Systolic blood pressure in relation to occupation and perceived work stress.

This paper outlines the methods used in a five-year prospective study of 4,607 Australian government employees and presents initial cross-sectional results examining associations between systolic blood pressure, occupation, and subjects' perceptions of work stress as ascertained in a questionnaire survey. The questionnaire items were grouped by means of principal component analysis into six factors representing adverse job characteristics. There were appreciable occupational differences in the mean scores for all six factors and in systolic blood pressure. The differences in systolic blood pressure among men from different occupational categories could not be explained by variation in the level of occupational stress perceived and reported by the subjects. Age, occupation, body mass index, and level of perceived stress arising from financial problems were significantly associated with systolic blood pressure level. A multiple linear regression model was used to determine the relationship between systolic blood pressure, age, and occupation after adjustment for the effects of the other variables.

Despite declining mortality, coronary hea rt disease (CHD) remains the leading cause of death in Australia (4,14). It is also the leading cause of premature retiremen t attributed to organic disease and of death prior to retirement (7). Al though CHD is not regarded as an o ccupational d isease, it ha s been sugg ested that aspect s of wo rk could co nt ribute to th e ris k (I , 11, 16, 17, 21 , 29) . This possibility is strengthened by occupational va riations in th e occurrence of CHD and in distributions of CHD risk facto r s, including blood pressure (10,16,21 ). The question there fore arises of whether the differences in CHD risk are due to occupational influences or to co nstitut ional a nd environmental factors distributed d ifferentially among occupations .
Union co ncerns about the possibl e link between wo rk stress and CHD led to the planning o f a blood pressure study within th e School of Public Health a nd Tropical M edicine in Sydney. While th ere is no generally accepted measure of harmful stress, the risk o f CHD has been found to va ry with oc cupation al level (16 ), responsibilit y a t wo rk (29), job control in rel ation to work load (I) , support for th e worker from supervisors, colleagues and family (17), and typ e A or type B behavior pattern (26) . This stud y has taken the approach of ranking subjects ' reactions to perceived work stress.
Occupational di fferences in CHD are important industrially and ha ve implications fo r th e prevent ion o f C H D . Worksit e intervention could be effective in reducin g CHD risk (8,27). The efficiency of int ervention may depend on the di stribution of risk in working populations.
Accordingly, the blood pressure stud y was planned with the foll ow ing three major objectives: (i) to id entify occupational factors associated with CHD risk, (ii) to observe the progression and community control of h ypertension a nd other CHD risk indicators in d iffer ent o ccupational groups, a nd in relation to occupa tional cha nge, a nd (iii) to co nd uc t a tria l o f riskfac to r modification based on counseling by occupational nurses.
This paper outlines the struc ture of and me thods used in the blood pressu re study . It th en presents results on the association o f occupation and blood pr essure a mo ng the men and on the co ntri butio n of subjects ' reactions to perceived work stress to this association.

Subjects and method
The first objecti ve was addressed by a cross-sectiona l survey whi ch provided th e base line for the prosp ective re search a pprop riate to the seco nd and third objecti ves. The sec o nd o bjective was addressed by a cohort study, and the third by a randomized controlled trial of intervention .
The study population con sisted of 4 607 Australian govern ment emplo yees workin g in Sydney. Th ey were recruited on a voluntary basis between April 1977 and Octob er 1980. Seventy-five percent of the subjects were recrui ted from Telecom Au str alia , 24 010 from the Au str alian Taxat ion Office, and 1 % from Au stralia Po st. Table 1 gives a breakdown o f the study population by employer and sex.
Th e initial data collection from each subject took place in worktime durin g two visits to the study center, located in Telecom premises in Sydne y. On the fir st visit subjects underwent preliminar y screening by a nur se and answer ed a self-administered wor k respon se qu estionnaire developed to record reaction s to perceived work stre ss. Thi s que stionnaire also included items on per son al relationships and domestic and financial problems, as well as items designed to identify the type A behavior pattern. The preliminary screening comprised interviewer-administered que stions seeking demo graphic and occupational information, mea suremen t of pul se rate and blood pres sure , a 12-lead electrocardiogram , anthr opometry, collection of a random venous blood sample for hematology and biochemical analysis, and collection of a mid stream urin e specimen for microscop y.
Th e subjects returned about two weeks late r fo r medi cal assessment. In the interim the y completed a biographical questionnaire which included items on personal health , medications, family history of cardiovascular disease , tob acco and alcohol consumption , diet , leisure-time activities, and additional occupational and demo graphic deta ils. Medical assessment consisted of a ph ysical examination focusing on the cardiovascular system and a simple test of heart rate response to Table 2. Criteria for risk categorizat io n. phy sical exertion. Mea sur ements of pulse rate and blood pressure were repeated und er conditions identical to tho se of the preliminary screening and followed the methods proposed by the WorId Health Organization for cardiovascular sur veys (22). Hawksley random-zero sphygmomanometers were used for blood pressure determination, the reading being made after the subject had been seated at rest for 15 min .
The physician conducting the medical assessment evaluat ed the screening data , relevant biographical data and laboratory results, discussed th ese result s and the physical findings with th e subject , a nd th en assigned the subject to one of four CHD risk categories depending on diastolic blood pressure , serum cholesterol level, mean skinfold thickness from three sites, heart rate response to exertion, cigarette smoking, and alcohol con sumption. Th e pres ence of significant electrocard iographic abnormalities and the use of hormonal contraception by women over the age of 40 years were also taken into account. The criteria for risk categorization ar e summarized in table 2.
Subj ects under tre atment for cardiovascula r and related disorders at enrollment were placed in a separat e gro up for observation, irrespecti ve of the assigned risk level. Subjects with risk levels II and 1lI entered the intervention trial and were randomized into intervention and reference groups. Intervention con sisted > 24 E a Pascal equivalent in parentheses. b One drin k equivalent to 10 g of alcohol. C Mean from thre e site s: subs capular, iliac cre st , and tr iceps folds . d Heart rate (HR) response to physical exercise: the subject was requ ired to step up w ith both feet onto a box 20 cm high at a rate of 30 steps per minute. HR was measured afte r each complete minute of stepping . Results were recorded as follows : A: HR :5120/min after 3 min of stepp ing , B: HR :5120/min after 2 min of stepping but > 120 after 3 min, C: HR :5120/min after 1 min of stepping but > 120 after 2 m in, D: HR "" 120/min after 1 min of stepping, and E: rest ing HR > 100/min or resting HR :5100/min but increasing by 10/min or mo re on standing. The test was omitted if the subject was taking beta-adrenergic blo cking drugs for any reason or had any contr aindic ati ng physical cond ition. It was abandoned when HR exceeded 120/min or if result E was assign ed.
of intensive counseling on risk factor modification by occupational nurses over a three-month period and did not include the prescription of drugs. Subjects with risk level IV not already receiving treatment were referred to their own practitioners or to clinics; many of them required drug therapy . No attempt was made to influence treatment received outside this study, but all subjects were eligible to enter the follow-up program whether or not they were receiving treatment. Clinical data collected on each subject were made available to his or her nominated medical attendant if the subject so wished . Follow-up routines depended on the assigned risk levels. All subjects were to be followed for five years, regardless of change of employer.
Those with risk levels III and IV, and others receiving treatment on enrollment, were approached annually for review. Those with risk levels I and II were scheduled for review three years after enrollment, and again two years later. At each review the procedure was the same as initially, screening being followed by separate medical assessment.
Apart from the intensive counseling during the three months after enrollment, the intervention and reference groups participated in the same follow-up routines as the subjects not in the intervention trial. Thus the intervention trial of selected subjects was nested within the cohort study of all the subjects , and conducted concurrently with it.
Any subject found to have elevated CHD risk or any other significant abnormality during any follow-up assessment was appropriately counseled or referred by the study physician, irrespective of base-line risk level or status in the intervention trial. Thus all subjects were eligible to receive some intervention in the study. The desigr.ated intervention and reference groups were distinguished by their risk factor levels on enrollment and by their random allocation to the intervention or reference group. This arrangement satisfied the ethical requirement to inform subjects of any clinical abnormalities detected during follow-up. Moreover, it simulated the operation of an occupational health service which provided counseling for employees within the context of numerous community source s of health advice.
The work response questionnaire was developed specifically for the blood pressure study, as no established instrument could be found to estimate response to physical and mental work stressors satisfactorily. The questions addressed recent perceived mental stress , with items on personal relationships, money and health (4 questions); job or people at work (1 question); frequency (14 questions) and amount (16 questions) of feelings about work; support (3 questions); physical symptoms experienced in reaction to work (12 questions); and behavior pattern (15 questions). The items were selected from many literature sources (2,5,12,18,19) and from earlier studies (6) as being potential indicators of work distress. Subjects were asked to rank 478 their responses to each item on a five-point scale in which the lowest rank represented no distress and the highest rank a high level of distress.
The present analysis deals with the 30 items on feelings about work , the three items on support , and one item on recent perceived mental stress referring to the job or people at work. These 34 questions, which are listed in appendix I, were grouped by mean s of principal component analysis using the Statistical Package for the Social Sciences (20). Six factors, together accounting for 50.2 % of the total variance, were identified . These factors accorded with prior logical grouping of the information sought in the questionnaire.
Details of the principal component analysis are given in appendix 2. Factor I represents boredom. It places weight on the subject's degree of interest in, involvement in, and satisfaction derived from work. Factor II represents lack of support. It places weight on the quality of communication surrounding the subject in the workplace, ie, support from supervisors and colleagues, consultation by management and supervisors, understanding of job responsibilities, and availability of counseling. Factor III represents quantitative overload, ie, time pressure, work load, and the imposition of conflicting demands. Factor IV represents qualitative overload, for example, level of responsibility, complexity of work, and contact with the public. Factor V represents inadequacy ofremuneration and prospects, ie, perceived adequacy of salary in relation to the nature of the job, present and future appreciation of skills by the employer, present and future relevance of skills, and opportunities for advancement. Factor VI represents unsatisfactory physical work conditions, including noise, ventilation, and workspace.
Based on the pnncipal component analysis, a total score for each of the six factors was computed for each subject as outlined in appendix 2. A high score on any factor represented perception of, and a strong reaction to, the adverse job characteristics identified in the factor.
Because of the need to disentangle any possible effect of non work distress, responses to the following three questions from the questionnaire were also examined: (i) how much, if any, stress or worry have you had lately with your spouse or other person you are intimately involved with, (ii) how much, if any, stress or worry have you had lately with other members of your family or e10se friends, and (iii) how much, if any, stress or worry have you had lately with money or home finance. (These items were not included in the principal component analysis .)

Results
The results described have been confined to men employed by Telecom and the Taxation Office. Data collection was complete for 3 246 (98.2 0J0) of the 3 306 men, who were divided into the following eight occu -pational categories (table 3): (i) executives and administrators, (ii) professionals, (iii) technicians, (iv) clerks, (v) tax investigators and valuers, (vi) tax assessors, (vii) skilled tradesmen, and (viii) other blue-collar workers. One hundred and eighty-six men were receiving drug treatment for hypertens ion (5.7 070); in 115 of these, systolic blood pressure was less than 160 mm Hg (21 331 Pa) and diastolic pressure less than 95 mm Hg (12665 Pal. The association between scores for the questionnaire factors , age, and occupational category was examined. Table 4 gives the mean score for each factor by age. An anal ysis of variance demonstrated that score was significantly related to age for factors I, II, III, V and VI (p <O .OOI) and to occup ation for all six factor s (p<O.OOl). There was no interaction effect between age and occupation. Table 5 gives th e mean factor scores for the occupat ional categories, adjusted for age with the use of partial regression coefficients.
Thus concerns related to factor I (boredom) were prominent in the tax assessor category; to factor II (lack of support) in the clerical cate gor y; to factor III (quantitative overl oad) in the adm inistrative and professional categories; to factor IV (qualitative overload) in the tax investigator, technical, skilled trades and other blue-collar categories; to factor V (inadequacy of remuneration and prospects) in the technical, skilled trades, and other blue-collar categories; and to factor VI (unsatisfactory physical work conditions) in the skilled trades category. The next step was to explore associations between occupation and systolic blood pressure and to determine whether work stress reported in the questionnaire was related to occupational variations in systolic blood pressure. The following extraneous variables were considered likely to complicate relationships between the occupational variables, including work stress, and systolic blood pressure: Age.   Education: The level of education might have influenced responses to the self-administered questionnaire, which relied heavily on subjects' language comprehension; some parallels between occupational category and level of education are to be expected.
Reported distress unassociated with work: perceived stress in relation to spouse, family members, and close friends and to financial problems.
For each of the six factors, scores were classed into low, medium, and high tertile ranges. Analyses of variance were done comparing the mean systolic pres-sures across the terti Ie ranges of the six factors, the age groups, the body mass index categories, level of education categories, and tertile ranges of responses to the three nonwork distress items. There was no significant association between the six work stress factors and systolic blood pressure. The only variables to have significant associations with systolic blood pres-Sure were age group (p < 0.001), occupational category (p < 0.01), body mass index (p < 0.001), and perceived stress arising from financial problems (p < 0.001). In addition the interaction of age with occupation had a significant effect (p<O.Ol).
The mean level of systolic blood pressure was found to increase with age in all occupations, but to the least extent among professionals (table 6), and to increase with increasing body mass index (table 7). However, the mean systolic blood pressure level decreased with increasing levels of perceived stress arising from financial problems (table 8). On the basis of th ese results a mul tiple linear regression model was fitte d with systolic blood pressure as the dependent variable. The model cont ained four main effects (age group, occup ational category, bod y mass index, and level o f perceived stress arising from financial problems), and the int era ction effect of age with occupation . The categorical independent variables were recorded as indicator variables. The model explained 16.7 070 of the variance.
The mean level of systolic blood pressure within each tertile score ra nge of the six work stress factors was adjusted for age, bod y mass index, and perceived stress arising from financial problems with the use of the coefficients derived from the regression equation. After adjustment, the mean systolic blood pressure did not vary appreciabl y across the three score levels of an y work stress fact or (table 9).
In order to confirm th at there was no work stress effect on systolic blood pr essure, and no effe ct of the interaction of work stress and occupation, the model was recon stru cted , each of the work stress factors and its interaction with occupational category being added. Th e resultant change in each case was negligible and non significant.
Finall y th e mean systolic blood pressure , adjusted for bod y mass index and the effect of perceived stress arising from financ ial problems, was calculated for each age group within each occupational category, again with the use of regression coeff icients. The results are given in table ro. In the 45-to 54-year age group, the adjusted mean systolic blood pressure ranged from 127 mm Hg (16 932 Pa) for techni cians to 135 mm Hg (17 998 Pa) for administrators. In the > 54-year age group, the range was from 126 mm Hg (16798 Pa) for professionals to 139 mm Hg (18 531 Pa ) for tax investigators. The variation was not so marked in the younger age groups. Overall , the professional and techn ical categories had the lowest systolic blood pressures. The increase with age was least in the professional category and greatest in the skilled trades, other blue-collar, and tax investigator categories.

Discussion
Initial cross-sectional results from the blood pre ssure study indicate appreciable dif ferences in systolic blood pressure among men of different occupational categori es. The se differen ces could not be explained by variation in the level of occupational stre ss perceived and reported by the men.
The se conclusions are presented with two reservation s. First, the result s pertain to a cross-sectional survey of a volunteer population. While the study was not intend ed to establish prevalence, the possibility of a volunteer bias cannot be overlooked ; relationships between occupation al variables and systolic blood pressur e could differ between volunteers and nonvolunteers . Th e base population from which the subjects were dra wn cannot be enumerated with any accura cy because major reorganization was taking place within Telecom and the Taxation Offi ce during the recruitment period (1977)(1978)(1979)(1980). Telecom employs about 92 000 people throughout Australia , about 30 000 o f whom are in New South Wales. How ever , the 3 390 male and female Telecom subjects in this study all came fro m Sydney metropolitan establishments, and the respon se rate from these establi shments was estimated to be about 45 %. The response rate from Sydney branches of the Au stralian Taxation Office was estimated to be about 35 %. There are few contemporary data on CHD risk factor distribution s in repr esentative samples of emp loyed per sons in Australia. It is Table 9. Mean systol ic blood pressure (in mm Hg with pa~cal equ ivalent in parent heses) by tert ile lev~ls of the percE'!ved work stre ss adjusted for age , body mass Inde x and perceived stress arising from f inancial problems.  therefore difficult to assess whether this large stud y population is representative .
The second reservation concerns the definition and measurement of occupa tional stress. Despite much investigation, the concept of stress remains nebulous. The experience of stress ran ges from the positive notion of challenge to something much more unpleasant, o ften surmised to be harmful. Such experience is difficult to define and measure. The measu rement of perceived stress in this study follows the precedent of identifying potentially stressful elements of worklife and then asking subjects to rank their react ions (if any) to these elements. The stressful elemen ts are adverse job characteristics about which employed people often complain. They apply in a wide variety of workplaces but cannot be comprehensive for all workplaces, The self-report process itself intro duces difficulties of interpretation. There is evidence that personal variables associated with CHD risk in general, and with elevated blood pressure specifically, can influence reaction s to and self-reported perception s of adverse element s in the work environment (28). For example, elevated resting blood pressure in you ng men has been found to be associated with high levels o f work satisfaction and low levels of aggression (indicated by willingness to reason in conflict situations) (13). Highly motivated people with low levels of verbal aggression may underreport adverse or stressful elements in their work environment. In contrast, peopl e with a known histo ry of hypertension may seek environmental " causes" for their disorder and overreport stressful clement s in their work when given th e opportunity to do so on a questionnaire . Hypertensives who are aware of their dia gnosis may have lower levels of psychological well-being (as measured by selected items from standard psychologi cal test batt eries) than normotensives or hypertensives unaware of their diagno sis, and labeled hypertensives may ha ve higher rates of absenteeism than hypertensives who do not know that they have elevated blood pressure (15). The level of psychological well-being may influence responses to questions such as those posed in th e work response qu estionnaire. The relationship betwe en psychological well-being and perceived work stress in hypertensives is not clear. It is also unclear whether absenteeism following labeling as a hypert ensive is related to perceptions of stress in the wor k environment.
Th e measurement of non work distress causes similar difficulties of interpr etation . In this study increasing levels of reported stress arising from financial problems were associated with decreasing systolic blood pressur e. The reason for this inverse relationship is a matter for speculation. Subje cts with higher systolic blood pressure levels were less inclined to report concern over mone y matters. The possible simple explanation that they were wealthier cannot be tested from the avai lable data.
Since the setting of protocols for this study , other approaches to the evaluation of occupational stress, 482 such as the measurement of physiological markers, have been tried (9,23,24,25). It rema ins to be seen whether th ey will offer greater obj ectivity.
Results from this study on blood pr essur e reinforce other observation s (II , 16, 21) that there arc dif ferences in CHD risk levels among occupational gro ups. The results suggest that , for systolic blood pressure , a major risk factor, the differences are unlik ely to be due to perceived work stress. This conclu sion was drawn fro m cross-sectional da ta on which it was not possible to examine the influence of such impo rtant variables as the length of time in the job . Forthcoming prospective data from the study will prov ide an opportunity for exploring relationships between occupational change, perceived work stress, and CHD risk. order of importance, and then it mad e a varimax rotation on six factors. The six-factor solution was chosen becau se it accounted for a sat isfactory proportion (50 .2 al o) of the tot al variance while at the same time allowing a coherent interpretation consistent with the intent of the work response que stionnaire. For the purpose of interpretation an item was considered to be represented in any given factor if its factor load ing was at least 0.33. From table A, which gives the factor loadings and communalities, it is evident that the six factors concentrate upon the following items: Factor IV: items 5, 10, 14, 22, 23, 24 Factor V: items 4,15,17,18,20,25 Factor VI: items 28, 29 The program also generated factor score coefficients for each item in the six factors. In each item the coefficient was multiplied by the corresponding standardized respon se value of the individual subject. The resulting products were then summed over the 34 items to produce six factor scores for each subject.