Epidemiology

Effect of diet and physical exercise interventionprogrammes on coronary heart disease risk insmoking and non-smoking men in Sweden

Gorel K Naslund, Mats Fredrikson, Mai-Lis Hellenius, Ulf de Faire

AbstractStudy objective - To investigate differencesbetween smokers and non-smokers inhealth behaviour, cardiovascular risk fac-tors, coronary heart disease (CHD) risks,health knowledge, health attitudes, andcompliance with a CHD prevention pro-gramme.Design - Differences between smokers andnon-smokers were studied via medical ex-aminations, questionnaires, physical ex-ercise activity logs, and food record sheets.Data were analysed using univariate andmultivariate analyses. The five and 10 yearCHD risks were assessed using the Fram-ingham CHD risk estimate.Setting - The Karolinska Hospital, Stock-holm, and Sollentuna Primary HealthCentre, Sollentuna, Sweden.Participants - The analyses were basedon 158 healthy smoking and non-smokingmen aged 35-60 years with raised cardio-vascular risk factors who enrolled in con-trolled, randomised six month diet andexercise programmes.Main results - Discriminant analysis sug-gested that smokers, compared with non-smokers, were characterised by a higheralcohol energy percent, lower HDL cho-lesterol concentration, lower systolic bloodpressure, and a higher plasminogen ac-tivator inhibitor-i (PAI-1) value. Know-ledge of the risk factors for CHD was nota discriminating factor. Both smokers andnon-smokers increased the exercise taken,improved their diet, and lowered theirCHD risk. Before, as well as after theintervention, smokers had a higher CHDrisk than non-smokers.Conclusions - The best CHD preventionaction that could be taken by smokerswould of course be to quit smoking. Thosewho cannot stop should be encouraged toimprove their diet and increase theamount of physical exercise they take inorder to reduce the health hazards oftheirsmoking behaviour.

(J7 Epidemiol Community Health 1996;50: 131-136)

Numerous studies show differences in the dietof smokers compared with non-smokers. It has

been reported that smokers have lower intakesof fibre, vitamin C, vitamin A, polyunsaturatedfat, protein, and carbohydrates.'2 Smokershave been found to drink more coffee andalcohol,34 and to add more salt to their food.56It is possible that smoking influences a person'sfood preferences by an effect on taste7 or thatsmokers pursue a more "unhealthy" lifestylethan non-smokers. Cockburn, for example, hasreported less use of sun protection measuresin smokers than in non-smokers.8 With regardto physical exercise, data are conflicting. Anumber of studies show no association betweensmoking and physical exercise, while somestudies show a weak negative relationship orhave reported smoking as a characteristic fordrop outs in an exercise programme.91O Castroet all' found that moderate to heavy smokersexhibit lower levels of health consciousnessand lower commitment to enact healthy be-havioural changes than non-smokers and lightsmokers.Smoking is associated with a number of

physiological changes which increase theCHD risk. Compared with non-smokers,smokers have raised serum concentrations ofcholesterol, serum triglycerides, and VLDLtriglycerides, and attenuated HDL cho-lesterol. 12-5 It has also been suggested that thedecreased ingestion of fibre could contributeto the increased risks of CHD and cancer insmokers.4 Furthermore, it has been shown thata diet low in dietary fibre is linked to highplasminogen activator inhibitor-i (PAI-1) val-ues, thereby impairing the fibrinolytic capacityand adding to the CHD risk.'6

In this study, we wished to establish thedietary pattern, cardiovascular risk factors,CHD risk, health beliefs, and health knowledgein smoking and non-smoking middle aged men.Because smoking is linked to a different intakeof nutrients,2 as well as to a low commitmentto improve health behaviour," our aim was tostudy the result of six month diet and exerciseprogrammes among smokers and non-smokers.The diet and exercise intervention programmehas been described previously. Men in the inter-vention groups significantly reduced a numberof risk factors for cardiovascular disease andlowered their estimated CHD risk comparedwith the control group. Diet and exercise wereabout equally effective in reducing cardio-

Department ofClinical Neuroscience,Section of Psychology,Karolinska HospitalZ 6, S-171 76Stockholm, SwedenG K Naslund

Department ofClinical Psychology,Uppsala University,S-751 42 Uppsala,SwedenM Fredrikson

Department ofMedicine, KarolinskaHospital, S-171 76Stockholm, SwedenM-L Hellenius

Department ofEnvironmentalMedicine, KarolinskaInstitute, S-171 77Stockholm, SwedenU de Faire

Correspondence to:Mr K Naslund.Accepted for publicationSeptember 1995

131_7ournal of Epidemiology and Community Health 1996;50: 131-136

Ndslund, Fredrikson, Hellnius, de Faire

vascular risk.'7 The theoretical foundation forthe present study was the health belief model'8which stresses psychosocial factors as de-terminants of health behaviour.

MethodsSUBJECTSA CHD prevention programme which com-bined a population and individual high riskstrategy was started in 1988 in Sollentuna,Greater Stockholm.'9 All adult visitors to theSollentuna Primary Health Care Centre whowere under 60 years of age were given theopportunity to fill out a short questionnaireon possible CHD risk factors. Those whosequestionnaire indicated the presence of one ormore risk factors were offered a free checkup of physiological risk factors and possibletreatment. Between August 1988 and April1990, 3000 visitors to the Sollentuna Centre(11% of all visitors in the age group 15-60years) completed the questionnaire.

Letters with invitations to participate in arandomised, controlled study on the effects ofnon-pharmacological treatment on CHD riskfactors were sent to a consecutive sample of187 men aged 35-60 years with slightly tomoderately raised cardiovascular risk factorswho were newly registered in the preventionprogramme. Twenty seven men declined toenroll in the programme. Of the 160 men whoaccepted the invitation, two were excluded dueto chronic disease. The inclusion criteria were:no history of CHD, diabetes or other severechronic disease; no regular intake of phar-macological agents; serum cholesterol 5-2-7 8 mmol/1; serum triglycerides <5 6 mmol/1;fasting blood glucose <67 mmol/1; and dia-stolic blood pressure <100 mmHg (measuredafter five minutes' rest in the supine position).The subjects joined the study from May toDecember 1990, and were re-examined fromDecember 1990 to May 1991.Twenty six per cent of the subjects were

regular smokers, and this corresponded to theproportion of regular smokers among Swedesolder than 16 years of age.20 Fifty per centof the smokers were light smokers (mean 4cigarettes per day; range 1 to 10 cigarettes).Thirty per cent were moderate smokers (mean16 cigarettes; range 11 to 20 cigarettes), and20 per cent were heavy smokers (mean 30cigarettes; range 21 to 40 cigarettes). Of thenon-smokers, 38% were ex-smokers.

CARDIOVASCULAR RISK FACTORSBody weightBody weight was measured to the nearest 0 1 kgby the same nurse using a calibrated electronicweighing device (Seca delta model 707). Bodymass index (BMI) was calculated as bodyweight in kg divided by height in in.

Blood pressureBlood pressure was determined to the nearest2 mmHg by the same physician, with thepatient supine after five minutes' rest, and

using a standard auscultatory method with a12 x 35 cm2 cuff. The systolic blood pressure(SBP) was noted at the appearance ofKorotkoffsounds, and the diastolic blood pressure (DBP)at the disappearance of the sounds. Both SBPand DBP were calculated as the mean value oftwo determinations.

Lipoproteins and apolipoproteinsVenous blood samples were drawn from anantecubital vein after overnight fasting and withrecommendations on the avoidance of strenu-ous physical activities during the preceding 24hours. Serum was prepared by centrifugationat 3000 rpm and it was stored at - 70°C untilall the samples could be analysed at the end ofthe study. Since all samples could not be as-sayed in the same run, care was taken to includeboth samples from each individual in the samebatch to maximise the analytical precision. Themajor serum lipoproteins (VLDL, LDL, andHDL cholesterol) were determined fromfresh serum samples by preparative ultra-centrifugation, precipitation and quantitativelipid analysis.2' Apolipoprotein A (Apo A) andapolipoprotein B (Apo B) were determinedusing turbidity immunoassay methods (Roche,Basel, Switzerland).

PAI-IPAI-1 activity was determined by adding acertain amount of tissue-type plasminogen ac-tivator (t-PA) to diluted plasma and measuringresidual t-PA activity.22 The results were ex-pressed in arbitrary units, one unit cor-responding to the amount that inhibited oneinternational unit of t-PA (calibrated with ref-erence preparation 83/517 for t-PA, NationalInstitute for Biological Standards and Control,London, UK).

THE HEALTH BELIEF QUESTIONNAIREA questionnaire on health beliefs and healthknowledge based on a protocol developed byNaslund et aP3 was used. The first section ofthis health belief questionnaire included Likertscale items on the person's interest in nutrition,belief in a link between food and health, beliefin diet and physical activity as preventive meas-ures against CHD (perception of benefits), fearof becoming a CHD victim (perception ofsusceptibility), and perception of the con-sequences of stroke and myocardial infarction(perception of severity).The second section covered health know-

ledge which was assessed by using a matrix ofseven health problems (diabetes, heart disease,lung cancer, mental illness, high blood pres-sure, raised blood lipids, and skin cancer) andnine potentially related factors (smoking, al-cohol, physical exercise, stress, heredity, animalfat, salt, dietary fibre, sunbathing). Subjectswere asked to mark the appropriate box if theythought that the health problem was influencedby a certain factor. A score on health knowledgewas computed by summing the correct answers

132

Smokers in CHD prevention programme

on which factors are associated to heart disease,high blood pressure, and high blood lipids.The beliefs in the importance of 16 health

related behaviours, including CHD preventivebehaviours such as not to smoke, eating fibre,avoiding dietary fat, avoiding excess coffee,avoiding excess sugar, taking regular exercise,avoiding stress, and having yearly blood pres-sure measurements, were assessed on 10 pointscales.

PROCEDUREAt baseline, and after the intervention, all sub-jects underwent a medical examination whichincluded measures of height, weight, bloodpressure, blood sampling, a physical exercisetest on a bicycle ergonometer, and a glucosetolerance test. Relative weight was estimatedwith BMI. The five and 10 yearCHD risks wereestimated at baseline and after the interventionusing the Framingham CHD risk estimate.It is based on the combined impacts of age,cigarette smoking, total and HDL cholesterol,systolic blood pressure, diabetes mellitus, andleft ventricular hypertrophy."4The subjects were given the health belief

questionnaire to be completed at the hospitalin connection with the baseline medical ex-amination. After a standardised interview in-cluding family situation, occupation, smokinghabits, family history of CHD and diabetes,and exercise habits, the men were randomisedto a diet group (D, n = 40), exercise group (E,n=39), diet and exercise group (DE, n=39),and a control group (C, n = 40). Stoppingsmoking was neither encouraged nor dis-couraged by the test leaders.

Diet interventionSubjects in the D and DE groups were givenindividual verbal and written dietary advice bythe physician at baseline. Two weeks later themen and their wives/partners, if present, met adietician who gave further individual dietaryadvice. The dietary recommendations were inaccordance with the National Cholesterol Edu-cation Program Step 125 and other consensusdocuments.""'8 The advice included energyintake to reach or maintain desirable weight,total fat <30 energy percent (E%), car-bohydrates (mainly complex) 50-60 E%, andcholesterol <300 mg/day.At baseline, and the week before the follow

up examination, the men in all groups com-pleted a precoded seven day food record sheetwith preprinted alternatives for foods con-sumed at meals. The sizes of cooked foodportions were estimated using a portion guide,and spreads on bread were estimated based ona photograph. Foods consumed between mealswere also recorded. The total daily energy in-take and intake of nutrients were assessed fromthe food record sheet and calculated with acomputer program.29

Physical activitySubjects randomised to E and DE groups weregiven individual verbal and written informationon exercise by the physician. Regular exercise

of an aerobic type (for example, walking andjogging) taken two to three times a week at anintensity of 60-80% of maximal heart rate, andlasting 30-45 minutes was recommended.30The men were told to maintain activity logsspecifying the date, type of activity, duration,and intensity3' of each exercise session. In ad-dition, they were given the opportunity to en-gage in aerobic sessions two to three times perweek.

ETHICSThe study conformed to the principles em-bodied by the Declaration of Helsinki and wasapproved of by the ethics committee at theKarolinska Hospital, Stockholm.

COMPLIANCEAll subjects, except for one man in the controlgroup, returned for the follow up examinationafter six months. All subjects completed thehealth belief questionnaire. Data on the intakeof energy and nutrients calculated from thefood record sheet completed at baseline andthe week before the follow up examination aftersix months served as measures ofdietary habits.Twenty eight men did not hand in completedfood record sheets. The activity logs maintainedby the subjects in E and DE groups duringthe intervention were used as a measure ofcompliance with increased physical exerciseaccording to given recommendations. In ad-dition, the physician undertook a standardisedinterview with the men at baseline and after sixmonths asking questions about the frequency ofexercise sessions, type of activity, duration, andintensity.

STATISTICSData were analysed using discriminant analysis,Student's t test, and X' statistics. Post hoc testsincluded Fisher PLSD and Tukey's HSD. A95% confidence interval was chosen as a meas-ure of uncertainty for estimates of within groupchanges in dietary factors and CHD risks.

ResultsStudent's t test showed no differences betweenex-smokers and never smokers with regard todietary pattern and cardiovascular risk factors.Because of this, the two groups have beenconsidered together as non-smokers in the sub-sequent analyses.

BACKGROUND CHARACTERISTICSThe x2 statistic showed that when smokers werecompared with non-smokers the former weremore likely to be blue collar workers (43-1%versus 20-6% workers; p<001). Smokers andnon-smokers were similar in terms of maritalstatus. Student's t test showed that the groupswere similar in age (mean age 46 years, range35-60).

133

Ndslund, Fredrikson, Hellenius, de Faire

Table 1 Diet of smokers and non-smokers at baseline. pvalues for differences between smokers and non-smokersbased on Student's t tests

Smokers Non-smokers p value(n = 30) (n = 100)

Energy (kJ/d) 8789 8770 0 97Total fat (E%) 36-6 35 7 0 43Polyunsaturated fat 5-7 5-5 0 49(E%)Fibre (g/d) 16-1 18 2 0 04Carbohydrates (E%) 42-4 45 5 0 007Saccharose (E%) 6-8 6-7 0 83Protein (E%) 14-1 14 0 0-82f carotene (mg/d) 1.9 1 9 0-91Vitamin C (mg/d) 57-2 64-5 0-26Vitamin E (mg/d) 7-1 7 0 0-87Alcohol (E%) 7-2 4-9 0 004

Table 2 Cardiovascular risk factors of smokers and non-

smokers at baseline. p values for differences betweensmokers and non-smokers based on Student's t tests

Smokers Non-smokers p value(n=41) (n= 117)

Body mass index 25-6 25-1 0 38Systolic blood pressure 126-0 131-8 0 03(mmHg)Diastolic blood 80-7 81-7 0-51pressure (mmHg)Serum cholesterol 6-0 6-2 0 41(mmol/l)HDL cholesterol 1-2 1-4 0 002(mmol/l)Serum triglycerides 1-7 1 3 0-01(mmol/l)VLDL triglycerides 1.1 0 08 0-02(mmol/l)PAI-1 (U/ml) 15 4 10-9 0 03Apo A (g/l) 1-4 1-5 0 004

PAI- I = plasminogen activator inhibitor- 1; Apo A = apo-lipoprotein A; HDL= high density lipoprotein; VLDL =verylow density lipoprotein.

HEALTH BELIEFSThe x2 statistic showed no differences betweensmokers and non-smokers with regard to in-terest in nutrition, belief in a link between foodand health, belief in diet and physical activityas CHD preventive actions, perceptions of se-

verity of illness, and susceptibility to illness.

HEALTH BEHAVIOURThe dietary pattern of all smokers and non-

smokers (n = 130) at baseline is presented intable 1. Data from 28 men are missing becauseof incomplete food record sheets. Student's t

test showed that the diet of smokers, comparedwith non-smokers, was characterised by a lowerintake of fibre, a lower carbohydrate E%, and a

higher alcohol E%. Smokers and non-smokersengaged in a similar number ofphysical exercisesessions per month (4 5 and 5-6, respectively).

CARDIOVASCULAR RISK FACTORS

Data on cardiovascular risk factors for allsmokers and non-smokers (n = 158) at baselineare presented in table 2. Student's t test showedthat smokers had lower levels of HDL cho-lesterol and Apo A, lower SBP, and higher levelsofserum triglycerides, VLDL triglycerides, andPAI-I than non-smokers. No subjects had dia-betes mellitus or left ventricular hypertrophy.

Table 3 Variables significantly discriminating betweensmokers and non-smokers in stepwise discriminant analysis(n= 130)

F value* p value

Subgroup: dietary variablesAlcohol (E%) 6 5 <0-001

Subgroup: cardiovascular riskfactorsHDL 5-3 <0 01SBP 6-9 <0 001PAI-I 5-6 <0 001

Subgroup: knowledge andattitudes

Attitude to smoking 27-6 <0 001

* F values indicated in the tables are the F values when allselected variables had been entered in the equation (F toremove).

Discriminant analysisTo discriminate between smokers and non-smokers, three separate stepwise discriminantanalyses were performed. The variable with thehighest F-value was entered at each step aslong as a substantial contribution to the dis-criminating function could be found, F>4. Theresults are presented in table 3. The alcoholE% was found to discriminate between smokersand non-smokers in the stepwise discriminantanalysis in which dietary variables were entered.HDL cholesterol, SBP, and PAI-1 were thephysiological variables found to discriminatebetween smokers and non-smokers. A thirddiscriminant analysis was performed using thescore on health knowledge and the ratings onthe importance of different positive health prac-tices. Attitude to smoking was found to be theonly discriminating variable. Thus, the dis-criminant analyses suggested that smokers werecharacterised by a higher alcohol E%, lowerHDL cholesterol level, lower SBP, higher PAI-1 level, and a more favourable attitude to smok-ing than non-smokers.

DIETARY CHANGESTable 4 presents the dietary changes made bysmokers and non-smokers in the D and DEgroups during the intervention. Men in the Eand C group were not told to change their diet,and are therefore not included. Both smokersand non-smokers significantly increased theirintake of fibre, carbohydrate E%, betacarotene,and protein E%. Smokers and non-smokersalso decreased the fat E%. Smokers sig-nificantly increased their intake of vitamin Cand reduced the saccharose E%, while non-smokers decreased energy intake and poly-unsaturated fat E% and alcohol E%.

CHANGES IN SMOKING BEHAVIOUROne man in the diet and exercise group stoppedsmoking and one man in the exercise groupstarted smoking during the intervention pro-gramme.

CHANGES IN PHYSICAL EXERCISEBoth smokers and non-smokers in E and DEgroups increased physical exercise. The two

134

Smokers in CHD prevention programme

Table 4 Differences in dietary intake after six months of intervention among men in thediet and dietlexercise groups. Mean (95% confidence intervals*). No differences betweensmokers and non-smokers were found.

Smokers Non-smokers(n = 20) (n = 45)

Energy (kJ) -717 (-1504, 70) -597 (-1152, -43)Fat (E%) -3-8 (-6-5, -1-1) -2-8 (-44, -1 32)Polyunsaturated fat (E%) -0 5 (-1 2, 0 2) -0-5 (-1-04, -0-02)Fibre (g/d) 3-1 (0 9, 5 3) 1-8 (0 6, 3 1)Carbohydrates (E%) 4-0 (0-5, 7-4) 2-0 (0 4, 3 6)Saccharose (E%) -2-4 (-3-8, -09) -0 9 (-1-9, 0-1)Protein (E%) 1-6 (0 5, 2 7) 1-5 (0 8, 2-1)3 carotene (mg/d) 1-3 (0 4, 2 3) 1-2 (0 5, 1-9)Vitamin C (mg/d) 23 9 (8-0, 39 9) 5 2 (-6 2, 16 6)Vitamin E (mg/d) -1-0 (-2-2, 0-2) -0-7 (-1-5, 0-9)Alcohol (E%) -2-1 (-4 4, 0-2) -0 8 (-1 4, -01)

* A confidence interval which does not include zero indicates a significant change within thegroup.

Table 5 Estimated five and 10 year coronary heart disease (CHD) risks (% CHDevents during five and 10 years, respectively) for smokers and non-smokers inintervention groups. Mean (95% confidence intervals). p values for between groupcomparisons based on Student's t test.

Smokers Non-smokers p value(n = 32) (n = 86)

At 5yCHD risk at baseline 4-4 3-2 0-003CHD risk after 3 9 2-7 0-008interventionMean change in CHD -0-66 (-1-16, -0 15) -0 44 (-0 75, -0 14) 0 47riskAt 10yCHD risk at baseline 9 9 7-2 0-002CHD risk after 8-8 6 2 0 001interventionMean change in CHD -1-10 (-2 05, -0 14) -1 0 (-1 56, -0-46) 0-89risk

* A confidence interval which does not include zero indicates a significant change within thegroups.

groups engaged in a similar number of exercisesessions after intervention (9 0 and 11h4 ses-sions per month, respectively).

CHD RISKTable 5 presents the estimated CHD risk ofsmokers and non-smokers in the interventiongroups (n=118) at baseline and after inter-vention, and mean changes. Smokers had sig-nificantly higher estimated five year and 10 yearCHD risk before, as well as after, interventioncompared with non-smokers. Both smokersand non-smokers lowered their five and 10 yearCHD risks during the intervention period. AFramingham risk estimate showed that if thesmokers would quit smoking, their five and 10year CHD risks after intervention would bereduced to 2-8 and 6-7%, respectively, or sim-ilar to the risks of the non-smoking group.

DiscussionDiscriminant analyses showed that smokerswere characterised by a significantly higher al-cohol E%, lower level of HDL cholesterol,lower SBP, and a higher level of PAI-I com-pared with non-smokers. The higher alcoholintake in smokers is well documented.3A mod-erate alcohol intake, compared with a lowerintake, is associated with increased HDL cho-lesterol.32 A possible explanation for the lowerHDL cholesterol concentration in the smokersdespite the higher alcohol consumption may

be that the negative effect of smoking on HDLcholesterol is stronger than the positive effectof alcohol. Previous studies have also reportedlower HDL cholesterol in smokers.'2 With re-gard to SBP, results from epidemiological stud-ies have generally shown that smokers havelower blood pressure than non-smokers.33

It has previously been reported that smokershave lower intakes of fibre, vitamin C, vitaminA, polyunsaturated fat, protein, and car-bohydrates,'2 as well as higher concentrationsof serum cholesterol, serum triglycerides, andVLDL triglycerides.'-5 Student's t test showeda lower intake of fibre and carbohydrates, aswell as higher serum triglycerides, VLDL tri-glycerides, and Apo A among smokers, but thiswas not confirmed in the discriminant analysis.It cannot be excluded that the lack of differencein this study between smokers and non-smokerswith regard to dietary and physiological factorscould be due to a type II error because of thesmall number of smokers.Both smokers and non-smokers increased

the amount of physical exercise they took andmanaged a number of positive dietary changesduring the intervention. For example, bothgroups increased their intake of dietary fibreand ( carotene. The subjects were advised tolower their fat intake to 30 E%. Neither thesmoking nor the non-smoking group managedthis reduction but both made a significant re-duction in fat E% consumption. With respectto vitamin C, only smokers ate significantlymore.

It is interesting that smokers and non-smokers reported the same belief in a linkbetween diet and health, and had similar beliefsin diet and physical exercise as preventivehealth actions. Despite having a higher CHDrisk, smokers perceived that they had the samesusceptibility to stroke and myocardial in-farction as non-smokers. This cannot easily beexplained by lack of knowledge, since know-ledge of risk factors for cardiovascular disease,including the hazards of smoking, was notdifferent in smokers and non-smokers. It ispossible that smokers systematically under-estimate the CHD risk associated with smok-ing. With regard to attitudes towards positivehealth practices, the discriminant analysisshowed that smokers and non-smokers sharedsimilar attitudes except that smokers had amore favourable attitude towards smoking.Both smokers and non-smokers lowered their

five and 10 year CHD risks by effecting changesin cardiovascular risk factors due to the dietand physical exercise programme.'7 However,both initially and after the intervention,smokers were characterised by a higher CHDrisk than non-smokers. The most importantbehavioural change for smokers would be toquit smoking. Stopping smoking would notonly lower the estimated CHD risk but possiblyalso reduce the impact of risk mechanisms suchas oxidative damage.2 Furthermore, stoppingsmoking may change and improve the diet ofthe smokers. It has been reported that after afew years, the diet of ex-smokers is similar tothat of never smokers.3435 Quitters also tend tohave an improved health consciousness.36 In

135

Ndslund, Fredrikson, Hellenius, de Faire

this study, ex-smokers and never smokers weresimilar both in respect of dietary factors andcardiovascular risk factors. This result suggeststhat to manage smoking cessation among

middle aged men, an approach aimed atchanges of attitude towards smoking may bemore effective than an intervention aiming atimproved health knowledge.

It is concluded that both smoking and non-

smoking middle aged men may improve theirdiet and exercise pattern during a six monthintervention programme and thus lower theirCHD risk. However, the best CHD preventivemeasure for smokers is to quit smoking.Smokers who will not quit smoking should bepersuaded to improve their eating habits withregard to dietary fibre'6 and antioxidants,2 andto increase the amount of physical exercise theytake in order to reduce the health injuriousimpact of their smoking behaviour.

This study was supported by funds from the Bank of SwedenTercentenary Foundation and the Swedish Lung-Heart Found-ation.

1 Fehily AM, Phillips KM, Yarnell JWG. Diet, smoking, socialclass and body mass index in the Caerphilly heart diseasestudy. Am Clin Nutr 1984;40:827-33.

2 Margetts BM, Jackson AA. Interactions between people'sdiet and their smoking habits: the dietary and nutritionalsurvey of British adults. BMJ 1993;307:1381-4.

3 Wichelow MJ, Erzinclioglu SW, Cox BD. A comparison ofthe diets of non-smokers and smokers. Br Addict 1991;86:71-81.

4 Midgette AS, Baron JA, Rohan TE. Do cigarette smokershave diets that increase their risks ofcoronary heart diseaseand cancer? Am Jf Epidemiol 1993;137:521-9.

5 Fulton M, Thomson M, Elton RA, Brown S, Wood DA,Oliver MF. Cigarette smoking, social class and nutrientintake: relevance to coronary heart disease. Eur ClinNutr 1988;42:797-803.

6 HostmarkAT, BergJ, Brudal S, Berge R, KierulfP, BjerkedalT. Coronary risk factors in middle-aged men as related tosmoking, coffee intake and physical activity. Scand Jf SocMed 1992;20:196-203.

7 Grunberg NE. The effects of nicotine and cigarette smokingon food consumption and taste preferences. Addict Behav1982;7:317-31.

8 Cockburn J, Hennrikus D, Scott R, Sanson-Fisher R. Ad-olescent use of sun-protection measures. Medj7Aust 1989;151:136-40.

9 Oldridge MB. Compliance of post myocardial infarctionpatients to exercise programs. Med Sci Sports 1979;11:373-5.

10 Blair SN, Jacobs DR, Powell KE. Relationship betweenexercise or physical activity and other health behaviors.Public Health Rep 1985;100:172-80.

11 Castro FG, Newcomb MD, McCreary C, Baezconde-Gar-banati L. Cigarette smokers do more than just smokecigarettes. Health Psychol 1989;8: 107-29.

12 Craig WY, Palomaki GE, Haddow JE. Cigarette smokingand serum lipid and lipoprotein concentrations: an analysisof published data. BMJ 1989;298:784-8.

13 Williams A, Robinson D, Bailey A. High density lipoproteinsand coronary risk factors in normal men. Lancet 1979;i:72-5.

14 Willett W, Hennekens CH, Castell W, Rossner B, Evans D,Taylor J, Kass EH. Effects of cigarette smoking on fastingtriglyceride, total cholesterol and high density lipoproteinsin women. Am HeartJt 1983;105:417-21.

15 Freedman DS, Srinivasan SR, Shear CL, Hunter SM, CroftJB, Webber LS, Berenson GS. Cigarette smoking initiationand longitudinal changes in serum lipids and lipoproteinsin early adulthood: the Bogalusa heart study. Am J7 Ep-idemiol 1983;124:207-19.

16 Nilsson TK, Sundell I, Hellsten G, Hallmans G. Reducedplasminogen activator inhibitor activity in high consumersof fruits, vegetables and root vegetables. J Intern Med1990;227:267-7 1.

17 Hellenius ML, de Faire U, Berglund B, Hamsten A, KrakauI. Diet and exercise are equally effective in reducingrisk for cardiovascular disease. Results of a randomized,controlled study of men with slightly to moderately raisedcardiovascular risk factors. Atherosclerosis 1993;103:81-91.

18 Becker MH, Rosenstock IM. Compliance with medicaladvice. In: Steptoe A, Mathews A, eds. Health care andhuman behaviour London: Academic Press, 1984:180-3.

19 Hellenius ML, de Faire U, Krakau I, Berglund B. Preventionof cardiovascular disease within the primary health caresystem - feasibility of a prevention programme within theSollentuna primary health care catchment area. Scand JPrim Health Care 1993;11:68-73.

20 Socialstyrelsen. Folkhalsorapport 1991 (public health report;in Swedish). Stockholm: Socialstyrelsen, 1992:168.

21 Carlson K. Lipoprotein fractionation. J Clin Pathol 1973;26:32.

22 Chiemelewska J, Ranby M, Wiman B. Evidence for a rapidinhibitor to tissue plasminogen activator in plasma. ThrombRes 1983;31:427-36.

23 Naslund GK, Wardle J, Steptoe A, Fredrikson M. Healthbehavior, knowledge and attitudes among Swedish uni-versity students. ScandJ Psychol 1993;34: 197-211.

24 Anderson KM, Wilson PWF, Odell PM, Kannel WB. Anupdated coronary risk profile. AHA Statement. Circulation1991;83:356-62.

25 Expert Panel of the National Cholesterol Education Pro-gram. Report of the National Cholesterol Education Pro-gram Expert Panel on detection, evaluation, and treatmentof high blood cholesterol in adults. Arch Intern Med 1988;148:36-69.

26 Grundy SM, Arky R, Bray GA, et al. Coronary risk factorstatement for the American public. A statement of theNutrition Committee, American Heart Association. Cir-culation 1985;72:1135-9.

27 European Atherosclerosis Study Group. Strategies for theprevention of coronary heart disease: a policy statementof the European Atherosclerosis Society. Eur HeartJ 1987;8:77-88.

28 WHO Study Group. Diet, nutrition and the prevention ofchronic diseases. WHO Technical Report Series 797. Ge-neva: World Health Organization, 1990.

29 Becker W. The Swedish household survey 1989. Presentedat the Sixth European Nutrition Conference, Athens,Greece. (Abstract) in Nutritional science, new developmentof consumer concern, 1991.

30 Gordon NF, Scott CB, Wilkinson WJ, Duncan JJ, Blair SN.Exercise and mild essential hypertension. Re-commendations for adults. Sports Med 1990;10:390-404.

31 Borg G. Perceived exertion as an indicator of somatic stress.Scand_J Rehab Med 1970;23:92-6.

32 Gaziano JM, Buring JE, Breslow JL, et al. Moderate alcoholintake, increased levels of high-density lipoprotein and itssubfractions, and decreased risk of myocardial infarction.N Engl JTMed 1993;329: 1829-34.

33 Green MS, Jucha E, Luz Y. Blood pressure in smokers andnonsmokers: epidemiological findings. Am Heart J 1986;111:932-40.

34 Morabia A, Wynder EL. Dietary habits of smokers, peoplewho never smoked and ex-smokers. Am J Clin Nutr 1990;52:933-7.

35 Bolton-Smith C, Woodward M, Brown CA, Tunstall-PedoeH. Nutrient intake by duration of ex-smoking in theScottish heart health study. BrJ7Nutr 1993;69:315-32.

36 West DW, Graham S, Swanson M, Wilkinson G. Five yearfollow-up of a smoking withdrawal clinic population. AmJ Public Health 1977;67:536-44.

136