Leisure-Time Physical Exercise: Prevalence, Attitudinal Correlates, and Behavioral Correlates among Young Europeans from 21 Countries

Leisure-Time Physical Exercise: Prevalence, Attitudinal Correlates, andBehavioral Correlates among Young Europeans from 21 Countries

Andrew Steptoe, D.Sc.,*,1 Jane Wardle, Ph.D.,* Raymond Fuller, Ph.D.,† Arne Holte, Ph.D.,‡Joao Justo, Ph.D.,§ Robbert Sanderman, Ph.D.,¶ and Lars Wichstrøm, Ph.D.\

*University of London, London, United Kingdom; †Trinity College, Dublin, Republic of Ireland; ‡University of Tromsø, Tromsø, Norway;§University of Lisbon, Lisbon, Portugal; ¶University of Groningen, Groningen, The Netherlands; and \Norwegian University of Science

and Technology, Norway

Background. Increasing leisure time physical exer-cise is a major target of public health programsthroughout the developed world, but few interna-tional comparisons of exercise habits among peoplefrom diverse cultures have been published. The objec-tives of this study were to assess the prevalence ofexercise among young adults from 21 European coun-tries, to analyze associations with health beliefs andrisk awareness, and to investigate relationshipsamong exercise, other health-related behaviors, andemotional well-being.

Methods. The European Health and Behaviour Sur-vey, a questionnaire survey of 7,302 male and 9,181female university students ages 18–30 years from 21countries, was analyzed.

Results. Age-adjusted prevalence of physical exer-cise in the past 2 weeks averaged 73.2% among menand 68.3% among women, but varied markedly frommore than 80% to less than 60% across countrysamples. Beliefs in the health benefits of exercise wereconsistently associated with physical exercise, as wasdesire to lose weight. Awareness of the influence ofexercise on heart disease averaged 52% among menand 54% among women, but was not strongly associ-ated with engagement in exercise. Associations amongexercise, lack of smoking, and sleep time were ob-served, but results for alcohol consumption were in-consistent. Social support and depression were inde-pendently associated with physical exercise.

Conclusions. Physical exercise levels are highly vari-able across samples of relatively privileged young Eu-ropeans from different countries. Associations withother health behaviors and with emotional well-beingsuggest that regular physical exercise is consistentwith a healthy lifestyle. Links with health beliefs areconsistent despite sociocultural differences, but defi-

cient knowledge of the health consequences of a sed-entary lifestyle remains a cause for concern. © 1997 Aca-

demic Press

Key Words: physical exercise; health beliefs; healthbehavior; depression; social support.

INTRODUCTION

Lack of physical activity increases risk of coronaryheart disease [1] and other diseases of major socioeco-nomic impact [2]. The low prevalence of regular leisuretime physical exercise is a cause for concern in manycountries, and encouragement of more active lifestylesis an important component of both national and inter-national public health recommendations [3,4]. Therehave been few international studies of leisure timephysical exercise. Data collected through surveys inindividual countries are difficult to compare because ofvarying time frames and methods of assessment. Thispaper describes results from a survey in which youngEuropean students from 21 countries were questionedabout physical activity using identical methods.

We assessed the prevalence of leisure time physicalexercise over previous 2 weeks in well-educated youngadults across Europe. Gender differences have fre-quently been reported, with young men exercisingmore than young women [5], but it is not known wheth-er this pattern is maintained across cultures. Sincephysical activity is negatively associated with per-ceived health status [6], the impact of persistent healthproblems was also analyzed.

Three supplementary issues were investigated. Thefirst concerns the associations among exercise, healthbeliefs, and risk awareness. Regular physical activityis positively correlated with the perceived health ben-efits of exercise in some [7] but not all studies [8]. Riskawareness, operationalized here as knowledge of theinfluence of exercise on coronary heart disease, is con-sidered a prerequisite of behavior change in manymodels of health promotion and disease prevention [9].International studies provide a unique opportunity to

1 To whom correspondence should be addressed at Department ofPsychology, St. George’s Hospital Medical School, University of Lon-don, Cranmer Terrace, London SW17 ORE, UK. Fax: (44) 181 7672741.

PREVENTIVE MEDICINE 26, 845–854 (1997)ARTICLE NO. PM970224

845

0091-7435/97 $25.00Copyright © 1997 by Academic Press

All rights of reproduction in any form reserved.

assess the robustness of associations between thesecognitive factors and exercise in different socioculturalcontexts.

The second objective was to study the relationshipbetween exercise and other health-related behaviors.The evidence concerning associations with smoking, al-cohol consumption, and sleep patterns is inconclusive[10,11], and the extent to which active people engage inother health behaviors in different cultural settings isnot known. Body weight and the desire to lose weightmay also be influential, so they were assessed in thisstudy. Finally, we studied the associations among ex-ercise, perceived social support, and depressed mood.Regular physical activity is thought to have a positiveimpact on psychological well-being [12] and is nega-tively associated with depression [13]. Social support isalso protective of depression and is positively related toadherence to physical activity programs [14]. The issueof whether associations between physical exercise anddepression are independent of social support wastherefore studied in a subsample of the cohort.

METHOD

These analyses were based on data from the Euro-pean Health and Behavior Survey, a study of healthbehaviors and health beliefs among college studentsfrom 21 European countries carried out between 1989

and 1992 and supported by the Medical and HealthResearch Program of the European Commission [15].The survey instrument has been described elsewhere[16], and only details relevant to the present report willbe presented.

Data were collected from 7,302 men and 9,181women ages 18–30 studying non-health-relatedcourses at universities in the countries listed in Table1. Data collection was based on a common assessmentprotocol that was translated and back-translated into15 languages, with a standard scoring and data man-agement system. Questionnaires were typically com-pleted in classes, although in some countries (France,The Netherlands, Sweden) postal surveys and indi-vidual recruitment were involved. Administration ofthe questionnaire in classes allowed failures of comple-tion to be counted, and response rates varied from 85 to95% in most countries. No information was availableconcerning nonresponders. Participants were told thatthey survey concerned activities related to health.

Physical exercise was assessed by responses to thequestion ‘‘Over the past 2 weeks have you taken anyexercise (e.g., sport, physically active pastime)?’’ Thosewho responded positively were asked what activitythey had carried out and how many times they hadexercised. In subsequent analyses, participants weredivided into those who had exercised one to four timesand those who had exercised five or more times in the

TABLE 1Age-Adjusted Prevalence of Exercise in the Past 2 Weeks: Proportion of Men and Women Who Were Physically Active,

with 95% Confidence Intervals

Country

Men Women

NPhysically

active 95% CI NPhysically

active 95% CI

Austria 368 75% 70–79% 349 75% 70–79%Belgium 501 74% 70–78% 606 62% 58–65%Denmark 392 75% 71–79% 379 81% 76–85%Finland 242 84% 79–90% 458 92% 88–96%France 284 67% 62–72% 354 58% 53–62%Germany (E) 327 76% 71–80% 346 72% 67–76%Germany (W) 386 80% 76–85% 371 79% 74–83%Greece 296 55% 50–60% 343 29% 25–43%Hungary 363 87% 82–91% 361 95% 90–99%Iceland 391 83% 79–87% 381 75% 70–78%Ireland 295 70% 65–75% 458 66% 62–70%Italy 388 74% 69–78% 383 58% 54–62%The Netherlands 246 83% 78–89% 493 83% 79–87%Norway 429 74% 70–78% 733 78% 75–81%Poland 380 76% 71–80% 378 73% 69–77%Portugal 313 52% 47–56% 492 35% 31–38%Spain 354 58% 54–63% 412 36% 32–40%Sweden 305 77% 73–82% 376 85% 80–89%Switzerland 235 83% 78–89% 340 78% 73–82%

UK England 293 66% 61–71% 394 62% 57–65%UK Scotland 117 64% 56–71% 261 63% 57–67%

Overall 6905 73.2% 8668 68.3%

STEPTOE ET AL.846

past 2 weeks. Smoking was measured by questionsadapted from the screening instrument used by theOffice of Population Censuses and Surveys in theUnited Kingdom [17]. In the present analysis, thesample was divided into smokers and nonsmokers onthe basis of reporting any current smoking. Sleepinghabits were measured by asking participants howmany hours of sleep they had on average. It was foundin the Alameda County Study that average sleep timewas an independent predictor of morbidity and mortal-ity [18]. Consequently, sleep times were divided intosatisfactory (7–8 hr) and unsatisfactory (less than 7 hror more than 9 hr per night). Alcohol consumption wasassessed by requesting participants to classify them-selves as nondrinkers, occasional drinkers, or regulardrinkers. Those who drank alcohol indicated how manydays they had consumed alcohol during the previous 2weeks and how many drinks they had consumed onaverage. They were subsequently divided into threecategories: nondrinkers; fewer than one unit per day;and one or more units per day. Desire to lose weightwas assessed on a yes/no format.

Health beliefs about physical exercise were assessedby asking subjects to rate their belief in the importanceto health of taking regular exercise on a 10-point scalefrom 1 4 low importance to 10 4 very great impor-tance. Risk awareness was assessed as part of a largersection of the survey concerned with knowledge of arange of lifestyle factors relevant to health [19].Subjects were asked on a yes/no format whether theybelieved that heart disease was influenced by exer-cise.

Data were collected in two phases with a 1-year in-terval. Preliminary analyses indicated no significantdifferences between the phases, so the cohorts werecombined. In the second phase of data collection, socialsupport and depression were assessed in participantsfrom all countries except Austria, Belgium, Hungary,and Italy. The short form of the Social Support Ques-tionnaire (SSQ) was administered, from which a mea-sure of social support availability was derived from theaverage of the number of people who could be calledupon to support the individual in a range of difficultsituations [20]. A rating of satisfaction with social sup-port was also obtained from the SSQ, but was not in-cluded in the analysis since the distribution was highlyskewed. Depression was assessed using the short ver-sion of the Beck Depression Inventory (BDI) [21]. Thismeasure consists of 13 of the 21 items of the full BDIand correlates highly with the full instrument [22].Analyses of social support and depression were basedon 5,633 and 5,529 individuals, respectively. Healthstatus was assessed by asking participants whetherthey suffered from any persistent health problems, andrespondents also provided information about heightand weight, from which body mass index (BMI) wascalculated [23].

RESULTS

Prevalence of Physical Exercise among Menand Women

Data concerning exercise were available from 6,905men and 8,668 women distributed across countries asshown in Table 1. The mean age of the sample variedsignificantly across countries from 19.1 ± 1.3 years inBelgium to 23.5 ± 2.60 in West Germany. The preva-lence of leisure time physical exercise was thereforeadjusted for age, and the levels in each country sampleare detailed in Table 1. Prevalence varied substantiallyacross samples with the lowest levels among men andwomen being recorded in Greece, Spain, and Portugaland the highest in Hungary and Finland. The confi-dence intervals summarized in Table 1 indicate thatprevalence differed from the overall population meanfor men in Finland, West Germany, Hungary, Iceland,the Netherlands, and Switzerland (with high levels)and in France, Greece, Portugal, Spain, England, andScotland (low levels). Among women, the levels in Aus-tria, Denmark, Finland, West Germany, Hungary, Ice-land, the Netherlands, Norway, Poland, Sweden, andSwitzerland were significantly higher than the overallpopulation mean, while the levels in Belgium, France,Greece, Italy, Portugal, Spain, England, and Scotlandlay significantly below the population average. Thesedata suggest a disappointingly low prevalence of lei-sure time physical exercise among relatively privilegedyoung adults in many European countries.

Men were more likely than women to have exercisedin the previous 2 weeks, but this difference was signifi-cant only in six countries (Belgium, Greece, Iceland,Italy, Portugal, and Spain). Prevalence was somewhathigher among women than men in Denmark, Finland,Hungary, and Sweden, albeit with overlapping confi-dence intervals. Men were more likely than women toengage in frequent physical activity. Overall, 35.5% (CI34.7, 36.9) of men had exercised on five or more occa-sions during the previous 2 weeks, compared with29.7% (CI 28.7, 30.6) of women (P < 0.0001). Whencountry samples were analyzed separately, this trendwas significant in seven cases (Belgium, France,Greece, Iceland, Italy, Portugal, and Spain). In Finlandalone, the reverse trend was apparent (P < 0.0001),since 52.7% of women had exercised five or more timescompared with 35.7% of men.

Overall, 17.0% of men and 22.8% of women reportedpersistent health problems, the most common beingallergies and skin problems. There was no significantassociation between health problems and physical ac-tivity among either sex.

Health Beliefs and Risk Awareness

Mean ratings of beliefs in the importance of takingregular exercise for health maintenance were high, av-

PHYSICAL EXERCISE AMONG YOUNG EUROPEANS 847

eraging 8.14 ± 2.0 among men and 8.19 ± 1.8 amongwomen. Across country samples, average belief ratingsranged from 8.96 ± 1.6 (Hungary) to 7.66 ± 2.4 (Poland)among men and from 9.15 ± 1.3 (Hungary) to 7.73 ± 2.0(Belgium) among women.

A marked association between physical exercise andbeliefs in the health benefits of exercise was observedamong men and women. This is illustrated in Fig. 1,which summarizes the proportion of physically active

individuals among people making ratings at each levelof the 10-point belief scale. There was a gradual in-crease in the prevalence of exercise as beliefs becamestronger.

The proportion of individuals who were aware of theassociation between exercise and heart disease in eachcountry sample is shown in Fig. 2. Overall, 52% of menand 54% of women were aware of the role of exercise,with striking country differences. More than 70% were

FIG. 1. Health beliefs and physical exercise. Age-adjusted proportion of physically active men and women in the sample of respondentsat each of the 10 points of the scale on which beliefs in the importance of regular exercise for health were rated from 1 4 low importanceto 10 4 very great importance.

FIG. 2. Exercise and heart disease. Age-adjusted proportion of men and women aware for the influence of exercise on heart disease ineach country sample. Au, Austria; Be, Belgium; Dk, Denmark; Fi, Finland; Fr, France; eG, former DDR; wG, Germany; Hu, Hungary; Ic,Iceland; Ir, Republic of Ireland; It, Italy; Nl, The Netherlands; No, Norway; Po, Poland, Pt, Portugal; Sp, Spain; Sw, Sweden; Sz, Switzerland;Eng, UK–England; and Sc, UK–Scotland.

STEPTOE ET AL.848

aware of the association in Denmark, Finland,the Netherlands, and Norway, but fewer than 40% inBelgium, Greece, Italy, and Poland. No consistent sexdifferences were observed. There was a modest asso-ciation between physical exercise and risk awareness,in that across all country samples 54.7% of the exer-cisers compared with 51.0% of sedentary individualsknew of the link with heart disease (P < 0.0001). How-ever, this association was not significant in any coun-try sample individually and was small in public healthterms.

Multiple logistic regression was carried out to assessthe independent contribution of belief ratings and riskawareness to participation in physical exercise. Ageand sex were included in the model, which was testedfor each country sample separately. For the purposes ofthis analysis, belief ratings were divided into low (1–5)and high (6–10) categories, and the results are detailedin Table 2. Findings for beliefs were very consistent. Ineach country, the estimated odds of physical exercisewere increased for those with a high as opposed to alow belief in the importance of exercise for health, ad-justed for age, sex, and risk awareness. All the beliefodds ratios were significant (P < 0.025 to P < 0.0001).By contrast, the odds ratios of physical exercise amongindividuals who were aware of the association betweenexercise and heart disease were significant in only onecountry sample (Portugal, P 4 0.03).

Physical Exercise and Other Health Behaviors

Lack of physical exercise was associated with ciga-rette smoking among both men and women; 40.7% ofinactive men were smokers, compared with 29.8% ofactive individuals (P < 0.0001), and the correspondingproportions were 34.4 and 26.8% in women (P <0.0001). The association was significant in nine indi-vidual country samples for men and in five samples forwomen. The proportion of smokers among respondentswho had exercised one to four times and five or moretimes over the previous two weeks was also calculated.These data are summarized in Fig. 3 and indicate thatamong men a consistent trend was present (P <0.00001), with fewer smokers among those who exer-cised frequently. The pattern was less striking amongwomen, although the trend remained statistically sig-nificant (P < 0.0001). In individual country samples,the relationship was significant in seven samples formen and in two samples for women.

Overall, reported sleep time was outside the 7- to8-hr ‘‘satisfactory’’ category among 18.4% of men and16.1% of women. A significantly larger number of ex-ercisers than sedentary individuals were categorizedas having satisfactory sleep levels among both men(83.0% vs 77.7%, P < 0.00001) and women (85.0% vs81.4%, P < 0.0001). In the analysis of individual coun-try samples, effects were significant in only two cases

TABLE 2Predictors of Exercise Behavior: Estimated Odds Ratios (OR) and 95% Confidence Intervals (CI) of Likelihood of Exercise forParticipants with High vs Low Beliefs in the Importance of Regular Exercise, and High vs Low Awareness of the Influenceof Exercise on Heart Disease, Controlling for Sex and Age

Country

Belief in the importance of regularexercise for health

Risk awareness(exercise and heart disease)

OR 95% CI OR 95% CI

Austria 6.46 (3.82, 10.9) 1.07 (0.899, 1.28)Belgium 5.04 (3.59, 7.08) 1.08 (0.936, 1.25)Denmark 7.99 (5.39, 11.9) 1.09 (0.863, 1.37)Finland 6.06 (2.60, 14.2) 1.14 (0.875, 1.50)France 7.89 (4.62, 13.5) 1.14 (0.962, 1.36)Germany (East) 9.78 (6.00, 15.0) 0.85 (0.699, 1.02)Germany (West) 4.00 (2.39, 6.66) 0.96 (0.941, 1.01)Greece 3.74 (1.99, 7.04) 0.99 (0.833, 1.17)Hungary 8.53 (3.70, 19.6) 0.82 (0.651, 1.11)Iceland 3.81 (2.25, 6.45) 1.04 (0.861, 1.27)Ireland 4.50 (2.67, 7.60) 0.92 (0.786, 1.08)Italy 2.54 (1.12, 5.76) 0.93 (0.782, 1.10)The Netherlands 6.11 (2.83, 13.2) 1.09 (0.876, 1.35)Norway 6.20 (4.25, 9.04) 1.00 (0.857, 1.17)Poland 2.75 (1.82, 4.17) 0.82 (0.673, 1.01)Portugal 6.00 (2.81, 12.8) 1.18 (1.02, 1.36)Spain 5.93 (3.03, 11.6) 0.96 (0.825, 1.14)Sweden 9.90 (5.60, 17.5) 0.99 (0.812, 1.21)Switzerland 4.56 (2.62, 7.95) 0.83 (0.668, 1.03)UK

England 5.54 (3.26, 9.41) 0.93 (0.771, 1.12)Scotland 18.25 (6.27, 53.2) 0.84 (0.699, 1.06)

PHYSICAL EXERCISE AMONG YOUNG EUROPEANS 849

for each sex, but the small proportion of unsatisfactorysleep times gave these analyses limited power. Therewere no clear associations with the frequency of physi-cal exercise.

Alcohol consumption levels were typically low: only27.3% of men and 12.5% of women drank an average ofone or more units per day while 34.5% of men and49.2% of women were nondrinkers. There was a sig-nificant association between exercise and alcohol con-sumption among women in five country samples (Hun-gary, Ireland, Poland, Portugal, and Scotland). In eachcase, exercisers were less likely to drink alcohol thansedentary women. This association was not consistentamong men and was statistically significant in only onecountry sample (Poland).

Overall, the BMI of men was significantly greaterthan that of women (means 22.4 ± 2.5 vs 20.8 ± 2.4, P< 0.0001). Associations with physical exercise wereanalyzed by dividing participants into low (BMI <20),average (BMI 20–25), and high (BMI > 25) categories.There was no association with physical activity amongwomen, since similar proportions of women in each cat-

egory exercised. An association was observed amongmen, with an average of 75.6% (CI 77.4, 76.7) of thosein the average weight category being physically active,compared with 70.4% (CI 67.3, 73.5) of the high BMIgroup and 64.2% (CI 61.2, 67.3) of the low BMI group.This association was significant in 8 of the 21 countrysamples of men when analyzed separately.

A desire to lose weight was reported among 17.4% ofmen and 44.1% of women. Across the entire cohortthere were strong associations between physical exer-cise and desire to lose weight among both sexes as canbe seen in Fig. 3 (P < 0.0001). The association wassignificant in seven of individual country samples forwomen but only in one case for men.

Physical Exercise, Social Support, and Depression

The SSQ was completed by 2,042 men and 3,591women, while BDI scores were available from 2,091men and 3,438 women. The analysis of these instru-ments was limited to the combined sample, since theindividual country samples were not large enough for

FIG. 3. Proportion of current cigarette smokers (top left), individuals wanting to lose weight (bottom left), people enjoying high socialsupport (top right), and those with depression ratings >4 (bottom right) among inactive respondents (None), people exercising between oneand four times in the past 2 weeks (1–4), and five or more times (5+). Age-adjusted means and 95% confidence intervals. m, men; d, women.

STEPTOE ET AL.850

separate analysis. The number of supportive individu-als averaged 3.01 ± 1.83 among men and 3.73 ± 1.78among women. For the purposes of analysis, the popu-lation was divided into those who reported 0–2 sup-ports (low support) and >3 (high support). Figure 3shows the significant association between physical ex-ercise and social support, with physically active indi-viduals being more likely to enjoy high social support(P < 0.00001).

Scores on the BDI were skewed toward low values,averaging 3.7 ± 4.4 among men and 3.9 ± 4.3 amongwomen. The population was divided into those with low(0–4) and moderate or high (>4) BDI ratings for analy-sis. Associations with exercise are summarized in Fig.3. It is apparent that among both men and women, thefrequency of moderate depression scores declined withincreasing levels of physical exercise (P < 0.0001).Among men, 24.1% of those exercising five or moretimes had moderate or high depression scores, com-pared with 33.6% of inactive males. The correspondingproportions for women were 28.0% vs 42.2%.

Social support and depression were negatively cor-related among men and women (r 4 −0.222 and−0.202, respectively, P < 0.0001). Logistic regressionwas therefore carried out to assess whether the asso-ciations with physical exercise were independent of oneanother. Age and sex were included in the model. Thisanalysis revealed significant independent associationsfor both variables with exercise, with the estimatedodds adjusted for age, sex, and depression of high vslow social support being 1.37 (CI 1.29, 1.46) and theodds of exercise for low vs moderate or high depressionbeing 1.19 (CI 1.12, 1.27).

DISCUSSION

International comparisons of behaviors relevant tohealth are important for a number of reasons. First,there are wide variations in Europe in the incidence ofpremature morbidity and mortality from coronaryheart disease and other disorders [24], and it is valu-able to determine the extent to which these arematched by lifestyle factors. Knowledge about healthbehaviors and their correlates is essential to the plan-ning of services and preventive programs. The issue isparticularly pertinent to Europe, where the Single Eu-ropean Community Act has reduced barriers to popu-lation mobility. Proposals to extend the EuropeanUnion to Eastern European countries where the pat-tern of morbidity is very different from that in WesternEurope [25] further emphasizes the need for interna-tional data.

The European Health and Behavior Survey was car-ried out with university students. The data cannot ofcourse be considered representative for each country,since university students are typically healthier andbetter educated than other sectors of society [26]. Al-though students of health-related topics were ex-

cluded, the national samples may vary considerably inother characteristics.Nevertheless, the sample has dis-tinct advantages for international comparisons. Uni-versity students are a relatively homogeneous group interms of education and socioeconomic status, andphysical activity is positively associated with both fac-tors [27]. The use of student samples limits the vari-ance in physical activity associated with illness anddisability [28]. Furthermore, university students forma privileged sector of society, and social diffusion pro-cesses may operate by which their behavior and atti-tudes are gradually transmitted to less advantagedgroups.

The response rate to this survey was high, averagingabout 80% in the different country samples. Unfortu-nately, we were not able to collect information aboutnonresponders. It is unlikely that students who re-fused to participate had any particular orientation inrelation to physical exercise, since exercise was onlyone of many issues addressed in the survey. However,since the study was introduced to potential partici-pants as a survey of activities related to health, it ispossible that nonresponders had less interest than re-sponders in health issues. Inasmuch as university stu-dents associate physical exercise with health, the datamay therefore by biased toward more favorable atti-tudes to exercise.

The results concerning age-adjusted prevalence ofphysical exercise over the previous 2 weeks showedwide variations across country samples (Table 1).There was a marked similarity in the proportion ofmen and women who were physically active in eachcountry sample, since the correlation between age-adjusted percentages was high (r 4 0.92, P < 0.001).Low levels of physical exercise were recorded in thesouthern European countries, with Portugal, Spain,and Greece having especially low prevalence amongboth men and women. A high prevalence was morecommon in Scandinavia, notably in Finland and Swe-den. These differences do not reflect seasonal effects,since data were collected over several months in eachcountry. High levels of physical exercise were recordedin Eastern European countries (particularly in Hun-gary), reflecting a tradition of sport and exercise par-ticipation in this part of the world. Levels in Englandand Scotland were disappointing in view of the effortsthat have been made to publicize the value of physicalactivity [29].

Direct comparisons with other data sets cannot bemade with confidence. The closest comparison is withthe WHO cross-national study of health behaviors inschool-age children, which assessed physical activityamong 15-year-olds in 11 countries [30]. The highestprevalence of exercise out of school was recorded inAustria, Hungary, and Norway, with low levels inSpain and Poland. However, 15-year-olds are moreconstrained by school activities and family habits than

PHYSICAL EXERCISE AMONG YOUNG EUROPEANS 851

are independent university students. Other interna-tional comparisons have been confined to elderly popu-lations [31].

Leisure time physical exercise is one of the fewhealth-related behaviors that are typically more preva-lent among men than women. Our results endorse thispattern, but suggest that the difference is not univer-sal. In four country samples, the age-adjusted preva-lence was higher among women than men, indicatingthat cultural factors may influence these differences.Discussion of the types of physical exercise carried outis beyond the scope of this paper, but the most fre-quently reported activities in most countries were run-ning or jogging, football, aerobics, and swimming. Thehigher prevalence among men may in part reflect agreater involvement in college-based sports.

The association between beliefs in the health ben-efits of regular exercise and physical activity was veryreliable, with a consistent trend of increasing partici-pation with stronger beliefs (Fig. 2). Beliefs in thehealth benefits of exercise predicted physical exercisein every country sample, independently of age, sex, andrisk awareness (Table 2). These associations confirmedthe relationships that have been observed for a widerange of behaviors and beliefs in the European Healthand Behavior Survey [15] and in other studies [32]. Theresult is notable for two reasons. First, robust associa-tions between health beliefs and behaviors were main-tained across a range of diverse cultures, with theprevalence of the behavior varying from 35 to 95%. Thecorrelation with health beliefs does not therefore de-pend on physical exercise exceeding any particularprevalence threshold. Second, health does not rankhigh among the motives for exercise among youngpeople. Factors such as having fun (30), leading an ex-citing life [33], appearance, and socializing [34] are en-dorsed more frequently as reasons for exercise thanhealth benefits. A causal interpretation of the associa-tion between health beliefs and exercise cannot bemade from this cross-sectional survey. Nevertheless, ina follow-up on a subgroup of participants, we havefound that beliefs in health benefits predicted changesin physical exercise over a 12-month period [35]. Con-sequently, health beliefs may be a legitimate target forintervention.

By contrast, the association between awareness ofthe health consequences of inactivity and behavior wasweak. Health knowledge was assessed in this analysisby awareness that physical exercise influences risk ofheart disease, since this association is well establishedin longitudinal studies [1]. The results show a disap-pointing level of awareness among this cohort of edu-cated young Europeans. Fewer than 60% of men andwomen were aware of the influence of exercise in themajority of country samples, and even in the mostknowledgeable sample (Denmark) some 20% were evi-dently unaware of the link (Fig. 2). These data high-

light a serious shortfall in knowledge about basichealth risks associated with a sedentary lifestyle. Theresult is of particular concern in that data were ob-tained from a well-educated sector of the population.At the same time, it should be noted that the associa-tion between risk awareness and engagement in physi-cal exercise is modest and not statistically significantin the majority of country samples (Table 2). Thesefindings suggest that although there is room for sub-stantial improvement in knowledge about the specifichazards of inactivity, deficient knowledge is not in it-self a major determinant of physical exercise amongyoung adults.

Observations concerning the associations betweenphysical exercise and other health behaviors have beenmixed. For example, many studies have shown thatexercise is inversely associated with cigarette smoking[10,36], but this was not confirmed in the U.S. NationalAdult Fitness Survey [37] and other surveys [38,39].Associations between exercise and smoking may beconfounded by socioeconomic status, since occupationalactivity is greater among lower status groups in whichsmoking is also more prevalent. The relationship be-tween exercise and alcohol consumption is variableacross studies and between sexes [11,40]. Many au-thorities have emphasized that there are distinct di-mensions of health behavior and that physical activitydoes not load on the same factor as smoking, alcoholconsumption, or sleep time [41–43]. In the presentstudy, regular physical exercise was associated with alower likelihood of smoking and with sleeping 7–8 hrper night. Physical exercise was also negatively relatedto alcohol consumption among women but not men.Team sporting events are frequently followed by drink-ing in the bar, and this may offset any salutagenic ef-fects of exercise.

Overweight individuals are more likely to drop out ofexercise programs and are less responsive to publichealth interventions than others [14,44]. In the pre-sent study, no simple association between body massindex and physical exercise was observed. Amongwomen, body mass index was not related to physicalexercise; instead desire to lose weight was a strongpredictor (Fig. 3). Weight control has previously beenidentified as a common motive for physical activityamong young women [34] and is not confined to thosewho are overweight by objective criteria. Among men, acurvilinear association emerged between body mass in-dex and physical exercise, with higher levels of engage-ment among individuals with a BMI in the averagerange (20–25). The lower prevalence of exercise in sub-jects with high BMI may reflect the pattern docu-mented in earlier studies. Men with low BMI may beless active because they are less likely to become in-volved in team sport than others.

The final set of analyses confirmed the associationbetween lack of physical exercise and depressed mood

STEPTOE ET AL.852

documented in other studies (Fig. 3). We also foundthat this was independent of the influence of socialsupport. The latter observation is significant, since so-cial isolation may heighten depressed mood while alsoreducing the likelihood of regular vigorous activity.Unfortunately, the limited numbers completing thispart of the survey prevented separate analysis of coun-try samples, so it is not possible to draw conclusionsabout the consistency of this finding across cultures.The result is consistent with longitudinal studiesshowing that regular physical activity may enhanceemotional well-being [45].

The limitations of this study should be noted [15].The cross-sectional design precludes any causal infer-ences from being drawn. The results are based on self-report, and the physical exercise measure included arange of sports and recreations that vary in their en-ergy demands. Sampling was restricted to a privilegedsector of young adults and to two or three centerswithin each country. It is possible that differences inthe type of student included in each country samplecontributed to the pattern of results. Nevertheless, theresults point to wide variations between socioculturalgroups that are a cause of concern and that merit fullerinvestigation. The associations with health-related be-haviors and emotional well-being suggest that regularphysical exercise is a behavior that is consistent with ahealthy lifestyle across cultures. The consistency of as-sociations between physical exercise and health beliefsfurther justifies attention to attitude change in preven-tive programs.

ACKNOWLEDGMENTS

This research was carried out within the Concerted Action onBreakdown in Human Adaptation: Quantification of Parameters,part of the Commission of the European Communities Biomedicaland Health Research Programme. Statistical analysis was supportedwith a grant from the Economic and Social Research Council, UK.The following colleagues contributed to the European Health andBehavior Survey: Austria, Professor Margit Koller and Professor Eli-sabeth Groll-Knapp (Vienna); Belgium, Professor Jan Vinck (Diep-enbeek); Denmark, Dr. Donald Smith (Århus); Finland, Dr. MarttiTuomisto (Tampere) and Dr. Raimo Lappalainen (Kuopio); France,Dr. France Bellisle and Dr. Marie-Odile Monneuse (Paris); East Ger-many, Dr. Konrad Reschke (Leipzig); West Germany, Dr. ThomasKohler (Hamburg), Professor Gudrun Sartory (Wuppertal) and Dr.Claus Vogele (Marburg); Greece, Professor Nicola Paritsis (Iraklion)and Dr. Bettina Davou (Ioanina); Hungary, Professor Maria Koppand Dr. Arpad Skrabskı (Budapest); Iceland, Professor ErlendurHaraldsson (Reykjavik); Italy, Dr. Anna Maria Zotti and Dr. GiorgioBertolotti (Veruno); Poland, Dr. Zbigniew Zarczynski (Krakow) andProfessor Andrzej Brodziak (Bytom); Spain, Professor Jaime Vila(Granada); Sweden, Professor Mats Fredrikson (Uppsala); Switzer-land, Professor Ruth Burckhardt and Dr. Laurent Rossier (Lau-sanne). We are also grateful for the assistance of Heather Smith inthe analysis of these data.

REFERENCES

1. Whaley MH, Blair SN. Epidemiology of physical activity, physi-cal fitness and coronary heart disease. J Cardiovasc Risk 1995;2:289–95.

2. Bouchard C, Shephard RJ, Stephens T. Physical activity, fitness,and health. Champaign: Human Kinetics Publications, 1994.

3. WHO/FIMS Committee on Physical Activity for Health. Exercisefor health. Bull World Health Org 1995;73:135–6.

4. Healthy People 2000: national health promotion and diseaseprevention objectives. Boston: Jones and Bartlett, 1992.

5. Allied Dunbar National Fitness Survey. London: The SportsCouncil and the Health Education Authority, SC/81/3M/6/92,1992.

6. Dishman RK, Sallis JF, Orenstein D. The determinants of physi-cal activity and exercise. Pub Health Rep 1985;100:158–71.

7. Sallis JF, Hovell MF. Determinants of exercise behavior. Exer-cise Sports Sci Rev 1990;18:307–30.

8. Lee C. Attitudes, knowledge, and stages of change: a survey ofexercise patterns in older Australian women. Health Psychol1993;12:476–80.

9. Bettinghaus EP. Health promotion and the knowledge—attitude–behavior continuum. Prev Med 1986;15:475–91.

10. Shephard RJ. Exercise and lifestyle change. Br J Sports Med1989;23:11–22.

11. Blair SN, Jacobs DR, Powell KE. Relationships between exerciseor physical activity and other health behaviors. Pub Health Rep1985;100:172–80.

12. Stephens T. Physical activity and mental health in the UnitedStates and Canada: evidence from four population surveys. PrevMed 1988;17:35–47.

13. Camacho TC, Roberts RE, Lazarus NB, Kaplan GA, Cohen RD.Physical activity and depression: evidence from the AlamedaCounty Study. Am J Epidemiol 1991;134;220–31.

14. Dishman RK. Determinants of participation in physical activity.In: Bouchard C, Shephard RJ, Stephens T, Sutton JR, McPher-son BD, editors. Exercise, fitness, and health. Champaign: Hu-man Kinetics Books, 1990:75–101.

15. Steptoe A, Wardle J. The European Health and Behaviour Sur-vey: the development of an international study in health psy-chology. Psychol Health 1996;11:49–73.

16. Wardle J, Steptoe A. The European Health and Behavior Sur-vey: rationale, methods and initial results from the United King-dom. Soc Sci Med 1991;33:925–36.

17. Steptoe A, Wardle J, Smith H, Kopp M, Skrabski A, Vinck J, etal. Tobacco smoking in young adults from 21 European coun-tries: association with attitudes and risk awareness. Addiction1995;90:571–82.

18. Wingard DL, Berkman LF, Brand RJ. A multivariate analysis ofhealth-related practices—a nine-year mortality follow-up of theAlameda County study. Am J Epidemiol 1982;116:765–75.

19. Steptoe A, Wardle J. What the experts think: a European studyof expert opinion about the influence of lifestyle on health. Eur JEpidemiol 1994;10:195–203.

20. Sarason IG, Sarason BR, Shearin EN, Pierce G. A brief measureof social support—practical and theoretical implications. J SocPers Relationships 1987;4:497–510.

21. Beck AT, Beck RW. Screening depressed patients in family prac-tice. Postgrad Med 1972;52:81–5.

22. Reynolds WM, Gould JW. A psychometric investigation of thestandard short form Beck Depression Inventory. J Consult ClinPsychol 1981;49:306–7.

23. Bellisle F, Monneuse M-O, Steptoe A, Wardle J. Weight concernsand eating patterns: a survey of university students in Europe.Int J Obes 1995;19:723–30.

24. Holland WW. European community atlas of ‘‘avoidable death.’’2nd ed. Oxford: Oxford Univ. Press, 1991.

PHYSICAL EXERCISE AMONG YOUNG EUROPEANS 853

25. Lopez AD. Assessing the burden of mortality from cardiovascu-lar diseases. World Health Stat Q 1993;46:91–96.

26. Patrick K, Grace TW, Lovato CY. Health issues for college stu-dents. Ann Rev Public Health 1992;13;253–68.

27. Stephens TS, Schoenborn CA. Adult health practices in theUnited States and Canada. Washington: National Center forHealth Statistics, PHS 88-1479, 1988.

28. Gottlieb NH, Green LW. Life events, social network, life-styleand health: an analysis of the 1979 National Survey of PersonalHealth Practices and Consequences. Health Educ Q 1984;11:91–105.

29. Health Promotion Research Trust. Fit for life. Cambridge:Health Promotion Research Trust, 1989.

30. King AJC, Coles B. The health of Canada’s youth. Ministry ofSupply and Services, H39-239/1992E, 1992.

31. Osler M, de Groot LC, Enzi G. Life-style: physical activities andactivities of daily living. Eur J Clin Nutr 1991;45:139–51.

32. Sallis JF, Simons-Morton BG, Stone EJ, Corbin CB, Epstein LH,Faucette N, et al. Determinants of physical activity and inter-ventions in youth. Med Sci Sports Exerc 1992;24:S248–57.

33. Eiser JR, Gable P. Health behavior as goal directed action. JBehav Med 1988;11:523–35.

34. Silberstein LR, Striegel-Moore RH, Timko C, Rodin J. Behavior-al and psychological implications of body dissatisfaction: do menand women differ? Sex Roles 1988;19:219–32.

35. Steptoe A, Sanderman R, Wardle J. Stability and changes inhealth behaviours in young adults over a one year period. Psy-chol Health 1995;10:155–69.

36. Conway TL, Cronan TA. Smoking, exercise, and physical fitness.Prev Med 1992;21:723–34.

37. President’s Council on Physical Fitness and Sports. Nationaladult physical fitness survey. Phys Fitness Res Digest 1974;4:1–27.

38. Marti B, Tuomilehto J, Salonen JT, Puska P, Nissinen A. Rela-tionship between leisure-time physical activity and risk factorsfor coronary heart disease in middle-aged Finnish women. ActaMed Scand 1987;222:223–30.

39. Perrier. The Perrier study: fitness in America. New York: Per-rier, 1979.

40. Heath GW, Kendrick JS. Outrunning the risks: a behavioral riskprofile of runners. Am J Prev Med 1989;5;347–52.

41. Bausell CR, Bausell RB. The internal structure of health-seeking behavior. Eval Health Professions 1987;16:460–75.

42. Sobal J, Revicki D, DeForge BR. Patterns of interrelationshipsamong health-promotion behaviors. Am J Public Health 1992;8:351–9.

43. Norman RMG. Studies of the interrelationships amongst healthbehaviors. Can J Public Health 1985;76:407–10.

44. Siegel PZ, Brackbill RM, Heath GW. The epidemiology of walk-ing for exercise: implications for promoting activity among sed-entary groups. Am J Public Health 1995;85:706–10.

45. Steptoe A, Edwards S, Moses J, Mathews A. The effects of exer-cise training on mood and perceived coping ability in anxiousadults from the general population. J Psychosom Res 1989;33:537–47.

STEPTOE ET AL.854