BrJ Sports Med 1996;30:84-89

REVIEW ARTICLE

A systematic review of physical activity promotionstrategies

Melvyn Hillsdon, Margaret Thorogood

London School ofHygiene and TropicalMedicine,Health PromotionSciences Unit,Department ofPublicHealth and Policy,Keppel Street,London WC1E 7HT,United KingdomMelvyn Hillsdon,research assistantMargaret Thorogood,senior lecturerCorrespondence to:Melvyn HillsdonAccepted for publication27 March 1996Key terms: exercise; physicalactivity; randomisedcontrolled trial; walking

Regular physical activity can play an importantrole in both the prevention and treatment ofcardiovascular disease, hypertension, non-insulin-dependent diabetes mellitus, stroke,some cancers, osteoporosis and depression, aswell as improving the lipid profile.'q A meta-analysis of the relation between physicalactivity and coronary heart disease reportedthat the relative risk of coronary heart diseasedeath in the least active compared with themost active was 1 9-fold.9 The magnitude ofthis relative risk is similar to that of the otherimportant cardiovascular disease risk factors,cigarette smoking, hypertension, and hyper-lipidaemia.'0

Despite this evidence, it is estimated that70% of the English population takes in-adequate physical activity11 compared to 31%who smoke, 30% with a raised serum choles-terol concentration, and 15% who are hyper-tensive. 2

In 1995 the Centers for Disease Control andPrevention (USA) and the American Collegeof Sports Medicine recognised the importanceof physical activity and published a publichealth message recommending that "everyadult should accumulate 30 minutes or moreof moderate-intensity physical activity onmost, preferably all days of the week".13 InMarch of this year the Health EducationAuthority also recognised the public healthpotential of physical activity, by embarking ona three year national campaign (Active for Life)at promoting the same message.Although a large body of evidence exists

about the health benefits of physical activity,far less is known about the effectiveness ofstrategies to achieve the increases in physicalactivity necessary to acquire these benefits.

In this paper we report a revised and updatedversion of a previous systematic review ofrandomised controlled trials ofphysical activitypromotion in apparently healthy, free livingadults.14 The aim of the paper is to providerecent and reliable information on theeffectiveness of physical activity promotion.There are randomised, controlled trials

using exercise as an intervention to study thephysiological effects of exercise and in themanagement of health problems, notablyhypertension, hyperlipidaemia, and over-weight. These show the effects of exercise onvarious physiological and biological outcomesand demonstrate the importance of exercise inthe management of disease. However, because

the main outcome of such trials is not physicalactivity, they do not help us understand theeffectiveness of physical activity promotionstrategies. For these reasons they were notconsidered for this review.

MethodsComputerised searches were carried out usingMedline, Excerpta Medica, Sport, andSCISearch from 1966-1996. The methoddescribed by Dickersin and colleagues'5 wasused to search for randomised controlledtrials on Medline. Key words for searching in-cluded "exercise", "physical activity", and"Randomised-Controlled-Trial". The searchwas limited to English language journals.Additional searching was carried out using thereferences from both existing reviews1618 andthe papers identified during the search. Inaddition to the studies described previously, afurther 10 studies were found. Those studiesincluded in the previous review were reread byboth of us independently, as were the newstudies identified during this search. Eachpaper was read and assessed using a shortenedversion of the EPI-Centre Review Guide-lines.19The criteria for inclusion of trials in the

review were:

* a control group* subjects assigned to control or interventionby a process of randomisation

* trials testing single factor interventions toincrease activity

* interventions tested on apparently healthy,free living adults

* minimum of 12 weeks duration* exercise behaviour was the dependent

variable

ResultsTen new trials were identified, with threemeeting the inclusion criteria. Two of the 10trials in the earlier review were excluded. Oneof these20 did not meet the new criterion of12 weeks minimum duration and we decidedon rereading that the other did not describethe exercise level of the control grouppostintervention.21 The 11 trials which areincluded in this review are described in table1A and B (studies 5 and 6 are from the samepaper and are reported separately forconvenience). All the trials were from the USA.We did not find any from the United Kingdom

84

on 14 July 2018 by guest. Protected by copyright.

http://bjsm.bm

j.com/

Br J S

ports Med: first published as 10.1136/bjsm

.30.2.84 on 1 June 1996. Dow

nloaded from

Review ofphysical activity promotion strategies

that met the inclusion criteria, although we areaware of some that are in progress. Subjectswere mainly white, middle aged, and welleducated. Most subjects were volunteers,recruited through local advertisements. Thetrials include an even mix of males and femaleswith an age range of 18-72 years (meanapproximately 49).

INTERVENTIONSTable 1 summarises the main exercise com-ponents of the trials and table 2 the results.Both tables are sorted by location (home orfacility) of exercise and then by outcome.Intervention periods ranged from five weeks totwo years. Seven of the trials includedpostintervention follow up periods whichranged from two months to 12 years. Mostoutcomes were analysed on an intention totreat basis. In the trials, subjects were asked toexercise between three and five times per weekfor 20-60 minutes. Few studies described theexercise intensity, but when it was describedthere was a mixture of moderate and vigorousintensities.

Location of exerciseThe location of the prescribed exercise was thehome for seven of the trials (table 1). By "homelocation" we refer to exercise that can takeplace in proximity to the subjects' homes ratherthan within their homes. Five of the homebased trials (studies 1-5) reported a positiveoutcome of the intervention. One of the trials(study 6) not showing a significant differencebetween groups was a comparison betweensubjects receiving telephone contact and thosenot receiving it. All of the subjects weresedentary at baseline and significantlyincreased their exercise level during theintervention. Those subjects receiving tele-phone support exercised more than thosewho did not, but the difference did not reachsignificance. Study 7, the other home basedtrial which did not show a significant differencepostintervention, did not involve giving specificadvice to subjects about increasing theirexercise. Subjects in this trial were given eithera fitness test, a health appraisal, or both andwere given feedback on the results. None of thethree intervention groups exercised more thanthe control group.

Facility based trials normally required thesubjects to attend specific sessions or groups ata local fitness centre or indoor track. Only twoof the five facility based trials showed asignificant difference between interventionsubjects and controls.Study 3 compared home based and facility

based exercise. After one year, subjectsassigned to the two home based armscompleted significantly more of the prescribedexercise sessions than subjects assigned toexercise at a facility (79%, 75%, and 53%respectively), with no significant differencebetween the two home based arms.

Components ofprescribed exerciseIn half of the trials in table 1, walking was theprescribed mode of exercise. All of the trials

showed a significant increase in exercise whencompared to controls. In one study (study 1),80% of subjects were walking an average of atleast five miles per week, with 61% of subjectsadhering to the prescribed level of seven milesper week at two years. Those trials in whichwalking was not recommended (studies 9-12)included exercise to music classes, gym based"endurance activity", and jogging. Only one ofthese trials (study 9) showed an increase inexercise. Subjects were females aged 18-20years, who may have tolerated the prescribedjogging better than the older groups in theother trials.Although the prescribed frequency of ex-

ercise averaged three to five times per week,most subjects were reporting lower frequencyat follow up, with an average two to three timesper week. Study 3 assigned subjects to threeintervention arms of varying frequencies. Oneof the two home based arms prescribed threesessions per week for 40 minutes at a highintensity, while the other home based armprescribed five sessions per week at a lowintensity. The third arm, where subjectsexercised at a local community hall, prescribedthree sessions per week. At one year there wasno significant difference between the two homebased arms on the percentage of prescribedsessions completed, with both completingsignificantly more than subjects in the facilitybased arm. Second year33 follow up data showthat subjects in the three times per week homebased arm were able to maintain significantlyhigher levels of adherence than those in the fivetimes per week home based arm who hadreduced to a level similar to that of the facilitybased arm (68%, 49%, and 36% of prescribedsessions respectively). Although the two homebased arms were prescribed differing intensitylevels, analysis of heart rate data showed thatboth arms actually exercised at an intensitynormally described as moderate.

Strategies for improving complianceA range of behavioural methods was employedto improve compliance. It is difficult tomeasure the effect ofsome ofthese as they wereoften part of multifaceted interventions taughtto all groups. Methods included reinforcement(rewarding subjects for successful completion),self monitoring (keeping personal records ofexercise performed), and relapse preventiontraining (learning to cope with situations thatprompt inactivity and preventing a missedsession leading to a return to preinterventionexercise levels). Some trials investigated theimpact of such strategies with varying results.

In study 4, subjects were randomly assignedto self monitoring, reinforcement, and controlarms. After 18 weeks, subjects in the twobehavioural treatment arms were exercisingsignificantly more than those in the controlarm. Study 11 found no difference in exerciselevels between subjects instructed in selfmonitoring and control subjects. Study 5 tooksubjects from an earlier trial and randomisedthem to two "maintenance" groups withdifferent frequencies of self monitoring.Subjects completing daily self monitoring

85

on 14 July 2018 by guest. Protected by copyright.

http://bjsm.bm

j.com/

Br J S

ports Med: first published as 10.1136/bjsm

.30.2.84 on 1 June 1996. Dow

nloaded from

Hillsdon, Thorogood

Table IA Summary of interventions: home based

Study Authors, year ofpublication, stated Length of Location of Authors'description Prescribedfrequency, intensity, and Controlsobjectives intervention exercise (home ofexercise duration of exercise

orfacility)1 Kriska2 - To examine factors 2 years Home Walking 3 X wk/3 miles per session briskly Assessment only

associated with exercise followingcompliance in post menopausal initial 8women weeks

2 Lombard23 - To determine the 12 weeks Home Walking (group 3 X wk at least 20 min per session Initial instructioneffect of frequency and structure of walkingtelephone prompts on frequency of encouraged)walking

3 King24 - To determine the 1 year 2 groups Walking and Two groups 3 X wk at 73-88% Assessment onlyeffectiveness of group v home home, 1 jogging peak heart rate for 40 min perbased training of higher and lower group facility session, one group 5 X wk atintensities 60-73% peak heart rate for 30 min

each session4 Noland2" - To assess effects of 18 weeks Home Walking, jogging 3 X wk at 30-40% or 60-70% Assessment and advice about

behavioural techniques on and swimming as VO2max for 30 min exercise; no behaviouraladherence to unsupervised exercise preferred treatment

5 King26 - To evaluate strategies for 6 months Home Walking and 4 X wk at 65-77% peak heart rate Same as intervention groupenhancing the maintenance of jogging for 30 min per session but reduced level of selfexercise training by healthy middle monitoringaged men and women (also seeNo 6)

6 King26 - To evaluate strategies for 6 months Home Walking and 4 X wk at 65-77% peak heart rate Same as intervention groupenhancing the adoption of exercise jogging for 30 min per session less regular telephone contacttraining by healthy middle agedmen and women (also see No 5)

7 Godin27 - To investigate the 3 months Home Physical activity None prescribed Assessment onlyeffectiveness of fitness testing and lasting 20-30 minhealth appraisal on exercise per sessionintention and behaviour

forms performed 35% more exercise sessions training who were jogging alone were stillthan subjects completing forms weekly. exercising at three months, compared with

Relapse prevention training was compared 36% (5/12) of those without such training. Bywith reinforcement strategies in a study of contrast, in the two group jogging arms relapsefemales attending exercise classes (study 12). prevention training did not increase joggingSubjects in the relapse prevention arm frequency at follow up.attended weekly lessons on relapse prevention Study 3 investigated the effect of subjects'immediately following an exercise class, while perceptions of whether they had achievedsubjects in the reinforcement group received expected physical or psychological benefitsT shirts and other rewards for successful after six months on subsequent exerciseattendance at a number of classes. Control adherence.34 Those subjects who reported theysubjects simply attended the exercise classes. had achieved expected benefits completedAt 18 weeks there was no difference between more exercise sessions in the next six monthsgroups on number of exercise sessions than those who did not achieve theirattended, with 72% of subjects attending less expectations. It seems that to maintainthan the prescribed three classes per week. adherence in the long term, subjects need to

In a trial of jogging alone or in a group, and perceive a physical or psychological gain fromof jogging with and without relapse prevention exercise.training (study 9), the impact of relapse Perhaps more important than any of theseprevention varied. Eighty three per cent behavioural methods in achieving high rates of(10/12) of subjects with relapse prevention compliance is ongoing follow up.

Table 2A Summary of results: home based

Study Data analysed No in Subjects Post Actualfrequency, intensity, and duration Main outcomes P< 005 Outcomeby "intention to Study intervention of exercise intervention group + or 0treat" follow up

I Yes 229 Post menopausal Annually Mean miles walking/wk = 8-4 Self reported walking level +women aged 50-65 significantly higher at years 1 and 2

compared to controls2 Yes 135 University staff and 12 weeks 46% of frequent prompt groups Frequent telephone contact improved +

faculty members, mean walking 3 X 20 min per week; 13% of adherence to walking prorammeage 40, mainly female low frequency prompts; 4% controls

3 Yes 357 160 women and 197 Ongoing Mean frequency = HIG - 1-2 X wk, Significant difference between +men aged 50-65. HIH -2 X wk, LIH - 3 X wk intervention and control groups plusPredominantly white significant difference between homeand well educated based and facility based groups

4 No 77 28 men (mean age 40) Nil Self monitoring group = mean of Behavioural interventions increased +and 49 women (mean 2 4/week for 26 min frequency of exercise compared toage 36) Reinforcement group = mean of controls

2-5/week for 29 min5 Yes 51 Male and female Nil 11-4 sessions/month for daily self Significant difference in number of +

middle aged subjects monitoring group; 7-5 sessions/month exercise sessions/month betweenfor weekly self monitoring group groups

6 Yes 52 Male and female Nil 124 sessions/month for 32 min in No significant difference in mean 0middle aged subjects telephone group; 9-8 sessions/month number of exercise sessions/month

for 28 min in comparison group between groups. Both groupsincreased exercise frequency overbaseline

7 No 200 Average age 39 (±9) None -2-3 sessions/month No difference between groups 0

+, positive significant difference; 0, no significant difference.

86

on 14 July 2018 by guest. Protected by copyright.

http://bjsm.bm

j.com/

Br J S

ports Med: first published as 10.1136/bjsm

.30.2.84 on 1 June 1996. Dow

nloaded from

Review ofphysical activity promotion strategies

Table lB Summary of interventions: facility based

Study Authors, year ofpublication, stated Length of Location of Authors'description Prescribedffrequency, intensity, and Controlsobjectives intervention exercise (home of exercise duration ofexercise

orfacility)

8 McAuley"' - To determine the 20 weeks Facility Walking 3 X wk, 40 min Initial instruction + exerciseutility of an efficacy based information classesintervention on exerciseparticipation

9 King2' - To study the effect of two 5 weeks Facility Jogging 4 X wk, individualised time and Instructed to jog alonelow cost methods of increasing the distance goalsnumber of participant controlledjogging episodes

10 MacKeen' - To study the effects 18 months Facility and Jogging, swimming, 3 X wk minimum, 35-75 minutes Assessment onlyof an 18 month exercise Home games per sessionintervention on adherence

11 Reid3' - To assess the effectiveness 1 hour Facility Endurance activity Advice about frequency, intensity, Assessment and writtenof physician prescribed exercise duration given but not described exercise adviceproramme with health educationand self monitoring components

12 Marcus32 - To assess effectiveness 18 weeks Facility Exercise to music 35-50 minutes 3 X wk Attendance at exercise group,of a relapse prevention programme class no behavioural techniqueand reinforcement programme inincreasing exercise adherence andshort term maintenance

Follow upTelephone calling was a common method forfollowing up clients in home based trials afteran initial instruction session. All of the homebased trials where researchers maintainedcontact with clients by telephone reportedpositive outcomes. Studies 2 and 6 investigatedthe effect of telephone prompting. Study 2randomised subjects to four levels of telephoneprompting or to a control arm. All subjectsreceived 15 minutes of instruction on walking.At six months there was a significant differencein numbers of subjects still walking betweenthe three prompted arms and the control arm,and between prompt frequency (once per weekversus once every three weeks). Study 6randomly assigned subjects who were waitinglist controls from a previous trial35 to twointerventions, one of which received telephonecontact (10 times during six months). Allsubjects received instructions in behaviouralmethods to improve compliance. Subjects inthe telephone prompting arm exercised morefrequently and for longer than those in thecontrol arm (12A4 sessions/month for 32minutes versus 9-8 sessions/month for 28minutes). This difference did not achievesignificance. Only subjects in the telephonearm significantly increased their fitness.

DiscussionWe have not attempted a formal meta-analysisof the trials in this review since this would be

inappropriate in view of the incompatible dataand varying quality of the trials described. Thisis in accordance with the criteria for attemptinga meta-analysis described by Eysenck.36 Theimportant public health question is whetherevidence exists to guide policy makersconsidering strategies to increase the activitylevels of a sedentary population. Trials thatwere able to demonstrate significant increasesin activity involved exercise that was homebased, ofmoderate intensity, involved walking,and had regular follow up.Walking from home was more successful

than exercise which relied on attendance atstructured exercise sessions. Only two facilitybased trials reported increases in exercise,compared with six of the home based trials. Allthose trials prescribing walking reportedincreases in activity. Moderate intensityactivity was also associated with highercompliance rates. Walking on level ground ata brisk pace would be a moderate intensityactivity for most people.

In Britain, walking is the most popularleisure time physical activity.37 Approximatelyhalf the subjects in a recent national survey"walked continuously for at least a mile at leastonce in the past week. However, only 26% ofmen and 21% of women walked at a brisk orfast pace, and only 14% of men and 17% ofwomen aged 55-74 walked at this pace. The1993 Health Survey for England'2 confirmedthese findings, reporting that 20% of women

Table 2B Summary of results: facility based

Study Data analysed No in Subjects Post Actualfrequency, intensity, and duration Main outcomes P < 005 Outcomeby "intention to Study intervention of exercise intervention group + or 0treat" follow up

8 Yes 125 Previously sedentary, None Not stated Intervention subjects exercised more +45-64 year olds frequently and for longer than controls

9 Yes 58 18-20 year old 2 months Mean frequency JAR and 83% of jogging alone + relapse +previously sedentary G = 2-4/week; GR = 1-4/week subjects still exercising at follow upfemale psychology compared to 36% of control subjectsstudents

10 No 171 Males aged 40-59 with 12 years Mean hours jogging/week at year No difference between exercise and 0CHD risk factors 13 = 0 3 hours control conditions at follow up on

jogging hours per week11 No 124 Male firefighters aged 6 months Not stated No significant difference between 0

24-56 groups at follow up12 Yes 120 Previously sedentary, 2 months Percentage of classes attended during No significant difference in attendance 0

female university the 18 weeks RP = 5 1/%, R = 49%, at 18 weeks or 2 month follow upemployees with a mean Controls = 44%age of 35 years

HIG, high intensity group; HIH, high intensity home; LIH, low intensity home; JAR, jogging alone + relapse prevention; G, group jogging; GR, groupjogging + relapse prevention; RP, relapse prevention; R, reinforcement.

87

on 14 July 2018 by guest. Protected by copyright.

http://bjsm.bm

j.com/

Br J S

ports Med: first published as 10.1136/bjsm

.30.2.84 on 1 June 1996. Dow

nloaded from

Hillsdon, Thorogood

and 30% of men were classified as moderatewalkers (fast or brisk pace), and 38% ofwomenand 32% of men classified as light intensitywalkers (slow or average pace). Brisk walkingis recommended for improving populationactivity levels by the American College ofSports Medicine and the Centres for DiseaseControl and Prevention(USA).'3 In England,the Health Education Authority's Active forLife campaign emphasises the importance ofbrisk walking for improving one's health.A United States survey has shown that

people in lower income groups, older people,women, blacks and Hispanic people par-ticipated in less exercise.38 These dif-ferences were not seen in the numbers whowere walking, which indicates that walkingmay be more universally accessible than othertypes of physical activity. In England, physicalactivity participation is lower in older people,women, those living in council properties,lower education groups,"1 and lower socio-economic groups.39Walking is also associated with a lower injury

rate than other forms of physical activity.40Injuries are reported as a barrier to exerciseparticularly in older age groups." Reviews ofthe determinants of physical activity reportfewer barriers to walking than other types ofphysical activity.Some younger men and most other adults

would improve their physical fitness if theytook up regular brisk walking- (fig 1).42Increases in cardiovascular fitness have beenassociated with reductions in cardiovascularand all-cause mortality.43 A report on thehealth benefits of walking which reviewed theimpact of walking on various cardiovasculardisease risk factors concluded that "regularwalking has the potential to lower bloodpressure, improve the lipid profile, reduce bodyfat, enhance mental well-being and reduce therisk of coronary heart disease."44This review has shown that when walking is

recommended and attendance at a facility isnot required, significant increases in activity

80 r

70 1-

60

50 F-

40 F-

Men Women

30 K-----

20 F-

10'

016-24 25-34 35-44 45-54 55-64 65-74

Age (yee16-24 25-34 35-44 45-54 5

Lines A andB correspond to 45 ml -kg-' min-' and 30 ml -kg-' -min-'. They dirange of values for aerobic fitness which would permit individuals to perform actcosting between S and 7.5 kcal -min-' (moderate intensity) at about 50% of thei

can be achieved. When subjects are followedup regularly the increases can be maintained.Our findings do not support the current

trend in physical activity promotion in thiscountry. There has been a rapid growth ingeneral practitioner (GP) prescription forexercise schemes. Estimates suggest thathundreds of such schemes exist in all parts ofthe country. A 1994 report45 found that a largeproportion of such schemes are leisure centremanaged, and involve GPs referring patients atreduced or no cost for an average period of 10weeks. The report estimated that less than 1%of a GP's patient list was referred into theschemes and also highlighted the fact that "noexamples of good evaluation" were found,preventing any conclusions about effec-tiveness. Although we have been informed ofongoing trials of such schemes, we were unableto find any results published in the scientificliterature. The emphasis placed on attending aleisure facility and the neglect of walking as aform of exercise is inconsistent with thefindings of this review.Most of the studies used volunteers

responding to advertisements to take part in aphysical activity programme. One study (study3) that used random digit dialling as a methodof recruitment only randomised 27% of thoseactually contacted, suggesting a high degree ofself selection.46 These recruitment methods tellus little about how to increase the physicalactivity levels of the vast majority of peoplewho are unlikely to respond to advertisements.The findings of this review should be viewed

with caution as they are based on only 12 trialsall of which were carried out in the USA.

FUTURE RESEARCHThere is an urgent need for experimentalresearch. In particular:* there should be trials undertaken in the

United Kingdom;* trials should include groups other than the

middle aged, middle class, and white;* there is a need for trials specifically dealing

with physical activity in the over 75s;* there is a need for evaluation of GP

prescription schemes by randomised con-trolled trials;

* there is a need to evaluate the effect of GPsadvising their patients to exercise.

CONCLUSION

Levels of physical activity can be increased and--------- the increase can be maintained for at least two

years. Interventions that encourage walkingand do not require attendance at a facility aremost likely to lead to sustainable increases inoverall physical activity. Regular follow up,which need not be time consuming andexpensive, improves the proportion of peopleable to maintain initial increases.

Brisk walking has the greatest potential forincreasing the overall activity levels of a

L-6465 sedentary population and meeting currentpublic health recommendations. It is also the

efine the kind of exercise most likely to be adopted byivities a range of ages, socioeconomic, and ethnicir T0W2max. groups as well as both sexes.

0-

a

E

.

0

a_

Eco

88

on 14 July 2018 by guest. Protected by copyright.

http://bjsm.bm

j.com/

Br J S

ports Med: first published as 10.1136/bjsm

.30.2.84 on 1 June 1996. Dow

nloaded from

Review ofphysical activity promotion strategies 89

In order to increase the attractiveness ofwalking for recreational purposes or as a modeof transport, attention will need to be paidto environmental factors which influencepersonal safety and convenience.

SummaryWe have reviewed randomised controlled trialsofphysical activity promotion to provide recentand reliable information on the effectiveness ofphysical activity promotion. Computeriseddatabases and references of references weresearched. Experts were contacted and askedfor information about existing work. Studiesassessed were randomised controlled trials ofhealthy, free living, adult subjects, whereexercise behaviour was the dependent variable.Eleven trials were identified. No UnitedKingdom based studies were found. Interven-tions that encourage walking and do not requireattendance at a facility are most likely to lead tosustainable increases in overall physical activity.Brisk walking has the greatest potential forincreasing overall activity levels of a sedentarypopulation and meeting current public healthrecommendations. The small number of trialslimits the strength of any conclusions and high-lights the need for more research.

1 Powell KE, Thompson PD, Caspersen CJ, Kendrick JS.Physical activity and the incidence of coronary heartdisease. Annu Rev Public Health 1987;8:253-87.

2 American College of Sports Medicine. Physical activity,physical fitness and hypertension. Med Sci Sports Exerc1993;25:i-x.

3 Helmrich SP, Ragland DR, Leung RW, Paffenbarger RS.Physical activity and reduced occurrence of non-insulindependent diabetes mellitus. N Engl Jf Med199 1;325:147-52.

4 Shinton R, Sagar G. Lifelong exercise and stroke. BMJ1993;307:231-4.

5 Lee I, Paffenbarger RS, Hsieh C. Physical activity and riskof developing colorectal cancer among college alumni. JNad Cancer Inst 199 1;83:1324-9.

6 Drinkwater BL. Physical activity, fitness and osteoporosis.In: Bouchard C, Shephard RJ, Stephens T. Physicalactivity, fitness and health: international proceedings andconsensus statement 1992. Champaign: Human KineticsPublishers, 1994:724-36.

7 North TC, McCullagh P, Vu Tran Z. Effect of exercise ondepression. Exerc Sports Sci Rev 1990;18:379-416.

8 Wood PD, Stefanick ML, Williams PT, Haskell WL. Theeffects on plasma lipoproteins ofprudent weight-reducingdiet, with or without exercise, in overweight men andwomen. NEnglJMed 1991;325:461-6.

9 Berlin JA, Colditz GA. A meta-analysis of physical activityin the prevention of coronary heart disease. Am JEpidemiol 1990;132:639-46.

10 Pooling Project Research Group. Relationship of bloodpressure, serum cholesterol, smoking habit, relativeweight and ECG abnormalities to incidence of majorcoronary events: final report of the pooling project. JChronic Dis 1978;31:202-6.

11 Allied Dunbar National Fitness Survey. Main findings.London: Sports Council and Health EducationAuthority, 1992:46.

12 Office of Population Censuses and Surveys. Health surveyfor England 1993. London: HMSO, 1995.

13 Pate RR, Pratt M, Blair SN, Haskell WL, Macera CA,Bouchard C, et al. Physical activity and public health: arecommendation from the Centers for Disease Controland Prevention and the American College of SportsMedicine. JAMA 1995;273:402-7.

14 Hillsdon M, Thorogood M, Anstiss TA, Morris J.Randomised controlled trials of physical activitypromotion in free living populations: a review. J EpidemiolCommunity Health 1995;49:448-53.

15 Dickersin K, Scherer, Lefebvre C. Identifying relevantstudies from systematic reviews. In: Chalmers I, AltmanDG, eds. Systematic reviews. London: BMJ PublishingGroup, 1995:64-74.

16 Dishman RK. Determinants of participation in physicalactivity. In: Bouchard C, Shephard RJ, Stephens T,Sutton JR, McPherson BD, eds. Exercise, fitness andhealth: a consensus of current knowledge 1988. Champaign:Human Kinetics Publishers, 1990:75-101.

17 King AC, Blair SN, Bild DE, Dishman RK, Dubbert PM,Marcus BH, et al. Determinants of physical activity andinterventions in adults. Med Sci Sports Exerc 1992;24(suppl):S221-36.

18 Dishman RK, Sallis JF. Determinants and interventions forphysical activity and exercise. In: Bouchard C, ShephardRJ, Stephens T, eds. Physical activity, fitness and health:international proceedings and consensus statement 1992.Champaign: Human Kinetics Publishers, 1994:214-38.

19 EPI-Centre. EPI-Centre review guidelines. London: SocialScience Research Unit, London University of Education,London, 1996.

20 Hoyt MF, Janis IL. Increasing adherence to a stressfuldecision via a motivational balance sheet procedure: afield experiment. J Personality Soc Psychol 1975;31:833-9.

21 Suter E, Marti B. Predictors of exercise adoption andadherence of middle-aged sedentary men in a randomisedcontrolled trial. Clin 7 Sports Med 1992;2:261-7.

22 Kriska AM, Bayles C, Cauley JA, Laporte RE, Sandler RB,Pambiano G. A randomised exercise trial in older women:increased activity over two years and the factors associatedwith compliance. Med Sci Sports Exerc 1986;18:557-62.

23 Lombard DN, Lombard TN, Winett RA. Walking to meethealth guidelines: the effect of prompting frequency andprompt structure. Health Psychol 1995;14: 164-70.

24 King AC, Haskell WL, Barr Taylor C, Kraemer HC,DeBusk RF. Group vs home based exercise training inhealthy older men and women. JAMA 1991 ;266:1535-42.

25 Noland MP. The effects of self monitoring andreinforcement on exercise adherence. Res Q Exerc Sport1989;60:216-24.

26 King AC, Barr Taylor C, Haskell WL. DeBusk RF.Strategies for increasing early adherence to and long termmaintenance of home based exercise training in healthymiddle aged men and women. Am J Cardiol 1988;61:628-32.

27 Godin G, Deshamais, Jobin J, Cook J. The impact ofphysical fitness and health age appraisal upon exerciseintentions and behaviour. J Behav Med 1987;10:241-50.

28 McAuley E, Courneya KS, Rudolph DL, Lox CL.Enhancing exercise adherence in middle-aged males andfemales. Prev Med 1994;23:498-506.

29 King AC, Frederiksen LW. Low-cost strategies forincreasing exercise behaviour. Behav Modif 1984;8:3-2 1.

30 MacKeen PC, Rosenberger JL, Slater JS, Channing NW,Buskirk ER. A 13 year follow up of a coronary heartdisease risk factor screening and exercise programme for40-59 year old men: exercise habit maintenance andphysiologic status. y Cardiopulm Rehab 1 985;5:5 10-23.

31 Reid EL, Morgan RW. Exercise prescription: a clinical trial.Amy Public Health 1979;69:591-5.

32 Marcus BH, Stanton AL. Evaluation of relapse preventionand reinforcement interventions to promote exerciseadherence in sedentary females. Res Q Exerc Sport1993;64:447-52.

33 King AC, Haskell WL, Young DR, Oka RK, Stefanick ML.Long term effects of varying intensities and formats ofphysical activity on participation rates, fitness andlipoproteins in men and women aged 50-65 years.Circulation 1995;91:2596-604.

34 Neff KL, King AC. Exercise programme adherence in olderadults: the importance of achieving one's expectedbenefits. Med Exerc Nutr Health 1995;4:355-62.

35 Juneau M, Rogers F, De Santos V, Yee M, Evans A,Bohn A, et al. Effectiveness of self monitored, home basedmoderate intensity exercise training in middle aged menand women. AmJrCardiol 1987;60:60-70.

36 Eysenck HJ. Problems with meta-analysis. In: Chalmers I,Altman DG, eds. Systematic reviews. London: BMJPublishing Group, 1995:64-74.

37 Office of Population Censuses and Surveys. Generalhousehold survey 1990. London: HMSO, 1992.

38 Siegel PZ, Brackbill RM, Heath GW. The epidemiology ofwalking for exercise: implications for promoting activityamong sedentary groups. Am J Public Health 1995;85:706-9

39 Cox BD, Huppert FA, Whichelow MJ. The health andlifestyle survey: seven years on. Aldershot: Dartmouth,1993.

40 Pollock M, Carroll JF, Graves JE, Leggett SH, Braith RW,Limacher M, et al. Injuries and adherence to walk/jog andresistance training programs in the elderly. Med Sci SportsExerc 1991;23:1194-200.

41 Hovell MF, Hofstetter CR, Sallis JF, Rauh MJD,Barrington E. Correlates of change in walking forexercise: an exploratory analysis. Res Q Exerc Sport1992;63:425-34.

42 Killoran AJ, Fentem, Caspersen C. Moving on: internationalperspectives on promoting physical activity. London: HealthEducation Authority, 1994.

43 Blair SN, Kohl HW, Barlow CE, Paffenbarger RS, GibbonsLW, Macera CA. Changes in physical fitness and all-cause mortality. JAMA 1995;273:1093-8.

44 Davidson RRC, Grant S. Is walking sufficient for health?SportsMed 1993-16:369-73.

45 Biddle 5, Fox K, Edmunds L. Physical activity promotion inprimary health care in England. London: Health EducationAuthority, 1994.

46 King AC, Harris RB, Haskell WL. Effect of recruitmlentstrategy on types of subjects entered into a primaryprevention clinical trial. Ann Epidemiol 1994;4:312-20.

on 14 July 2018 by guest. Protected by copyright.

http://bjsm.bm

j.com/

Br J S

ports Med: first published as 10.1136/bjsm

.30.2.84 on 1 June 1996. Dow

nloaded from