All(n = 15)

312216

11

Effective*(n = 3)

0100112

Partly effective*(n = 8)

0022048

Ineffective*(n = 4)

3000011

* Effectiveness Is based on the reviewer's Judgementb This category overlaps with the previous categories and thus n does not add up to 15 l/.e., the Interventions which used 'partnerships'In their development also appear In one of the other categories according to Its Intervention type. Details of which Interventions weredeveloped using 'partnerships' can be found In the text.)

advice/education; skill development; environmental modi-fication and comprehensive. Comprehensive interven-tions were defined as those involving components whichtargeted both individual and organizational/environ-mental levels of change.7 In addition, an attempt wasmade to illuminate those interventions which were devel-oped in partnership with employees, a key principle forthe organization of such programmes found in muchof the literature on workplace health promtoion.6>9>28>29

The table indicates whether the outcome evaluationshowed an intervention to be effective, partly effective(i.e., effective for some outcomes, ineffective for others)or ineffective. In this way the analysis has attempted tolink the principles of developing workplace programmeswith evidence of effectiveness.

Eleven of die interventions were found to be effectiveor partly effective (i.e., effective for some outcomes, inef-fective for others) and four were found to be ineffective.Comprehensive programmes seem to be a promisingstrategy for effective interventions widi five out of the sixsoundly evaluated interventions of this type shown to beeffective or partly effective. There is also a relationshipbetween effectiveness and employee partnership with 10out of 11 interventions shown to be effective or partlyeffective. Because of the small number of studies, how-ever, this analysis should be considered only as tentativeat this exploratory stage.

Given the small number of studies in each category itis important to interpret these results within the contextof each study's specific evaluation and interventiondetails. Effective interventions were: an educational pro-gramme, developed and delivered in partnership withemployees, for reducing occupational cancer;48 a pro-gramme providing an environmental change within work-place cafeterias to increase the sales of low fat meals49

and a comprehensive programme targeting individualand organizational levels of change in the context ofcardiovascular disease prevention.50

Parkinson et ol. evaluated the 'Coke Oven InterventionProgram' for reducing occupational cancer in 28 cokeplants in Canada and the USA using a study sample ofmostly male and highly educated employees.48 The inter-vention was developed in conjunction with the union of

steel workers; it combined the expertise of workers anduniversity personnel and consisted of health educationsessions, delivered by staff and union members, andincluded health and safety, occupational cancer surveil-lance and regulations for personal practice. Compared tocontrol plants, participants improved significantly intheir knowledge of the Coke Oven Standard (guidelinesdeveloped by die Occupational Safety and Healdi Admin-istration to address the hazards of coke oven emissions)and in workplace safety behaviours.

Levin evaluated a low-effort, low-budget interventionwhich attracted attention to the low-fat selections avail-able in workplace cafeterias by means of an appropriatelabel, a poster and a prize draw.49 It was found that salesof targeted low-fat meals increased significantly in theintervention cafeteria as compared to a carefully matchedcontrol cafeteria over a four-week assessment period(from 4.3% of total entrees sold to 11.9%). This increasein sales was maintained at six months follow-up. Thestudy sample consisted of a predominantly Hispanicpopulation of government employees in an urban setting.A process evaluation suggested that one reason for pro-gramme success was the use of materials accessible toemployees with a range of literacy skills.

The 'Live for Life' programme consisted of healdiscreening, lifestyle improvement programmes and organ-izational changes planned in conjunction widi a taskforce of employees.50 A high participation rate wasachieved through the provision of a regular and conven-ient programme backed up with organization changessuch as on-site exercise facilities, nutritious food in thecafeteria and vending machines and a no-smoking policy.Intervention worksites showed improvement in exercise,smoking, stress management and weight loss comparedto control worksites at 2 years follow-up. The studysample was relatively young and predominandy highlyeducated.

The seven partly effective interventions consisted of:two interventions which aimed to increase healdiy eating,one of which used personalized delivery of informa-tion,51 me other involving the development of specificskills;52 one intervention which aimed to encouragehealthy eating specifically in die context of osteoporosis

at Mem

orial Univ. of N

ewfoundland on A

ugust 1, 2014http://occm

ed.oxfordjournals.org/D

ownloaded from

544 Occup. Med. Vol. 49, 1999

prevention using personalized education;53 two alcoholabuse prevention programmes, one involving skill devel-opment54 and the other a comprehensive programmecombining education with changes to workplace culture55

and three comprehensive programmes targeting individ-ual level and organizational factors to change a variety ofhealth practices.56"63

The intervention which used personalized delivery ofinformation on dietary intake of fat, fruit and vegetables51

found that predominantly male, highly educated em-ployees randomly assigned to a tailored nutrition edu-cation intervention showed a greater decrease in fatconsumption at 1 month follow-up as compared withemployees assigned to a general nutrition educationintervention. However there was no difference found forfruit and vegetable consumption. The investigators useda computerized system to generate letters with feed-backmessages. A concurrent process evaluation suggestedthat tailored information was an important part of theintervention, with more people in this group rating theletter as being of personal relevance, containing newinformation and increasing their motivation to makedietary changes.

White-collar employees randomly assigned to askills-based healthy eating intervention (education, dietanalysis, supermarket tours, group walks) showed lessagreement with attitudinal barriers to eating healthy food(e.g., healthy food is costly) and greater self-efficacy thana control group.52 However, although the interventionwas based on a needs assessment and focus groupswithin the target population, it did not change intentionsto adopt healthy eating or increase positive beliefs aboutthe relationship between a healthy diet and health.

Similar mixed results were found for the healthy eatingintervention developed in the context of osteoporosisprevention.53 Female employees introduced to four learn-at-home lessons about osteoporosis prevention througha 30 minute motivational meeting (group-delivery) werecompared to female employees who received the samefour lessons with no introduction (impersonal delivery).The delivery approaches were adapted according tofeedback from a sample of employed women. Femaleemployees maintained a higher behaviour habits score at4 months follow-up as compared with those who receivedimpersonal delivery but there were no changes in calciumintake. However, both groups showed improvements ascompared with a group receiving cancer informationonly. The attrition rate in this study was relatively highand non-completers in the gTOup-delivery condition hada significantly better score for behaviour/habits at pre-test. Thus caution has to be exercised in terms of thegeneralizability of these findings.

Skill development was a central feature of one of thealcohol abuse prevention programmes which was partlyeffective: 'The Working People Program' which was field-tested with 108 predominantly blue-collar workers.54Therelevance and acceptability of materials used in the inter-vention had previously been modified in light of workers'feedback. Immediately after the intervention there wasa decrease in the number of drinking days and heavydrinking, but not for the number experiencing problems

at work due to alcohol, or the average number of drinksconsumed on drinking days. A low participation rate andrelatively high drop-out rate were major problems in thestudy. A process evaluation revealed that most of thoseremaining in the study rated the programme as 'verygood' or 'excellent', although management support forthe programme was felt to be extremely limited.

An alcohol abuse prevention programme which tar-geted individual and organizational levels of change wasa mandatory programme which aimed to alter workplaceculture and to promote employee ownership of the pro-gramme.55 The study population consisted of equalproportions of males and females and included bothblue and white collar employees. Components of theprogramme were: a supervisory training programme formanagers and supervisors aimed at culture change, pre-vention and early intervention; an interactive educationprogramme for employees and a peer-helper programme.At 3 months follow-up, the intervention was judged in-effective for smoking and marijuana use but effectivefor measures of alcohol use. Several factors were deemedto have contributed to the success of the programme,including strong management support, the involvementof an employee task force, the positive emphasis on car-ing about co-workers and the linking of the programmeto existing plant safety and productivity initiatives.

Employee representation for planning interventionactivities in a 'Wellness Co-ordinating Committee' wasused in one of the partly effective programmes aiming toincrease healthy behaviours by targeting individual andorganizational levels of change (screening, health educa-tion, follow-up counselling, a menu of intervention typesand change in plant organization).56"59 Using a studypopulation of mostly hourly paid, male workers it wasfound that, as compared with workplaces receiving lesscomprehensive interventions, there were greater reduc-tions in smoking and greater control over blood pressure.However weight loss achieved by some people was offsetby the weight gained by others. An integral processevaluation found that engaging the 'eager' employeesinto wellness programmes was easy if programmes wereprovided on-site; engaging the 'reluctant' employeesrequired one-to-one approaches and the provision ofincreased choice through a menu approach.

Similar mixed results were found for state employeesin worksites where employee committees selected a seriesof health promotion programmes.60 These typicallyincluded: weight control and nutrition, stress manage-ment, exercise, smoking cessation, alcohol education andsafety education. Changes at the individual level weresupported by changes at the organizational level, suchas providing healthy food choices in the cafeteria andsmoking regulations. Ten months after implementation,employees showed significantly greater positive changesin smoking and alcohol use when compared with work-sites not exposed to the programme, but no change forexercise behaviour, diet and stress.

The final intervention which was partly effective was acancer prevention and control programme which tar-geted both individual and organizational levels of changeand was developed in partnership with employees. It was

at Mem

orial Univ. of N

ewfoundland on A

ugust 1, 2014http://occm

ed.oxfordjournals.org/D

ownloaded from

A. Haitian et a/.: Review: health promotion Interventions In the wonVptece 545

evaluated in the 'Working Well Trial' which tested theintervention with a predominantly male and blue collarpopulation.61"63 The intervention was implemented in114 workplaces which all targeted eating patterns and/orsmoking and occupational exposure to carcinogens. Theproject had three structural levels: (1) an overarchingdecision-making structure to establish and co-ordinateadministration; (2) a management structure within eachof the four study centres to implement the interventionand (3) an employee advisory board and a co-ordinatorin each workplace to help tailor and implement the inter-vention. Members on the advisory board were trained inthe goals and content areas of the project, and in theirroles and responsibilities. The intervention had threemain goals: to increase motivation to change throughincreased awareness; to provide skills training for indi-viduals ready to take action and to maintain behaviourchange through supportive changes at the organizationallevel.

At 1 year follow-up, there were small but significantincreases in fruit and vegetable consumption and reduc-tions in fat consumption in the intervention group ascompared with the control group, but no differences indietary fibre intake or smoking prevalence. In addition,the authors highlighted that although the decrease in fatconsumption in the intervention group was greater thanin the control group, fat consumption also notablydecreased in the control group. This suggests either amodest secular trend in the reduction of fat consumptionor the possibility that the control group may havebecome contaminated through awareness of some of themessages of the intervention. In the context of the mixedfindings of this study the authors highlighted somechallenges to the use of 'partnerships' and suggestedeconomic hardships, lay offs and change in managementmay have disrupted intervention strategies which reliedon worker participation.

Of the four ineffective interventions, three were con-cerned with weight loss. These interventions all used apay-roll based incentive system which involved partici-pants agreeing to have a fixed sum withheld from theirpay cheque. This could be earned back depending onprogress made with weight loss. Any money remainingwas divided amongst those who reached their personalweight loss goals. This system was combined with healtheducation classes to control weight amongst mixed sex,blue and white collar workers at a university or at amedium-sized workplace within a metropolitan area. Onestudy did not find any effect on weight64 whereas the othertwo showed that most of the people who lost weightregained it within one year after the intervention.65"67

The fourth ineffective intervention targeted both indi-vidual and organizational levels of change and was devel-oped in partnership with employees. The 'Take HeartProject' highlighted some methodological challenges inthe use of 'partnerships'.68-69 Employee Steering Com-mittees at each site aimed to enhance ownership andinvolvement, and chose brief, low-intensity health educa-tion and environmental change activities best suited totheir work site from a menu. There were no differences

in smoking, fat intake or cholesterol levels between work-places randomly assigned to control and interventionsites at 2 year follow-up. A concurrent process evaluationhighlighted the importance of giving 'employee steeringgroups' greater direction; ensuring active participationfrom all representatives on the steering group andencouraging greater interaction between sites.69

DISCUSSION

Effective workplace health promotion partly depends onthe interest and willingness of employers to support suchprogrammes and of employees to participate. Conse-quently two important factors are collecting evidencethat convinces employers of the beneficial effects ofhealth promotion and finding ways to make programmesrelevant and acceptable to employees. From this reviewof 110 outcome evaluations studies in the workplace, itcan be concluded that a lot of progress still needs to bemade on both accounts.

In terms of methodological quality of the outcomeevaluations, few outcome evaluations met the four 'core'criteria necessary to be considered as providing reliableevidence of effectiveness. The most common shortcom-ings were lack of employment of an equivalent control/comparison group and a failure to report pre-interventiondata. Most studies in this review cannot be considered tobe sufficiently rigorous to make a strong case for work-place health promotion in terms of the benefits for healthand/or other outcomes. Further, most interventions wereevaluated with respect to short-term effects only. Of thefifteen 'sound' outcome evaluations identified in thisreview, follow-up intervals ranged from immediatelyafter the intervention to three years, with the majority ofevaluations only employing follow-up intervals of sixmonths or less. Thus, there is an urgent need to invest-igate the longer-term maintenance of these short-termeffects. This resonates with other systematic reviews ofworkplace health promotion programmes which havefound that conclusive evidence of effectiveness is not yetavailable.70"77 In addition, the effectiveness of differentapproaches to workplace health promotion is likely to beinfluenced by different organizational structures. There isvery little discussion of this in the available literature.

This review found that health promotion interventionsin the workplace more often address disease preventionissues guided by epidemiological data than needs identi-fied by the recipients themselves: only 21% of the outcomeevaluations reported that the evaluated interventions werebased on what employees said they wanted or what theythought were the problems that needed addressing. Inaddition, only about one in five studies included someattempt to involve the target population in the develop-ment of the programme. Most programmes were tar-geted at the individual level; supportive organizationalmodifications were scant. These findings indicate a sub-stantial discrepancy between the recommendations ofwhat counts as 'good practice' within workplace healthpromotion and what is reported in the evaluation ofeffectiveness literature. 'Good practice' may well exist,

at Mem

orial Univ. of N

ewfoundland on A

ugust 1, 2014http://occm

ed.oxfordjournals.org/D

ownloaded from

546 Occup. Med. Vol. 49, 1999

but such programmes are either not rigorously evaluatedfor their effectiveness and/or many of the evaluationfindings may remain outside the public domain.

The limited overall attempt to ensure that programmesare based on the needs of employees could also explainvariation in the reported participation rates which variedfrom 2% to nearly 100% [many studies (42%) did notstate their participation rate]. Some studies also sufferedfrom high drop-out rates. These findings undermine theclaim that workplaces provide easy access to large cap-tive audiences and suggest that any effort invested inworkplace health promotion should be more responsiveto the needs of workers. If participation in workplacehealth promotion is to be increased, research into thebarriers and facilitators for participation needs to informprogramme development and implementation. Studieswithin the context of an evaluation of a cancer controland prevention programme in the USA, the 'WorkingWell Trial',61"63 indicated that low-risk non-participantswere already participating in health promotion activitiesoutside the workplace, whereas moderate-risk and high-risk non-participants were preoccupied with past failuresin health behaviour change, were suspicious of themotives of health promotion programmes and found thework environment to be non-supportive. A common bar-rier to participation regardless of risk status was lack oftime. Participation was more likely, however, when work-ers were aware of changes their employer had made toreduce occupational hazards.78'79These findings suggestthat participation in workplace health promotion may beincreased if interventions also take into account healthrisks arising from work activities.

Taken together the findings about the effectiveness ofworkplace health promotion and employee partnershipssuggest

• there should be visible and enthusiastic support for,and involvement in, the intervention from top man-agement;

• there should be involvement of employees at all organ-izational levels in the planning, implementation andactivities of the intervention;

• a focus on a definable and modifiable risk factor,which constitutes a priority for the specific workergroup, can make an intervention more acceptable tothat group of workers and increase their participationand

• interventions should be tailor-made to the character-istics and needs of the recipients.

These recommendations give a very important role toemployee involvement and the use of partnerships forplanning and implementing health promotion in theworkplace. Whilst the findings of this review suggest thatthe use of partnerships is a promising strategy forincreasing the effectiveness of programmes, processevaluations of the use of such strategies suggest somechallenges to their use, including economic hardshipsand trying to strike a balance between the promotion ofemployee ownership and giving employees enough sup-port.68'79 This highlights the need for further research

to determine factors to facilitate the success of suchpartnerships.

Though these recommendations are very general andcertainly have been advocated before, our review hasclearly indicated that they are far from commonlyapplied in practice. Health promotion programmes havebecome increasingly popular, but persuading companiesto implement relevant programmes in the appropriateway, and to evaluate them properly has proved difficult,mainly due to lack of financial and human resources, aswell as time constraints.13

Given the small number of soundly evaluated inter-ventions identified in this review, specific recommenda-tions for effective interventions have yet to be identified.The 'sound' outcome evaluations identified in this reviewtested interventions with specific populations in terms ofage, ethnicity, social class, education level and gender.Such sociodemographic detail is crucial for evaluatingthe generalizability of research findings and for examin-ing any selective effects of an intervention.80 Similarly,the specific context in which the intervention is imple-mented is important. For example the partly effectiveskill development intervention to reduce alcohol anddrug consumption reported by Stoltzfus et al. was amandatory programme.55 This creates a very differ-ent context compared with programmes which involvevoluntary participation.

Of particular concern is the lack of any soundly evalu-ated outcome evaluations identified in the UK. Sincemany employers in the USA bear the health care costsof their workforce, workplace health promotion pro-grammes arise in a very different context from that in theUK and other European countries. This again warrantscaution in the generalizability of the findings of thisreview. Although many workplace health promotion pro-grammes are in progress in the UK, many programmeshave either not been formally evaluated or much of theinformation is unpublished. This highlights the need formore rigorous research on the effectiveness of workplacehealth promotion in the UK.

A further issue highlighted by this review is the lack ofoutcome evaluations with integral process evaluations.Only one in five studies in this review included bothprocess and outcome measures. Springett and Dugdilladvocate the use of a more comprehensive, holisticapproach to the evaluation of workplace health promo-tion and the need to develop new research paradigms.13

Such calls are made throughout health promotion andhave often led to fierce debates about what constitutes'evidence'. The issue is not one of competition betweendifferent methods, but about collecting a range of goodquality data, which increases the validity of conclusionsand provides us with a fuller picture of the extent towhich programmes work, and the reasons why.

CONCLUSION

Although the workplace has enormous potential as a set-ting for improving the health of the adult population,many programmes seem to ignore the needs and views

at Mem

orial Univ. of N

ewfoundland on A

ugust 1, 2014http://occm

ed.oxfordjournals.org/D

ownloaded from

A. Harden et al.: Review: health promotion Interventions In the workplace 547

of the target population in the planning and implementa-tion of workplace health promotion programmes. Inaddition, the general lack of good quality evaluationstudies is a matter for concern. Evaluation should beincluded as an integral part of any new intervention pro-gTamme and include both a range of outcome and processmeasures.

REFERENCES

1. Clark N, McLeroy K. Creating capacity through healtheducation: what we know and what we don't know. HealthEdu Quar 1995; 22: 273-389.

2. Green L, Kreuter M. Health Promotion Planning. An Edu-cational and Environmental Approach. Mountain View, CA(USA): Mayfield Publishing Company, 1991.

3. Hawe P, Degeling D, Hall J. Evaluating Health Promotion. AHealth Worker's Guide. Sydney, Australia: MacLennan andPetty, 1995.

4. Pencak M. Workplace Health Promotion Programs. Nurs-ing Clinic of North America 1991; 26: 233-240.

5. Wong ML, Alsagoff F, Koh D. Health promotion — afurther field to conquer. Singapore Med J 1992; 33:341-346.

6. Health Education Authority. Action on Health at Work.London, UK: HEA, 1993.

7. Wilson MG, Holman PB, Hammock A. A comprehensivereview of the effects of worksite health promotion on health-related outcomes. Am J Health Prom 1996; 10: 429-435.

8. Faculty of Public Health Medicine Committee on HealthPromotion. Health Promotion in the Workplace. Guidelinesfor Health Promotion No. 40. London, UK: Royal Collegeof Physicians, 1995.

9. Tones K, Tilford S. Health Education: Effectiveness andEfficiency and Equity. London, UK: Chapman and Hall,1994.

10. Department of Health. On the State of Public Health 1994.The Annual report of the Chief Medical Officer of theDepartment of Health. London, UK: HMSO, 1995.

11. Health Education Authority. Health Promotion in the Work-place — a Summary. London, UK: HEA, 1993.

12. Sanders D, Crowe S. Overview of health promotion in theworkplace. In: Scriven A, Orme J, eds. Health PromotionProfessional Perspectives. 1996: 199-209.

13. Springett J, Dugdill L. Workplace health promotion pro-grammes: towards a framework for evaluation. Health EduJ 1995; 54: 88-98.

14. Department of Health. Our Healthier Nation. London,UK: HMSO, 1998.

15. Department of Health. On the State of Public Health 1996.The Annual report of the Chief Medical Officer of theDepartment of Health. London, UK: HMSO, 1997.

16. Alexy B. Factors associated with the participation or non-participation in a workplace wellness centre. Res NursingHealth 1991; 14: 33-40.

17. Goetzel RZ, KahrTY, Aldana SG, Kenny GM. An evalu-ation of Duke University's Live for Life Health Promotionprogram and its impact on employee health. Am J HealthProm 1996; 10: 340-341.

18. Knight KK, Goetzel RZ, Fielding JE, et al. An evaluationof Duke University's Live For Life health promotion pro-gram on changes in worker absenteeism. J Occup EnvironMed 1994; 36: 533-536.

19. Sapolsky HM, Airman D, Greene R, Moore JB. Corporateattitudes towards health care costs. Milbank Mem FundQuart Health Soc 1981; 59: 561-585.

20. Heaney CA, Inglish P. Are employees who are at risk forcardiovascular disease joining worksite fitness centers? JOccup Environ Med 1995; 37: 718-724.

21. Conrad P. Who comes to worksite wellness programs? Apreliminary review. J Occup Med 1987; 29: 317-320.

22. Gebhardt DL, Crump CE. Employee fitness and well-ness programs in the workplace. Am Psychol 1990; 45:262-272.

23. Greene GW, Strychar I. Participation in a worksite choles-terol education program in a university setting. J Am DietAssoc 1992; 92: 1376-1381.

24. Sorensen G, Pechacek T, Pallonen U. Occupational andworksite norms and attitudes about smoking cessation. AmJ Public Health 1986; 76: 544-549.

25. Spilman MA. Gender differences in worksite health pro-motion activities. Soc Sci Med 1988; 26: 525-535.

26. Stunkard AJ, Cohen Y, Felix MRJ. Weight loss competitionsat the worksite: how they work and how well. Prev Med1989; 18: 460-474.

27. Zavela KJ, Davis LG, Cottrell RR, Smith WE. Do only thehealthy intend to participate in worksite health promotion?Health Edu Quar 1988; 15: 259-267.

28. Weston R. What is health promotion? How does it relate tothe workplace? A background paper to support the Con-ference in Berlin on 29-30 September 1994. EurJ CancerPrev 1995; 4: 115-122.

29. Wynne R. Action for Health at Work: The Next Steps. Euro-pean Foundation for the Improvement of Living andWorking Conditions, 1993.

30. Carlaw RW, Mitdemark MB, Bracht N, Luepker R.Organisation for a community cardiovascular healthprogram: Experiences from the Minnesota Heart HealthProgram. Health Edu Quar 1984; 11: 243-252.

31. Heaney CA, Van Ryan M.The implications of status, classand cultural diversity for health education practice: thecase of worksite stress reduction programs. Health Edu Res1996; 11: 57-70.

32. Israel BA, Checkoway B, Schulz A, Zimmerman M.Health education and community empowerment: Con-ceptualizing and measuring perceptions of individual,organisational and community control. Health Edu Quar1994; 21: 149-170.

33. France-Dawson M, Holland J, Fullerton D, Kelly P, ArnoldS, Oakley A. Review of Effectiveness of Workplace HealthPromotion Interventions. London, UK: Social ScienceResearch Unit, 1994.

34. Peersman G, Harden A, Oliver S. Effectiveness of HealthPromotion Interventions in the Workplace: A Review. London,UK: Heakh Education Authority, 1998.

35. Weston R. A Systematic Review of Cancer Health Promotion.London, UK: Health Education Authority, 1999.

36. Cochrane Library [database on disk and CD-ROM;updated quarterly]. The Cochrane Collaboration. Oxford,UK: Update Software, 1998.

37. Cochrane Collaboration. Report. Oxford, UK: TheCochrane Centre, 1994.

38. Biglan A, Severson H, Ary D, et al. Do smoking preventionprograms really work? Attrition and the internal and ex-ternal validity of an evaluation of an evaluation of a refusalskills training program. J Behav Med 1987; 10: 159-171.

39. Chalmers I, Haynes B. Reporting, updating, and correct-ing systematic reviews of the effects of health care. Br MedJ 1994; 309: 862-865.

40. Dickersin K, Lefebvre C. Identifying relevant studies forsystematic reviews. Br Med J 1994; 309: 1286-1291.

41. Knipschild P. Systematic reviews: Some examples. Br MedJ 1994; 309: 719-721.

at Mem

orial Univ. of N

ewfoundland on A

ugust 1, 2014http://occm

ed.oxfordjournals.org/D

ownloaded from

548 Occup. Med. Vol. 49, 1999

42. Loevinsohn BP. Health education interventions in develop-ing countries: a methodological review of published articles.Int J Epidemiol 1990; 4: 788-794.

43. MacDonald G, Sheldon B, Gillespie J. Contemporarystudies of the effectiveness of social work. BrJ Social Work1992; 22: 615-643.

44. Mulrow CD. Rationale for systematic reviews. Br Med J1994; 309: 597-599.

45. Schnaps E, De Bartolo R, Moskowitz J, Pally CS, ChurginS. A review of 127 drug prevention program evaluations. JDrug Issues 1981; 11: 17-43.

46. Schwartz D, Flamant R, Lellouch J. ChnicalTrials. London,UK: Academic Press, 1980.

47. Peersman G, Oliver S, Oakley A. EPI-Centre ReviewGuidelines. London, UK: EPI-Centre, 1997.

48. Parkinson DK, Bromet EJ, Dew MA, Dunn LO, BarkmanM, Wright M. Effectiveness of the united steel workers ofAmerica coke oven intervention program. J Occup Med1989; 31: 464-472.

49. Levin S. Pilot study of a cafeteria program relying pri-marily on symbols to promote healthy choices. J Nutr Edu1996; 28: 282-285.

50. Wilbur CS, Hartwell TD, Piserchia PV. The Johnson andJohnson Live for Life Program: its organization and evalu-ation plan. In: Cataldo M, Coates T, eds. Health and Indus-try. New York, NY (USA): John Wiley & Sons, 1986.

51. Brug J, Steenhuis I, Van Assema P, De Vries H. The impactof a computer-tailored nutrition intervention. Prev Med1996; 25: 236-242.

52. Sheeshka JD, Woolcott DM. An evaluation of a theory-based demonstration worksite nutrition promotion pro-gram. Am J Health Prom\994; 8: 263-264.

53. Brown JL. Effect of delivery method on impact of learn-at-home lessons at worksites. J Nutr Edu 1996; 28:140-148.

54. Cook RF, Back AS, Trudeau J. Preventing alcohol useproblems among blue-collar workers: a field test of theWorking People program. Subst Use Misuse 1996; 31:255-275.

55. Stoltzfus JA, Benson PL. The 3m alcohol and other drugprevention program: description and evaluation. J PrimaryPrev 1994; 15: 147-159.

56. Erfurt JC, Foote A, Heirich MA, Gregg W. Improving par-ticipation in worksite wellness programs: Company healtheducation classes, a menu approach, and follow-up coun-seling. Am J Health Prom 1990; 4: 270-278.

57. Erfurt JC, Foote A, Heirich MA. Worksite wellness pro-grams: Incremental comparison of screening and referralalone, health education, follow-up counseling, and plantorganisation. Am J Health Prom 1991; 5: 438-448.

58. Erfurt JC, Foote A, Heirich MA. The cost-effectivenessof work-site wellness programs for hypertension control,weight loss, and smoking cessation. J Occup Med 1991; 33:962-970.

59. Erfurt JC, Foote A, Heirich MA. The cost-effectivenessof worksite wellness programs for hypertension control,weight loss, smoking cessation and exercise. Pen Psychol1992; 45: 5-27.

60. Kronenfeld JJ, Jackson K, Blair SN, et al. Evaluating healthpromotion: a logitudinal quasi-experimental design. HealthEdu Quar 1987; 14: 123-139.

61. Abrams DB, Boutwell WB, Grizzle J, Heimendinger J,Sorensen G, Varnes J. Cancer control at the workplace: TheWorking Well Trial. Prev Med 1994; 23: 15-27.

62. Heimendinger J, Feng Z, Emmons K, et al. The WorkingWell Trial: baseline dietary and smoking behaviors ofemployees and related worksite characteristics. Prev Med1995; 24: 180-193.

63. Sorensen G, Thompson B, Glanz, K, et al. Work site-basedcancer prevention: Primary results from the Working WellTrial. Am J Public Health 1996; 86: 939-947.

64. Jeffery RW, Forster JL, French SA, et al. The HealthyWorker Project: A work-site intervention for weight controland smoking cessation. Am J Public Health 1993; 83:395-401.

65. Forster JL, Jeffery RW, Sullivan S, Snell MK. A work-siteweight control program using financial incentives collectedthrough payroll deduction. J Occup Med 1995; 27:804-808.

66. Forster JL, Jeffery RW, Sncll MK. One-year follow-upstudy to a worksite weight control program. Prev Med1988; 17: 129-133.

67. Jeffery RW, Forster JL, Snell MK. Promoting weight con-trol at the worksite: A pilot program of self-motivationusing payroll based incentives. Prev Med 1985; 14:187-194.

68. Glasgow RE, Terborg JR, Hollis JF, et al. Modifying dietaryand tobacco use patterns in the worksite: the take heartproject Health Edu Quar 1994; 21: 69-82.

69. Glasgow RE, Terborg JR, Hollis JF, Severson HH, BolesSM. Take Heart Results from the initial phase of work-siteWellness Program. Am J Public Health 1995; 85:209-216.

70. Anderson DR, Staufacker MJ. The impact of worksite-based health risk appraisal on health-related outcomes: Areview of the literature. Am J Health Prom 1996; 10:499-508.

71. Terborg JR, Hibbard J, Glasgow RE. Behavior change atthe worksite: Does social support make a difference? Am JPublic Health 1995; 10: 125-131.

72. Dugdill L, Springett J. Evaluation of workplace health pro-motion: A review. Health Edu J 1994; 53: 337-347.

73. Glanz K, Sorensen G, Farmer A. The health impact ofworksite nutrition and cholesterol intervention programs.Am J Health Prom 1996; 10: 453-470.

74. Hennrikus DJ, Jeffery RW. Worksite intervention for weightcontrol: A review of the literature. AmJ Health Prom 1996;10: 471-498.

75. Roman PM, BlumTC. Alcohol: A review of the impact ofworksite interventions on health and behavioral outcomes.Am J Health Prom 1996; 11: 136-149.

76. Shephard RJ. Worksite fitness and exercise programs: areview of methodology and health impact Am J HealthProm 1996; 10: 436-452.

77. Wilson M G, Jorgensen C, Cole G. The health effects ofworksite HTV/AIDS interventions: A review of the researchliterature. Am J Health Prom 1996; 11: 150-157.

78. Emmons K, Linnan L, Abrams D, Lovell H. Women whowork in manufacturing settings: factors influencing theirparticipation in worksite health promotion programs.Women's Health Issues 1996; 6: 74-81.

79. Sorensen G, Stoddard A, Ockene JK, Hunt MK, Young-strom R. Worker participation in an integrated healthpromotion/health protection program: Results from theWeUWorks Project Health Edu Quar 1996; 23: 191-203.

80. Oakley A, Peersman G, Oliver S. Social characteristics ofparticipants in health promotion effectiveness research:Trial and error? Edu Health 1996; 11: 305-317.

at Mem

orial Univ. of N

ewfoundland on A

ugust 1, 2014http://occm

ed.oxfordjournals.org/D

ownloaded from