I"I/ma! oJ !II/allal ,'v/rdieillr 1997: 2-l2: 1-l9-15h

Chronic work stress is associated with atherogenic lipids andeIe va ted fibrinogen in middIe-aged men

J. SfEGRIST". R. PETER". P. CREMER" & D. SEfDELh1'/"01111111' '!lIstilll/I' oJ Medical ""eio!O!l!!. Heillricll-Hl'illl' Ullil'l'fsi/!J "f DiisseitlOl}: 1IIl/1'!llStill/le "f (lillica! CIlI'lIIislr!J. l.l/d\\li!1 JHaxillliliallS

Ullil'l'rsit!J of .\11/1Iic'11; Cer/IIII/I!J

Abstract. Siegrist J. Peter R. Cremer E Seidel D.(University 01' Düsseldorf and Ludwig MaximiliansUniversity 01' Munieh. Germany). Chronic workstress is associated with atherogenie lipids and elevat­ed fibrinogen in middle-aged men. J Intern Med1997: 242: 149-56.

Objectives. '1'0 examine the association between amodel 01' chronie work stress (high efforts in combi­nation with low rewards) and two risk factors 01'

coronary heart disease. low-density-Iipoprotein(LDL)-cholesterol and fibrinogen.Design. A cross-sectional study in a group 01' 179healthy middle-aged (48.5 ~ 4.5) male middle man­agers.Setting. A large car-producing enterprise in Germany.Results. After adjustment for relevant covariates.logistie regression analysis showed independent

lntroduction

An impressive number 01' prospective epidemiologicalstudies documented elevated relative risks 01' fatal or

nonfatal acute myocardial infarction in middle-agedmen suffering from chronic psychomental or socio­emotional stress at work [for overview see ]-71. Onlypart 01' this elevated risk was attributable to health­detrimental behaviours associated with stressful

experience. such as heavy cigarette smoking. lack 01'

physical exercise or faulty diet [3. 5. 81. Experimentaland clinical evidence indicates that pathophysiologi­cal pathways linking chronically-stressful experienceto coronary artery disease involve enhanced activa­tion 01' the autonomie nervous system and 01' differ­ent neuroendocrine and immune responses. whieh inthe long run adversely affect the cardiovascular sys­tem [9-121. Partly. these pathways affect important

1) 1997 Blaekwell Seien er Ltd

effects 01' a composite measure 01' high cffort andlow reward at work on the prevalence 01' clevated(upper tertile. Le. 2: 160 mg dL ') LDL-cholesterol(prevalence odds ratio (POR) = 3.57: 95'Yc,confi­dence intervals (Cl): 1.24-10.20) and on elevated

(upper quintile. Le. 2:420 mg dL") plasma fibrinogen(POR = 6.71 (Cl: 1.57-28.76). Apart from thiscore measure. cigarette smoking. overweight andalcohoI consumption were the covariates with therclativcly strongest contributions to the multivariatemodel.

Conclusions. Results give preliminary evidence on anindependent association 01' chronie work stress with

atherogenic lipids and with elevated fibrinogen in anoccupationally homogeneous group 01' healthy mid­dle-aged men.

Keywords: fibrinogen. LDL-cholesterol. work stress.

coronary risk factars such as atherogenic lipids [13.141. elevated plasma fibrinogen [9.151 or high bloodpress ure [12.16J. Yet. the epidemiological demon­stration 01' associations between chronic work stress

and elevated low-density-Iipoprotein (LDL)-choles­terol[17. 18]. plasma fibrinogen [19-231 or hyper­tension [24. 25] so far is limited.

This study was undertaken to test these associationsin a socio-economically homogeneous group 01' mid­dle-aged men experiencing different levels 01' chronicstress at work. Whereas results concerning the preva­lence 01' hypertension were published elsewhere [261.this report presents the original findings with regardto low density lipoprotein (U)L)-cholesterol and plas­ma fibrinogen. two coronary risk factors whose clini­cal significance is widely accepted [27-291.

'1'0 introduce the notion 01' chronic work stress intoa scientific framework. a standardized measurement

]49

150 J. SIEGRIST etal.

approach based on a theoretical concept is needed. Inthis respect the present study is focused on the modelof effort-reward imbalance at wark developed andtested in our group 151. In summary. this modeldefines chronically stressful experience at work interms 01' amismatch between high costs spent andlow gains received in occupationaJ life. Two sources01' effort are distinguished. an extrinsic source. thedemands on the job. and an intrinsic source. themotivations 01' the individual worker in ademandingsituation. In this laUer regard. we introduced theconcept 01' 'need for contro!' as a personal paUern 01'

coping with the demands at work (see 'Data andmethods·). These efforts are spent as part 01' a social­Iy-organized exchange process where rewards aredistributed by three transmitter systems: money.esteem. and job security or promotion prospects (soca lied occupational status control). Lack of reciproci­

ty between costs and gains defines astate 01' emotion­al distress. with special propensity to autonomiearousal and cardiovascular risk.

In view 01' the rather high prevalence 01' these con­ditions in modern working life. it is important toknow whether they are associated with the preva­lence of coronary risk factors such as elevatedatherogenic lipids or elevated plasma fibrinogen.

Data and methods

Swdy slllllple

A cross-sectional study was conducted in a large car­producing enterprise as part of a cardiovascularscreening offered to the total group 01' middle man­agers by the company's Occupational HealthDepartment. Based on epidemiological informationon prevalence 01' cardiovascular risk factors. werestricted our investigation to men aged 40-55 years.During the study period a total 01' 189 middIe man­agers in this age category were screened. Roughly95% (n = ] 79) of them agreed to participate in our

investigation (48.5 ± 4.5 years). 01' the remaining10 men 6 refused to participate and 4 suffered fromlong-term sickness absence.

In this study middle managers have been chosentor two reasons. First. we decided to study a group as

homogeneous as possible in terms 01' education.training. and occupational status to rule out the con­fou nding effects 01' these variables [1. 2. 3. 4.17-231. Secondly. duc to exposure to specific job

characteristics. we expected a rather high proportion01' middle managers to exhibit high chronic workstress in terms of high etTort and/or low reward.

The present analysis is restricted to middle man­

agers with complete data. This leaves ] 75 subjectsfor analysis with regard to LDL-cholesterol and 143subjects for multivariate analysis with respect to plas­ma fibrinogen (97% and 79%. respectively. 01' thetotal study sam pIe). When considering possible sam­pie bias within the reduced study group we did notfind systematic differences on core variables between

the full and the reduced sampie.

Biollledicillmeasllres

Independent blood press ure reudings were taken byoccupational physicians as part 01' the cardiovascularscreening programme using sphygmanometryaccording to WHO criteria. In addition. body weightund height were assessed. Manifest hypertension wasdefined according to WHO criteria (160 mmHg sys­tolic and/ar 95 mmHg diastolie blood pressure).

Blood sampies were obtained from subjects understandardized conditions. Le. after 30 min 01' rest in

supine position at tixed diurnal time. A fasting time 01'

12 h was not required for our measurel11entapproach 130 I. Plasma fibrinogen was determined byan immunochemical method using monospecincantisera and a nephelometer analyser BNA fromßehring Werke. Marburg. Germany. The interassayvariation coefficient 01' librinogen measures was aslow as 2.5%. The fibrinogen measurement in thisstudy was based on a standard material (Behring AG.Marburg. Germany) wh ich. duc to its former declara­tion by the manufacturer at that time. providedabout 40% higher values as compared to the cur­rently used standardization procedure 01' this methodor to functional tests according to the Clauss method[3] I.

Total serum cholesterol and triglycerides weredetermincd enzymatically. Lipoproteins were mea­sured by commercially availablc techniqucs(lmmuno Diagnostica. Heidelberg: Boehringer.Mannheim): quantitative lipoprotein electrophoresis130. 321: precipitation techniques with dextrane sul­phate and sodium phosphotung-state/MgCI2.respectivcly. All analyses were perfonned blind by theDepartment 01' C1inieal Chemistry. In addition. heartrate (resting EKG) and conventionallaboratory para­meters were assessed 1321.

t? 1997 Blackwcll Scicncc Lld1011/'11(/101llllmllli Ml'liicil/(' 142: 149-1 :;1)

C!-IRONIC WORK STRESS IN MlDDLE-AGED MEN ] 5]

In this study. data analysis is based on eontinuousand on eategorieal dinicaI data. We are aware 01' thefact that international standardization 01' cut-off

points defining abnormal levels 01' LOL eholesteroland 01' plasma fibrinogen is still under diseussion.Therefore. our decisions eoncerning cut-off points 01'

upper limits are justified by referenee to recentprospeetive epidemiologieal studies indicating thatthese cut-off points deHne thresholds 01' dearly ele­vated incidenee 01' eoronary events: they are theupper quintile eorresponding to 2:4.:20 mg &' withrespeet to plasma fibrinogen [31. for discussion seealso 27. 33. 341 and the upper tertile (2: 160 mg dL I)with respeet to LDL-cholesterol[3 51.

Psychosorial and belJavio/lral measlIres

The variables 01' the model 01' effort-reward imbal­

anee at work were measured by a set 01' Likert-sealeditems whieh were previously tested and whieh wereassessed by means 01' a struetured interview and aself-administered questionnaire [36.37]. The experi­enee 01' extrinsic effort (demands) was measured by

assessing the frequency and intensity 01' being dis­tressed by time pressure. frequent interruptions.ineonsistency 01' demands. and severity 01' workproblems. Faetor scores based on these items werecomputed (Cronbaeh's alpha: ranging from .76 to.69).

Intrinsic effort was assessed by a self-administeredquestionnaire measuring 'need for contro!'. a eritiealpersonal style 01' coping with work demands [37].The questionnaire contains 29 dichotomous itemsdeseribing exeessive job involvemenl. positive andnegative feelings and attitudes related to work eom­mitment as measured by four unidimensional scales:(i) need for approval: (ii) eompetitiveness and latenthostility: (iii) impatienee and disproportionate irri­tability: and (iv) inability to withdraw from workobligations. These four subseales were repeatedlyfound to load on one latent factor (Cranbach's alpha

ranging from .70 to .63). In a previous study wefound that persons scoring in the uppcr tertile 01' thisseale were at risk 01' suffering from Heute myocardial

infarction [38].Measures 01' oecupational rewards covered the

three dimensions 01' income. esteem and reciprocaI

support. and 01' oecupational status contral (job inse­curity. poor promotion praspects. forced job change).Again. frequeney and intensity were assessed by

© 1997 ßlackwcll Science Ud Juumal ur hHrmeli Mrdicinr 242: 149-156

Likert-scaled ilems. and respective scores were com­pu ted (Cranbach's alpha ranging from .82 to .48).

Aecording to our theoretical assumption.effort-reward imbalance is defined as the simultane­ous manifestation 01' at least one dimension measur­

ing high extrinsic or intrinsic effort and 01' at least

one dimension measuring low reward. We do notexpect all dimensions to be significantly related to the

crileria under study. but we postulate independenteffects 01' a three-eategorical variable eombininginformation on high elTort and low reward on LDL­cholesterol and on fibrinogen. after adjusting for rele­vant confoundcrs (see 'Statistical analyses·).

Information on health-damaging behaviours(ciga rette smoking. alcohol consumption. lack 01'

physical exercise. lack 01' weight contral. faulty diel.noncompliance with medication) was recorded fromthe structured interview using standardized. well­tested questions [32.361.

Stlltistical analyses

In addition to bivariate analysis. multivariate statisti­cal methods include analysis 01' variance (anova) for

continuous data and logistic regression analysis forcategorical data. In analysis 01' variance. F-test andmean va!ues are displayed. In logistic regressionanalysis. the model fit 01' the most parsimoniousmodel is tested by the likelihood ratio difference testin a bottom up pracedure 1391. Regression coeffi­eients. standard errors. multivariate prevalance oddsratios and the 95% confidence intervals are indicated

in respective tables. For more detailed information onthe statistical methods see [371. A level 01' signifi­eanee 01' p ::=: 0.05 was accepted in all analyses.Ca\culations were performed on personal computersusing SPSS PC for Windows for data analysis.

Results

In Table 1. the study sampie is characterized withrespeet to the main biomedical ami behavioural vari­ables (means and frequencies). The prevalcnce asweil as mcan valucs and standard devialions of co[O­

nary risk factars in this sampie are weil comparablcto thosc obscrvcd in large. representative sampies 01'

the male German population in this age group [32.40.41].

According to the cut-off points mcnlioned in the'Data and mcthods' seetion. the two subgroups 01'

J. SIE G R IST cl al.

Table 1 ßiobehaviouraf eoronary risk faetors in I 79 middle managers

Variable

Age ( years)

BMI (kg/m')

Systolie blood pressure

(mmHg)Diastolie blood pressure(mmHg)Total eholesterol

(mgdL ')I.DL-eholesterol

(mgdL ')HDL-eholesterol

(mgdL ')Fibrinogen (mgdL ')

Mean :!: SD

4S.5:!:4.5

27.1 :!: 3.4

136.1 :!: IS.2

SS.7:!: llA

233.3 :!: 40.2

141.! :!: 37.6

55.3:!: 13.S

372.7:!: 63.7

Regular cigarelte

smoking (> 10 cig. day ')Lack of physical

exereise «onee week ')Hypertension"

Hyperlipidaemia'

Elevated I1brinogen"

Frequeney of alcohol

eonsumption (regular)"

24.0

59.0

30.0

32.0

22.3

29.6

'SBI': > I 00 mmHg and/or DBI' >95 mmHg. "LDL-eholesterol: upper wrtile. Le. > 100 mg dL '. 'Fibrinogen: upper quintile. Le. >420 mgdL '. 'At least one 'regular"answer on three three-eategorieal seales (ne"er. seldom. regular) measuring the frequency of heer. wine and spirit

eonsumption. Values represent means and standard deviations (SDI Odt side) and pereentages (right side).

middle managers with normal or abnormal values 01'

LDL-cholesterol and 01' fibrinogen. respectively. are

compared in Table 2. From this table it is obvious thatregular cigarette smoking is more prevalent in bothsubgroups at risk. Regular alcohol consumption isfound to be more prevalent in middle managers withelevated LDL-cholesterol whereas overweight and

younger age are more frequent in the group with ele­vated fibrinogen. Turning to the psychosocial data.the composite measure 01' effort-reward imbalanceindicates that at least twice as many middle man­agers with abnormal values 01' LDL-cholesterol orplasma fibrinogen suffer from chronic work stress ascompared to the group with normal values (20.0%

Table 2 ßiobehavioural and psychosoeial faetors associated with elevated IJ)L-cholesterol and elevated I1brinogen levels in middIe-managers

Elevated LDL-eholesterolElevated I1brinogen(upper tertile)

(upper quintile)

Yes

NoYes No

Variable(n = 55)(n = 115)(n = 30)(n = 113)

Age (> 50 years)

34.5%-l1.6'Xl26.7tyo'":; ] .3j}{1

LDL-eholesterol (upper tertile)

33.3%21\.3%

Fibrinogen (upper quintile)

32.0%10.41}{)

Hypertension (> 160/95 mmHg)

34.6%17 ....JtYo23.3%29.2%

Regular cigarette smoking (> 10 cig. day-I)

3().9'1019.1%40.0%'21.2'J{1

Frequency of alcohol eonsumption (regular)

4()A%27,4%20.7%31.6'}{)

Laek or physical exereise «onee week-I )

(,6.7%~:;.8fM,66.7%55.S'K,

Body wdght (;:,28.8 kg/m2)

26.9%19.5%33.3%17.7%

HiUll (?:tril!sic ('jJort workload lupper terlile)

4 3.(i'~{)3 3,0%5(J.O%61.6%

Hiull illlrillSic l'ßilrl Need for approval (uppcr terlile)

2'5. '1(}'f\27.0%30,0%'14.2%

tOll' rell'(/rd Lack of reciproeal support (upper tertile)

4 5. "~}{I'"30,4%70.()%:;5.X1X,

Low promotion prospects (upper tertile)

30.9%23.9%33.3%22.3%

EjJort-rl'II'(/rd im/will/lfl' imlex Neither high efforta nor Iow reward'

30.9%4h.l%20.0%31\.1%

High effort' OR low reward"

49.1%44.23%60.0%54.0%

High crfort' AND low reward"

20.0%9.0%2(>.()% 8.<)1:1.,

'P:S; .05::Workload· (LDL-eholesterol), 'need for appro\'al' (fibrinogen): "Lack of reciproeal support (LDL-cholesterol and fibrinogen).

© 1997 Blackwell Science Ltd fOlIrl/II14 III//'mal Mrdil'illl' 142: J 49-1 ,)/i

CHRONIC WORK STRESS IN MIDDLE-AGED MEN 153

Table 3 Logistic regression analyses: eHort-reward im balance associated with clevated LDL-cholesterol and elevated fibrinogen in middlc­managers

ModellVariable Elevated LDL-cholesterol

Multivariate pre\'alence odds-ratio

Outcome (sampie) 95%CI

Regular cigarelle smoking (> 10 cig. day-I )Frequency of alcohol consllmption (regular)

E//orl-reward imbalalICe illdex

Neither high cffort" nor low reward'Either high effort" OR low reward'

High effort" AND low reward'

2.19*1.83

1.001.623. S7*

1.01-4.75

0.89-3.74

0.77-3.42

1.24- 10.20

Model 11

Variable Elevated fibrinogen ,"Iliitivariate prcvalcnce odds-ratio

Regular cigarette smoking (> I 0 cig. day ')Body wcight (>28.8 kg/m2)EjJorl-mvard imbalallce illdex

Ncither high effort' nor low reward'Either high eHort' OR low reward"

High effort' AND low reward"

3.46**3.14*

1.00

2.566.71**

I. 34-8.89

1.18-8.37

0.87-7.391.57-28.76

**P =::; .0 I: *P ::; .05. "Workload (upper tcrtile): "lack of reciprocal support (upper tertilc): 'need I<Jrapproval (lipper tertile).

3Workload; blaCk of reciprocal support. low promotion prospects.

Fig. 1 mfort-rcwürd imbaluncc und mcan LDL-cholesterollc\'cls

amongst 168 middle managers.

even more restricted as we included only thosemiddle managers who suffered from lack of re­ciprocal support and from lack of promotionprospects simultaneously. thus defining a high-riskgroup in psychosocial terms. As can be seen fromFigure 1. mean LDL-cholesterol. adjusted for theeffect of age. is clearly eieva ted in this psycho­social high-risk group (174.8 mg dL-' as comparedto the grand mean 01' 141.7 mg dL '; F = 3.9:P = 0.(2).

Respective analyses with continuous data on fib­rinogen were not indicated due to the fact that thedistribution of fibrinogen values precluded the appli­cation of anova.

(174.8 mg dL-')

1-j

High effortaAND low rewardb

F = 3.9 P = 0.02

1

Elther high effortaOR low rewardb

Nelther high effortanar low rewardb

~•..-~ '-.J 20CI",E 0>

,g.sc_-

~r~ 0

~ ~ I (1386 mg dL-')~::~ :: -20:0-'"Li>

CI« -40

vs. 9.6% and 8.0%). Under both conditions. lowreward is measured by the same indicator ('lack ofreciprocal support'). However. in the case of LDL­cholesterol. high effort is measured by the factorscore of extrinsic demands (termed 'workload')whereas in the case of fibrinogen. intrinsic effort as

measured by the subscale 'need for approval' is partof the composite measure. In an inductive approach.these two different indicators of effort were chosen

according to their discriminatory power.Results of a statistical test of the study hypothesis

are demonstrated in Table 3. using logistic regressionanalysis. In both logistic regression models. preva­lence odds ratios (POR) of being at coronary risk are

elevated in the subgroup of middle managers whosuffer from effort-reward imbalance at work. This

holds true after adjusting for relevant biomedical andbehavioural confounders (e.g. cigarette smoking andalcohol consumption with regard to LDL-cholcsterol.cigarette smoking and body weight with regard tofibrinogen). As not all possible confounders l42. 43 J

passed the Likelihood-ratio-difference-Iest of modelfit (sec 'Methods') age. hypertension, and lack of

physical exercise had to be excludcd from analysis.Finally, as indicated in 'Data and methods·. we

present an analysis of continuous data. usingLDL-cholesterol as outcome criterion. Again, a

three-categorical variable is constructed usingindicators of extrinsic effort and of low reward. Inthis case. the definition criterion of low reward is

© 1997 ßlackwel\ Science Ud Journal 01 Illternal Medicille 242: 149-156

154 J. SIEGRIST etal.

Discussion

In this cross-sectional study 01' a group 01' economi­

caHy-active middle-aged male middle managerschronic work stress in terms 01' high effort and lowreward is associated with significantly increased rela­tive risks 01' having abnormal levels 01' LDL-cholesteroland plasma fibrinogen. Results are neither explainedby important confounders such as cigarette smoking.overweight. age. lack 01' physical exercise or alcoholconsumption 142. 43]. nor are they influenced bysocial dass. a variable which was shown to be inverse­

ly associated with fibrinogen [19-22 J and. partly.with atherogenic lipids 181. Current stress physiologi­cal evidence on epinephrine-induced platelet activa­tion and increase in plasma fibrinogen [9. 15 J and onsympatho-adrenergically enhanced endogenous lipidmetabolism [12-141 offers a possible frame 01' inter­pretation 01' reported findings. Clearly. this study lacksdirect information on an association between chronicwork stress. catecholamine concentrations and level

01' fibrinogen or IJJL-cholesterol.Additional support in favour 01' a stress-physiolog­

ical interpretation 01' the observed association isgiven by further results obtained from the same studypopulation: middle managers suffering from higheffort. in terms 01' time pressure and severe problems.in combination with anticipated forced job change(Iow reward) exhibited an increased relative risk 01'

being hypertensive (multivariate odds ratios rangingfrom 1.86-2.94: [26]). Moreover. in a previousprospective investigation on cardiovascular risk anddisease in a cohort 01' male blue-collar workers. indi­cators 01' effort-reward imbalance at work were asso­

ciated with a high ratio 01' low-density-lipoprotein(LDL)- to high-density-lipoprotein (HDL)-cholesterol[171. and they predicted incident-fatal or non fatalcardiovascular events [381.

Studies from other research teams lIsing a somc­what different concept of measuring chronic workstress. thc demand-control model 01' job strain 12. 3J.reported similar findings with regard to fibrinogen[10.13. but for negative results see 441.

LimitIIUolls of t!Je stud!J

This cross-sectional study is based on a limited sam­

pfe size and a limited range 01' measurements.Though unlikely. the observed statistical effects couldbe attribllted to some covarying risk factor not mea­sured in our protocol. such as coffee consumption

145J or genetic risk 129J. Alternatively. one mayarglle that persons suffering from cardiovascular riskmay express higher levels 01' chronic work stress. Thislatter interpretation. however. is unlikely for two rea­sons: first. the majority 01' middle managers with ele­vated level 01' fibrinogen or LDL-cholesterolconsidered themselves as being healthy amI were notpharmacologically treated.

Moreover. the robustness 01' reported findings maybe limited as evidenced by large confidence intervalsin logistic regression analyses. The fact that variablesentering the composed measure 01' high effort andlow reward were not exactly identical in all analysesneeds some further commen\. As mentioned in thc

'Data and methods' section. we do not expect everycore variable measuring the model to be a predictor01' the criteria under study. Rather. the simultancousmanifestation 01' at least one condition 01' high effort(either extrinsic or intrinsic) and 01' at least one con­

dition 01' low reward is hypothesized to account forthe specific quality and intensity 01' stressful work­related experience which may underlie the observcdcardiovascular risk. Nevertheless. in future studies

composite measures 01' effort-reward imbalance will

be required which define critical thresholds at studyonset and wh ich will be comparable across differentinvestigations.

A further argument refers to the fact that somemiddle managers have suffered from high LDL-cho­lesterol and elevated fibrinogen simultaneously. andthat accounting for this association may threaten therobustness 01' the reported lindings. Therefore allanalyses were repeated taking into account the 10subjects who were at risk from both conditions. No

significant changes were observed.Due to deviation from normal distribution 01' crite­

rion variables and due to unequal size 01' subgroups.multivariaie staiistical analysis in ihis rcport wasconfined to logistic regression analysis. The respec­tive loss 01' information on a possible dose-responserelationship between work stress and cardiovasclllarrisk must be considered a major limitation. Someadditional analyses uSing anova procedure wereperformed. and they give preliminary support infavollr 01' such an assumption. at least whenintroducing continuous data on I1)L-cholesterol(see Figure 1).

A further limitation 01' this study concerns the cut­off points on which analyses are based. For two rea­sons. thc case 01' plasma fibrinogen may be

© 1997 BJackwcll Scicncc LId journal of Internal A1l'dil'il1l' 242: 149- J ;6

CHRONIC WORK STRESS IN MIDDLE-AGED MEN 155

particularly critieal. First. international consensuson clinically relevant thresholds is not yet asadvaneed as in the ease of LDL-eholesterol. Seeondly.standardization of measurement techniques used inepidemiologieal studies is still insufficient. thus ren­dering diffieult direct eomparisons across differentstudy populations.

Despite these Iimitations. we maintain that rcporl­ed statistical associations between theoretically­based measures of ehronic work stress and elevated

atherogenic lipids. as weil as plasma fibrinogen.deserve further attention in epidemiologieal and clin­ieal studies on cardiovascular risk and disease. This

holds true especially in view of the high prevalenceof stressful working conditions in terms ofeffort-reward imbalanee in the current global econo­my. and in view of the high prevalence and clinicalsignifieanee of the two cardiovaseular risk factorsunder study. LDL-eholesterol and plasma fibrinogen.

It may weil be that future attention 10 the psychoso­cial conditions specified above eontributes to furthersuceesses in reducing the burden 01' cardiovascularrisk and disease.

References

Kaplan CA. KeHlE. Soeioeeonomic faetors and eardiovaseulardisease: a review 01' the literaturc. Circullllioll 199~: 88:1973-97.

2 Schnall PI.. Landsbergis PA. Backer D. Job strain and cardio­vasculardisease. AIIIIURev Publ Hellllh 1994: 15: 381-411.Karasek RA. Theorelll: Helllllr!l H\Jrk: Slress. Produclh·il!l. emd

clre Recc>llsCn/c!ie>llof Workillil UJ~.New York. USA: Basic Books.1990.

4 \Iarmot M. Elliott I~ eds. Corolmry Hellrl Disellse El'idellliolol!Y.

()xlc)rd: Oxford lIniversity Press. 1992.5 Siegrist J. Advcrse health effeets 01' high-effortllow-rcward

conditions. J OCCUI'Heallir PS!fC11011996: 1: 27-41.6 Kasl S. Inllllence 01' the work environment on eardiovaseular

health: historical. conceptual. and methodological prospce­

tive.1 Oc('//I' HI'lI!II1PS!fC1101199h: 1:42-56.7 Krislcnsen TS. L'ardiovaseular diseases and thc work environ·

ment. In: CheremisinolT PN. cd. EIlCyclol'l'dill of EllvirOlIIll('//W!Ccmlrol TecJlll%Y!I. Vol. 7. 1994; 217-43.

8 ~tarmot MG. Davcy Smith C. Stansfield S. Patel C. North F.

Head J. White J. eC ul. Health ineljuality among ßritish eivil ser­vimls: the Whitehallll study. uI/Jcel 1991: 337: 138 7-9 ~.

9 ßeamish RE. Dhalla NS. Singul TK. Stress emd Ekart Disease.

Boston. USA: :vtartinus Nijhoff. 1985.10 Henry JP. Biological basis 01' the stress response. Illte!//, PI1ysio/

Bell(/\' Sci 1992: 27: 66-83.1 ] Manuck SB. Marsland AL. Kaplan JR. Williams JK. The patho­

genieity 01' behavior and il~ neuroendocrine mediation: anexample from coronary artery disease. PS!lcllOS0111,vlrd 1995;57: 27:;-83.

J 2 Weiner H. l't>rlllrbillg 1111'Or!Jel/JislII. TI/(' Biolo!J!I of Slres~/ill

Q 199 i Blaekwc\l SeieHec Ud joumal oJ Illternal Medicille 242: 149-1 :;6

Exprri/'llce. Chicago. USA: The University 01' Chicago Press.1992.

13 Dimsdale JE. Ikrd JA. Variability 01' plasma lipids in respons('to emotional arousal. PS!lclwso/ll Med J 982: 44: 413-30.

14 Sapolski RM. Mott t;E. Social subordinance in wild baboons is

associated with suppressed high density lipoprotein choles­

terol: the possible role 01' ehronical social stress. ElIllocrillOlo!J!I1987: 121: 1605-10.

I 5 MarkO\'itz JD. Matthews KA. Platclets and coronary heart dis­ease: potential psychophysiologic mechanisms. PS!lclwsolJIAlrd [991: 53: 643-68.

16 lIenry 11'. Liu J. !\Ieehan \VI'. Psychosocial stress and experi­mental hypertension. In: Laragh JH. Brenner BI\I. eds.Hypertensioll: l'atlwphysiolo!JY. Diagllosis. allli MmlCl!J(·/IU'1I1.i\:ew

York: Raven Press. 1995: 905-21.J 7 Siegrist J. Mal,ehinger H. Cremer 1'. Seidel D. Atherogenic risk

in men suffering from oecupational stress. Arherosclerosis1988; 69: 211-18.

18 Mattiasson I. Lindgärde F. :'I:ilsson JA. Theordl T. Threat 01'

unemployment and cardiovaseular risk factors: longitudinalstudy 01' ljuality on sleep and serum cholesterol concentra­tions in men threatened with redundancy. Br Med J 1990:301:461-6.

19 Rosengren A. Wilhelmsen L. Welin L. Tsipogianni A. Teger­Nilsson AC. Wedel H. Social inlluences and cardiovascular risk

faetors as determinanls 01' plasma fibrinogen concentration in

a general population sampIe 01' middle aged men. BI' Med J

J 990: 300; 634-8.20 Brunner E. Davey Smith C. Marrl10t M. Call11er R. Bekinska .\1.

O'lkien J. Childhood soeial cireumstanees and psychosocialand behavioural factors as determinants 01' plasma fibrinogen.Le/llcel 1996: 347: 1008-1.3.

21 Markowe HL. Marmot MG. Shipley MI. Bulpitt C). Meade TW.Sterling Y et al. Fibrinogen: a possible link between social dass

and coronary IICart discase. BI' MeeE1198 5: 291: 1312-4.22 Wilson nv, Kaplan GA. Kauhanen J. Cohen RD. Wu .\1.

Salonen R. Salonen JT. Associalion betwcen plasma librinogenconcentration and five socioeconomie indices in the Kuopio

Ischemie Heart Disease Risk Factor Study .. 1/11 I Epid/'llliol1993: 137: 2n-300.

23 /.Ioller I.. Kristensen TS. Plasma librinogen and ischemic heartdiscasc risk faetors. Arterioscler TlmJ//lb 1991: 2: 344-50.

24 Schnall PI.. Pieper C. Schwartz JE. Karasek RA. Schussel Y.

Devereux RP. Pickering TC. The relationship between 'jobstrain·. workplace diastolic biom! pressure. and lert ventricularmass index: reslJlts 01' a ease-control study. IllMt!1990; 263: 1929-35.

25 Ragland D. Winklcby M. Schwalbe J. lIolman B. Prevalence 01'

hypertension in busdrivers. 1111 J El'idellliol I 'IR 7: ] 6: 201;-14.26 Siegrist J. Peter R. Threat to oeeupational status control and

cardiovaseular risk. Isrl II,Jed Sci 1996: 32: 179-84.27 Wilhelmscn L. Svärdsudd K. K()rsan-ßengl~en K. Larsson ß.

\Velin L. Tibblin G. Fibrinogen as a risk factor 1'01' stroke and

myocardial infarction. N Ellyll MeeE 191\4: 3] J: 50] .28 Ross R. The pathogcncsis 01' atherosderosis - an update. N

Ellyll MeeE1986: 314; 488-500.29 Fuster V. Badimon 1.. Badimon Jj. Chescbro JH. The pathogene­

sis 01' coronary artery discase and the acute eoronary syn­dromes. N Ellgl J MeeE 1992: 326: 242-50: 3](}-18.

30 Wieland H. Seidel D. Quantitive Lipoprotein Electrophoresis.In: Lewis LA. cd. CRC Hmldlwok oI Eh-ctrol",on'sis W. BocaRaton. USA: CRC Press. 1983: 8 ~-j()2.

156 J. SIEGRIST etal.VAN BARLINGEN etal.

31 Cremer P. Nagel D. Labrot B. Mann H. Muche R. Elster H.

Seidel D. Lipoprotein Lp(a) as predictor of myocardial infarc­

tion in comparison to fibrinogen. LDL cholesterol and otherrisk factors. Ellr J ClilllllVt'st 1994: 24: 444-53.

32 Cremer P. Nagel O. Labrot B. Muche R. Elster H. Mann H.

Seidel O. Göttinger Risiko-. Inzidenz- und Prävalenzstudie(GRIPS) [Göttinger Risk. Incidence. and Prevalence Study·1Berlin. Heidelberg. New York. London. Paris. Tokyo: Springer.1991.

33 Baker IA. Eastham R. Elwood PC. Etherington M. O'Brien JR.Sweetnam PM. Haemostatic factors associated with ischemic

IJeart disease in men aged 45-64 years: the Speed weIl Study.

Br Hearl J 1982: 47: 490-94.34 Polsom AR. Wu KK. Davies CE. Conlan MG. Sorlie PD. Szklo M.

Population correlatcs 01' plasma tibrinogen and factor VII:

putative cardiovascular risk factors. All1ernsclernsis 199 I: 91:19]-205.

35 Expert Panel on Detection. Evaluation. and Treatment 01' HighBlood Cholesterol in Adults. Summary 01' thc Second Report01' the National Cholesterol Education Program (NCEP) ExpertPanel on Oetection. Evaluation. and Treatment 01' High BloodCholcsterol in Adults (Adult Treatment Panel II). JAMA 1993:269: 30 15-23.

36 Siegrist J, Peter R. Job stressors and coping characteristics inwork-related disease: issues 01' validity. Work [, Stress 1994: s:130-40.

37 Peter R. Sieg rist J. Chronic work stress. sickness absence. and

hypertension in middle managers - general or specific socio­

logical explanations? SOl' Sei Med 1997. in press.38 Siegrist J. Peter R. Motz W. Strauer BE. The role of hyperten­

sion. lefl ventricular hypertrophy and psychosocial risks incardiovascular disease: prospective evidence from blue-collar

men. Ellr Hellrl J] 992: 13 (Suppl.): 89-95.39 Efron B. The efficiency of logistic regression compared to nor­

mal discriminanl analysis. JAM/I1975: 70: 892-8.40 Hoffmeister H. Mensink GBM. Stolzenberg H. Hocltz J. Kreuter

H. Laaser U et al. Reduction 01' coronary hearl disease risk fae­

tors in the German cardiovascular prevention study. Prel' Med1996: 25: 135--45.

41 Assmann G. Schulte H. Relation 01' high density lipoprotein

cholestero! and triglycerides to incidence of atherosclerotic

coronary artery disease (the PROCAM experience). AIII J

Cardio/1992: 70: 733-7.42 Fehily AM. Milbank JE. Yarnell jWG. Hayes TM. Kubiki AJ.

Eastham RD. Oietary determinants 01' lipoproteins. total cho­

leslerol. viscosity. fibrinogen and blood pressure. AIII J Clill Nw] 982: 36: 890-96.

43 Krobot K. Hense HW. Cremer P. Eberle E. Keil U. Oelerminants

01' plasma fibrinogen: relation to body weight, waist-lo-hipratio. smoking. alcohol. age and sex. Artl'rioscla Tlr/'OIII/J 1992:7: 780-88.

44 Folsom AR. Qamhieh HT. Flack JM. Hilner IE. Liu K. HowardBV cl al. Plasma fibrinogen: levels and correlates in youngadults (CARDlA Study). AIII J Epidellliol199 3: 138: 1023-36.

45 Bak AA. van Vliel HH. Grubbe DE. Coffee. caffeine and hemo­stasis: resulls from two randomized sludies. AtlJe/'Osclernsis

1990: 83: 249-55.

Received 12 November 1996: accepted II March 1997.

CorrespOlldellee: Professor Dr Johannes Siegrist. Institut fürMedizinische Soziologie. Heinrich Heine Universität. Postfach 101007. D-4000] Oüsseldorf, Germany (fax: +49-211-811-2390:e-mai!: siegrist@uni-duesscldorr.de).

1')')7 nlackwcl/ Scil'/lcl' LId liJll/'IlIIl (Ir IlIli'/'/1II1 Mi'dirirll' 241: 149-1 :;"