- Email: [email protected]
Women, Employment Status, and Hypertension
Women, Employment Status, and Hypertension: Cross-Sectional and Prospective Findings from the Atherosclerosis Risk in Communities (ARIC) Study KATHRYN M. ROSE, PhD, BETH NEWMAN, PhD, HERMAN A. TYROLER, MD, MOYSES SZKLO, MD, DrPH, DONNA ARNETT, PhD, AND NARAIN SRIVASTAVA, MD
PURPOSE: This study examined the cross-sectional and prospective associations between employment status and hypertension among middle-aged, African-American (AA) and European-American (EA) women participating in the Atherosclerosis Risk in Communities Study. METHODS: Employed women and homemakers from the baseline examination (1987–89) were included in the cross-sectional study (n 5 7351). Associations between employment and the incidence of hypertension ascertained at visit 2 (1990–92) were determined among those who at baseline, had low-normal blood pressure (not hypertensive and systolic blood pressure (SBP) < 120 mm Hg systolic and diastolic blood pressure (DBP) < 80 mm Hg (n 5 3194). Logistic regression analysis was used to examine the association between employment status and hypertension by ethnicity, taking into account covariates. RESULTS: At baseline, employed women were less likely to be hypertensive (SBP > 140 mm Hg or DBP > 90 mm Hg or current use of antihypertensive drugs) than were homemakers (prevalence odds ratio (POR) 5 0.70; 95% confidence interval (CI) 5 0.62–0.79), controlling for age, body mass index, and education. Among the subgroup who had low-normal blood pressure at baseline, employed women were less likely to develop hypertension during the three-year time period than were homemakers (odds ratio (OR) 5 0.68; 95% CI 5 0.44–1.05). The inverse association was stronger among AA (RR 5 0.37; 95% CI 5 0.16–0.88) than EA (OR 5 0.83; 95% CI 5 0.50–1.38) women. CONCLUSIONS: These findings suggest that the inverse association between hypertension and employment status is not due to a healthy worker effect, and that employment may confer protection against incident hypertension in women. Ann Epidemiol 1999;9:374–382. 1999 Elsevier Science Inc. All rights reserved. KEY WORDS:
African-American, Employment Status, Hypertension, Women.
INTRODUCTION The steady increase in women’s labor force participation between the 1950s and 1990s (1) gave rise to the prediction that women would experience higher rates of coronary heart disease (CHD) (2, 3). Investigators also hypothesized that the increased CHD risk would be most pronounced among women employed in traditionally “male” occupations. Based on the notion that higher rates of CHD in men, in part, were related to stressors encountered in the workplace, these
From the Department of Epidemiology, University of North Carolina at Chapel Hill, School of Public Health (K.M.R., B.N., H.A.T.); Department of Epidemiology, Johns Hopkins School of Hygiene and Public Health, Baltimore, MD (M.S.); Division of Epidemiology, University of Minnesota School of Public Health, Minneapolis, MN (D.A.); and Veterans Administration Medical Center, Jackson, Mississippi (N.S.). Address reprint requests to: Kathryn M. Rose, Ph.D., Department of Epidemiology, Cardiovascular Disease Program, University of North Carolina at Chapel Hill, 137 E. Franklin Street, NationsBank Plaza, Suite 306, Chapel Hill, NC 27514. Received May 19, 1998; accepted April 2, 1999. 1999 Elsevier Science Inc. All rights reserved. 655 Avenue of the Americas, New York, NY 10010
hypotheses implicitly assumed that the stress associated with childcare, housework, and even “women’s” paid work was relatively benign. At the population level, while women’s labor force participation has continued to increase, CHD mortality has decreased (4, 5). An increased risk of CHD has been noted in a few studies, restricted to select subgroups of employed women (6, 7), whereas others have found no association between employment status and cardiovascularrelated outcomes (8, 9) or have observed a protective employment effect (10). The female labor force is largely comprised of younger and middle-aged women among whom manifest CHD is a relatively rare event. In contrast, elevated blood pressure, which is consistently associated with cardiovascular morbidity and mortality (11–13), is a condition with a relatively high prevalence (14). Although the evidence is equivocal, stress is considered a risk factor for high blood pressure (15–17). If employment does, in fact, adversely influence the cardiovascular health of women, an association with hypertension may be more apparent. A few studies have examined this association. Evidence from one study suggests 1047-2797/99/$–see front matter PII S1047-2797(99)00015-0
AEP Vol. 9, No. 6 August 1999: 374–382
Selected Abbreviations and Acronyms BMI 5 body mass index DBP 5 diastolic blood pressure POR 5 prevalence odds ratio RR 5 relative risk SBP 5 systolic blood pressure WHR 5 waist-hip ratio
that the employment-hypertension association may have shifted across time. In the 1960–62 National Health Examination Survey (NHES), employed women were more likely to be hypertensive than were homemakers, while in the 1976–80 Second National Health and Nutrition Examination Survey (NHANES-II), employed women were less likely to be hypertensive (18). Two other studies conducted in the late 1970s and early 1980s also have reported lower risks of hypertension among employed European-American (EA) women (19, 20); however associations were attenuated after controlling for lifestyle variables. All studies to date have utilized cross-sectional data, precluding the ability to rule out an influence of hypertension on employment status. The current study examines the association between employment status and hypertension in a large, bi-racial cohort of women. A major purpose is to determine if employmenthypertension associations noted cross-sectionally are also seen prospectively, and if variations in hypertension between employed women and homemakers are uniform across categories of workers. Because labor force participation rates (21), employment opportunities (22), and overall socioeconomic conditions (23, 24) have historically varied between African-American (AA) and EA women, and because the relationships between employment status and hypertension may be stronger among AA women (18), a second purpose of this study is to determine if employment–hypertension associations vary by ethnicity.
MATERIALS AND METHODS Participants Participants included in the current study were from the baseline examination (1987–89) of the Atherosclerosis Risk in Communities (ARIC) Study, which was designed to investigate the natural history and etiology of atherosclerosis. Probability samples of men and women ages 45 to 64 were selected from four US communities: Forsyth County, NC; northwest suburbs of Minneapolis, MN; Washington County, MD; and Jackson, MS. AA were over-sampled in Forsyth County and sampled exclusively in Jackson to ensure that ethnicity-specific estimates were possible. Response rates for the initial interview and physical examination were 46% in Jackson and between 65 and 67 percent in the other three communities. A comparison of respon-
Rose et al. WOMEN, EMPLOYMENT STATUS, AND HYPERTENSION
375
dents to non-respondents has previously been described (25), and a more detailed account of the design and procedures has been published (26). In the current study, we were interested in contrasting employed persons to homemakers. Of the 8685 AA and EA women, we excluded those with missing data on employment status (n 5 18) as well as those who were temporarily out of the work force (n 5 99), unemployed (n 5 237), or retired (n 5 945). Also excluded were those with missing data that precluded determination of hypertensive status (n 5 35). Thus, 7351 of the 8685 (85%) baseline participants were included in our cross-sectional analyses. The prospective portion of our study focused on the association between baseline employment status and the incidence of hypertension between baseline and visit 2, hence, participants were limited to the subset of women who attended the three-year (approximate) follow-up examination (n 5 6750). To reduce the probability of misclassifying hypertension at baseline, women who had either a systolic blood pressure (SBP) . 120 mm Hg or a diastolic blood pressure (DBP) . 80 mm Hg or who were taking antihypertensive medications also were excluded (n 5 3532). Thus, the women included in the prospective analyses are referred to in this paper as ‘low-normotensive’. A total of 3194 women (43% of the cross-sectional participants) remained. Baseline and Follow-up Measurements During the baseline home interview, participants were queried about their current occupational status. A woman was classified as employed if she indicated that she was currently working at a job for pay, either full- or part-time. Homemakers were defined as those currently not working outside of the home. Employed persons were asked to indicate their most recent job, which was classified into one of six groups based on the Standard Occupational Classification used in the 1980 Census (27): managerial and professional specialties; technical, sales, and administrative support; service; farming, forestry, and fishing; precision production, craft and repair; and operators, fabricators, and laborers. Only 15 of the employed women were classified into the farming, forestry, and fishing occupations; thus, in occupation-specific analyses these women were omitted. Three sitting blood pressure measurements were taken after the participant had been resting for at least five minutes. Measurements were taken on the right arm with a random-zero sphygmomanometer of the appropriate cuff size. SBP and fifth-phase DBP were measured to the nearest 2-mm Hg. The average of the second and third measurements was used in our analyses. Hypertension was defined as SBP > 140 mm Hg or DBP > 90 mm Hg or current (within the last two weeks) use of anti-hypertensive medications. Blood pressure was measured and hypertension status was ascertained in a comparable manner during baseline and
376
Rose et al. WOMEN, EMPLOYMENT STATUS, AND HYPERTENSION
follow-up examinations. Thus, classification to the incident hypertension category at follow-up required a transition from < 120/80 mm Hg to SBP > 140 and/or DBP > 90 mm Hg, and/or use of antihypertensive drugs. Sociodemographic and health-related information also was obtained during the baseline examination. Age was measured in years. Educational attainment in ARIC was available as a six-level categorical variable and family income as an eight-level categorical variable. However, after determining that use of trichotomous variables did not lead to residual confounding, we categorized education as less than high school graduate, high school graduate with or without vocational training, and at least some college, and family income as , $16,000, $16,000–24,999, and $25,0001. Participants were classified by diabetes status, with diabetes defined as one or more of the following: non-fasting glucose > 200 mg/dl, fasting glucose > 140 mg/dl, self-reported history of diabetes, or current treatment for diabetes (insulin, oral hypoglycemic agents). Participants were asked to rate their health in comparison to individuals of the same age and categorized as excellent/very good or fair/poor. Cigarette smoking status was defined as current, former, and never smoker. Alcohol intake was ascertained from a series of 13 close-ended questions inquiring about past and current consumption of beer, wine, and liquor. This information was used to estimate grams of ethanol consumed per week. Family history of hypertension was ascertained based on self-report of parental (mother, father) history of hypertension. Height was measured to the nearest centimeter and weight to the nearest pound; this information was used to calculate body mass index (BMI) (kg/m2). Waist-hip ratio (WHR), the ratio of the circumference of the waist, at the level of the umbilicus, to the circumference of the hip, at the level of the maximum protrusion of the gluteal muscles was used to measure body fat distribution. Physical activity was measured using a questionnaire developed by Baecke (28); the sports index, ranging from one (low) to five (high), was calculated. Statistical Analysis In both the cross-sectional and prospective studies, logistic regression analyses were used to calculate the odds ratios (ORs) and 95 percent confidence intervals (CIs) for hypertension, comparing employed women (serving as the “exposed”) to homemakers (referent group). Age-adjusted models were run overall and within ethnic groups (AA, EA). Coefficients from these models were used to calculate ageadjusted hypertension prevalence estimates, overall and by employment status. Multiple logistic regression models contained other known risk (protective) factors for hypertension, including ethnicity (overall model only), BMI, WHR, education, income, sports physical activity, alcohol intake, diabetes, family history of hypertension, and smoking status. Final estimates, however, were based on parsimonious mod-
AEP Vol. 9, No. 6 August 1999: 374–382
els, i.e., those that included the minimum subset of variables which produced risk estimates of the employment statushypertension relationship that were comparable (i.e., equal to first decimal place) to those obtained in the more inclusive models. In the prospective study, two additional baseline variables were considered: baseline SBP and perceived health status (which served as an effect modifier of the employment-hypertension association in an earlier study (29)).
RESULTS Participant Characteristics In the cross-sectional study, the average age was 53.1 (6 5.5) years. The average BMI was 27.8 (6 6.1), and the average WHR was 0.89 (6 0.08). Thirty percent of the women were AA. Forty-two percent had family incomes below $25,000, and 23% had less than a high school education. More than 70% were employed. The subset of lownormotensive women included in the prospective study were slightly younger (51.9 6 5.2 years), leaner (BMI 5 25.6 6 4.7), had more favorable fat distribution (WHR 5 0.87 6 0.08), were less likely to be AA (15%), more highly educated (only 14% with less than a high school education), and had higher family incomes (72% with incomes greater than $25,000). These women were also more likely to be employed (78%). The age-adjusted prevalence of hypertension was 32% (Figure 1), but varied considerably by ethnicity, with AA women having a prevalence (57%) two times greater than EA women (23%). Among those who were low-normotensive at baseline, approximately 3% were classified as hypertensive three years later. As with prevalence, incidence varied by ethnicity, being three times higher among AA (9%) than EA (3%) women. Table 1 presents characteristics of participants by employment status, overall and by ethnicity. Homemakers were, on average, between two and three years older and had slightly higher BMIs and WHRs. Homemakers were more likely to have a lower level of education and to have lower family income than were employed women. Among AA homemakers, 81% had family incomes below $16,000, and 63% had less than a high school education; in contrast, among EA homemakers, 19% had family incomes below $16,000 and 25% had less than a high school education. Alcohol intake was similar among AA employed women and homemakers, while among EA women, alcohol intake was higher among those who were employed. Although level of sports physical activity was higher for EA than for AA women, within ethnic groups, sports physical activity levels were similar for employed women and homemakers. Likewise, the prevalence of smoking did not vary much by employment status. The prevalence of diabetes and perception of fair or poor health status were markedly higher
AEP Vol. 9, No. 6 August 1999: 374–382
Rose et al. WOMEN, EMPLOYMENT STATUS, AND HYPERTENSION
377
FIGURE 1. Age- and ethnicity- (overall model only) adjusted prevalence of hypertension at baseline examination (1987–89) and percentage classified as hypertensive at visit 2 (1990–92), overall and by ethnicity, among women participants in the ARIC study. AfricanAmerican (AA); European-American (EA).
among AA than EA women, and within ethnic groups, were considerably higher among homemakers than among employed women. The proportion of women reporting a family history of hypertension did not vary by employment status. Cross-Sectional Findings To determine whether employment in different types of occupations influenced the employment-hypertension asso-
ciation, the age- and ethnicity- (overall model only) adjusted prevalence of hypertension was calculated for each group, and each category of employed women was contrasted to homemakers (Table 2). Overall, and within ethnic groups, homemakers had a significantly higher prevalence of hypertension than did any of the groups of employed women. For AA women, 68% of the homemakers had prevalent hypertension; in contrast, the percentage of hypertensives ranged from 43% (precision production, craft, and
TABLE 1. Selected characteristics of women participants at baseline examination (1987–1989), by employment status, overall, and by ethnicity, ARIC study European-American
Mean age, years (6 SD) Mean BMI, kg/m2 (6 SD) Mean WHR (6 SD) Family income (%) , $16,000 $16–24,999 $25,0001 Education (%) , High school (HS) HS or vocational College or higher Mean alcohol, gm/wk (6 SD) Mean sports index (6 SD) % current smokers % with diabetes % fair/poor health % family Hx hypertension a
African-American
Total
Employed n 5 3607a
Homemaker n 5 1527a
Employed n 5 1792a
Homemaker n 5 425a
Employed n 5 5399a
Homemaker n 5 1952a
52.5 (5.2) 26.4 (5.4) 0.88 (0.08)
55.3 (5.5) 27.2 (5.8) 0.91 (0.08)
52.3 (5.3) 30.5 (6.2) 0.90 (0.08)
54.5 (5.9) 31.6 (7.4) 0.93 (0.08)
52.4 (5.2) 27.7 (6.0) 0.89 (0.08)
55.1 (5.6) 28.1 (6.4) 0.91 (0.08)
11.6 13.6 74.9
19.3 17.4 63.2
50.1 20.0 29.9
80.8 10.9 8.4
23.9 15.6 60.5
31.7 16.1 52.2
11.6 50.3 38.2 26.2 (54.0) 2.4 (0.8) 24.7 4.9 7.3 66.2
24.9 51.5 23.6 21.6 (53.0) 2.5 (0.8) 22.9 9.6 17.7 66.4
33.6 31.1 35.3 11.0 (39.4) 2.1 (0.6) 24.0 14.3 30.8 76.1
63.1 27.4 9.5 11.2 (53.0) 2.1 (0.7) 25.0 30.5 56.9 75.5
18.9 43.9 37.2 21.2 (50.2) 2.3 (0.7) 24.5 7.9 15.1 69.4
33.2 46.3 20.6 19.4 (53.2) 2.4 (0.8) 23.3 14.1 26.3 68.2
n’s for some variables are slightly lower due to missing data.
378
Rose et al. WOMEN, EMPLOYMENT STATUS, AND HYPERTENSION
AEP Vol. 9, No. 6 August 1999: 374–382
TABLE 2. Age- and ethnicity- (overall model only) adjusted prevalence of hypertension at baseline examination (1987–1989), by occupational category among women participants of the ARIC Study Total Employment group
African-American
n
Managerial and professional Technical, sales, administrative support Service Precision production, craft, repair Operators, fabricators, laborers Homemakersa
1530 1948 1227 141 537 1952
%
n b
23.9 31.1b 33.4c 21.8b 30.0b 39.8
472 288 765 49 218 425
European-American
% b
45.0 56.7d 58.8d 43.2c 51.4b 68.1
n
%
1058 1660 462 92 319 1527
17.5b 22.3b 24.0e 15.4d 23.6f 28.8
a
Referent Group. p , 0.0001. c p , 0.001. d p , 0.01. e p , 0.05. f p , 0.10. b
repair occupations) to 59% (service) among those employed. Among EA women, 29% of homemakers were classified as hypertensive, compared to 15% (precision production, craft, and repair occupations) to 24% (operators/laborers and service occupations) of those employed. Because there was a lack of a clear gradient by type of occupation, we classified all employed women into one group in subsequent analyses. Table 3 presents prevalence odds ratios (PORs) and 95% CIs from age- and ethnicity- (overall model only) adjusted multiple logistic regression models of the association between employment status and hypertension, overall and by ethnicity. When age-adjusted, employment showed an inverse association with hypertension for both AA (POR 5 0.54; 95% CI 5 0.43–0.69) and EA (POR 5 0.66; 95% CI 5 0.58–0.76) women, with the inverse association being marginally stronger for AA. Models that additionally controlled for ethnicity (overall model only), BMI, and education yielded associations that were only modestly attenuated (POR 5 0.61; 95% CI 5 0.48–0.78, for AA women, and POR 5 0.74; 95% CI 5 0.63–0.85, for EA women). These
results were similar to those obtained from models additionally controlling for WHR, diabetes, physical activity, alcohol intake, smoking, family income, and family history of hypertension (data not shown). Prospective Findings Table 4 presents results from the prospective analyses of women who were low- normotensive at baseline. During the three years of follow-up, among AA, 18% of homemakers developed hypertension compared to only 8% of employed women (OR 5 0.42; 95% CI 5 0.18–0.95). For EA women, the three-year incidence of hypertension was much lower, and the difference between employed women and homemakers was more modest, with confidence intervals including 1.0 (2.5% vs. 3.4%, respectively; OR 5 0.74, 95% CI 5 0.45–1.21). Controlling for BMI and education in addition to age and ethnicity (overall model only) only modestly influenced risk estimates. As in the cross-sectional analysis, controlling for the larger set of covariates did not substan-
TABLE 3. Prevalence odds ratios (POR) and 95% confidence intervals (CI) for association between employment status and hypertension at baseline examination (1987–1989), overall and by ethnicity, among women participants of the ARIC Study n Total Employed Homemaker African-American Employed Homemaker European-American Employed Homemaker a b
Age-adjusteda POR 95% CI
7351
n
Multivariate modelb POR 95%CI
7336 0.63 0.56–0.71 1.00
2217
0.70 0.62–0.79 1.00 2209
0.54 0.43–0.69 1.00 5134
0.61 0.48–0.78 1.00 5127
0.66 0.58–0.76 1.00
Overall model also adjusts for ethnicity. Models control for age, ethnicity (overall model only), BMI, and level of education.
0.74 0.63–0.85 1.00
AEP Vol. 9, No. 6 August 1999: 374–382
Rose et al. WOMEN, EMPLOYMENT STATUS, AND HYPERTENSION
379
TABLE 4. Age- and ethnicity- (overall model only) adjusted incidence of hypertension at visit 2 (1990–1992) and odds ratios (OR) and 95% confidence intervals (CI) for association between employment status and the incidence of hypertension at visit 2, overall and by ethnicity, among women participants of ARIC study n Total Employed Homemaker African-American Employed Homemaker European-American Employed Homemaker a b
Age-adjusted Incidencea
Multivariate modelb
Age-adjusted OR
95% CI
3.0 4.6
0.64
0.42–0.98 1.00
8.3 17.9
0.42
0.18–0.95 1.00
2.5 3.4
0.74
0.45–1.21 1.00
3194
n
OR
95%CI
0.68
0.44–1.05 1.00
0.37
0.16–0.88 1.00
0.83
0.50–1.38 1.00
3191
481
480
2713
2711
Per 100 women; overall model also adjusts for ethnicity. Models control for age, ethnicity (overall model only), BMI, and level of education.
tially alter results, and they were not retained in the final models. The additional inclusion of baseline SBP and perceived health status at baseline to the multivariate model presented in Table 4 modestly influenced the employment status-hypertension associations (OR 5 0.68; 95% CI 5 0.44–1.07 for all women; OR 5 0.86; 95% CI 5 0.51–1.44 for EA; OR 5 0.36; 95% CI 5 0.15–0.87 for AA).
DISCUSSION After controlling for sociodemographic and behavioral risk factors, employed women had a lower prevalence of hypertension than did homemakers. This finding is consistent with, and stronger in magnitude than, previous reports from cross-sectional studies (18–20). However, by their nature, cross-sectional designs cannot distinguish whether hypertension status, at least in part, determines employment status. Thus, it is of greater interest to determine if these associations persist in prospective analyses. In aggregate, we found an inverse association between employment status and incident hypertension of a magnitude similar to that noted cross-sectionally; after controlling for covariates, this association approached, but did not reach, statistical significance. The strength of this inverse association between employment status and incident hypertension among women varied strongly by ethnicity: it was more notable and statistically significant among AA, in contrast to a modest and not statistically significant association for EA. Differences in the strength of the employment-hypertension association by ethnicity have been reported in an earlier cross-sectional study of women participants of NHES and NHANES-II (18). Our findings do not support the hypothesis that employment among women is associated with higher rates of hypertension. This prediction arose during the post World War II “baby boom” years, when there was a resurgence in domes-
ticity among women and a strong societal norm for married women, particularly during the childbearing years, to remain in the home (30). Women who worked during this time were more likely to do so out of economic need. Under such circumstances, it is not surprising that theories favored the premise that stress contributed to higher rates of disease among employed women. Circumstances are now more favorable for employed women: strong societal proscriptions against married women’s employment have eroded; in fact, the majority of women are now employed (1). Additionally, due in part to the Civil Rights and Women’s Movements, a wider range of employment opportunities and more equitable salaries exist. Newer theories focusing on the potential for women’s employment to benefit health have developed. One, the role accumulation model (31), predicts that employed women have a health advantage over full-time housewives, as employment is a source of gratification and self-esteem beyond that obtainable from non-paid work; other postulated benefits include financial remuneration and independence, as well as increased social support. Strong inverse gradients between cardiovascular health outcomes and indicators of socio-economic status (SES) are now consistently reported (4, 32, 33). Thus, the potential to generate income and positively influence socio-economic circumstances may reflect one important mechanism whereby employment favorably influences blood pressure levels. However, evaluating such associations is not straightforward, due to the complexity of interrelationships between socioeconomic resources and employment, particularly among women. To a certain degree, a dual-peak pattern of employment existed for women in the past: labor force participation was highest among young women, lowest among those during their child rearing years, and then somewhat higher among those who were older (34). However, this was largely a middle-class ideal. Women who were
380
Rose et al. WOMEN, EMPLOYMENT STATUS, AND HYPERTENSION
socially disadvantaged (e.g., AA, unmarried women) were more likely to participate in the paid labor force. AA women historically have had higher labor force participation rates and have been more likely to work full-time than EA women (35). AA women also are more likely to be unmarried (36), and consequently serve as the sole breadwinner; however, due to the unfavorable status of AA men in the US labor force, even married AA women are more likely to contribute substantially to the total family income (24). The participants in the current study first entered the labor force at a time when proscriptive employment norms were operational. Thus, if employment does have a protective effect on hypertension, one might expect to see stronger associations between employment and blood pressure among socio-economically disadvantaged groups whose employment would have been more consistent across the adult years. The relatively strong inverse association between employment and hypertension found for AA compared to EA women is consistent with such an explanation. Information on marital status at the baseline examination was not included in the ARIC baseline examination data set. Thus, marital status information from the three-year follow-up examination was used as a proxy to impute baseline marital status for an exploratory analysis to determine if the crosssectional employment-hypertension associations varied between married and unmarried women (these analyses were not repeated in the smaller group of women included in the prospective analysis because of the limited number of unmarried women). Among EA women, the magnitude of the inverse association was markedly stronger for unmarried (POR 5 0.42; 95% CI 5 0.28–0.64) than for married women (POR 5 0.77; 95% CI 5 0.65–0.91), again consistent with the idea that employment may be more influential among those who are socio-economically disadvantaged. In contrast, among AA women, age-adjusted employment status-hypertension associations were virtually equivalent among married (OR 5 0.62; 95% CI 5 0.43–0.88) and unmarried (OR 5 0.60; 95% CI 5 0.39–0.91) women. This lack of variation among AA women also can be considered consistent with prevailing explanation, given that married women’s earnings contribute more substantially to the total family income than do their EA counterparts. We contrasted homemakers to employed women stratified by type of occupation. While homemakers had a consistently higher prevalence of hypertension than did each occupational group, there was no consistent variation in hypertension prevalence among those in different occupational categories, despite reports of associations between cardiovascular outcomes and occupational status (e.g., 33). A lack of association between occupation and cardiovascular-related outcomes among women has been reported elsewhere (8, 9, 29). The lack of a graded association between a women’s type of occupation and hypertension may reflect that the women’s occupations, per se, are not necessarily
AEP Vol. 9, No. 6 August 1999: 374–382
good markers of their socioeconomic status; this topic has been discussed previously (37, 38). An extensive literature describes the healthy worker effect (39), a phenomenon whereby employed persons have more favorable health than those not working due to selection for health at time of hire and differential persistence in the work force, again on the basis of health status. However, non-employed groups included in such studies (e.g., unemployed, disabled) are not necessarily similar to homemakers, who presumably are not involved in the paid labor force because of non-paid work responsibilities (e.g., child rearing, housework). Over the past decades, the profile of a typical homemaker has shifted considerably: labor force participation among married (and thus, less financially disadvantaged) women has increased dramatically (35), hence a presumably healthy subgroup of homemakers has transferred to the employed group. In the current study, it is obvious (Table 1) that homemakers had less favorable health and socio-economic profiles than did their employed counterparts. This suggests that women “choosing” to be homemakers may be disproportionately comprised of women who, as a group, are prone to higher rates of adverse health outcomes; also, their socio-economically disadvantaged state may be reflective of barriers to gaining or maintaining employment. Thus, it was surprising that the magnitude of inverse associations noted in both our prevalence and incidence studies were not strongly effected by controlling for SES and other factors related to cardiovascular health. It is possible that other adverse health conditions, not measured in our study, may have contributed to the observed inverse association. Likewise, education and family income may not have adequately captured SES, leading to residual confounding. A major strength of our study is the inclusion of prospective analyses, which eliminate the antecedent-consequent ambiguity characteristic of cross-sectional studies. However, by restricting participants to those who were low-normotensive at baseline we limited the pool of newly diagnosed hypertensives at the three-year follow-up examination and reduced statistical power. This was particularly an issue for EA women, as only a small portion of those who were lownormotensive at baseline subsequently developed hypertension. While the OR for EA women was consistent with an inverse association, confidence intervals were wide and included the null value. We re-ran analyses using a less restrictive inclusion criteria (< 130/85 mm Hg at baseline) and obtained a comparable OR; the confidence interval, while narrower, still included 1.0 (OR 5 0.86; 95% CI 5 0.62–1.21). These findings are based on a large, community-based cohort, with a substantial number of AA women that were selected from two southern communities located in an area often referred to as the “stroke belt”. Their employment experiences, health status, and socioeconomic circum-
AEP Vol. 9, No. 6 August 1999: 374–382
stances may not reflect those of AA women living in other areas. Thus, future research based on a more geographically diverse population of AA women is necessary to see if our findings are generalizable to the broader population. ARIC participants were all middle-aged, including women who first entered the labor market during a distinct time period that preceded both the Civil Rights and Women’s Movements. This influenced their initial choices and opportunities in the labor force, and probably at least to some extent, their current status. Findings based on a younger, more recent cohort of women operating under a different set of social norms and expectations might differ. Above, we argued that the potential health effects of employment might be greater among women whose participation in the labor force has been more consistent. There was some indirect, empirical support for this notion in our data. However, employment status was defined dichotomously (employed, homemaker) and limited to current status, thus precluding a direct quantification of amount of time employed. Thus, in future studies investigating the impact of women’s employment status on health, it would be optimal to collect information about extent of labor force participation (e.g., hours worked per week, number of weeks worked per year) and longevity of employment (e.g., number of entrances into/exits from the labor force, percentage of adult years spent in labor force), and other sources of family income. Such information would help clarify if women who spend more time in the labor force do in fact demonstrate stronger associations with health outcomes than those with lower levels of participation. Similarly, many of the psychosocial factors theorized to adversely or positively influence the health of employed women (e.g., social support, self-esteem, stress) are not often measured in epidemiologic investigations. Inclusion of such measures in studies of employment status and health could help verify the validity of the hypotheses. In summary, the inverse association between employment status and hypertension noted at baseline was also true for incidence of hypertension among AA women who were low-normotensive at baseline. A more modest, although not statistically significant, inverse association was also suggested for EA women. The mechanisms whereby employed women have lower rates of hypertension are not known, although we suspect that employment’s influence on a woman’s socioeconomic circumstances is probably at least partially responsible. Positive consequences of employment as well as unfavorable socioeconomic and other circumstances of homemakers as a group both probably contribute. To better clarify the role of psychosocial factors as potential mechanisms, more extensive measurement is needed. As employment continues to become a more prominent part of women’s lives, its potential to influence health outcomes increases. Thus, it is important to consider em-
Rose et al. WOMEN, EMPLOYMENT STATUS, AND HYPERTENSION
381
ployment status as a potentially important exposure in epidemiologic investigations of women’s health. This article was supported in part by National Heart, Lung, and Blood Institute Contracts N01-HC-55015, N01-HC-55016, N01-HC-55018, N01-HC-55019, N01-HC-55020, N01-HC-55021, and N01-HC-55022. The work was completed while the lead author was a post-doctoral fellow in the Cardiovascular Disease Epidemiology Training Program supported by National Institutes of Health, National Heart, Lung, and Blood Institute NRSA grant 5T32HL07055. The authors thank the staff at the ARIC centers: University of North Carolina at Chapel Hill—Phyllis Johnson, Marilyn Knowles, Catherine Paton; University of North Carolina, Forsyth County—Kay Burke, Wilhelmenia Cheeks, Revitha Cook, Shirley Cothern; University of Mississippi Medical Center, Jackson—Betty Warren, Dorothy Washnigton, Mattye Watson, Nancy Wilson; University of Minnesota, Minneapolis—Greg Feitl, Chris Hunkins, Ellie Justiniano, Laura Kemmis; Johns Hopkins University, Baltimore—Joyce Chabot, Carol Christman, Dorrie Costa, Patricia Crowley; University of Texas Medical School, Houston—Valarie Stinson, Pam Pfile, Hogan Pham, Teri Trevino; The Methodist Hospital, Aterosclerosis Clincial Laboratory, Houston— Wanda R. Alexander, Doris J. Harper, Charles E. Rhodes, Selma M. Soyal; Bowman-Gray School of Medicine, Ultrasound Reading Center, WinstonSalem—Anne Safrit, Melanie Wilder, Linda Allred, Carolyn Bell; University of North Carolina at Chapel Hill, Coordinating Center—Peter DeSaix, La Sonya Goode, Thomas Goodwin, Steve Hutton.
REFERENCES 1. Bureau of the Census. Statistical Abstracts of the US: 1990. 110th ed. Washington, DC: Government Printing Office; 1991. 2. Baruch GK, Biener L, Barnett RC. Women and gender in research on work and family stress. Am Psychol. 1987;42:130–136. 3. Friedman M, Rosenman RH. Type A Behavior and Your Heart. New York: Knopf; 1974. 4. Eaker ED, Packard B, Thom TJ. Epidemiology and risk factors for coronary heart disease in women. Cardiovasc Clin. 1989;19:129–145. 5. Feldman JJ, Makuc DM, Kleinman JC, Cornoni-Huntly J, et al. National trends in educational differentials in mortality. Am J Epidemiol. 1989;129:919–933. 6. Haynes SG, Feinleib M. Women, work, and coronary heart disease: Prospective findings from the Framingham Heart Study. Am J Public Health. 1980;70:133–141. 7. Pasanante MR, Nathanson CA. Female labor force participation and female mortality in Wisconsin, 1974–1978. Soc Sci Med. 1985; 21:655–668. 8. Eaker ED, Pinsky J, Castelli WP. Myocardial infarction and coronary death among women: Psychosocial predictors from a 20-year followup of women in the Framingham study. Am J Epidemiol. 1992; 135:854–864. 9. Kotler P, Wingard DL. The effect of occupational, marital and parental roles on mortality: The Alameda County Study. Am J Public Health. 1989;79:607–612. 10. Hibbard JH, Pope CR. Effect of domestic and occupational roles on morbidity and mortality. Soc Sci Med. 1991;32:805–811. 11. Stamler J, Neaton JD, Wentworth DN. Blood pressure (systolic and diastolic) and risk of fatal coronary heart disease. Hypertension. 1989;13(suppl):I2–I12. 12. Fiebach NH, Herbert PR, Stampfer MJ, Coldit GA, Wille HWC, Rosner B, et al. A prospective study of high blood pressure and cardiovascular disease in women. Am J Epidemiol. 1989;130:646–654. 13. Rutan GH, Kuller LH, Neaton JD, Wentworth DN, McDonald RH, Smith WM, et al. Mortality associated with diastolic hypertension
382
Rose et al. WOMEN, EMPLOYMENT STATUS, AND HYPERTENSION
AEP Vol. 9, No. 6 August 1999: 374–382
and isolated systolic hypertension among men screened for multiple risk factor intervention trial. Circulation. 1988;77:504–514. 14. Drizd T, Dannenberg AL, Engel A. Blood Pressure Levels in Persons 18–74 Years of Age in 1976–80, and Trends in Blood Pressure from 1960 to 1980 in the United States. (Vital and health statistics, Series 11: Data from the National Health and Examination Survey, the National Health, Nutritional and Examination Survey I and National Health, Nutrition and Examination Survey II, no. 234) (DHHS publication no. (PHS) 86–1684). Hyattsville, MD: National Center for Health Statistics; 1986. 15. Ostfeld A, D’Atri D. Rapid sociocultural change and blood pressure. Adv Pyschosom Med. 1979;9:20. 16. James GD, Jenner DA, Harrison GA, Baker P. Differences in catecholamine excretion rates, blood pressure, and lifestyles among young western Somoan men. Human Biol. 1985;57:635–47.
26.
27.
28.
29.
17. Pickering TG, James GD, Schnall PL, Schlussel YR, Pieper CF, Gerin W, et al. Occupational stress and blood pressure. In: Frankenhaeuser M, Lundberg U, Chesney M, eds. Women, Work, and Health: Stress and Opportunites. New York: Plenum; 1992:171–186.
30.
18. Rose KM, Newman B, Bennett T, Tyroler HA. Employment status and high blood pressure in women: Variations by time and by sociodemographic characteristics. Ann Epidemiol. 1997;7:107–114.
31. 32.
19. Kritz-Silverstein D, Wingard DL, Barrett-Conner E. Employment status and heart disease risk factors in middle-aged women: The Rancho Bernardo Study. Am J Public Health. 1992;82:215–219. 20. Hazuda HP, Haffner SM, Stern MP, Knapp JA, Eifler CW, Rosenthal M, et al. Employment status and women’s protection against coronary heart disease. Am J Epidemiol. 1986;123:623–640. 21. Amott T, Mathaei TL. The growth of wage work. In: Amott T, Mathaei TL, eds. Race, Gender and Work: A Multicultural Economic History of Women in the United States. Boston: South End Press; 1991:291–314. 22. Amott T, Mathaei TL. We specialize in the wholly impossible. In: Amott T, Mathaei TL, eds. Race, Gender and Work: A Multicultural Economic History of Women in the United States. Boston: South End Press; 1991:141–192. 23. Deckard BS, ed. The Women’s Movement: Political, Socioeconomic, and Psychological Issues. New York: Harper and Row; 1975. 24. Bureau of the Census. Current Population Reports. Series P-60. Washington, DC: Government Printing Office; 1980. 25. Jackson R, Chambless LE, Yang K, Byrne T, Watson R, Folsom A, et al. Differences between respondents and nonrespondents in a multi-
33. 34. 35.
36.
37. 38. 39.
center community-based study vary by gender and ethnicity. J Clin Epidemiol. 1996;12:1373–1379. The ARIC Investigators. The Atherosclerosis Risk in Communities (ARIC) study: Design and objectives. Am J Epidemiol. 1989;129: 687–702. U.S. Bureau of Census: 1980 Census of Population. Classified Index of Industries and Occupations. Final Ed. Washington, DC: U.S. Government Printing Office; 1982. Baecke JAH, Burema J, Frijters JER. A short questionnaire for the measurement of habitual physical activity in epidemiological studies. Am J Clin Nutr. 1982;36:936–942. Rose KM. Women, work and blood pressure: Variations in the employment status hypertension relationship across time, and by age, race and education. (Doctoral Dissertation). University of North Carolina at Chapel Hill; 1994. Amott T, Mathaei TL. Whatever you fight, don’t be ladylike. In: Amott T, Mathaei TL, eds. Race, Gender and Work: A Multicultural Economic History of Women in The United States. Boston: South End Press; 1991:1495–141. Arber S, Gilbert G, Dale A. Paid employment and women’s health: A benefit or a source of role strain? Soc Health Illness. 1985;7:375–400. Liu K, Cedres MS, Stamler J, Dyer A, Stamler R, Dyer A, Stamler R, Nanas S, et al. Relationship of education to major risk factors and death from coronary heart disease, cardiovascular diseases and all causes: Findings of three Chigago epidemiologic studies. Circulation. 1982;66:1308–1314. Rose G, Marmot MG. Social class and coronary heart disease. Br Heart J. 1981;45:13–19. Sorensen G, Verbrugge LM. Women work and health. Am Rev Public Health. 1987;8:235–251. Ries P, Stone AJ. Employment. In: Ries P, Stone AJ, eds. The American Women 1992–1993: A Status Report. New York: Norton and Company; 1992:306–349. Ries P, Stone AJ. Family and Household Structure. In: Ries P, Stone AJ, eds. The American Women 1992–1993: A Status Report. New York: Norton and Company; 1992:244–262. Libaratos P, Link BG, Kelsey J. The measurement of social class in epidemiology. Epidemol Rev. 1988;10:87–121. Wright EO. Women in the class structure. Polit Soc. 1989; 17:1–34. Sterling TD. The healthy worker effect on morbidity rates. J Occup Med. 1985;27:477–482.
PURPOSE: This study examined the cross-sectional and prospective associations between employment status and hypertension among middle-aged, African-American (AA) and European-American (EA) women participating in the Atherosclerosis Risk in Communities Study. METHODS: Employed women and homemakers from the baseline examination (1987–89) were included in the cross-sectional study (n 5 7351). Associations between employment and the incidence of hypertension ascertained at visit 2 (1990–92) were determined among those who at baseline, had low-normal blood pressure (not hypertensive and systolic blood pressure (SBP) < 120 mm Hg systolic and diastolic blood pressure (DBP) < 80 mm Hg (n 5 3194). Logistic regression analysis was used to examine the association between employment status and hypertension by ethnicity, taking into account covariates. RESULTS: At baseline, employed women were less likely to be hypertensive (SBP > 140 mm Hg or DBP > 90 mm Hg or current use of antihypertensive drugs) than were homemakers (prevalence odds ratio (POR) 5 0.70; 95% confidence interval (CI) 5 0.62–0.79), controlling for age, body mass index, and education. Among the subgroup who had low-normal blood pressure at baseline, employed women were less likely to develop hypertension during the three-year time period than were homemakers (odds ratio (OR) 5 0.68; 95% CI 5 0.44–1.05). The inverse association was stronger among AA (RR 5 0.37; 95% CI 5 0.16–0.88) than EA (OR 5 0.83; 95% CI 5 0.50–1.38) women. CONCLUSIONS: These findings suggest that the inverse association between hypertension and employment status is not due to a healthy worker effect, and that employment may confer protection against incident hypertension in women. Ann Epidemiol 1999;9:374–382. 1999 Elsevier Science Inc. All rights reserved. KEY WORDS:
African-American, Employment Status, Hypertension, Women.
INTRODUCTION The steady increase in women’s labor force participation between the 1950s and 1990s (1) gave rise to the prediction that women would experience higher rates of coronary heart disease (CHD) (2, 3). Investigators also hypothesized that the increased CHD risk would be most pronounced among women employed in traditionally “male” occupations. Based on the notion that higher rates of CHD in men, in part, were related to stressors encountered in the workplace, these
From the Department of Epidemiology, University of North Carolina at Chapel Hill, School of Public Health (K.M.R., B.N., H.A.T.); Department of Epidemiology, Johns Hopkins School of Hygiene and Public Health, Baltimore, MD (M.S.); Division of Epidemiology, University of Minnesota School of Public Health, Minneapolis, MN (D.A.); and Veterans Administration Medical Center, Jackson, Mississippi (N.S.). Address reprint requests to: Kathryn M. Rose, Ph.D., Department of Epidemiology, Cardiovascular Disease Program, University of North Carolina at Chapel Hill, 137 E. Franklin Street, NationsBank Plaza, Suite 306, Chapel Hill, NC 27514. Received May 19, 1998; accepted April 2, 1999. 1999 Elsevier Science Inc. All rights reserved. 655 Avenue of the Americas, New York, NY 10010
hypotheses implicitly assumed that the stress associated with childcare, housework, and even “women’s” paid work was relatively benign. At the population level, while women’s labor force participation has continued to increase, CHD mortality has decreased (4, 5). An increased risk of CHD has been noted in a few studies, restricted to select subgroups of employed women (6, 7), whereas others have found no association between employment status and cardiovascularrelated outcomes (8, 9) or have observed a protective employment effect (10). The female labor force is largely comprised of younger and middle-aged women among whom manifest CHD is a relatively rare event. In contrast, elevated blood pressure, which is consistently associated with cardiovascular morbidity and mortality (11–13), is a condition with a relatively high prevalence (14). Although the evidence is equivocal, stress is considered a risk factor for high blood pressure (15–17). If employment does, in fact, adversely influence the cardiovascular health of women, an association with hypertension may be more apparent. A few studies have examined this association. Evidence from one study suggests 1047-2797/99/$–see front matter PII S1047-2797(99)00015-0
AEP Vol. 9, No. 6 August 1999: 374–382
Selected Abbreviations and Acronyms BMI 5 body mass index DBP 5 diastolic blood pressure POR 5 prevalence odds ratio RR 5 relative risk SBP 5 systolic blood pressure WHR 5 waist-hip ratio
that the employment-hypertension association may have shifted across time. In the 1960–62 National Health Examination Survey (NHES), employed women were more likely to be hypertensive than were homemakers, while in the 1976–80 Second National Health and Nutrition Examination Survey (NHANES-II), employed women were less likely to be hypertensive (18). Two other studies conducted in the late 1970s and early 1980s also have reported lower risks of hypertension among employed European-American (EA) women (19, 20); however associations were attenuated after controlling for lifestyle variables. All studies to date have utilized cross-sectional data, precluding the ability to rule out an influence of hypertension on employment status. The current study examines the association between employment status and hypertension in a large, bi-racial cohort of women. A major purpose is to determine if employmenthypertension associations noted cross-sectionally are also seen prospectively, and if variations in hypertension between employed women and homemakers are uniform across categories of workers. Because labor force participation rates (21), employment opportunities (22), and overall socioeconomic conditions (23, 24) have historically varied between African-American (AA) and EA women, and because the relationships between employment status and hypertension may be stronger among AA women (18), a second purpose of this study is to determine if employment–hypertension associations vary by ethnicity.
MATERIALS AND METHODS Participants Participants included in the current study were from the baseline examination (1987–89) of the Atherosclerosis Risk in Communities (ARIC) Study, which was designed to investigate the natural history and etiology of atherosclerosis. Probability samples of men and women ages 45 to 64 were selected from four US communities: Forsyth County, NC; northwest suburbs of Minneapolis, MN; Washington County, MD; and Jackson, MS. AA were over-sampled in Forsyth County and sampled exclusively in Jackson to ensure that ethnicity-specific estimates were possible. Response rates for the initial interview and physical examination were 46% in Jackson and between 65 and 67 percent in the other three communities. A comparison of respon-
Rose et al. WOMEN, EMPLOYMENT STATUS, AND HYPERTENSION
375
dents to non-respondents has previously been described (25), and a more detailed account of the design and procedures has been published (26). In the current study, we were interested in contrasting employed persons to homemakers. Of the 8685 AA and EA women, we excluded those with missing data on employment status (n 5 18) as well as those who were temporarily out of the work force (n 5 99), unemployed (n 5 237), or retired (n 5 945). Also excluded were those with missing data that precluded determination of hypertensive status (n 5 35). Thus, 7351 of the 8685 (85%) baseline participants were included in our cross-sectional analyses. The prospective portion of our study focused on the association between baseline employment status and the incidence of hypertension between baseline and visit 2, hence, participants were limited to the subset of women who attended the three-year (approximate) follow-up examination (n 5 6750). To reduce the probability of misclassifying hypertension at baseline, women who had either a systolic blood pressure (SBP) . 120 mm Hg or a diastolic blood pressure (DBP) . 80 mm Hg or who were taking antihypertensive medications also were excluded (n 5 3532). Thus, the women included in the prospective analyses are referred to in this paper as ‘low-normotensive’. A total of 3194 women (43% of the cross-sectional participants) remained. Baseline and Follow-up Measurements During the baseline home interview, participants were queried about their current occupational status. A woman was classified as employed if she indicated that she was currently working at a job for pay, either full- or part-time. Homemakers were defined as those currently not working outside of the home. Employed persons were asked to indicate their most recent job, which was classified into one of six groups based on the Standard Occupational Classification used in the 1980 Census (27): managerial and professional specialties; technical, sales, and administrative support; service; farming, forestry, and fishing; precision production, craft and repair; and operators, fabricators, and laborers. Only 15 of the employed women were classified into the farming, forestry, and fishing occupations; thus, in occupation-specific analyses these women were omitted. Three sitting blood pressure measurements were taken after the participant had been resting for at least five minutes. Measurements were taken on the right arm with a random-zero sphygmomanometer of the appropriate cuff size. SBP and fifth-phase DBP were measured to the nearest 2-mm Hg. The average of the second and third measurements was used in our analyses. Hypertension was defined as SBP > 140 mm Hg or DBP > 90 mm Hg or current (within the last two weeks) use of anti-hypertensive medications. Blood pressure was measured and hypertension status was ascertained in a comparable manner during baseline and
376
Rose et al. WOMEN, EMPLOYMENT STATUS, AND HYPERTENSION
follow-up examinations. Thus, classification to the incident hypertension category at follow-up required a transition from < 120/80 mm Hg to SBP > 140 and/or DBP > 90 mm Hg, and/or use of antihypertensive drugs. Sociodemographic and health-related information also was obtained during the baseline examination. Age was measured in years. Educational attainment in ARIC was available as a six-level categorical variable and family income as an eight-level categorical variable. However, after determining that use of trichotomous variables did not lead to residual confounding, we categorized education as less than high school graduate, high school graduate with or without vocational training, and at least some college, and family income as , $16,000, $16,000–24,999, and $25,0001. Participants were classified by diabetes status, with diabetes defined as one or more of the following: non-fasting glucose > 200 mg/dl, fasting glucose > 140 mg/dl, self-reported history of diabetes, or current treatment for diabetes (insulin, oral hypoglycemic agents). Participants were asked to rate their health in comparison to individuals of the same age and categorized as excellent/very good or fair/poor. Cigarette smoking status was defined as current, former, and never smoker. Alcohol intake was ascertained from a series of 13 close-ended questions inquiring about past and current consumption of beer, wine, and liquor. This information was used to estimate grams of ethanol consumed per week. Family history of hypertension was ascertained based on self-report of parental (mother, father) history of hypertension. Height was measured to the nearest centimeter and weight to the nearest pound; this information was used to calculate body mass index (BMI) (kg/m2). Waist-hip ratio (WHR), the ratio of the circumference of the waist, at the level of the umbilicus, to the circumference of the hip, at the level of the maximum protrusion of the gluteal muscles was used to measure body fat distribution. Physical activity was measured using a questionnaire developed by Baecke (28); the sports index, ranging from one (low) to five (high), was calculated. Statistical Analysis In both the cross-sectional and prospective studies, logistic regression analyses were used to calculate the odds ratios (ORs) and 95 percent confidence intervals (CIs) for hypertension, comparing employed women (serving as the “exposed”) to homemakers (referent group). Age-adjusted models were run overall and within ethnic groups (AA, EA). Coefficients from these models were used to calculate ageadjusted hypertension prevalence estimates, overall and by employment status. Multiple logistic regression models contained other known risk (protective) factors for hypertension, including ethnicity (overall model only), BMI, WHR, education, income, sports physical activity, alcohol intake, diabetes, family history of hypertension, and smoking status. Final estimates, however, were based on parsimonious mod-
AEP Vol. 9, No. 6 August 1999: 374–382
els, i.e., those that included the minimum subset of variables which produced risk estimates of the employment statushypertension relationship that were comparable (i.e., equal to first decimal place) to those obtained in the more inclusive models. In the prospective study, two additional baseline variables were considered: baseline SBP and perceived health status (which served as an effect modifier of the employment-hypertension association in an earlier study (29)).
RESULTS Participant Characteristics In the cross-sectional study, the average age was 53.1 (6 5.5) years. The average BMI was 27.8 (6 6.1), and the average WHR was 0.89 (6 0.08). Thirty percent of the women were AA. Forty-two percent had family incomes below $25,000, and 23% had less than a high school education. More than 70% were employed. The subset of lownormotensive women included in the prospective study were slightly younger (51.9 6 5.2 years), leaner (BMI 5 25.6 6 4.7), had more favorable fat distribution (WHR 5 0.87 6 0.08), were less likely to be AA (15%), more highly educated (only 14% with less than a high school education), and had higher family incomes (72% with incomes greater than $25,000). These women were also more likely to be employed (78%). The age-adjusted prevalence of hypertension was 32% (Figure 1), but varied considerably by ethnicity, with AA women having a prevalence (57%) two times greater than EA women (23%). Among those who were low-normotensive at baseline, approximately 3% were classified as hypertensive three years later. As with prevalence, incidence varied by ethnicity, being three times higher among AA (9%) than EA (3%) women. Table 1 presents characteristics of participants by employment status, overall and by ethnicity. Homemakers were, on average, between two and three years older and had slightly higher BMIs and WHRs. Homemakers were more likely to have a lower level of education and to have lower family income than were employed women. Among AA homemakers, 81% had family incomes below $16,000, and 63% had less than a high school education; in contrast, among EA homemakers, 19% had family incomes below $16,000 and 25% had less than a high school education. Alcohol intake was similar among AA employed women and homemakers, while among EA women, alcohol intake was higher among those who were employed. Although level of sports physical activity was higher for EA than for AA women, within ethnic groups, sports physical activity levels were similar for employed women and homemakers. Likewise, the prevalence of smoking did not vary much by employment status. The prevalence of diabetes and perception of fair or poor health status were markedly higher
AEP Vol. 9, No. 6 August 1999: 374–382
Rose et al. WOMEN, EMPLOYMENT STATUS, AND HYPERTENSION
377
FIGURE 1. Age- and ethnicity- (overall model only) adjusted prevalence of hypertension at baseline examination (1987–89) and percentage classified as hypertensive at visit 2 (1990–92), overall and by ethnicity, among women participants in the ARIC study. AfricanAmerican (AA); European-American (EA).
among AA than EA women, and within ethnic groups, were considerably higher among homemakers than among employed women. The proportion of women reporting a family history of hypertension did not vary by employment status. Cross-Sectional Findings To determine whether employment in different types of occupations influenced the employment-hypertension asso-
ciation, the age- and ethnicity- (overall model only) adjusted prevalence of hypertension was calculated for each group, and each category of employed women was contrasted to homemakers (Table 2). Overall, and within ethnic groups, homemakers had a significantly higher prevalence of hypertension than did any of the groups of employed women. For AA women, 68% of the homemakers had prevalent hypertension; in contrast, the percentage of hypertensives ranged from 43% (precision production, craft, and
TABLE 1. Selected characteristics of women participants at baseline examination (1987–1989), by employment status, overall, and by ethnicity, ARIC study European-American
Mean age, years (6 SD) Mean BMI, kg/m2 (6 SD) Mean WHR (6 SD) Family income (%) , $16,000 $16–24,999 $25,0001 Education (%) , High school (HS) HS or vocational College or higher Mean alcohol, gm/wk (6 SD) Mean sports index (6 SD) % current smokers % with diabetes % fair/poor health % family Hx hypertension a
African-American
Total
Employed n 5 3607a
Homemaker n 5 1527a
Employed n 5 1792a
Homemaker n 5 425a
Employed n 5 5399a
Homemaker n 5 1952a
52.5 (5.2) 26.4 (5.4) 0.88 (0.08)
55.3 (5.5) 27.2 (5.8) 0.91 (0.08)
52.3 (5.3) 30.5 (6.2) 0.90 (0.08)
54.5 (5.9) 31.6 (7.4) 0.93 (0.08)
52.4 (5.2) 27.7 (6.0) 0.89 (0.08)
55.1 (5.6) 28.1 (6.4) 0.91 (0.08)
11.6 13.6 74.9
19.3 17.4 63.2
50.1 20.0 29.9
80.8 10.9 8.4
23.9 15.6 60.5
31.7 16.1 52.2
11.6 50.3 38.2 26.2 (54.0) 2.4 (0.8) 24.7 4.9 7.3 66.2
24.9 51.5 23.6 21.6 (53.0) 2.5 (0.8) 22.9 9.6 17.7 66.4
33.6 31.1 35.3 11.0 (39.4) 2.1 (0.6) 24.0 14.3 30.8 76.1
63.1 27.4 9.5 11.2 (53.0) 2.1 (0.7) 25.0 30.5 56.9 75.5
18.9 43.9 37.2 21.2 (50.2) 2.3 (0.7) 24.5 7.9 15.1 69.4
33.2 46.3 20.6 19.4 (53.2) 2.4 (0.8) 23.3 14.1 26.3 68.2
n’s for some variables are slightly lower due to missing data.
378
Rose et al. WOMEN, EMPLOYMENT STATUS, AND HYPERTENSION
AEP Vol. 9, No. 6 August 1999: 374–382
TABLE 2. Age- and ethnicity- (overall model only) adjusted prevalence of hypertension at baseline examination (1987–1989), by occupational category among women participants of the ARIC Study Total Employment group
African-American
n
Managerial and professional Technical, sales, administrative support Service Precision production, craft, repair Operators, fabricators, laborers Homemakersa
1530 1948 1227 141 537 1952
%
n b
23.9 31.1b 33.4c 21.8b 30.0b 39.8
472 288 765 49 218 425
European-American
% b
45.0 56.7d 58.8d 43.2c 51.4b 68.1
n
%
1058 1660 462 92 319 1527
17.5b 22.3b 24.0e 15.4d 23.6f 28.8
a
Referent Group. p , 0.0001. c p , 0.001. d p , 0.01. e p , 0.05. f p , 0.10. b
repair occupations) to 59% (service) among those employed. Among EA women, 29% of homemakers were classified as hypertensive, compared to 15% (precision production, craft, and repair occupations) to 24% (operators/laborers and service occupations) of those employed. Because there was a lack of a clear gradient by type of occupation, we classified all employed women into one group in subsequent analyses. Table 3 presents prevalence odds ratios (PORs) and 95% CIs from age- and ethnicity- (overall model only) adjusted multiple logistic regression models of the association between employment status and hypertension, overall and by ethnicity. When age-adjusted, employment showed an inverse association with hypertension for both AA (POR 5 0.54; 95% CI 5 0.43–0.69) and EA (POR 5 0.66; 95% CI 5 0.58–0.76) women, with the inverse association being marginally stronger for AA. Models that additionally controlled for ethnicity (overall model only), BMI, and education yielded associations that were only modestly attenuated (POR 5 0.61; 95% CI 5 0.48–0.78, for AA women, and POR 5 0.74; 95% CI 5 0.63–0.85, for EA women). These
results were similar to those obtained from models additionally controlling for WHR, diabetes, physical activity, alcohol intake, smoking, family income, and family history of hypertension (data not shown). Prospective Findings Table 4 presents results from the prospective analyses of women who were low- normotensive at baseline. During the three years of follow-up, among AA, 18% of homemakers developed hypertension compared to only 8% of employed women (OR 5 0.42; 95% CI 5 0.18–0.95). For EA women, the three-year incidence of hypertension was much lower, and the difference between employed women and homemakers was more modest, with confidence intervals including 1.0 (2.5% vs. 3.4%, respectively; OR 5 0.74, 95% CI 5 0.45–1.21). Controlling for BMI and education in addition to age and ethnicity (overall model only) only modestly influenced risk estimates. As in the cross-sectional analysis, controlling for the larger set of covariates did not substan-
TABLE 3. Prevalence odds ratios (POR) and 95% confidence intervals (CI) for association between employment status and hypertension at baseline examination (1987–1989), overall and by ethnicity, among women participants of the ARIC Study n Total Employed Homemaker African-American Employed Homemaker European-American Employed Homemaker a b
Age-adjusteda POR 95% CI
7351
n
Multivariate modelb POR 95%CI
7336 0.63 0.56–0.71 1.00
2217
0.70 0.62–0.79 1.00 2209
0.54 0.43–0.69 1.00 5134
0.61 0.48–0.78 1.00 5127
0.66 0.58–0.76 1.00
Overall model also adjusts for ethnicity. Models control for age, ethnicity (overall model only), BMI, and level of education.
0.74 0.63–0.85 1.00
AEP Vol. 9, No. 6 August 1999: 374–382
Rose et al. WOMEN, EMPLOYMENT STATUS, AND HYPERTENSION
379
TABLE 4. Age- and ethnicity- (overall model only) adjusted incidence of hypertension at visit 2 (1990–1992) and odds ratios (OR) and 95% confidence intervals (CI) for association between employment status and the incidence of hypertension at visit 2, overall and by ethnicity, among women participants of ARIC study n Total Employed Homemaker African-American Employed Homemaker European-American Employed Homemaker a b
Age-adjusted Incidencea
Multivariate modelb
Age-adjusted OR
95% CI
3.0 4.6
0.64
0.42–0.98 1.00
8.3 17.9
0.42
0.18–0.95 1.00
2.5 3.4
0.74
0.45–1.21 1.00
3194
n
OR
95%CI
0.68
0.44–1.05 1.00
0.37
0.16–0.88 1.00
0.83
0.50–1.38 1.00
3191
481
480
2713
2711
Per 100 women; overall model also adjusts for ethnicity. Models control for age, ethnicity (overall model only), BMI, and level of education.
tially alter results, and they were not retained in the final models. The additional inclusion of baseline SBP and perceived health status at baseline to the multivariate model presented in Table 4 modestly influenced the employment status-hypertension associations (OR 5 0.68; 95% CI 5 0.44–1.07 for all women; OR 5 0.86; 95% CI 5 0.51–1.44 for EA; OR 5 0.36; 95% CI 5 0.15–0.87 for AA).
DISCUSSION After controlling for sociodemographic and behavioral risk factors, employed women had a lower prevalence of hypertension than did homemakers. This finding is consistent with, and stronger in magnitude than, previous reports from cross-sectional studies (18–20). However, by their nature, cross-sectional designs cannot distinguish whether hypertension status, at least in part, determines employment status. Thus, it is of greater interest to determine if these associations persist in prospective analyses. In aggregate, we found an inverse association between employment status and incident hypertension of a magnitude similar to that noted cross-sectionally; after controlling for covariates, this association approached, but did not reach, statistical significance. The strength of this inverse association between employment status and incident hypertension among women varied strongly by ethnicity: it was more notable and statistically significant among AA, in contrast to a modest and not statistically significant association for EA. Differences in the strength of the employment-hypertension association by ethnicity have been reported in an earlier cross-sectional study of women participants of NHES and NHANES-II (18). Our findings do not support the hypothesis that employment among women is associated with higher rates of hypertension. This prediction arose during the post World War II “baby boom” years, when there was a resurgence in domes-
ticity among women and a strong societal norm for married women, particularly during the childbearing years, to remain in the home (30). Women who worked during this time were more likely to do so out of economic need. Under such circumstances, it is not surprising that theories favored the premise that stress contributed to higher rates of disease among employed women. Circumstances are now more favorable for employed women: strong societal proscriptions against married women’s employment have eroded; in fact, the majority of women are now employed (1). Additionally, due in part to the Civil Rights and Women’s Movements, a wider range of employment opportunities and more equitable salaries exist. Newer theories focusing on the potential for women’s employment to benefit health have developed. One, the role accumulation model (31), predicts that employed women have a health advantage over full-time housewives, as employment is a source of gratification and self-esteem beyond that obtainable from non-paid work; other postulated benefits include financial remuneration and independence, as well as increased social support. Strong inverse gradients between cardiovascular health outcomes and indicators of socio-economic status (SES) are now consistently reported (4, 32, 33). Thus, the potential to generate income and positively influence socio-economic circumstances may reflect one important mechanism whereby employment favorably influences blood pressure levels. However, evaluating such associations is not straightforward, due to the complexity of interrelationships between socioeconomic resources and employment, particularly among women. To a certain degree, a dual-peak pattern of employment existed for women in the past: labor force participation was highest among young women, lowest among those during their child rearing years, and then somewhat higher among those who were older (34). However, this was largely a middle-class ideal. Women who were
380
Rose et al. WOMEN, EMPLOYMENT STATUS, AND HYPERTENSION
socially disadvantaged (e.g., AA, unmarried women) were more likely to participate in the paid labor force. AA women historically have had higher labor force participation rates and have been more likely to work full-time than EA women (35). AA women also are more likely to be unmarried (36), and consequently serve as the sole breadwinner; however, due to the unfavorable status of AA men in the US labor force, even married AA women are more likely to contribute substantially to the total family income (24). The participants in the current study first entered the labor force at a time when proscriptive employment norms were operational. Thus, if employment does have a protective effect on hypertension, one might expect to see stronger associations between employment and blood pressure among socio-economically disadvantaged groups whose employment would have been more consistent across the adult years. The relatively strong inverse association between employment and hypertension found for AA compared to EA women is consistent with such an explanation. Information on marital status at the baseline examination was not included in the ARIC baseline examination data set. Thus, marital status information from the three-year follow-up examination was used as a proxy to impute baseline marital status for an exploratory analysis to determine if the crosssectional employment-hypertension associations varied between married and unmarried women (these analyses were not repeated in the smaller group of women included in the prospective analysis because of the limited number of unmarried women). Among EA women, the magnitude of the inverse association was markedly stronger for unmarried (POR 5 0.42; 95% CI 5 0.28–0.64) than for married women (POR 5 0.77; 95% CI 5 0.65–0.91), again consistent with the idea that employment may be more influential among those who are socio-economically disadvantaged. In contrast, among AA women, age-adjusted employment status-hypertension associations were virtually equivalent among married (OR 5 0.62; 95% CI 5 0.43–0.88) and unmarried (OR 5 0.60; 95% CI 5 0.39–0.91) women. This lack of variation among AA women also can be considered consistent with prevailing explanation, given that married women’s earnings contribute more substantially to the total family income than do their EA counterparts. We contrasted homemakers to employed women stratified by type of occupation. While homemakers had a consistently higher prevalence of hypertension than did each occupational group, there was no consistent variation in hypertension prevalence among those in different occupational categories, despite reports of associations between cardiovascular outcomes and occupational status (e.g., 33). A lack of association between occupation and cardiovascular-related outcomes among women has been reported elsewhere (8, 9, 29). The lack of a graded association between a women’s type of occupation and hypertension may reflect that the women’s occupations, per se, are not necessarily
AEP Vol. 9, No. 6 August 1999: 374–382
good markers of their socioeconomic status; this topic has been discussed previously (37, 38). An extensive literature describes the healthy worker effect (39), a phenomenon whereby employed persons have more favorable health than those not working due to selection for health at time of hire and differential persistence in the work force, again on the basis of health status. However, non-employed groups included in such studies (e.g., unemployed, disabled) are not necessarily similar to homemakers, who presumably are not involved in the paid labor force because of non-paid work responsibilities (e.g., child rearing, housework). Over the past decades, the profile of a typical homemaker has shifted considerably: labor force participation among married (and thus, less financially disadvantaged) women has increased dramatically (35), hence a presumably healthy subgroup of homemakers has transferred to the employed group. In the current study, it is obvious (Table 1) that homemakers had less favorable health and socio-economic profiles than did their employed counterparts. This suggests that women “choosing” to be homemakers may be disproportionately comprised of women who, as a group, are prone to higher rates of adverse health outcomes; also, their socio-economically disadvantaged state may be reflective of barriers to gaining or maintaining employment. Thus, it was surprising that the magnitude of inverse associations noted in both our prevalence and incidence studies were not strongly effected by controlling for SES and other factors related to cardiovascular health. It is possible that other adverse health conditions, not measured in our study, may have contributed to the observed inverse association. Likewise, education and family income may not have adequately captured SES, leading to residual confounding. A major strength of our study is the inclusion of prospective analyses, which eliminate the antecedent-consequent ambiguity characteristic of cross-sectional studies. However, by restricting participants to those who were low-normotensive at baseline we limited the pool of newly diagnosed hypertensives at the three-year follow-up examination and reduced statistical power. This was particularly an issue for EA women, as only a small portion of those who were lownormotensive at baseline subsequently developed hypertension. While the OR for EA women was consistent with an inverse association, confidence intervals were wide and included the null value. We re-ran analyses using a less restrictive inclusion criteria (< 130/85 mm Hg at baseline) and obtained a comparable OR; the confidence interval, while narrower, still included 1.0 (OR 5 0.86; 95% CI 5 0.62–1.21). These findings are based on a large, community-based cohort, with a substantial number of AA women that were selected from two southern communities located in an area often referred to as the “stroke belt”. Their employment experiences, health status, and socioeconomic circum-
AEP Vol. 9, No. 6 August 1999: 374–382
stances may not reflect those of AA women living in other areas. Thus, future research based on a more geographically diverse population of AA women is necessary to see if our findings are generalizable to the broader population. ARIC participants were all middle-aged, including women who first entered the labor market during a distinct time period that preceded both the Civil Rights and Women’s Movements. This influenced their initial choices and opportunities in the labor force, and probably at least to some extent, their current status. Findings based on a younger, more recent cohort of women operating under a different set of social norms and expectations might differ. Above, we argued that the potential health effects of employment might be greater among women whose participation in the labor force has been more consistent. There was some indirect, empirical support for this notion in our data. However, employment status was defined dichotomously (employed, homemaker) and limited to current status, thus precluding a direct quantification of amount of time employed. Thus, in future studies investigating the impact of women’s employment status on health, it would be optimal to collect information about extent of labor force participation (e.g., hours worked per week, number of weeks worked per year) and longevity of employment (e.g., number of entrances into/exits from the labor force, percentage of adult years spent in labor force), and other sources of family income. Such information would help clarify if women who spend more time in the labor force do in fact demonstrate stronger associations with health outcomes than those with lower levels of participation. Similarly, many of the psychosocial factors theorized to adversely or positively influence the health of employed women (e.g., social support, self-esteem, stress) are not often measured in epidemiologic investigations. Inclusion of such measures in studies of employment status and health could help verify the validity of the hypotheses. In summary, the inverse association between employment status and hypertension noted at baseline was also true for incidence of hypertension among AA women who were low-normotensive at baseline. A more modest, although not statistically significant, inverse association was also suggested for EA women. The mechanisms whereby employed women have lower rates of hypertension are not known, although we suspect that employment’s influence on a woman’s socioeconomic circumstances is probably at least partially responsible. Positive consequences of employment as well as unfavorable socioeconomic and other circumstances of homemakers as a group both probably contribute. To better clarify the role of psychosocial factors as potential mechanisms, more extensive measurement is needed. As employment continues to become a more prominent part of women’s lives, its potential to influence health outcomes increases. Thus, it is important to consider em-
Rose et al. WOMEN, EMPLOYMENT STATUS, AND HYPERTENSION
381
ployment status as a potentially important exposure in epidemiologic investigations of women’s health. This article was supported in part by National Heart, Lung, and Blood Institute Contracts N01-HC-55015, N01-HC-55016, N01-HC-55018, N01-HC-55019, N01-HC-55020, N01-HC-55021, and N01-HC-55022. The work was completed while the lead author was a post-doctoral fellow in the Cardiovascular Disease Epidemiology Training Program supported by National Institutes of Health, National Heart, Lung, and Blood Institute NRSA grant 5T32HL07055. The authors thank the staff at the ARIC centers: University of North Carolina at Chapel Hill—Phyllis Johnson, Marilyn Knowles, Catherine Paton; University of North Carolina, Forsyth County—Kay Burke, Wilhelmenia Cheeks, Revitha Cook, Shirley Cothern; University of Mississippi Medical Center, Jackson—Betty Warren, Dorothy Washnigton, Mattye Watson, Nancy Wilson; University of Minnesota, Minneapolis—Greg Feitl, Chris Hunkins, Ellie Justiniano, Laura Kemmis; Johns Hopkins University, Baltimore—Joyce Chabot, Carol Christman, Dorrie Costa, Patricia Crowley; University of Texas Medical School, Houston—Valarie Stinson, Pam Pfile, Hogan Pham, Teri Trevino; The Methodist Hospital, Aterosclerosis Clincial Laboratory, Houston— Wanda R. Alexander, Doris J. Harper, Charles E. Rhodes, Selma M. Soyal; Bowman-Gray School of Medicine, Ultrasound Reading Center, WinstonSalem—Anne Safrit, Melanie Wilder, Linda Allred, Carolyn Bell; University of North Carolina at Chapel Hill, Coordinating Center—Peter DeSaix, La Sonya Goode, Thomas Goodwin, Steve Hutton.
REFERENCES 1. Bureau of the Census. Statistical Abstracts of the US: 1990. 110th ed. Washington, DC: Government Printing Office; 1991. 2. Baruch GK, Biener L, Barnett RC. Women and gender in research on work and family stress. Am Psychol. 1987;42:130–136. 3. Friedman M, Rosenman RH. Type A Behavior and Your Heart. New York: Knopf; 1974. 4. Eaker ED, Packard B, Thom TJ. Epidemiology and risk factors for coronary heart disease in women. Cardiovasc Clin. 1989;19:129–145. 5. Feldman JJ, Makuc DM, Kleinman JC, Cornoni-Huntly J, et al. National trends in educational differentials in mortality. Am J Epidemiol. 1989;129:919–933. 6. Haynes SG, Feinleib M. Women, work, and coronary heart disease: Prospective findings from the Framingham Heart Study. Am J Public Health. 1980;70:133–141. 7. Pasanante MR, Nathanson CA. Female labor force participation and female mortality in Wisconsin, 1974–1978. Soc Sci Med. 1985; 21:655–668. 8. Eaker ED, Pinsky J, Castelli WP. Myocardial infarction and coronary death among women: Psychosocial predictors from a 20-year followup of women in the Framingham study. Am J Epidemiol. 1992; 135:854–864. 9. Kotler P, Wingard DL. The effect of occupational, marital and parental roles on mortality: The Alameda County Study. Am J Public Health. 1989;79:607–612. 10. Hibbard JH, Pope CR. Effect of domestic and occupational roles on morbidity and mortality. Soc Sci Med. 1991;32:805–811. 11. Stamler J, Neaton JD, Wentworth DN. Blood pressure (systolic and diastolic) and risk of fatal coronary heart disease. Hypertension. 1989;13(suppl):I2–I12. 12. Fiebach NH, Herbert PR, Stampfer MJ, Coldit GA, Wille HWC, Rosner B, et al. A prospective study of high blood pressure and cardiovascular disease in women. Am J Epidemiol. 1989;130:646–654. 13. Rutan GH, Kuller LH, Neaton JD, Wentworth DN, McDonald RH, Smith WM, et al. Mortality associated with diastolic hypertension
382
Rose et al. WOMEN, EMPLOYMENT STATUS, AND HYPERTENSION
AEP Vol. 9, No. 6 August 1999: 374–382
and isolated systolic hypertension among men screened for multiple risk factor intervention trial. Circulation. 1988;77:504–514. 14. Drizd T, Dannenberg AL, Engel A. Blood Pressure Levels in Persons 18–74 Years of Age in 1976–80, and Trends in Blood Pressure from 1960 to 1980 in the United States. (Vital and health statistics, Series 11: Data from the National Health and Examination Survey, the National Health, Nutritional and Examination Survey I and National Health, Nutrition and Examination Survey II, no. 234) (DHHS publication no. (PHS) 86–1684). Hyattsville, MD: National Center for Health Statistics; 1986. 15. Ostfeld A, D’Atri D. Rapid sociocultural change and blood pressure. Adv Pyschosom Med. 1979;9:20. 16. James GD, Jenner DA, Harrison GA, Baker P. Differences in catecholamine excretion rates, blood pressure, and lifestyles among young western Somoan men. Human Biol. 1985;57:635–47.
26.
27.
28.
29.
17. Pickering TG, James GD, Schnall PL, Schlussel YR, Pieper CF, Gerin W, et al. Occupational stress and blood pressure. In: Frankenhaeuser M, Lundberg U, Chesney M, eds. Women, Work, and Health: Stress and Opportunites. New York: Plenum; 1992:171–186.
30.
18. Rose KM, Newman B, Bennett T, Tyroler HA. Employment status and high blood pressure in women: Variations by time and by sociodemographic characteristics. Ann Epidemiol. 1997;7:107–114.
31. 32.
19. Kritz-Silverstein D, Wingard DL, Barrett-Conner E. Employment status and heart disease risk factors in middle-aged women: The Rancho Bernardo Study. Am J Public Health. 1992;82:215–219. 20. Hazuda HP, Haffner SM, Stern MP, Knapp JA, Eifler CW, Rosenthal M, et al. Employment status and women’s protection against coronary heart disease. Am J Epidemiol. 1986;123:623–640. 21. Amott T, Mathaei TL. The growth of wage work. In: Amott T, Mathaei TL, eds. Race, Gender and Work: A Multicultural Economic History of Women in the United States. Boston: South End Press; 1991:291–314. 22. Amott T, Mathaei TL. We specialize in the wholly impossible. In: Amott T, Mathaei TL, eds. Race, Gender and Work: A Multicultural Economic History of Women in the United States. Boston: South End Press; 1991:141–192. 23. Deckard BS, ed. The Women’s Movement: Political, Socioeconomic, and Psychological Issues. New York: Harper and Row; 1975. 24. Bureau of the Census. Current Population Reports. Series P-60. Washington, DC: Government Printing Office; 1980. 25. Jackson R, Chambless LE, Yang K, Byrne T, Watson R, Folsom A, et al. Differences between respondents and nonrespondents in a multi-
33. 34. 35.
36.
37. 38. 39.
center community-based study vary by gender and ethnicity. J Clin Epidemiol. 1996;12:1373–1379. The ARIC Investigators. The Atherosclerosis Risk in Communities (ARIC) study: Design and objectives. Am J Epidemiol. 1989;129: 687–702. U.S. Bureau of Census: 1980 Census of Population. Classified Index of Industries and Occupations. Final Ed. Washington, DC: U.S. Government Printing Office; 1982. Baecke JAH, Burema J, Frijters JER. A short questionnaire for the measurement of habitual physical activity in epidemiological studies. Am J Clin Nutr. 1982;36:936–942. Rose KM. Women, work and blood pressure: Variations in the employment status hypertension relationship across time, and by age, race and education. (Doctoral Dissertation). University of North Carolina at Chapel Hill; 1994. Amott T, Mathaei TL. Whatever you fight, don’t be ladylike. In: Amott T, Mathaei TL, eds. Race, Gender and Work: A Multicultural Economic History of Women in The United States. Boston: South End Press; 1991:1495–141. Arber S, Gilbert G, Dale A. Paid employment and women’s health: A benefit or a source of role strain? Soc Health Illness. 1985;7:375–400. Liu K, Cedres MS, Stamler J, Dyer A, Stamler R, Dyer A, Stamler R, Nanas S, et al. Relationship of education to major risk factors and death from coronary heart disease, cardiovascular diseases and all causes: Findings of three Chigago epidemiologic studies. Circulation. 1982;66:1308–1314. Rose G, Marmot MG. Social class and coronary heart disease. Br Heart J. 1981;45:13–19. Sorensen G, Verbrugge LM. Women work and health. Am Rev Public Health. 1987;8:235–251. Ries P, Stone AJ. Employment. In: Ries P, Stone AJ, eds. The American Women 1992–1993: A Status Report. New York: Norton and Company; 1992:306–349. Ries P, Stone AJ. Family and Household Structure. In: Ries P, Stone AJ, eds. The American Women 1992–1993: A Status Report. New York: Norton and Company; 1992:244–262. Libaratos P, Link BG, Kelsey J. The measurement of social class in epidemiology. Epidemol Rev. 1988;10:87–121. Wright EO. Women in the class structure. Polit Soc. 1989; 17:1–34. Sterling TD. The healthy worker effect on morbidity rates. J Occup Med. 1985;27:477–482.