Brisson, Chantal PhD; Laflamme, Nathalie PhD; Moisan, Jocelyne PhD; Milot, Alain MD, MSc, FRCPC; Masse, Benoit PhD; Vezina, Michel MD, MPh, FRCPC
BP = blood pressure.
To our knowledge, only two studies have evaluated the effect of family responsibilities on blood pressure [1,2]. Both studies were conducted in populations of employed women. The authors reported a significant positive association between the number of children and hypertension prevalence  and systolic and diastolic blood pressures at home . In addition, the work of Frankenhaeuser  has shown that among female managers, blood pressure remained elevated in the evening after work, unlike male managers who showed a decrease. In the Framingham Study, employed women who had three or more children had a higher incidence of cardiovascular disease than employed women who had no children or than housewives with three or more children . Increases in blood pressure have been found in women and men exposed to high psychological job strain [5,6].
An increasing proportion of women are professionally active in most industrialized countries  and have to assume both professional and family roles. Several studies showed that women assume more of the domestic work and child caring than men [8-10], even when both of them are working . The domestic workload was reported to increase with the presence of children [8,11]. Several studies from North America and Scandinavia found that women's jobs are more likely to involve high psychosocial constraints than men's jobs [4,12-14].
The objectives of the present study were to evaluate the association between large family responsibilities and ambulatory blood pressure among white-collar women and to assess the association between the combined exposure of large family responsibilities and high job strain, and blood pressure. The independent effect of job strain on blood pressure has been described in a previous study .
Study Design and Selection of the Study Population
A cross-sectional study was conducted. The selection of the study population has been described elsewhere . Subjects were identified using data from a previous investigation on work environment and cardiovascular disease (Medical Research Council of Canada MA-11364). A total of 3183 (76% of those eligible) white-collar women in eight organizations participated in this previous investigation conducted in 1991 to 1993 (Table 1). Of these, 2813 were between 18 and 64 years old and were working full time. Due to the high cost and large amount of time needed to monitor blood pressure, we limited the number of subjects to the minimum needed for sufficient statistical power. We, therefore, took advantage of our database to select only women exposed to high or low job strain (N = 983). Other selection criteria were also applied (Table 1) to maximize power by restricting the number of factors to be controlled for in the analysis. Women who had children in shared custody were excluded to obtain a more homogeneous exposure to family responsibilities. Eligible women (N = 689) were stratified into four groups according to the family environment (presence or absence of children) and the work characteristics (high or low strain), and a random sample of approximately 100 women was drawn from each of the four groups (N = 416) to reach the final optimal sample size of approximately 50 in each group (N = 199). Selected women were contacted by a trained interviewer between October 1993 and May 1994. Of these women, 19 (5%) could not be contacted, 24 (5%) refused to participate, and 163 (39%) conformed to at least one exclusion criteria (Table 1). Women who looked after a physically or mentally ill child or adult (particular stress) or had definite angina pectoris classified by the Rose questionnaire  (to minimize any possible influence of medication on blood pressure) were excluded. This left 210 women in the sample. Subsequent to this, 10 women with missing data on one hourly BP average in the diurnal period and 1 woman with missing data on oral contraceptive use were eliminated from the analysis, reducing the size of the study sample to 199. The 11 women excluded were comparable with respect to their family environment (presence or absence of children) and their work characteristics (high strain or low strain) to the 199 women used in the analysis. These 199 women ranged from 24 to 62 years of age with a mean +/- SD of 38.2 +/- 6.9. They were either managers (N = 4), professionals (N = 62), technicians (N = 59), clerical employees (N = 71), or other public service employees (N = 3).
Data were collected from two different sources. Eligibility criteria for the initial selection and data on personality were taken from the 1991 to 1993 database. All other data were collected between October 1993 and June 1994. First, women selected in the initial selection process were contacted by telephone. About 1 week after the telephone interview, subjects still eligible completed a self-administrated questionnaire on work and family characteristics, lifestyle, and sociodemographic variables. Weight and height were measured. In the 2 weeks after the questionnaire, ambulatory blood pressure was recorded using the Spacelabs 90207 monitor (Spacelabs Produits Medicaux Ltee, St-Laurent, Quebec, Canada). Subjects wore the monitor for 24 hours during a normal working day. Measures were taken every 15 minutes between 7:00 AM and 10:00 PM and every 30 minutes between 10:00 PM and 7:00 AM. Monday and Friday were excluded to eliminate possible variations due to the proximity of the weekend. Each woman also completed a diary indicating her posture, physical activity, and the occurrence of any stressful event before or at the time of each blood pressure reading.
Average diurnal (8:00 AM-9:00 PM) systolic and diastolic blood pressures were calculated, using mean hourly blood pressures. In addition, average systolic and diastolic BPs for the working period (9:00 AM-4:00 PM), the evening period (5:00-9:00 PM) and the night (12:00 midnight-6:00 AM) were calculated. The end of the evening period was brought forward to 9:00 PM because of incomplete information about the sleeping period (to assure that women would be awake during the period under examination).
Five measures were used to assess simple and more complex definitions of family responsibilities: a) living with children (yes/no), b) number of children living with the respondent, c) children load, d) proportion of domestic work, and e) family load. The last three measures were adapted from Tierney et al. . Adaptations were made to refine the weight given to young children and to provide a more comprehensive list of housework based on information provided in other studies [8,11,18].
The children load index is a measure of the quantity of housework. It is based on the observation that the number of hours of housework is proportional to the number of children and their ages [8,11]. Younger children were ascribed more weight than older children [8,11,18]. Children in the age category 0 to 5, 6 to 11, and >or=to12 years were, respectively, weighted 3, 2.5, and 2. The algorithm was: Children load = (3 x number of children from 0 to 5 years old) + (2.5 x number of children from 6 to 11 years old) + (2 x number of children 12 years old or older). The categories used were: scores of 0 = low; >0 to <or=to4 = medium; >4 = high.
Proportion of Domestic Work
The proportion of domestic work index quantifies the percentage of domestic work performed by the respondent. It was assessed by a 13-item questionnaire evaluating domestic work performed in the previous week . Ten tasks were related to housework: meal preparation, grocery shopping, laundry, ironing, fixing things inside the house, cleaning inside the house, working outside or around the house, paying the bills, taking care of car problems, and planning family life. Three tasks were related to child care: making the children understand what is wrong and what is right, taking the children to the doctor, and putting the children to bed. Subjects had to answer whether they primarily performed each task by themselves, with the help of another person (spouse, child, or someone else), primarily by another person, or whether nobody did the task, which respectively gave them a score of 1, 0.5, 0, and 0. The algorithm was: Proportion of domestic work = (summation of the score for the 13 tasks)/number of relevant tasks. A task was considered irrelevant if it was not performed by anyone. The total score was divided by the number of relevant tasks to evaluate the proportion instead of the quantity (which was already evaluated by the index of children load). The categories used were: scores of 0< to <53.85 = low; 53.85<or=to to <69.23 = medium; >or=to69.23 = high.
The family load index represents the total measure of both children load and percentage of domestic work. It is based on the children load adjusted by the proportion of domestic work performed by the respondent. The algorithm was: Family load = children load + 1) x proportion of domestic work (we added 1 to children load so that a woman with no children would not have "0," inasmuch as this could not be weighted by the proportion of domestic work). The categories used were: scores of 0<or=to to <112 = low; 112<or=to to <or=to323.05 = medium; >323.05 = high.
High job strain arises when a person facing high psychological demands has low decisional latitude or control over his or her work . Psychological demands and decisional latitude were assessed using the two 9-item scales of the Karasek's Job Content Questionnaire . The validity of a preliminary version of the instrument containing 14 of the 18 items used in our study has been documented [19,21]. The validity of the 18-item instrument (two 9-item scales) has also been documented for the original English version  and the French version used in this study . Psychological demands refer to the quantity of work, the intellectual requirements, and the time constraints of the job. Decisional latitude refers to the possibility of making decisions, of being creative on the job and of using and developing one's abilities. Scores of psychological demands and decisional latitude were dichotomized at the median of the distribution of a random sample of all workers in the province of Quebec (high psychological demands, score >or=to9, and low decisional latitude, score <or=to72) . Job strain was categorized as follows: Workers exposed to both high psychological demands and low decisional latitude composed the high-strain group, and all others composed the low-strain group (the sample size available did not allow more refined categories). Job strain data used in the analyses came from the 1993 to 1994 self-administered questionnaire. Between the initial and the second data collection, 48% of the study population had a variation of psychological demands or decisional latitude score greater than 10%.
Other Relevant Variables
Body mass index was defined by weight (kg)/height (m2). Smokers were defined as subjects smoking at least one cigarette each day. Alcohol use was measured by the average number of drinks of alcohol consumed per week in the last 12 months (a drink being a glass of beer, a glass of wine, or an ounce of hard liquor). Physical activity during leisure time was measured by the number of times per week the subject engaged in vigorous physical activity lasting at least 20 minutes. Oral contraceptive use was defined according to the current usage. Demographic and work-related variables measured by the questionnaire included age, degree of education completed, occupation, number of hours worked per week, and family earnings. Physical effort at work was defined with a question taken from the Quebec Health Study . Social support at work was measured with an 8-item scale from the Job Content Questionnaire . Four items were related to supervisor support and four were related to coworker support. The measure of social support outside work was composed of answers to four questions on available social support (advice or listening) that were used in the Quebec Cardiovascular Study . Personality was evaluated with a variable composed of five questions on hostility taken from the Cook-Medley Hostility Scale  and three questions on anger taken from the study by Haynes et al. . In addition, difficulty expressing feelings was measured using five questions adapted from the alexithymia construct validated by Taylor et al. .
All analyses were conducted separately for each systolic and diastolic BP and for each exposure definition. The analysis involved calculating crude mean diurnal BP grouped according to the different family responsibility indexes and comparing the mean BP of each exposed group to the unexposed group. The potentially modifying effect of age (25-39 vs. 40-64 years), education (with vs. without a university degree), size of organization (divided in three groups according to the number of workers in the organization), number of hours worked (35-40 vs. more than 40 hours), having worked for less than a year at this job (<1 year; >or=to1 year), social support at work (divided at the median), and difficulty expressing feelings (divided at the median) were first evaluated by stratified analysis. We then evaluated the statistical significance of the interaction using the F statistic obtained from an analysis of variance. The potentially confounding effect of all variables listed in the Methods section was evaluated using an analysis of variance. Each covariate was added one by one to the model, which contained only the exposure parameters. To be included in the model, the covariate had to have changed at least one exposure coefficient by more than 10%. Adjusted mean BP grouped according to the different family responsibility indexes and differences of adjusted mean BP between each exposed group and the unexposed group were calculated using an analysis of covariance. Student's t test was used to assess whether the observed BP differences between each exposed group and the unexposed group were statistically significant. For the analysis using the combined exposure of family responsibilities and job strain as independent variables, the previously described strategies to assess modifiers and confounders were also used.
The mean +/- SE systolic and diastolic BPs were, respectively, 119.4 +/- 8.0 mm Hg and 76.6 +/- 6.2 mm Hg during daytime (8:00 AM-9:00 PM), and 99.6 +/- 6.7 mm Hg and 59.1 +/- 5.6 mm Hg at night (12:00 midnight-6:00 AM). Systolic BP was higher among less educated women, smokers, and women taking oral contraceptives (Table 2), as observed in previous studies [29-32]. Systolic BP tended to be higher among the older women (N = 7). Leisure-time physical activity was not associated with BP. All but three women reported low physical effort at work.
Mean diurnal BP grouped according to family responsibility measures are presented in Table 3. Results are presented separately for women with and without a university degree, due to the modifying effect of education observed in our data. There was no modifying effect for age, size of organization, number of hours worked, and social support at work and outside work. The modifying effect of education was reflected in the multivariate analysis by a significant multiplicative interaction between family responsibility indexes, education, and BP (p < .05). The final models used to estimate the effect of family responsibilities and job strain on BP included the following covariates: age, education, smoking, oral contraceptive use, and the interaction between exposure and education. Among women with a university degree, women with large family responsibilities had higher diurnal systolic (2.7-5.7 mm Hg) and diastolic (1.8-4.0 mm Hg) BPs when compared with women with small family responsibilities, except for the measure relating to the proportion of domestic work performed by the respondent (Table 3). These differences reach statistical significance (p <or=to .05) for the three family responsibility indexes that take into account the number of children for systolic BP and for two of these for diastolic BP. BP was not higher among women with moderate family responsibilities when compared with those with small family responsibilities. Women with three or more children (N = 7) were also studied separately and yielded results similar to those of women with two children. Among women without a university degree, there was no consistent association between family responsibilities and systolic and diastolic BP.
Mean diurnal BP grouped according to job strain and the combination of family responsibilities and job strain is presented in Table 4. As in Table 3, results are presented separately for women with and without a university degree, due to the modifying effect of education. There was also a significant multiplicative interaction between job strain, education, and BP (p < .05) . Among women holding a university degree, there was a statistically significant elevation of systolic and diastolic BP of 5.9/4.3 mm Hg associated with high job strain (high psychological demands and low decisional latitude). The combination of large family responsibilities (defined by having children, high children load, or high family load) and high job strain was associated with an increase ranging from 8.1 to 10.9 mm Hg for systolic BP and 5.5 to 7.1 mm Hg for diastolic BP, these differences being statistically significant for all three measures of family responsibilities. Elevated BP was also found, but was lower and generally not statistically significant, for women exposed to either large family responsibilities or high job strain but not both. The multiplicative interaction of family responsibilities and job strain was tested (using the F statistic obtained from covariance analysis) and was not statistically significant (p > .15), whatever the definition of family responsibilities. Among women without a university degree, no statistically significant association was found.
Following these analyses, we examined the differences in systolic and diastolic BPs at work, in the evening, and during the night, grouped according to the combination of family load (the total measure of family responsibilities) and job strain exposure among women with a university degree (Table 5). The combination of high family load and high job strain was associated with statistically significant increases in systolic and diastolic BPs for all three periods (work, evening, and night).
Large family responsibilities were associated with significant increases in diurnal systolic and diastolic BPs among white-collar women holding a university degree. In these women, combined exposure of large family responsibilities and high job strain tended to have a greater effect on BP than the exposure to either one of these factors. No effect was found among women without a university degree.
Some strengths of the present study are worth noting. Ambulatory measures of BP are more representative of the true values than are casual measures and have twice the precision of a single measure . Associations between large family responsibilities, high job strain, and BP among women holding a university degree were significant after adjusting for most potential confounders. After controlling for multiple comparisons using the Bonferroni correction , all associations were still significant at alpha = .05 except for the simple effect of the number of children on diastolic BP, which remained near the significance level.
Limitations are also present. The cross-sectional design limits causal inference and may lead to selection or information bias. For example, the proportion of domestic work could have been overestimated by women, especially those who had previously been told that they had high BP. However, this is unlikely in our data, inasmuch as no association was found with the proportion of domestic work. The measures of family responsibilities also have limitations. The measure of the proportion of domestic work may not include all of the relevant tasks. However, the 10 housework tasks evaluated were pertinent in 77% of subjects (and only 6% had more than one irrelevant task). The three level answer scale of this measure may limit its discrimination power and the load related to each task may not be similar. Also, in the proportion of domestic work index, dividing the total score by the number of relevant tasks may theoretically lead to a high score in women who have less tasks. However, this is not a limitation of the index, inasmuch as its aim is to measure the proportion of work performed and not the load per se. Nevertheless, it requires caution when interpreting it as a simple index. Our measure of job strain in two categories (high vs. others) instead of more refined categories is also a limitation. These limitations may lead to a nondifferential information bias which, in general, leads to an underestimation of the true association. Night BP may have been influenced by the presence of the monitor and the difficulty some women had in sleeping. This could induce a bias that may underestimate the association observed with night BP.
In our data, the score of proportion of domestic work did not vary significantly according to the number of tasks. Indeed, among women with children, those having <or=to11 tasks and those having >11 tasks had comparable scores: 60.4 versus 60.9, p > .05. Among all women, those having <or=to11 tasks and those having >11 tasks had scores of 65.0 and 60.9, (p > .05). However, the score of children load and the score of family load were both significantly higher among women who had >11 tasks than among those with less tasks, both in women with children (p < .01) and in women without children (p < .001).
We evaluated whether the absence of association among less educated women could be explained by a differential selection factor. The proportion of women excluded for each exclusion criterion was similar when grouped according to the presence or absence of children at home and according to education level. However, as we documented in a previous report , there was a differential selection by job strain and education level. Indeed, among women without a university degree, more than twice as many women with high strain (25.2%) were excluded because they no longer worked in the same organization as were women with low strain (10.7%) (p = .003: Fisher's exact test). Among women holding a university degree, the proportion of exclusions for this criterion was similar when grouped according to job strain exposure (20.6 vs. 15.2, p = .50; Fisher's exact test). If women excluded also tended to have a higher BP than women who participated, this differential selection could lead to an underestimation of the true association between job strain and BP in less educated women and thus possibly explain the absence of an association in these women. Similar biases, related to the healthy worker selection effect, have been documented in occupational studies . In addition, if there is an interaction between family responsibilities, job strain, and BP, this differential selection could also lead to an underestimation of the effect of family responsibilities on BP among less educated women. We also evaluated other possible explanations for the modifying effect of education. Homogeneity of the exposure scores could not explain the absence of an association among less educated women, inasmuch as both family responsibility measures and job strain scales were distributed similarly, whatever the education level. Previous studies on family responsibilities and BP [1,2] and those on job strain and BP in women  did not evaluate the possible modifying effect of education. Two studies on job strain and heart disease among women that were stratified by social class showed mixed results [4,36]. Among men, contrary to our findings, two studies found a stronger effect of job strain on BP in lower socioeconomic status groups [37,38]. Thus, there is little prior evidence consistent with our finding of an effect only in highly educated women. However, Light et al.  have recently reported that the effect of effort coping at work on BP was higher among women with high job status than among other women. In addition, Frankenhaeuser et al.  suggested that female middle managers experience the strongest conflict between work and home life and reported that their BP remained elevated in the evening after work, unlike the female clerical workers who showed a decrease. These two studies are consistent with ours in suggesting that highly educated women, assuming that they have high job responsibilities, experience stronger conflicts between work and family than other women, thereby increasing their vulnerability to stressors.
Among women with a university degree, the increase in BP associated with large family responsibilities and with both large family responsibilities and high job strain are clinically significant. Indeed, it has been demonstrated in a meta-analysis of nine prospective studies conducted among women and men that a persistent elevation of 5 mm Hg of diastolic BP increases the risk of strokes by 34% and the risk of coronary heart disease by 21% . The association between daytime ambulatory BP and these end points was found to be stronger than that of casual blood pressure [41,42].
Our data suggests that the number of children may be a determinant of systolic and diastolic BP among women with a university degree. The age of the children may also be of some importance in this relationship. These results are consistent with earlier findings of a positive association between the number of children and hypertension prevalence  and systolic and diastolic BPs at home  in women. In our study, the increase in BP seems to be present only among women exposed to high children load (two or more children) rather than a gradient relationship as reported in the two previous studies [1,2]. Finally, the absence of a significant relationship between the proportion of domestic work performed by the women and BP is also consistent with earlier findings .
Women exposed to both large family responsibilities and high job strain tended to have greater increases in BP than women exposed to only one of these factors. The effect of this double exposure seemed to be the sum of both main effects. This increase in BP was present through work, evening, and night, suggesting a persistent effect beyond the work setting. Although this finding was based on a small number of women exposed to both factors (N = 5), it was statistically significant, independent of confounders, and was not due to the influence of a few single observations.
In conclusion, the present study showed increases in BP associated with large family responsibilities among white-collar women holding a university degree. In these women, the combined exposure to large family responsibilities and high job strain tended to have a greater effect on BP than the exposure to only one of these factors. Results provide no evidence of an effect among less educated women. This last finding could be due to an underestimation of the true effect in less educated women in our study or to a true difference in women with higher education. Additional studies are needed to evaluate the combined effect of large family responsibilities and high job strain on BP and the possible modifying effect of education on these associations.
This work was supported by the National Health Research and Development Program of Canada, Grant 6605-4095-60B, the Medical Research Council of Canada Grant MA-11364, the Heart and Stroke Foundation of Quebec, and the Saint-Sacrement Hospital Foundation. Chantal Brisson is a National Health Research Scholar from Health Canada.
The authors wish to thank Marie Begin and Carole Blanchette for data collection and survey work, and Celine Valin for her clerical support. The authors also wish to thank women who participated in this study and their employers and union representatives.
1. James GD, Cates EM, Pickering TG, Laragh JH. Parity and perceived job stress elevate blood pressure in young normotensive working women. Am J Hypertens 1989;2(8):637-9.
2. Zimmerman MK, Hartley WS. High blood pressure among employed women: a multi-factor discriminant analysis. J Health Soc Behav 1982;23:205-20.
3. Frankenhaeuser M, Lundberg U, Fredrikson M, Melin B, Tuomisto M, Myrsten A-L, Hedman M, Bergman-Losman B, Wallin L. Stress on and off the job as related to sex and occupational status in white-collar workers. J Organ Behav 1989;10:321-46.
4. Haynes SG, Feinleib M. Women, work and coronary heart disease: prospective findings from the Framingham heart study. Am J Public Health 1980;70(2):133-41.
5. Schnall PL, Landsbergis PA, Baker D. Job strain and cardiovascular disease. Ann Rev Public Health 1994;15:381-411.
6. Theorell T, Karasek RA. Current issues relating to psychosocial job strain and cardiovascular disease research. J Occup Health Psychol 1996;1(1):9-26.
7. Organisation de cooperation et de developpement economique. Eco-Sante, version 1.5. 1993.
8. Le Bourdais C, Hamel PJ, Bernard P. Le travail et l'ouvrage. Charge et partage des taches domestiques chez les couples quebecois. Sociol Soc 1987;19:37-55.
9. Wright EO, Shire K, Hwang S-L, Dolan M, Baxter J. The noneffects of class on the gender division of labor in the home: a comparative study of Sweden and the United States. Gender Soc 1992;6(2):252-82.
10. Biernat M, Wortman CB. Sharing of home responsibilities between professionally employed women and their husbands. J Pers Soc Psychol 1991;60:844-60.
11. Lundberg U, Mardberg B, Frankenhaeuser M. The total workload of male and female white collar workers as related to age, occupational level, and number of children. Scand J Psychol 1994;35:315-27.
12. Hall EM. Gender, work control, and stress: a theoretical discussion and an empirical test. Int J Health Services 1989;19(4):725-45.
13. Netterstrom B, Kristensen TS, Damsgaard MT, Olsen PO, Sjol A. Job strain and cardiovascular risk factors: a cross sectional study of employed Danish men and women. Br J Indust Med 1991;48:684-9.
14. Greenlund K, Liu K, Knox S, McCreath H, Dyer A, Gardin J. Psychosocial work characteristics and cardiovascular disease risk factors in young adults: the Cardia Study. Soc Sci Med 1995;41(5):717-23.
15. Laflamme N, Brisson C, Moisan J, Milot A, Masse B, Vezina M. Job strain and ambulatory blood pressure among female white-collar workers. Scand J Work Environ Health 1998;24:334-343.
16. Rose GA. Chest pain questionnaire. Mihbank Memorial Fund Q 1965;43(Pt 2):32-39.
17. Tierney D, Romito P, Messing K. She ate not the bread of idleness: exhaustion is related to domestic and salaried working conditions among 539 Quebec hospital workers. Women Health 1990;16(1):21-42.
18. National Institute of Child Health, and Human Development. Women's lifestyle during pregnancy study questionnaire. New York: Columbia University; 1988.
19. Karasek RA Jr. Job demands, job decision latitude, and mental strain: implications for job redesign. Adm Sci Q 1979;24:285-308.
20. Karasek RA. Job content questionnaire and user's guide. Los Angeles: Department of industrial and system engineering, University of Southern California; 1985.
21. Karasek RA, Theorell T. Healthy work: stress, productivity and the reconstruction of working life. New York: Basic Books; 1990.
22. Schnall PL, Schwartz JE, Landsbergis PA, Warren K, Pickering TG. Relation between job strain, alcohol, and ambulatory blood pressure. Hypertension 1992;19:488-94.
23. Larocque B, Brisson C, Blanchette C. Coherence interne, validite factorielle et validite discriminante de la traduction francaise des echelles de demande psychologique et de latitude decisionnelle du "Job Content Questionnaire" de Karasek. Rev Epidemiol Sante Publique 1998;46:371-381.
24. Sante Quebec. Et votre coeur, ca va? Rapport de l'enquete quebecoise sur la sante cardiovasculaire. Quebec: Gouvernement du Quebec; 1994.
25. Sante Quebec. Et la sante, ca va? Quebec: Rapport de l'enquete Sante Quebec; 1987.
26. Barefoot JC, Dodge KA, Peterson BL, Dahlstrom WG, Williams RB Jr. The Cook-Medley hostility scale: item content and ability to predict survival. Psychosom Med 1989;51:46-57.
27. Haynes. Psychosocial scales and items used in the Framingham heart study. Am J Epidemiol 1978;107(5):382-3.
28. Taylor GJ, Bagy M, Ryan DP, Parker JDA. Validation of the Alexithymia Construct: A measurement-based approach. Can J Psychiatry 1990;35:290-7.
29. Shapiro AP, Rutan GH. Hypertension in women: differences and implications. In: Eaker ED, Packard B, Wenger NK, Clarkson TB, Tyroler HA, editors. Coronary heart disease in women. New York: Haymark & Doyma Inc; 1987. p. 172-6.
30. Benowitz NL, Kuyt F, Jacob P III. Influence of nicotine on cardiovascular and hormonal effects of cigarette smoking. Clin Pharmacol Ther 1984;36:74-81.
31. Dyer AR, Stamler J, Shekelle R, Schoenberger J. The relationship of education to blood pressure. Circulation 1976;54:987-92.
32. The Joint National Committee on Detection Evaluation, and Treatment of High Blood Pressure. The fifth report of the Joint National Committee on detection, evaluation, and treatment of high blood pressure (JNC V). Arch Intern Med 1993;153:154-83.
33. Radaeli A, Piepoli M, Adamopoulos S, Pipilis A, Clark SJ, Casadei B, Meyer TE, Coats AJS. Effects of mild physical activity, atenonol and the combination on ambulatory blood pressure in hypertensive subjects. J Hypertens 1992;10(10):1279-82.
34. Kleinbaum DG, Kupper LL. Applied regression analysis and other multivariable methods. Boston: Dusbury Press; 1978.
35. Checkoway H, Pearce NE, Crawford-Brown DJ. Research methods in occupational epidemiology. New York: Oxford University Press; 1989.
36. LaCroix A, Haynes S. Occupational exposure to high demand/low control work and the incidence of CHD in the Framingham Cohort [PhD Thesis]. Chapel Hill: University of North Carolina; 1984.
37. Theorell T, Perski A, Akerstedt T, Sigala F, Ahlberg-Hulten G, Svensson J, Eneroth P. Changes in job strain in relation to changes in physiological state. Scand J Work Environ Health 1988;14:189-96.
38. Landsbergis PA, Schnall PL, Warren K, Pickering TG, Schwartz JE. Association between ambulatory blood pressure and alternative formulations of job strain. Scand J Work Environ Health 1994;20:349-63.
39. Light KC, Brownley KA, Turner JR, Hinderliter AL, Girdler SS, Sherwood A, Anderson NB. Job status and high-effort coping influence work blood pressure in women and blacks. Hypertension 1995;25(Pt 1):554-9.
40. MacMahon S, Peto R, Cutler J, Collins R, Sorlie P, Neaton J, Abbott R, Godwin J, Dyer A, Stamler J. Blood pressure, stroke, and coronary heart disease. Lancet 1990;335(Pt 1):765-74.
41. Perloff D, Sokolow M, Cowan RM, Juster RP. Prognostic value of ambulatory blood pressure measurements: further analyses. J Hypertens 1989;7(Suppl 3):S3-10.
42. Perloff D, Sokolow M, Cowan R. The prognostic value of ambulatory blood pressures. JAMA 1983;249(20):2792-8.