Ethnic Variation in Hypertension Among Premenopausal and Perimenopausal Women
Study of Women’s Health Across the Nation
Data are sparse regarding hypertension prevalence, treatment, and control among some ethnic groups of American women. Furthermore, the effects of ethnicity on hypertension, independent of other factors that vary with ethnicity, are poorly understood. We examined the prevalence of hypertension (defined as systolic ≥140 or diastolic ≥90 mm Hg or receiving treatment), treatment, and control (to <140/<90 mm Hg) in a multiethnic study of premenopausal and perimenopausal women. Stepwise multivariable logistic regression was used to select covariates associated with hypertension. Among 3292 women, 46.9% were white, 28.3% were black, 8.7% were Hispanic, 7.6% were Chinese, and 8.5% were Japanese. Among these 5 ethnic groups, respectively, there was substantial variation in prevalence of normal blood pressure levels (<120/<80 mm Hg; 59.9%, 35.4%, 16.8%, 67.2%, and 63.7%) and hypertension (14.5%, 38.1%, 27.6%, 12.8%, and 11.0%). After multivariable adjustment, hypertension prevalence was 2 to 3× higher among black and Hispanic women but similar among Chinese and Japanese women compared with white women. Among hypertensive participants, prevalence of antihypertensive treatment was highest among blacks (58.9%) and whites (55.2%) and lowest among Chinese (34.4%). Prevalence of control to goal blood pressure levels was highest among whites (43.0%) and Japanese (38.7%) and markedly lower among Hispanic women (11.4%). Compared with whites, black and Hispanic women have significantly higher prevalence of hypertension independent of other factors, whereas Chinese and Japanese women have similar prevalence. Treatment and control rates vary considerably across ethnicities. Greater efforts must be made to improve hypertension awareness, treatment, and control in all middle-aged women, particularly those in ethnic minority groups.
Ethnic differences in blood pressure levels and prevalence of hypertension have been described within certain segments of the United States population. Higher rates of hypertension among blacks compared with non-Hispanic whites are noteworthy and contribute to their increased risks of cardiovascular disease (CVD).1–6 However, comparative data are sparse regarding contributing factors that underlie hypertension in these and other ethnic populations, particularly among women at younger and middle ages. Data are especially sparse regarding the relative prevalence of hypertension and its associated factors among Chinese-American and Japanese-American women.7 Furthermore, the prevalences of treatment and control of blood pressure in women of different ethnicities are poorly characterized.
The Study of Women’s Health Across the Nation (SWAN)8,9 is a multicenter study of a multiethnic cohort of women designed to examine the menopausal transition and its effects on physical, metabolic, and psychosocial factors. This study provides a unique opportunity to examine the prevalence of hypertension, associated factors, and treatment and control among premenopausal and perimenopausal white, black, Hispanic, Chinese, and Japanese women in the United States. These data could help target public health efforts aimed at prevention of hypertension and CVD in specific populations of women of differing ethnic backgrounds.
SWAN is a multicenter, longitudinal cohort study being conducted in community-based groups of women. At baseline, 3302 women who belonged to 1 of 5 self-identified ethnic/racial groups were recruited: non-Hispanic whites (n=1550), blacks (n=935), Japanese (n=281), Chinese (n=250), and Hispanics (n=286). Eligibility criteria for entry into the SWAN longitudinal cohort were: 42 to 52 years of age; intact uterus and at least 1 ovary; no current use of estrogens or other medications known to affect ovarian function; at least 1 menstrual period in the 3 months before screening; and self-identification as a member of 1 of the 5 eligible ethnic groups. Cohort recruitment, enrollment, and study procedures have been described in detail previously.8,9 Recruitment techniques were designed to generate a representative sample of women at each of the 7 study sites. All 7 sites enrolled non-Hispanic white women, and each site also enrolled women belonging to 1 prespecified minority ethnic group. Black women were enrolled in Boston, Chicago, the Detroit area, and Pittsburgh, whereas Japanese, Chinese, and Hispanic women were enrolled in Los Angeles, Oakland, Calif, and Hudson County, NJ, respectively. Study procedures were approved by the institutional review board at each site, and all participants provided informed consent.
For the present cross-sectional analysis, we included all 3302 women who attended the baseline examination in 1996 to 1997. Ten women were excluded because of incomplete blood pressure data. At baseline evaluation, blood pressure was measured 3× in the right arm by an examiner using a mercury column sphygmomanometer after the subject had been at rest in the seated position for ≥5 minutes. Systolic blood pressure (SBP) level was defined as the first appearance of sound (Korotkoff phase 1), and diastolic blood pressure (DBP) level was defined as the disappearance of sound (phase 5). Measurements were separated by ≥2 minutes. The means of the second and third SBP and DBP measurements were used to derive the reported blood pressure for the examination.
Hypertension was defined as blood pressure ≥140 mm Hg SBP or ≥90 mm Hg DBP or receiving antihypertensive therapy, in accordance with the definitions used in the sixth and seventh reports of the Joint National Committee on Prevention, Detection, Treatment and Control of High Blood Pressure (JNC 610 and JNC 711) and the National Health and Nutrition Examination Surveys (NHANES).1,2 Antihypertensive medication use was obtained via self-report and verified by medication review at the examination. Demographic, anthropometric, and serological data were all obtained at the same examination as blood pressure as described previously.8,9
Initial descriptive analyses were designed to compare the prevalence of JNC 7 blood pressure stages and hypertension as well as rates of treatment and control between ethnicities. Next, we compared differences between nonhypertensive and hypertensive women at baseline, using χ 2 tests for categorical variables and t tests and Wilcoxon rank-sum tests for continuous variables, as appropriate. Using stepwise logistic regression with hypertensive status as the dependent variable, we next determined the covariates that were significantly associated with hypertension in multivariable analysis, with a P value <0.05 for entry and retention in the model. In all analyses, age, ethnicity, and examination site were forced into the models. All other covariates with P<0.05 in univariate analysis were eligible for selection in the stepwise model. We next examined the interactions between ethnicity and other covariates in association with hypertensive status to determine whether there were ethnic-specific patterns of covariates associated with hypertension. Finally, because weight varies greatly by ethnicity and is known to be a major determinant of blood pressure and hypertension, we performed a secondary analysis to examine the joint effects of body mass index (BMI) and waist/hip ratio (WHR), the 2 weight variables selected in our multivariable model, after adjustment for age, examination site, and ethnicity. For this analysis, participants were stratified by clinical levels of BMI (<25, 25 to 29.9, and 30 kg/m2) and by tertiles of WHR. The referent group was women who were in the lowest BMI stratum and the lowest tertile of WHR.
There were 3292 women in the study sample. Overall, 720 (21.9%) were classified as hypertensive at baseline. Characteristics of the study sample are shown in Table 1, separately for nonhypertensive and hypertensive participants. As expected, in unadjusted analyses, hypertensive participants were older and heavier, and they had more adverse lipid profiles, including higher total and LDL cholesterol and triglycerides, lower HDL cholesterol, and higher lipoprotein(a). Hypertensives were also more likely to be diabetic, and they were less physically active than normotensives.
After adjustment for other covariates in multivariable analysis (Table 2), black and Hispanic women were significantly more likely to have hypertension, with 2- to 3-fold odds of being hypertensive, compared with non-Hispanic white women. Chinese and Japanese women were slightly but not significantly more likely to be hypertensive after multivariable adjustment. There was 1 significant first-order interaction between ethnicity and triglyceride levels. However, inclusion of the interaction term did not materially alter the findings.
As expected, a number of other covariates were independently associated with hypertension, including variables for each of the other components of the metabolic syndrome (abdominal obesity, triglycerides, HDL cholesterol, and fasting glucose levels). Of note, even after controlling for age, ethnicity, and other factors, BMI and WHR entered the multivariable model predicting hypertensive status, independent of each other. In secondary analyses, we examined ethnic-specific models using the same covariates as shown in Table 2. Compared with the results shown in Table 2, the findings were remarkably similar in each ethnic group, with few exceptions. Among black women, current smoking was especially strongly associated with hypertension (odds ratio [OR], 1.81; 95% confidence interval [CI], 1.24 to 2.63). Among Chinese women, BMI was particularly associated with hypertension (OR, 1.20; 95% CI, 1.05 to 1.38), whereas among Japanese women, WHR was notable (OR, 1.17; 95% CI, 1.05 to 1.31).
It should be noted that the majority of the Japanese and Chinese women fall in the lower tertile of the BMI distribution, whereas the majority of the Hispanic and black women fall into the upper tertile of the BMI distribution. When we restricted the analysis to participants in the middle tertile of BMI only, in which there was greater overlap in BMI between ethnic groups, the results were essentially unchanged, with black (OR, 3.26; 95% CI, 2.08 to 5.12) and Hispanic (OR, 2.11; 95% CI, 1.17 to 3.81) women having significantly higher likelihood of hypertension.
To further explore the effects of BMI and WHR on hypertensive status, we examined their joint effect on hypertensive status in a multivariable model (Figure 1). Within each stratum of BMI, the odds of being hypertensive increased 2- to 3-fold with increasing tertile of WHR. Similarly, for any given WHR, the odds of being hypertensive increased 2- to 3-fold with increasing BMI stratum. For participants who were obese (BMI >30 kg/m2) and had the highest WHR, the adjusted odds of being hypertensive were almost 7× higher than for lean women (BMI <25 kg/m2) with the lowest WHR. However, the P value for the interaction term between BMI and WHR did not achieve statistical significance.
In secondary analyses, we examined whether dietary variables affected the associations between ethnicity and hypertension. Several dietary factors were associated with hypertensive status in univariate analyses, including dietary sodium, fat, saturated fat, and oleic acid intakes. When these covariates were added to the multivariable model shown in Table 2, only dietary sodium remained significantly associated with hypertension, and the ORs for the different ethnic groups were not altered materially.
Ethnicity and Hypertension Prevalence, Treatment, and Control
Compared with all non-Hispanic white women (mean blood pressure 114/74 mm Hg), unadjusted mean blood pressure levels were higher among black (126/77 mm Hg; P<0.001) and Hispanic (123/82 mm Hg; P<0.001) women, similar among Chinese women (113/73 mm Hg; P=NS), and lower among Japanese women (111/75 mm Hg; P<0.01). The prevalence of JNC 7 blood pressure stages, antihypertensive treatment, and control to goal blood pressure level of <140/<90 mm Hg are shown in Figure 2, according to ethnicity. Whereas approximately two thirds of white, Chinese, and Japanese women had normal blood pressure levels (<120/<80 mm Hg), only 1 in 3 black women and 1 in 6 Hispanic women had normal blood pressure levels. More than half of Hispanic women were classified as prehypertensive, compared with approximately one fourth of the other ethnic groups. In pairwise analyses comparing hypertension prevalence with non-Hispanic white women (14.5%), black (38.1%) and Hispanic (27.6%) women had significantly higher prevalences of hypertension (both P<0.001), whereas Chinese (12.8%) and Japanese (11.0%) women had similar prevalences (Figure 2).
Among the 720 hypertensive participants, 53.9% were receiving treatment, but only 32.2% were controlled to goal levels. We observed significant ethnic differences in the prevalence of treatment and control (Table 3). More than 55% of white and black women were receiving antihypertensive drug therapy. Japanese women had a slightly but not significantly lower prevalence of treatment, whereas Hispanic and Chinese women had significantly lower treatment prevalences. The prevalence of treatment to goal blood pressure levels (<140/<90 mm Hg) was highest among whites and Japanese, significantly lower in blacks and Chinese, and dramatically lower in Hispanic women (Table 3).
In the present analysis from a multiethnic national sample of premenopausal and perimenopausal women, we observed that even after adjustment for other covariates associated with ethnicity and hypertension, Hispanic women were twice as likely, and black women nearly 3× as likely to have hypertension as white women. In contrast, Chinese and Japanese women had lower crude prevalences of hypertension; after multivariable adjustment, they had a slightly but not significantly higher likelihood of being hypertensive compared with white women. We also observed an important joint effect of body mass and body shape on the risk of being hypertensive that was independent of age and ethnicity. Adjustment for dietary factors did not attenuate the odds of being hypertensive substantially for any ethnic group.
There were marked ethnic disparities in hypertension treatment and control. Treatment rates were suboptimal in all ethnic groups. Nonetheless, the majority of treated hypertensive white (78%), Chinese (73%), and Japanese (80%) women were controlled to goal blood pressure levels. Hypertensive black women had the highest prevalence of treatment (59%) of any ethnic group, but only about half of treated hypertensives were controlled. Most strikingly, hypertensive Hispanic women had substantially lower rates of treatment (38%), and only 30% of treated hypertensives were controlled to goal levels, indicating that only 11% of all hypertensive Hispanic women were controlled to goal levels.
Public Health Implications
On the basis of these findings, we suggest that urgent attention should be paid to public health and clinical efforts aimed at increasing awareness of the importance of hypertension prevention, treatment, and control among all middle-aged women, but particularly among black and Hispanic women. Our data suggest that one of the major barriers to achieving blood pressure control to goal levels is suboptimal rates of treatment among women of all ethnicities, with a contribution of insufficient treatment particularly seen among Hispanic women. Access to care and insurance issues may also be important contributors to low treatment rates.12
Data from large antihypertensive treatment trials indicate that most hypertensive individuals require ≥2 medications to achieve blood pressure reduction to goal levels of <140/<90 mm Hg. Nonetheless, hypertension control rates of >60% can be achieved with use of low-cost medications.13 Greater efforts need to be devoted to enhancing physician and patient awareness of the benefits of blood pressure lowering in reducing morbidity and mortality and ensuring adherence with national guideline recommendations.
Another factor that may contribute to the suboptimal rates of hypertension treatment is poor awareness among women with regard to the risks of CVD and the importance of cardiovascular risk factors. In a 2003 survey of a nationally representative sample of women,14 fewer than half (46%) correctly identified CVD as the leading cause of death among women in the United States, with substantial differences in knowledge between white (55%), black (30%), and Hispanic (27%) women. Further, only 13% of women identified CVD as their greatest personal health risk, whereas 51% identified cancer as their greatest health risk. Again, black and Hispanic women in particular underestimated their personal CVD risk.14
The high prevalence of prehypertensive blood pressure levels, particularly among Hispanic women, has potentially important implications for future development of hypertension and cardiovascular risk. Data suggest that individuals with these blood pressure levels are much more likely to progress to frank hypertension.15 Furthermore, the risks associated with prehypertensive blood pressure levels are not trivial, with approximately a doubling of risk for CVD events compared with normal blood pressure levels.16,17
The current analysis also extends our understanding of cross-sectional covariates associated with hypertension in a multiethnic sample of women at midlife. It is notable that all of the other components of the metabolic syndrome were independently associated with hypertension, indicating the role of this syndrome as a determinant of hypertension and cardiovascular risk in women. An unexpected result was the significant association between lipoprotein(a) level and hypertension even after adjustment for ethnicity, a finding that merits further investigation.
It should be noted that BMI and WHR entered the stepwise multivariable model independently; indeed, BMI was the first independent covariate to enter the model, followed next by WHR, indicating their important association with hypertension. As shown in Figure 2, when we examined BMI and WHR jointly, the joint association with hypertension was remarkably strong and was independent of age and race. Although cross-sectional associations cannot be used to demonstrate causation, our data strongly support the hypotheses that weight gain and development of visceral adiposity in women are associated with greater risk for hypertension.18–22 Furthermore, they support results of studies suggesting that weight loss may be an important public health and clinical measure for hypertension prevention and control in younger and middle-aged women.23
Current Study in Context
Our hypertension prevalence data are similar to those from nationally representative samples for white women. For example, in the NHANES III phase 1 sample (1988 to 1991), Burt et al1 used the same definition as the present study to describe prevalences of hypertension in several ethnic groups. Among 40- to 49-year-old women in their sample, hypertension prevalence was 11.3% among non-Hispanic whites, 33.2% among non-Hispanic blacks, and 10% among Mexican-Americans. In comparison, prevalence rates in the current sample were 14.5%, 38.1%, and 27.6% (for Hispanic-Americans), respectively. The higher rates observed in the present study may reflect different sampling techniques, slightly higher age of the SWAN cohort (42 to 52 years), or perhaps the higher rates of overweight and obesity in the SWAN cohort compared with NHANES III phase 1. The SWAN baseline data used for this report were obtained in 1996 to 1997, by which time the current epidemic of overweight and obesity had advanced significantly compared with 1988 to 1991.24 The further increase in obesity prevalence is likely to have worsened the data on hypertension.
In more recent data from NHANES 1999 to 2000,2 overall rates of awareness and treatment were higher, but rates of control were lower in women than in men. Notably, Mexican-American women had substantially lower rates of awareness, treatment, and control of hypertension than non-Hispanic white and non-Hispanic black women,2 consistent with our findings. The present data from SWAN extend previously available data pertaining to the prevalence of hypertension among Chinese and Japanese women in the United States. Whereas mean blood pressure levels among indigenous Asian (particularly Chinese) women tend to be higher than among white American women,25 we observed that unadjusted hypertension prevalence was not higher among these 2 ethnic groups in the United States compared with whites. Adjustment for other covariates did result in higher ORs for hypertension in Chinese-American and Japanese-American women compared with whites, although these were not statistically significant. The mechanisms underlying low rates of antihypertensive treatment among Chinese women merit further study. Cultural issues regarding Western medications and lack of physician awareness of the problem of hypertension in this ethnic group may be contributing factors.
We had limited power to detect modest differences in hypertension prevalence, treatment, and control, especially among Chinese and Japanese women. The difficulties of collecting a large cohort and following participants with intensive annual examinations precluded selection of the SWAN cohort as a national probability sample of women at midlife. As described in the Methods section, the SWAN design included sampling of only 1 ethnic minority at each study site, whereas all 7 study sites enrolled non-Hispanic white women. Given the regional specificity of ethnic group enrollment, it is possible that the women at a given site may not be representative of their ethnic group as a whole. However, as per SWAN study protocol, we attempted to account for regional variation by adjusting for study site in all analyses, providing standardization across the white and black groups. It is also possible that there was some bias in recruiting between ethnic groups in SWAN; but potential participants were identified by population lists and random-digit dialing, and participation rates were similar for all ethnicities except Hispanics.8 Nonetheless, SWAN provides a unique opportunity to study a contemporaneous sample of premenopausal and perimenopausal women, an understudied group in general, from multiple ethnicities in large population centers in the United States.
University of Michigan, Ann Arbor: MaryFran Sowers, principal investigator (PI; U01 NR04061); Massachusetts General Hospital, Boston: Robert Neer, PI, 1995 to 1999, Joel Finkelstein, PI, 1999 to present (U01 AG012531); Rush University, Rush-Presbyterian-St Luke’s Medical Center, Chicago, Ill: Lynda Powell, PI (U01 AG012505); University of California, Davis/Kaiser: Ellen Gold, PI (U01 AG012554); University of California, Los Angeles: Gail Greendale, PI (U01 AG012539); University of Medicine and Dentistry-New Jersey Medical School, Newark: Gerson Weiss, PI (U01 AG012535); and the University of Pittsburgh, Pennsylvania: Karen Matthews, PI (U01 AG012546).
National Institutes of Health Program Office
National Institute on Aging, Bethesda, Md: Sherry Sherman, 1994 to present, Marcia Ory, 1994 to 2001; National Institute of Nursing Research, Bethesda, Md: Janice Phillips, 2002 to present, Carole Hudgings, 1997 to 2002.
University of Michigan, Ann Arbor: Rees Midgley, PI, 1995 to 2000, Daniel McConnell, 2000 to present (U01 AG012495; Central Ligand Assay Satellite Services).
New England Research Institutes, Watertown, Mass: Sonja McKinlay, PI (U01 AG012553), 1995 to 2001; and/or the University of Pittsburgh, Pennsylvania: Kim Sutton-Tyrrell, PI (U01 AG012546), 2001 to present.
Chris Gallagher, chair 1995 to 1997; Jenny Kelsey, chair 1997 to 2002; Susan Johnson, chair 2002 to present.
The Study of Women’s Health Across the Nation (SWAN) is funded by the National Institute on Aging, the National Institute of Nursing Research, and the National Institutes of Health Office of Research on Women’s Health. We thank the study staff at each site and all the women who participated in SWAN.
- Received March 10, 2005.
- Revision received April 6, 2005.
- Accepted August 4, 2005.
Burt VL, Whelton P, Roccella EJ, Brown C, Cutler JA, Higgins M, Horan MJ, Labarthe D. Prevalence of hypertension in the US adult population: results from the Third National Health and Nutrition Examination Survey, 1988–1991. Hypertension. 1995; 25: 305–313.
Heart Disease and Stroke Statistics-2005 Update. Dallas, Tex: American Heart Association; 2004.
Sowers MF, Crawford SL, Sternfeld B, Morganstein D, Gold EB, Greendale GA, Evans D, Neer R, Matthews KA, Sherman S, Lo A, Weiss G, Kelsey J. SWAN: a multicenter, multiethnic, community-based cohort study of women and the menopausal transition. In: Lobo RA, Kelsey J, Marcus R, eds. Menopause: Biology and Pathobiology. New York, NY: Academic Press; 2000: 175–188.
Chobanian AV, Bakris GL, Black HR, Cushman WC, Green LA, Izzo JL Jr, Jones DW, Materson BJ, Oparil S, Wright JT Jr, Roccella EJ. Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure. Hypertension. 2003; 42: 1206–1252.
Mosca L, Ferris A, Fabunmi R, Robertson RM. Tracking women’s awareness of heart disease: an American Heart Association national study. Circulation. 2004; 109: 573–579.
White FM, Pereira LH, Garner JB. Associations of body mass index and waist:hip ratio with hypertension. CMAJ. 1986; 135: 313–320.