Popul Res Policy Rev (2011) 30:885–911 DOI 10.1007/s11113-011-9215-4
Social Networks and Support, Gender, and Racial/Ethnic Disparities in Hypertension Among Older Adults Bridget K. Gorman • Jeremy R. Porter
Received: 19 January 2011 / Accepted: 30 August 2011 / Published online: 9 October 2011 Springer Science+Business Media B.V. 2011
Abstract Using data from the 2001 NHIS and the 2005–2006 and 2007–2008 NHANES, we examine how self-reporting a previous diagnosis of hypertension among adults aged 65? differs by race/ethnicity for men and women; we explore the extent to which disparities are driven by group differences in social risk factors, particularly social support and integration; and last, whether these relationships mimic patterns seen for measured hypertension at interview. Findings indicate that rates of ever-diagnosed hypertension in both samples are highest among black seniors and older women and lowest among Mexican-American men, with the gender gap lowest among whites and substantially higher among blacks and Mexican-Americans. However, replication analyses of NHANES models using measured hypertension, instead of a self-report of having ever been diagnosed with hypertension, suggests that reporting bias and measurement error contribute to observed disparities, as racial/ethnic differences in hypertension rates are smaller when measured hypertension is examined, especially among women. Logistic regression models also show that while adjusting for group differences in measures of support and integration mediates some of the disparity in measured hypertension between Mexican-American and white seniors, adjusting for support and integration amplifies black-white disparities in both ever diagnosed and measured hypertension—driven primarily by adjustment for attendance at religious services, which reduces hypertension risk for all older adults but is more commonly reported among black seniors, especially women.
B. K. Gorman (&) Department of Sociology MS28, Rice University, 6100 Main St., Houston, TX 77005-1892, USA e-mail:
[email protected] J. R. Porter City University of New York, New York, NY, USA
123
886
B. K. Gorman, J. R. Porter
Keywords Hypertension Race/ethnicity High blood pressure Gender Social support Social integration
Introduction The burden of hypertension in the U.S. is high, affecting an estimated 65 million adults in 2005–2006 (Egan et al. 2010). Rates are also increasing because of a growing, aging population and rising obesity rates (Fields et al. 2004). While burdensome for many Americans, hypertension has been shown to occur more often in selected race/ethnic groups, especially African Americans. Compared to whites, their hypertension prevalence rate is almost 50% higher (Glover et al. 2005). Reducing disparities in hypertension requires, in part, an examination of the pathways that bind the two together, particularly for seniors since hypertension rates are quite high in later life (National Center for Health Statistics (NCHS) 2007). Yet, we know a lot less about racial/ethnic disparities in elder health than we do about problems among children and younger adults (Anderson et al. 2004), and few studies have considered how the relationship between race/ethnicity and hypertension differs between men and women, oftentimes because the data will not support gender-stratified analyses. Using data from two sources (the 2001 wave of the National Health Interview Survey (NHIS), and the 2005–2006 and 2007–2008 waves of the National Health and Nutrition Examination Survey (NHANES)), we examine how hypertension prevalence among older adults differs by race/ethnicity, separately for men and women. This analysis is guided by three questions. First, to what extent are disparities in self-reporting a previous diagnosis of hypertension driven by group differences in social risk factors, particularly social support and integration? While we examine several categories of risk, social support and integration are of particular interest due to their health-enhancing qualities and importance for successful aging (Berkman and Glass 2000), and since both differ by gender and race/ethnicity among seniors (Cornwell et al. 2008; Krause and Keith 1989; Mendes de Leon and Glass 2004). Second, does the relationship between support/integration and self-reporting a previous diagnosis of hypertension vary for white, black, and Mexican-American men and women? And, third, how sensitive are these relationships to how hypertension status is measured—specifically, do patterns established for self-reporting a previous diagnosis of hypertension mimic patterns seen if measured hypertension is considered instead?
Race, Ethnicity, and Hypertension Hypertension rates vary substantially across racial and ethnic groups, and appear particularly elevated among blacks (e.g., Crimmins et al. 2004). Hypertension among blacks also tends to start earlier and be more severe in that physical damage occurs more frequently at lower levels of blood pressure (Stewart et al. 2006). From 1999 to 2004, the hypertension rate among seniors aged 60? was 82% for black adults, compared to 66% for white and 68% for Mexican-American adults
123
Hypertension Among Older Adults
887
(Ostchega et al. 2007). Given their generally poor socioeconomic status (SES), the relatively low hypertension rate for Mexican-Americans is reflective of the generally healthy status of this population. Most often this status is attributed to the large proportion who are foreign born and therefore more likely to be selected on health when they migrate to the U.S., and who after arrival typically engage in less health-damaging behaviors (Markides and Eschbach 2005). Our focus on seniors is especially relevant given that racial/ethnic health differences among elders are under-studied, even though these disparities are of great concern given projected increases in the size of the elderly population in the coming years (Cohen 2004). While persistent across the life course, health disparities among elders tend to be of smaller magnitude than in younger age cohorts (Hummer et al. 2004). Hypertension, however, is one condition where disparities across groups increase with age—at least with regards to the black-white gap in hypertension. Geronimus et al. (2007) show that the black-white predicted odds of hypertension increase from 1.7 to 3.1 between the ages of 15 and 65, and this increase is steeper for women than men. They attribute the growing disadvantage of blacks to ‘‘weathering’’, whereby the health of blacks deteriorates at a faster pace than whites because of the repeated and cumulative experience of stress and social/economic adversity. We further consider how racial disparities in hypertension differ by gender. While gender differences in health are due, in part, to biological differences between men and women, social circumstances also influence norms and expectations regarding the behavior of men and women, and these in turn help to maintain and even exacerbate biological health differences (Bird and Rieker 2008). Studies show appreciable gender differences in morbidity, and variability in the size of the gender gap across racial and ethnic groups (Read and Gorman 2006). For hypertension, women (especially black women) report higher rates than men (Glover et al. 2005). Geronimus et al. (2007) show that among white adults aged 55–65 there is a 5-point gap in hypertension prevalence between men and women (44 and 49%), compared to a 12-point gender gap among blacks (62% for black men, and 74% for black women). Potential Mechanisms While typically less studied than other health determinants (e.g., SES), social relationships as an important indicator for health date back to the works of E´mile Durkheim, especially Suicide (Durkheim 1897/1951). Since then, studies of social networks have grown tremendously. In their review, Berkman and Glass (2000) conclude that social networks influence individual health through multiple pathways, two of which we examine: (a) the provision of social support (including emotional and financial support), and (b) social engagement and attachment (which we refer to as social integration, and test with measures of marital status, family size, contact with friends and family, and attendance at church and other group events). Social support, like SES, should be considered a fundamental cause of disease since it is a resource that can be used to avoid the risk of disease or minimize the consequences of illness (Link and Phelan 1995).
123
888
B. K. Gorman, J. R. Porter
Social networks are thought to affect health by influencing health-promoting or health-damaging behaviors, affecting cognitive and emotional states, and operating via physiological pathways related to stress (Berkman and Glass 2000). Social support and integration have been linked to hypertension rates (see review by Uchino et al. 1996; Bell et al. 2010), although studies suggest that aspects of both influence hypertension through psychological well-being (Bosworth et al. 2003; Gorman and Sivaganesan 2007). Yet, findings are mixed on the utility of support and integration for explaining race/ethnic disparities in elder health, since there has been little study of this issue at the national level or with samples that take an in-depth look at health disparities beyond simple black-white contrasts (see review by Mendes de Leon and Glass 2004). Perhaps the most consistent finding is that elder blacks report more religious and group activities than elder whites (e.g., Cornwell et al. 2008). Patterns are more mixed for other measures of social support and integration. For example, while white seniors report lower levels of neighborly socializing than Hispanic seniors, their overall network size is larger than either Hispanic or black seniors (Cornwell et al. 2008). Furthermore, the health benefits of social support appear to be distinct from social integration (Mendes de Leon and Glass 2004). However, since senior women appear consistently advantaged over men in terms of social support and network ties (Cornwell et al. 2008; Krause and Keith 1989), examining these relationships separately for men and women may help with assessments of their utility for shaping racial/ethnic differences in hypertension among older adults. Beyond social support and integration, other mechanisms should be considered as well. Racial groups in the U.S. tend to be strongly stratified along socioeconomic lines (DeNavas-Walt et al. 2007) and studies show a relationship between hypertension and SES (Crimmins et al. 2004). For example, hypertension increases with declining education (Gorman and Sivaganesan 2007) and when persons report financial difficulties (Matthews et al. 2002). Yet, evidence also suggests that the health benefit of high socioeconomic resources may not operate as strongly for blacks and Hispanics as it does for whites (Crimmins et al. 2004; Geronimus et al. 2007). Studies also show that blood pressure drops when adults engage in regular physical activity, when weight is reduced among persons who are overweight or obese, and when alcohol is consumed in moderate amounts (Chobanian et al. 2003). In addition, while whites have less healthy behavior than blacks and/or Hispanics on some measures (e.g., smoking), on others (e.g., physical activity and obesity) the opposite is true (Winkleby and Cubbin 2004). Studies have also shown that health behaviors can account for a portion of racial/ethnic disparities in hypertension (e.g., Read and Gorman 2007). Psychological characteristics may also shape race/ethnic disparities in hypertension, although findings are mixed, at best. Few studies have examined mental health issues among older adults, especially minorities, and the evidence is inconclusive regarding the burden of problems across racial/ethnic groups (Myers and Hwang 2004). However, longitudinal studies provide good evidence that, over time, depression is associated with an increased risk of hypertension (Bosworth et al.
123
Hypertension Among Older Adults
889
2003; Meyer et al. 2004). Additionally, Read and Gorman (2007) demonstrate that depression has a much stronger association with hypertension among black and Hispanic women than among white women. Hypertension Awareness and Measurement Assessments of hypertension among adults are typically based on self-reports, usually because studies are conducted using large, national health datasets, where the actual measurement of blood pressure can be cost-prohibitive. However, given the diminished access to health care (and quality health care) that selected racial minorities experience in the United States (e.g., Smedley et al. 2003), self-reports likely misrepresent the racial dimensions of hypertension status due to a lack of diagnosis, even if adults suffer from the condition. Studies show that Medicare insurance does not cover all the medical costs that seniors incur, and rates of supplemental insurance coverage are much higher among white than black or Hispanic seniors (Angel and Angel 2006). And according to Ostchega et al. (2007), hypertension awareness rates (i.e., those with measured high blood pressure who also self-report that they have been diagnosed with hypertension) are highest among black seniors, especially women (84%), compared to 74% of white and 70% of Mexican-American women. Rates are similar to women for white and Mexican-American men, but are lower for black men (77%). As such, biomarker data, including measured hypertension, provides a better hypertension measurement that is likely to be free of the reporting bias that confounds assessments based strictly on respondent self-report (see discussion by Shih et al. 2010). That being said, the differences associated with responses to ever being diagnosed with hypertension and actually measuring hypertension are not well known. This is particularly true in regards to the relationship of these two different indicators to levels of social support and integration, lifestyle differences, and SES/demographic variations.
Data and Methods Study Population This research draws on data from two sources. First, the 2001 wave of the NHIS is an annual survey conducted by the NCHS and the Centers for Disease Control and Prevention (CDC). NHIS uses a multi-stage, stratified, cluster sample design, and oversamples blacks and Hispanics. When weighted, they are nationally representative of the non-institutionalized U.S. civilian population aged 18 and above. The 2001 wave (n = 33,326) was chosen because a series of additional questions were asked regarding social support and integration. Because of the focus on seniors, we restrict the sample to adults aged 65 and older with valid information on hypertension, and who identify with a racial/ethnic group that is large enough to examine separately: non-Hispanic whites (n = 4,750), non-Hispanic blacks (n = 681) and Mexican or Mexican-American adults (n = 316).
123
890
B. K. Gorman, J. R. Porter
Second, we examined pooled data from the 2005–2006 and 2007–2008 waves of the NHANES, an ongoing, nationally representative survey of the health, functional, and nutritional statuses of non-institutionalized U.S. persons aged 12 and older, also conducted by the NCHS and CDC. NHANES uses a stratified, multistage probability design, and oversamples low-income persons, adolescents, adults over the age of 60, blacks, and Mexican-Americans. These two waves of data were chosen because of the inclusion of measures that permit a comparison with measures available in the 2001 NHIS (e.g., NHANES depression questions, asked for the first time in the 05–06 wave). As with the NHIS data, we limit the NHANES sample to seniors (aged 65?) and only those respondents identified as white (n = 2,414), black (n = 492), or Mexican-American (n = 263).1 Measures We examine two measures of hypertension status. First, both NHIS and NHANES respondents were asked whether they had ever received a hypertension diagnosis, which we refer to as self-reporting a previous diagnosis of hypertension (1 = yes, 0 = no). Second, for NHANES respondents, hypertension was also measured during the interview. Following the CDC’s guidelines for hypertension, an indicator of hypertension was created if the respondent had a measurement of 140 or greater systolic blood pressure or 90 or greater diastolic blood pressure (1 = measured hypertension, 0 = no measured hypertension), based on the average score across up to three blood pressure measurements or the self-reported identification of being on high blood pressure medication (see Geronimus et al. 2007; Glover et al. 2005; Ostchega et al. 2007). In addition to gender and racial/ethnic identity, both samples include detailed measures of social integration and support (see Table 1). For social network size, we include measures of the total number of persons living in the family, along with marital status. NHIS also includes questions (1 = yes, 0 = no) that measure aspects of social integration during the past two weeks; specifically, whether respondents got together socially with, or used the telephone to talk to friends, neighbors or relatives who lived outside their home. They were also asked whether they went to a religious place of worship for services or other activities; and wether they went to a show or movie, sports event, club meeting, class or other group event. For the NHANES, measures of the number of close friends and the frequency of church attendance (1 = never, and 5 = once a week or more) were included.
1
All analyses are run with STATA (StataCorp 2007), using Taylor series-approximate methods with SVY commands to adjust for the complex sample design of the NHIS and NHANES. Furthermore, all appropriate weights were included to make the samples nationally representative. For both samples, missing values on predictor measures were imputed using a regression based on demographic and socioeconomic sample characteristics. For imputed measures in the NHIS sample, the majority had a very low rate of missing values (under 5%). For the family income-to-poverty ratio, however, non-response was higher (34.5% of cases); therefore regression models include a dummy variable that flags missing cases on income (not shown). For imputed measures in the NHANES sample, again the majority of the data had a very low level of missing values (under 5%). The only exception is a slightly higher rate in regards to the social support variables (ranging from 7 to 13% missing).
123
Emotional support
45.1***
3.0*
Widowed
Never married
81.1*
85.1
60.1***
44.1*
Got together w/friends or neighbors
Got together with relatives
Went to a place of worship
Went to a group event
Married or cohabiting
Marital status
43.4**
88.1*
Has someone for emotional support
Has someone for financial support
60.1*
96.3
Ever diagnosed with hypertension
Panel B: NHANES sample
2937
93.2
Phoned relatives
Sample size
88.5
Phoned friends or neighbors
During the last 3 weeks…
1.7 (0.8)**
7.4***
Divorced or separated
Family size
44.4***
Married or cohabiting
Marital status
52.3***
4.3 (1.0)
Ever diagnosed with hypertension
Panel A: NHIS sample
White
a
Women
34.6*
87.1*
99.4*
79.6*
469
36.6
75.6
80.9*
75.5
91.8
89.5**
1.9 (1.3)
6.0
58.8***
13.7
20.8***
4.3 (1.2)
73.0*
Black
b
42.6
73.2**
86.2
55.8
200
30.2***
73.3***
90.3
71.3*
90.1
77.9**
2.0 (1.4)**
9.3*
37.8
13.3*
39.6
4.2 (1.4)
64.1**
Mexican
c
69.7***
82.1
92.3
55.4?
1813
41.7
54.0**
80.2
81.7
88.0
80.4
1.9 (0.6)
3.2
13.6
6.1***
77.0***
4.3 (0.9)
50.2**
Whitea
Men
Table 1 Self-reported hypertension prevalence, social support, and integration, adults aged 65?, NHIS and NHANES
49.8
79.4
92.5
63.6**
212
36.7
63.9
79.4
79.6
86.1
77.9
2.0 (1.0)*
6.0
16.5
15.6
60.6
4.3 (0.4)
63.8**
Blackb
64.5
68.9
82.4
42.4*
116
34.4
65.2
81.0
82.8
83.1
72.7
2.3 (1.3)**
3.2
16.9
9.6
70.3
4.3 (1.3)
43.1
Mexicanc
Hypertension Among Older Adults 891
123
123
3.2 (1.7)* 260
3.8 (1.2)*
3.8 (2.7)
2.6 (1.6)*
13.6*
35.1
17.2
Black
b
a
?
Significance tests relative to same-gender black adults;
b
145
3.6 (1.6)
3.6 (2.4)*
3.2 (1.9)***
7.4*
33.7*
14.7
Mexican
c
830
2.8 (1.7)*
10.1 (11.1)
1.9 (0.8)***
6.2***
15.7
8.7*
Whitea
Men
232
3.2 (1.7)*
6.5 (7.4)
2.7 (1.6)*
18.1
16.3
14.4
Blackb
relative to same-gender Mexican adults; c relative to same-gender white adults
p = .061; * p B .05; ** p B .01; *** p B .001 (two-tailed t-test)
Except for sample size, numbers presented are weighted (standard deviations)
829
Frequency of church attendance
Sample size
8.7 (7.7)***
Number of close friends
3.3***
1.8 (0.9)*
Never married
Family size
10.1*
43.1
Widowed
White
a
Women
Divorced or separated
Table 1 continued
118
2.9 (1.7)
6.8 (7.9)*
3.4 (1.8)***
10.5*
13.2
12.5
Mexicanc
892 B. K. Gorman, J. R. Porter
Hypertension Among Older Adults
893
For social support, both surveys include a measure of emotional support, measured somewhat differently. In NHIS, respondents were asked: ‘‘How often do you get the social and emotional support you need?’’, where 1 = never and 5 = always. In NHANES, respondents were asked two questions concerning whether they have someone they can rely on for emotional or financial support (1 = yes, 0 = no). Demographic control measures include age at interview and nativity status (1 = U.S. born, 0 = foreign born). We assess SES with two measures: (1) the highest level of school completed, and (2) the family’s income-to-poverty ratio, which is each respondent’s family income as a proportion of the income level that defines the federal poverty line. For both surveys, health behavior is captured with four measures, including smoking status and frequency of drinking alcohol. For NHIS, participation in physical activity is based on the averaged response to four questions regarding participation in vigorous activities, light or moderate activities, as well as musclestrengthening and stretching activities (1 = never, to 5 = 5? times per week; a = .73). For NHANES, exercise frequency is measured as the number of times the respondent walked or biked in the past week. Last, Body Mass Index (CDC 2007) is grouped into three categories: overweight (25.0–29.9), obese (30.0?), and neither overweight nor obese (\25.0), which hereafter we refer to as ‘‘normal weight.’’ Finally, we adjust for a measure of mental health. In NHIS, psychological distress is the averaged response to six questions that asked, during the last 30 days, how often the respondent felt sad, hopeless, restless, nervous, worthless, and that everything was an effort (a = .86), where 1 = none of the time and 5 = all of the time (see Kessler 2002). In NHANES, mental health was measured similarly through the use of an average response gauging interest in doing things, feeling down, trouble sleeping, having little energy, poor appetite, feeling bad about self, trouble concentrating, and thoughts of being better off dead (a = .73), where 1 = not at all and 4 = nearly every day.
Results Sample Characteristics Table 1 presents sample characteristics for self-reporting a previous diagnosis of hypertension, social support, and integration. For women, blacks report having ever been diagnosed with hypertension at a significantly higher rate than white and Mexican-Americans, although black NHANES women report a slightly higher hypertension rate than black NHIS women (79.6% vs. 73.0%, respectively). In both samples, we also see that hypertension rates are lower among white women (52.3% in NHIS and 60.1% in NHANES). However, the ranking of Mexican-American women differs by sample; in NHIS, 64.1% of older Mexican women report hypertension, compared to 55.8% in NHANES. Altogether, these patterns show that black women report the highest absolute rates of hypertension, with the lowest level of variation across the two samples.
123
894
B. K. Gorman, J. R. Porter
For black women, there are several potential explanations for their elevated rate. Black women report the lowest proportion of marriage/cohabitation of any group, although they report relatively high levels of social integration, including the highest rate of attendance at religious services (although in the NHANES black and Mexican-American women report far fewer close friends when compared to whites). While racial/ethnic differences in control measures are not presented in tabular form, the economic status of black women is significantly worse than that of white women (although not as bad as Mexican-American women). Black women also report the least exercise and the highest obesity rate, although more black than white women have never smoked. For Mexican-American women, their hypertension rate is significantly lower than black women, despite their lower education and income levels. However, relative to black women they also report higher rates of marriage/cohabitation, more exercise, and less smoking and drinking. For emotional support and social integration, differences are more variable (although as shown in Table 1, Mexican-American women report significantly larger families than black women, and report that they socialize with relatives more often, while black women report significantly more attendance at religious services). For senior men, hypertension rates are lower than women in every group. The highest rate occurs among black men, with rates near equivalent across samples (63.8 and 63.6%). Rates for white men are lower, at 50.2% in NHIS and 55.4% in NHANES. However, unlike senior women, Mexican-American men in both samples report the lowest hypertension rates (43.1% in NHIS and 42.4% in NHANES). Most of the racial/ethnic patterns in risk factors are the same for men and women, although there are some notable differences. Compared to Mexican-American women, Mexican-American men report more income (not shown), have higher rates of marriage/cohabitation, and bigger families. Mexican-American men also report fewer differences with white and black men in terms of support and social integration, and they report rates of drinking, obesity, and to some extent smoking that are more similar with white men than with these comparisons among women. Ever-Diagnosed Hypertension: NHIS Sample Table 2 presents odds ratios from logistic regression models predicting everdiagnosed hypertension among older NHIS women. For men and women separately, we begin with our baseline Model 1 that regresses hypertension on race/ethnicity, adjusting only for age and nativity. We then sequentially test the mediating influences of SES (Model 2), social support and integration (Model 3), and health behaviors and psychological distress (Model 4). For senior women, Model 1 shows that relative to whites, black women have over twice the odds (OR = 2.54) of being ever diagnosed with hypertension, compared to OR = 1.64 for Mexican-American women. Model 2 shows that SES explains only a small part of the black-white difference in self-reporting a previous diagnosis of hypertension (OR = 2.16), but these measures do explain the elevated rate among Mexican-American women (OR = 1.30). Model 3 adds support and integration, but this causes a very slight increase in the odds of hypertension for
123
Hypertension Among Older Adults
895
Table 2 Odds ratios: ever received a hypertension diagnosis, NHIS older adults Women (n = 3,606) Model 1
Model 2
Men (n = 2,141) Model 3
Model 4
Model 1
Model 2
Model 3
Model 4
Race/ethnicity (ref: white) Black
2.54***
2.21***
2.22***
2.05***
1.75**
1.54*
1.62*
1.65*
Mexican
1.64**
1.30
1.37
1.30
.73
.63
.67
.71
.89
.92
.94
.91
.84
.87
.88
.84
1.01*
1.02**
1.01
1.01
1.01
1.02*
U.S. born Age
1.03***
1.02***
Education level (ref: \high school) High school graduate
.79*
.81*
.86
.90
.91
.94
Some college
.81
.85
.92
.91
.92
.98
College degree
.55***
.58***
.65**
.86
.86
.96
.98
.98
.99
.97
.97
.97
.99
1.00
1.02 .70
Income-to-poverty ratio Emotional support
.96*
Marital status (ref: Married/cohabiting) Divorced or separated
1.02
.94
.71
Widowed
1.27*
1.22
.75
.74
Never married
1.29
1.30
1.19
1.22
1.10
1.06
.93
.91
1.01
1.02
1.17
1.25
Family size During the last 2 weeks… Phoned friends or neighbors Phoned relatives
1.21
1.25
.99
.88
Got together with friends/neighbors
1.14
1.19
1.03
.99
Got together with relatives
.95
.93
.97
1.01
Went to a place of worship
.78**
.79**
.84
.86
Went to a group event
.96
.98
.95
.95
Smoking status (ref: current smoker) Former smoker
1.47**
1.25
Never smoked
1.23
1.24
Former drinker
1.03
1.44**
Current drinker
.86
1.35*
0.92
1.01
Overweight
1.47***
1.56***
Obese
2.72***
2.58***
Alcohol use (ref: lifetime abstainer)
Frequency of physical activity Body mass index (ref: normal weight)
Psychological distress Pseudo R2
1.18** .02
.03
.03
.06
1.20* .01
.01
.01
.06
* p B .05; ** p B .01; *** p B .001
123
896
B. K. Gorman, J. R. Porter
both black and Mexican-American women. These models also show that the odds of hypertension are significantly higher among widows, and lower among women who have recently attended religious services. Last, adjusting for health behaviors (particularly smoking and BMI) and psychological distress further reduces the odds ratio for black women, but they remain over twice as likely as white women to report hypertension (OR = 2.05). While Table 2 showed little evidence of mediation, it may be that measures of support and integration operate differently in their association with hypertension for white, black, and Mexican-American women. To explore this, we tested a series of interactions between race/ethnicity and each measure of social support and integration, separately for men and women (results not shown); one term was significant and improved model fit: race/ethnicity * family size among women. This interaction showed that family size is positively associated with ever-diagnosed hypertension, but only among Mexican-American women. Among MexicanAmerican women who live alone, 33% are predicted to report hypertension; this rises to 70% of women living with 3? people at interview. For senior men, Model 1 shows that black men are 75% more likely to report hypertension than white men, while Mexican-American men are less likely (but not significantly so) to have been diagnosed with hypertension. Unlike the findings for women, Table 2 shows few significant associations between social risk factors and hypertension among men. Measures of support and integration are unrelated to everdiagnosed hypertension (although adjusting for these measures does result in a modest increase in the odds ratio for black men), and the income-to-needs ratio emerges as significant, and negative, in Model 4 after we adjust for health behaviors and psychological distress. And, while alcohol use, BMI, and psychological distress significantly predict ever-diagnosed hypertension among senior NHIS men, they do not explain any of the remaining black-white gap in hypertension; black men remain 65% more likely to report hypertension than white men. Interaction tests for men between race/ethnicity and each social support and integration measure were not significant. Ever-Diagnosed Hypertension: NHANES Sample Turning first to senior NHANES women, we find that, as with NHIS women, we cannot explain the black-white hypertension gap in self-reporting a previous diagnosis with an adjustment for control measures (Table 3). While the odds ratio for black women decreases very slightly with adjustment for SES, the odds ratio increases in Model 3 after we adjust for measures of support/integration—where, surprisingly, more close friends increases the odds of reporting a hypertension diagnosis. In Model 4, the odds ratio for black women increases a bit more (OR = 2.66) when we adjust for health behaviors (especially exercise and BMI) and psychological distress. These adjustments also cause church attendance and family size to emerge as significant and negative predictors of ever being diagnosed with hypertension, while adults who report that they have someone they can rely on for financial support emerges as significant and positive. For Mexican-American women, we see a similar pattern to NHIS women in that our fully adjusted model
123
.83
.84
1.01
Mexican
U.S. born
Age
1.06 1.16 .97
Has someone for emotional support
Has someone for financial support
Frequency of church attendance
.53* .66
Never smoked
.91*
1.23*
.97
1.06*
.81*
.61
.77
1.19
.99
.53*
.73
.81
1.01
1.15
.68
2.66***
Former smoker
Smoking status (ref: current smoker)
.97 1.06*
Never married
Number of close friends
.62
Widowed
Family size
1.21 1.04
Divorced or separated
Marital status (ref: married/cohabiting)
.94
.69*
.97
.68*
Income-to-poverty ratio
College degree
.74
.98 .83
.83
1.01
1.02
.73
2.58***
Some college
.97
.81
.69*
2.21***
High school graduate
Education level (ref: \high school)
2.27**
Black
Race/ethnicity (ref: white)
Model 4
.98
1.04
.62*
1.42*
.95
1.31
1.12
1.11
.98
.77
.61*
1.49**
Model 2
Model 1
Model 3
Model 1
Model 2
Men (n = 1,180)
Women (n = 1,234)
Table 3 Odds ratios: ever received a hypertension diagnosis, NHANES older adults
.92*
1.09
.89
.98*
1.11*
.88
.86
.94
.96
1.17
1.01
1.03
.99
.96
.56*
1.82**
Model 3
.82
.66
.92*
1.04
.91
.98*
1.16*
1.20
.94
1.37
1.03
.97
.92
.87
.99
.53
.44**
1.61*
Model 4
Hypertension Among Older Adults 897
123
123 Model 4
* p B .05; ** p B .01; *** p B .001
Pseudo R
2
.04
.05
.06
.17
1.68*
2.77***
Obese
Psychological distress
1.72*
Overweight
Body mass index (ref: normal weight)
.96*
.78
Frequency of physical activity
.81
Current drinker
.02
.02
Model 2
Model 1
Model 3
Model 1
Model 2
Men (n = 1,180)
Women (n = 1,234)
Former drinker
Alcohol use (ref: lifetime abstainer)
Table 3 continued
.04
Model 3
.12
.95
3.25***
2.12***
.99
1.23*
1.86*
Model 4
898 B. K. Gorman, J. R. Porter
Hypertension Among Older Adults
899
(Model 4) shows no significant difference in comparison to white women (although the odds ratio is always lower than for white adults). For senior NHANES men, odds ratios for ever being diagnosed with hypertension are elevated among blacks, and as with NHIS men we do not explain this difference with adjustment for controls. As with NHANES women, accounting for support and integration amplifies the white-black hypertension gap somewhat, and in Model 4 we find that black senior men experience 1.61 times the odds of a previous diagnosis of hypertension as white men—nearly identical to NHIS black men in Table 2 (OR = 1.65). For Mexican-American men, we also see a similar pattern to NHIS men in that they report lower diagnosis of hypertension odds than senior white men in all models. However, these differences are somewhat larger in the NHANES sample, and are significantly different from white seniors, with Mexican-American men reporting .44 times the odds of hypertension diagnosis in comparison to white men in Model 4. However, an important difference is that, unlike models for NHIS men, here we see direct effects of support and integration measures on diagnoses in hypertension. In particular, their odds of ever being diagnosed with hypertension decline significantly as church attendance and the number of close friends increases, but increases with family size. Interaction tests between race/ethnicity and social support/integration measures, for NHANES men and women, were not significant. Measured Hypertension: NHANES Seniors Lastly, we explore the sensitivity of hypertension disparities to the method of measurement, replicating analyses for NHANES men and women using hypertension measured at interview (see Table 4). In Panel A, we present prevalence rates by gender and race/ethnicity, and see that measured hypertension rates remain highest among black seniors (72.3% of women and 68.0% of men). For white and MexicanAmerican women (71.5 and 66.0%, respectively), only the latter rate differs significantly from black women. Yet among white and Mexican-American men (62.4 and 48.6%, respectively), both groups have significantly lower prevalence rates than black men. However, unlike findings for self-rated hypertension, measured rates for white and black women do not significantly differ (having dropped from a 19.6 point difference to just a 0.8 point difference), and the gap between black and Mexican-American women, while significant, is more than three times smaller than for self-reporting a previous diagnosis of hypertension (due to a 10.2 point increase in hypertension among Mexican-American women and a corresponding 7.3 point drop among black women). As a result, prevalence rates are now significantly lower among Mexican-American women in comparison to white women (whose measured prevalence rate is 11.4 points higher than for self-reported diagnosis). For men, however, while racial/ethnic disparities are smaller than the odds of ever being diagnosed, the decline is not nearly as great as for women, so measured hypertension among black men is significantly higher than white and Mexican-American men. In Panel B, regression models show that, adjusting for age and nativity status in Model 1, measured hypertension rates show no significant differences among black
123
123
.62*
U.S. born
.98 .96
College graduate?
Income-to-poverty ratio
.98 .99 .91
Number of close friends
Has someone for emotional support
.77
Never married
Family size
.82 .93
Widowed
.94
.57*
.64*
.76
.56*
1.03**
.69
1.87**
Divorced or separated
Marital status (ref: married/cohabiting)
1.49* 1.16
Some college
.54*
1.04***
.63*
1.08
High school graduate
Education (ref: No HS Grad)
1.05***
.63*
Mexican
Age
1.01
Black
Race/ethnicity (ref: white)
.83
.99
.92
.71
.99
.65
.92
.52*
.70
.85
.17*
1.03*
.45*
1.63*
Model 4
.97
.99
.51***
1.03
Model 1
Model 3
Model 1
Model 2
Men (n = 1,180)
62.4*
Whitea 68.0***
Blackb
.98
1.29
1.23
1.41*
.68*
1.00
.55**
1.11
Model 2
Men (n = 1,180)
Women (n = 1,234)
66.0*
Mexican
c
Panel B: odds ratios
72.3**
Black
b
71.5
White
a
Women (n = 1,234)
Measured hypertension
Panel A: Prevalence (%)
Table 4 NHANES older adults, measured hypertension: prevalence rates, and odds ratios from logistic regression models
.86
.95*
1.05
1.06
.65*
.75
.95
.95
.92
1.09
.85
1.01
.69
2.01**
Model 3
.77
.94*
.98
.62
.77
.88
.94
.92
.76
.94
.55
1.01
.49*
2.02**
Model 4
48.6***
Mexicanc
900 B. K. Gorman, J. R. Porter
.06
.21**
.15
Pseudo R2
c
b
a
Significance tests relative to same-gender white adults
Significance tests relative to same-gender Mexican adults
Significance tests relative to same-gender black adults
* p B .05; ** p B .01; *** p B .001
1.13
1.86*
Obese
Psychological distress
1.37
Overweight
Body mass index (ref: normal weight)
.73*
.65
Current drinker
Frequency of physical activity
.49*
Former drinker
Alcohol use (ref: lifetime abstainer)
1.23
Never smoked
.79*
.99
Former smoker
.04
.98
Frequency of church attendance
.03
.96
Smoking status (ref: current smoker)
Model 4
.02
Model 1
Model 3
Model 1
Model 2
Men (n = 1,180)
Women (n = 1,234)
Has someone for financial support
Panel B: odds ratios
Table 4 continued
.02
Model 2
.05
.89**
.94
Model 3
.13
.65*
3.34***
1.64*
1.01
1.43*
.77
.67
.77
.88**
.91
Model 4
Hypertension Among Older Adults 901
123
902
B. K. Gorman, J. R. Porter
men or women when compared to whites, while rates for Mexican-American men and women are significantly lower than whites. While age shows a strong, positive association with measured hypertension among women, and being born in the U.S. is protective against measured hypertension at interview, models among men show no such direct effects. Adjusting for SES in Model 2 has minimal effect on the race/ ethnicity odds ratios, and we see a similar effect of education on hypertension for men and women, with the highest odds of hypertension occurring among men and women who only have a high school diploma. Adjusting for measures of social integration and support in Model 3 does result in two changes in the odds ratios for race/ethnicity, seen among men and women: (1) Odds ratios for black men and women increase substantially in size and now differ significantly from whites (OR = 1.87 for women and 2.01 for men), and (2) odds ratios for Mexican-American men and women no longer differ significantly from whites. While no measure of integration or support shows a direct relationship with hypertension among older women, adjusting for these measures does allow education to emerge as strong, and protective, against measured hypertension. For senior men, on the other hand, Model 3 shows a significantly lower odds of measured hypertension among men who are widowed (compared to those who are currently married or cohabiting), among men with a greater number of close friends, and among men who attend religious services more frequently. Adjusting for health behaviors and mental health in Model 4, however, causes the odds ratios for Mexican-American men and women to decline again, and both groups experience about half the odds of having hypertension at interview in comparison to white adults. For black men, adjusting for these measure shifts their hypertension odds little, and they remain 2.02 times as likely to have hypertension as white men, while the odds ratio for black women declines somewhat in size but remains significant and positive (OR = 1.63). While obesity increases hypertension risk for both older men and women, women alone benefit from not smoking—older women who have never smoked are 79% less likely to have hypertension at interview, compared to women who currently smoke. Men experience no such health benefit. Women also benefit from regular exercise, and from giving up alcohol use, while among men hypertension odds is higher among those who currently drink, and surprisingly, declines significantly with increasing psychological distress. For measures of support and integration, church attendance continues to show a significant, negative association with hypertension in Model 4, for senior men and women. While robust for men, as we saw for self-reporting a previous diagnosis of hypertension, this effect for women emerges as significant only in Model 4. In additional modeling (not shown), once we adjust away the suppressing influence of less exercise among those who attend church frequently (especially for black women), church attendance emerges as significant. Comparing Hypertension Measures To explore these patterns further, in Table 5 we examine group differences in both ever-diagnosed and measured hypertension across gender and racial/ethnic groups. Results show that in the total sample, 57% of respondents reported hypertension as a
123
Hypertension Among Older Adults
903
Table 5 Cross-tabulation (% values), NHANES older adults: measured hypertension by self-reported diagnosis prevalence rates, 2005–2008 Measured hypertension
Self-reported hypertension No
Total sample White female White male Black female Black male Hispanic female Hispanic male
Yes
No
29
3
Yes
11
57
No
25
2
Yes
14
58
No
34
2
Yes
10
53
No
18
9
Yes
3
70
No
30
1
Yes
7
62
No
17
3
Yes
16
53
No
48
3
Yes
10
39
For the total sample and each of the six sub-groups, differences between self-reported and measured hypertension were significant at p \ 0.001, v2
medical condition, and were simultaneously categorized as being hypertensive on the basis of their blood pressure measurements and medication status. Likewise, about 29% of the sample reported not having hypertension and were simultaneously categorized as not being hypertensive through the collection of biomarker data. In terms of instances where ever being diagnosed and measured hypertension do not concur, about 3% of the sample self-reported having ever been diagnosed as hypertensive but then failed to be classified as hypertensive by either their blood pressure reading or medication use. More commonly, almost 11% of respondents indicated that they had never received a diagnosis of hypertension, but were then identified as being hypertensive on the basis of their blood pressure readings at interview. These patterns show that when there is a disjuncture in measured and self-reporting of past diagnoses, it appears that it is almost always in the direction of under-diagnosis, as nearly all self-reported hypertension is confirmed via current blood pressure measurements and medication use. Indeed, when we examine patterns by race/ethnicity and gender, and in terms of ‘‘over-diagnosis’’ (mis-report through error or because they may no longer suffer from hypertension), we see the highest rate among black females (about 9%), compared to only about 1% of black males and about 2% of whites (of either gender). More interesting, however, is the 11% of the total sample that is underdiagnosed, with the highest rates occurring among Hispanic females (about 16%) and White females (about 14%). Least likely to be under-diagnosed are blacks, especially women, with about 3% of women and 7% of men falling into this
123
904
B. K. Gorman, J. R. Porter
category. White and Hispanic males are the intermediate category, reporting lower under-diagnosis rates than white and Hispanic women, but higher rates than blacks. Thus, black seniors have both the highest absolute rates of hypertension, but are also the most aware of the existing condition. In contrast, white and Hispanic females are the most likely to have hypertension at interview (per blood pressure measurement) and to not know/report the condition when queried at interview.
Conclusion We began by examining self-reported, ever-diagnosed hypertension status, which represents the typical manner that hypertension is operationalized in health disparity studies. These findings resemble past research in that rates are higher among women than men, and the size of the gender gap differs across racial/ethnic groups (e.g., Geronimus et al. 2007). In both NHIS and NHANES samples, the gender gap in ever-diagnosed hypertension is lowest among whites (2.1 and 4.7 points, respectively), and substantially higher among blacks (9.2 and 16.0 points) and Mexican-Americans (21.0 and 13.4 points). While white seniors (male and female) in both samples report lower rates than blacks of having ever been diagnosed with hypertension, the pattern relative to Mexican-Americans is more variable across samples—compared to whites, self-reported rates of diagnosis for MexicanAmericans are lower for NHIS men and NHANES men and women (but significantly so only for NHANES men). For women especially, the NHIS and NHANES Mexican-American samples do differ in important ways—particularly for education, as 78.5% of NHIS women did not complete high school, compared to 64.8% of NHANES women (not shown). And, the elevated rate of self-reported diagnosis among NHIS Mexican-American women is reduced to non-significance once we adjust for education, indicating that socioeconomic differences across samples likely contribute to these differences. We also find mixed evidence that social and behavioral risk factors mediate disparities in reporting a hypertension diagnosis among senior women or men. In both samples, odds ratios among black adults remain elevated in all models, while for Mexican-Americans men and women generally experience an equal (or lower) odds of self-reporting a diagnosis of hypertension relative to whites. However, we do see some evidence for a mediating effect of SES—especially for MexicanAmerican women, who experience an equal or significantly lower odds of hypertension diagnosis in relation to white women (depending on the sample), but only after adjustment for SES differences between groups. The odds of reporting a hypertension diagnosis is also significantly lower among Mexican-American than white men, but this lowered odds is somewhat more stable across models, which also supports the ‘‘Hispanic Paradox’’ (Markides and Eschbach 2005), whereby the Mexican-American hypertension rate is lower than would be expected, given their lower income and schooling. For NHIS, ancillary analyses (not shown) indicated that this advantage is not driven by diminished access to or utilization of medical
123
Hypertension Among Older Adults
905
care services, as reported patterns change little with adjustment for whether they have talked to a doctor about their health in the last year.2 Our results also show that adjusting for group differences in measures of support and integration, among NHIS but especially among NHANES adults, amplifies the black-white gap in hypertension—regardless of how we operationalized hypertension. Our finding that black seniors consistently report more frequent church attendance than white seniors supports prior research showing that elderly blacks are more embedded in more formal (including church) support networks (Taylor and Chatters 1986). Research by Krause (2002, 2006) also suggests that older blacks have better developed social support systems from the church than do older whites, resulting in more stress-buffering effects from church-based support. Yet, our interaction tests failed to find evidence of racial/ethnic difference in the effect of church attendance on hypertension rates, suggesting that seniors of all races (but especially women) benefit from attendance and involvement in services and related activities. However, since attendance is higher than whites, controlling for this influence amplifies the risk of hypertension among black relative to white seniors. These findings are supported by others showing that religious involvement reduces mortality risk and numerous physical health problems, including hypertension (see reviews by George et al. 2000 and Hummer et al. 2004), and the consistency we show across samples, for both ever-diagnosed and measured hypertension, provides strong evidence regarding the health benefits of religious membership and activity for seniors. In contrast to the findings for black seniors, our models also show that adjusting for measures of social support/integration amplifies the Mexican-American-white gap in reporting a diagnosis of hypertension, but here the findings are weaker and do not apply to NHANES women. Furthermore, when measured hypertension is examined instead, our models show that adjusting for support and integration actually mediates a portion of the hypertension gap between Mexican American and white seniors, both male and female. Studies examining health outcomes among Mexican American migrants often attribute part of their robust health profile to cultural ties and norms that emphasize healthier behaviors and strong family support networks (see discussion by Jasso et al. 2004; Vega and Amaro 1994). Given the high proportion of Mexican Americans in our NHANES sample who are foreign born (*42%), this finding supports such an interpretation—although it is unclear why this pattern is only revealed for measured hypertension. Results also provide limited evidence that measures of social support/integration operate differently across racial/ethnic groups for women, and no evidence for men. Specifically, the probability of ever having been diagnosed with hypertension increases for senior NHIS women who live in large families, but interaction tests showed this occurs only among Mexican Americans. This pattern might reflect the health costs that can accompany social roles and obligations. Mexican American men and women live in larger families than black and white seniors (see Table 1), and research finds that while the presence of family members can be positive, they 2
We only test this for the NHIS sample, as this health care information is not available in our NHANES sample.
123
906
B. K. Gorman, J. R. Porter
come with stresses (Mendes de Leon and Glass 2004). However, NHANES models showed no such interaction among women, finding instead that older men (regardless of race/ethnicity) experience a significant elevation in their odds of reporting a hypertension diagnosis as family size increases, and that senior women’s risk increases with their number of close friends. This positive effect may indicate that seniors become aware of their risk for hypertension through their friends and family, who may themselves be hypertensive and/or have encouraged respondents to seek medical care. Alternatively, since older NHANES men appear to benefit from an increasing number of close friendships, while women’s risk for reporting a hypertension diagnosis increases, this may also support a more gender-specific interpretation. Studies show that women’s social networks include more ties with higher levels of interaction than men (Pugliesi and Shook 1998), and that their relationships more often depend on emotional closeness and are more reciprocal in nature, as they typically provide help to a greater number of kin and friends than men (Gallagher 1994; Liebler and Sandefur 2002). In a related vein, studies also show that care-giving expectations and responsibilities of social relationships also appear to be associated with disproportionate health costs for women (see Pavalko and Woodbury 2000). However, it is important to remember the selectivity associated with the use of cross-sectional data for respondents over the age of 65, whereby the unhealthiest members of each group are excluded due to prior mortality. Furthermore, while our findings for reporting a previous diagnosis of hypertension mirror the conclusion of Mendes de Leon and Glass (2004) that the benefits of social support and integration for senior health do not appear disproportionate across racial/ethnic groups, we were unable to examine the full range of pathways through which they impact health (e.g., instrumental and informational support; see Berkman and Glass 2000). This makes it likely that we have underestimated the role of social support and integration to some extent. More generally, regression models for both samples were able to account for only some of the black-white difference in self-reporting a previous diagnosis of hypertension, especially for women. This may be due to the inability of our study to fully capture the ‘‘weathering’’ phenomenon proposed by Geronimus et al. (2007) due to the lack of survey information regarding early life experiences. However, it is also likely that our findings are skewed due to the use of hypertension data based on respondent self-report, and the temporal problems introduced because of question wording (i.e., ‘‘ever’’ diagnosed). Although an individual may have been diagnosed with hypertension at an earlier date, they may now be actively involved in a healthy lifestyle which has brought the condition under control, resulting in a normal blood pressure reading at interview. As such, we replicated our NHANES models using an alternate version of hypertension, based on measured high blood pressure at interview and reported use of blood pressure medicines. Findings from these models suggest that reporting bias and measurement error may be driving some (but not all) of the patterns discussed for hypertension based on self-reported diagnosis. Hypertension rates increase for all groups when measured hypertension is examined—with the exception of black women, whose measured hypertension rate is lower than their self-reported diagnosis rate. These
123
Hypertension Among Older Adults
907
deviations in under-/over-diagnoses of hypertension are interesting to consider, and results show that measured hypertension parallels trends based on self-report, as black seniors, but especially women, are the least likely to have undiagnosed hypertension. In contrast, Hispanics and white men are most likely to be overrepresented. This suggests an unequal likelihood of being diagnosed as hypertensive based on gender and racial/ethnic identity. More specifically, those that we expect to have the greatest odds of hypertension (blacks seniors) are also the least likely to be hypertensive without knowing it, while race/gender combinations that are less associated with high levels of hypertension are most likely to be underrepresented. Overall, when measured hypertension is considered, the gender gap in hypertension is much smaller for black NHANES seniors (4.3 vs. 16.0 point gap for ever-diagnosed hypertension), and larger for white (9.1 vs. 4.7) and Mexican American seniors (17.4 vs. 13.4). This smaller gap for black seniors occurs because only among black women is their measured hypertension rate smaller than their ever-diagnosed rate. As Ostchega et al. (2007) have shown, also using NHANES data, older black women (aged 60?) have a higher rate of being aware that they have hypertension when compared to black men, whites, or Mexican Americans. Racial differences in medical care treatment have been shown to occur through stereotyping, prejudice, and bias during the medical encounter (Smedley et al. 2003), and lower status individuals (e.g., black women) experience heightened and harmful stress levels compared to higher status individuals—stresses that can negatively impact health and functioning across the life course (Pearlin et al. 2005). Blood pressure evaluation is one of the most routine components of a medical check-up or encounter, and can even be done by an individual outside of a formal medical setting through the use, for example, of automated blood pressure-reading machines located at selected stores. This suggests that heightened exposure to stress and discriminatory treatment, combined with a disproportionate awareness of the potential for hypertension among health care providers and black adults themselves (especially women), contributes substantially to the very high rate of self-reported hypertension diagnosis among senior black women reported in Table 1. Since prevalence rates for measured hypertension in Table 4 show that black-white rates for senior women are near equivalent, while black-white differences for senior men remain significant but are substantially smaller in relation to reporting a hypertension diagnosis, this indicates that hypertension prevalence rates among women are more sensitive to measurement format than men. Among Mexican American seniors, for measured hypertension we again find a pattern of equal-to-or-better-than hypertension risk in relation to white adults. However, the gender gap increases for Mexican Americans when measured hypertension is considered, with the rate among Mexican American senior men is the lowest of any group, at 48.6%. As one-quarter of these men are born outside the United States, this suggests that Mexican American men may be more selected on health than women (see Gorman et al. 2010), leading to a more robust health profile when compared to women. Beyond measures of support and integration, explanatory measures across Tables 3 and 4 (across models for self-reported diagnosis and measured
123
908
B. K. Gorman, J. R. Porter
hypertension in NHANES) show both similarity and differences in how predictor measures relate to hypertension status. In terms of similarities, one can see that the controlled effects of race are the same across tables, as blacks are more likely and Hispanics less likely to be hypertensive in reference to whites. Health behaviors also show similar effects for both ever diagnosed and measured hypertension, where only women benefit from not smoking, drinking is harmful to men, and overweight and obesity increase hypertension (especially for men). Indeed, our analysis shows that senior women who are obese are almost twice as likely to be hypertensive than lower weight women, and obese men are over three times more likely to be hypertensive than their normal weight peers—indicating that policy efforts aimed at reducing obesity should also result in a substantial reduction in risk for high blood pressure among seniors. However, we also see that, only among older women, age and nativity are significant predictors of measured but not self-reported diagnosed hypertension. Other research has found evidence that measures of nativity and acculturation are more strongly related to medical conditions among women than men (e.g., Gorman et al. 2010), although it is unclear why this effect would only emerge for measured hypertension, and not reporting a diagnosis of the condition. We also find, for both measures of hypertension, that education is more protective against hypertension for women than men, a finding which supports the work of Ross and Mirowsky (2010), who found a similar effect for functional problems. In addition, and somewhat unexpectedly, we see a flipped effect between Tables 3 and 4 for psychological distress. In Table 3 (ever-diagnosed hypertension), psychological distress increases hypertension risk for women, but is not significant for men—but in Table 4 (measured hypertension), higher psychological distress is associated with less hypertension, but the relationship is not significant for women. While numerous studies document the physical health consequences of depression, to date most studies that examine how depression affects morbidity and mortality do not explore gender differences, and among the few that do, the pattern is not clear (see review by Sevick et al. 2000). This finding is a topic in need of additional study. In conclusion, while this paper demonstrates substantial racial/ethnic and gender difference in hypertension among seniors, we find evidence that group differences in social support/integration suppress disparities in hypertension status, regardless of how hypertension is operationalized, but only when considering patterns for black relative to white seniors. For Mexican American seniors, the role of social support and integration for shaping hypertension disparities relative to white seniors is less clear, varying by gender and whether a self-reported diagnosis or measured hypertension is considered; more study of these relationships, with larger samples, is needed to replicate and expand upon our findings. Furthermore, our findings for reporting a diagnosis versus measured hypertension among NHANES seniors suggests that while bias confounds racial/ethnic and gender patterns of hypertension prevalence rates, regression models for both hypertension measures show substantial similarity in how related social and behavioral factors shape racial/ethnic differences in hypertension among seniors, particularly church attendance among women. However, that the gender gap for black seniors is substantially smaller when measured hypertension is used suggests that gender differences in the utilization of
123
Hypertension Among Older Adults
909
medical care services may also be skewing observed hypertension rates. Studies consistently show that women utilize health care services more than men, even after accounting for female-specific services like gynecology (e.g., Koopmans and Lamers 2006), and that men initiate fewer preventative and self-care practices, use fewer medications, make fewer physician visits (Courtenay 2000), and when ill, are more likely to delay the help-seeking process (Cameron and Bernardes 1998). As such, while our findings illustrate how social factors, including SES and social support and integration, contribute to racial/ethnic differences in hypertension rates, our findings also suggest that assessments based solely on self-reported diagnosis of hypertension as an indicator of actual hypertension status misstates the racial/ethnic and gender composition of the hypertensive population in the United States. This has important implications for health care policy and planning, as other work has shown how gender differences in medical care use can skew gender patterns for self-reported medical conditions among racial minorities (see Gorman et al. 2010). While self-reported health data is often useful, our findings illustrate the sensitivity of social health research to the manner in which we operationalize our terms, particularly for more ‘‘silent’’ medical conditions, likely hypertension, and highlight the need for future work to supplement research that studies self-reported racial, ethnic, and gender disparities in health with measures based on direct clinical assessment. Acknowledgement We gratefully thank the Program in Health Economics, James A. Baker III Institute for Public Policy at Rice University, for their generous grant given in support of this research.
References Anderson, N. B., Bulatao, R. A., & Cohen, B. (Eds). (2004). Critical perspectives on racial and ethnic differences in health in later life. In N. B. Anderson, R. A. Bulatao, & B. Cohen (Eds.), Washington, DC: The National Academies Press. Angel, R. J., & Angel, J. L. (2006). Diversity and aging in the United States. In H. B. Robert & L. K. George (Eds.), Handbook of aging, the social sciences (6th ed., pp. 94–100). San Diego, CA: Academic Press. Bell, C. N., Thorpe, R. J., Jr., & LaVeist, T. A. (2010). Race/ethnicity and hypertension: The role of social support. American Journal of Hypertension, 23(5), 534–540. Berkman, L. F., & Glass, T. (2000). Social integration, social networks, social support, and health. In L. F. Berkman & K. Ichiro (Eds.), Social epidemiology (pp. 137–173). USA: Oxford University Press. Bird, C. E., & Rieker, P. P. (2008). Gender and health: The effects of constrained choices and social policies. New York: Cambridge University Press. Bosworth, H. B., Bartash, R. M., Olsen, M. K., & Steffens, D. C. (2003). The association of psychosocial factors and depression with hypertension among older adults. International Journal of Geriatric Psychiatry, 18, 1142–1148. Cameron, E., & Bernardes, J. (1998). Gender and disadvantage in health: Men’s health for a change. Sociology of Health & Illness, 20, 673–693. Centers for Disease Control and Prevention. (2007). BMI—body mass index: About BMI for adults. Retrieved January 14, 2007, from http://www.cdc.gov/nccdphp/dnpa/bmi/adult_BMI/about_adult_ BMI.htm#Why. Chobanian, A. V., Bakris, G. L., Black, H. R., Cushman, W. C., et al. (2003). The seventh report of the joint national committee on prevention, detection, evaluation, and treatment of high blood pressure. Journal of the American Medical Association, 289(19), 2560–2572.
123
910
B. K. Gorman, J. R. Porter
Cohen, B. (2004). Introduction. In N. B. Anderson, R. A. Bulatao, & B. Cohen (Eds.), Critical perspectives on racial and ethnic differences in health in later life (pp. 1–22). Washington, DC: National Academies Press. Cornwell, B., Laumann, E. O., & Philip Schumm, L. (2008). The social connectedness of older adults: A national profile. American Sociological Review, 73(2), 185–203. Courtenay, W. H. (2000). Behavioural factors associated with disease, injury, and death among men: Evidence and implications for prevention. Journal of Men’s Studies, 9(1), 81–142. Crimmins, M. E., Hayward, M. D., & Seeman, T. E. (2004). Race/ethnicity, socioeconomic status, and health. In N. B. Anderson, R. A. Bulatao, & B. Cohen (Eds.), Critical perspectives on racial and ethnic differences in health in later life (pp. 310–352). Washington, DC: The National Academies Press. DeNavas-Walt, C., Proctor, B. D., & Smith, J. (2007). Income, poverty, and health insurance coverage in the United States: 2006. U.S. Census Bureau, current population reports, P60-233. Washington, DC: US Government Printing Office. Durkheim, E´. (1897/1951) Suicide: A study in sociology. Glencoe, IL: Free Press. Egan, B. M., Zhao, R., & Axon, R. N. (2010). US trends in prevalence, awareness, treatment, and control of hypertension, 1998–2008. Journal of the American Medical Association, 303(20), 2043–2050. Fields, L. E., Burt, V. L., Cutler, J. A., Hughes, J., Roccella, E. J., & Sorlie, P. (2004). The burden of adult hypertension in the United States 1999 to 2000: A rising tide. Hypertension, 44, 398–404. Gallagher, S. K. (1994). Doing their share: Comparing the patterns of help given by older and younger adults. Journal of Marriage and the Family, 56, 567–578. George, L. K., Larson, D. B., Koenig, H. G., & McCullough, M. E. (2000). Spirituality and health: What we know and what we need to know. Journal of Social and Clinical Psychology, 19, 102–116. Geronimus, A. T., Bound, J., Keene, D., & Hicken, M. (2007). Black-white differences in age trajectories of hypertension prevalence among adult women and men, 1999–2002. Ethnicity and Disease, 17(1), 40–48. Glover, M. J., Greenlund, K. J., Ayala, C., & Croft, J. B. (2005). Racial/ethnic disparities in prevalence, treatment, and control of hypertension B United States, 1999–2002. Morbidity and Mortality Weekly Reports, 54(1), 7–9. Gorman, B. K., Read, J. G., & Krueger, P. M. (2010). Gender, acculturation, and health among Mexican Americans. Journal of Health and Social Behavior, 51(4), 440–457. Gorman, B. K., & Sivaganesan, A. (2007). The role of social support and integration for understanding socioeconomic disparities in self-rated health and hypertension. Social Science and Medicine, 65(5), 958–975. Hummer, R. A., Benjamines, M. R., & Rogers, R. G. (2004). Racial and ethnic disparities in health and mortality among the U.S. elderly population. In N. B. Anderson, R. A. Bulatao, & B. Cohen (Eds.), Critical perspectives on racial and ethnic differences in health in later life (pp. 53–94). Washington, DC: The National Academies Press. Jasso, G., Massey, D. S., Rosenzweig, M. R., & Smith, J. P. (2004). Immigrant health: Selectivity and acculturation. In B. A. Norman, R. A. Bulatao, & B. Cohen (Eds.), Critical perspectives on racial, ethnic differences in later life (pp. 227–266). Washington, DC: National Academies Press. Kessler, R. C. (2002). Short screening scales to monitor population prevalences and trends in non-specific psychological distress. Psychological Medicine, 32, 959–976. Koopmans, G. T., & Lamers, L. M. (2006). Gender and health care utilization: The role of mental distress and help-seeking propensity. Social Science and Medicine, 64, 1216–1230. Krause, N. (2002). Exploring race differences in a comprehensive battery of church-based social support measures. Review of Religious Research, 44(2), 126–149. Krause, N. (2006). Exploring the stress-buffering effects of church-based and secular social support on self-rated health in late life. Journal of Gerontology, 61B(1), S35–S43. Krause, N., & Keith, V. (1989). Gender differences in social support among older adults. Sex Roles, 21(9/ 10), 609–628. Liebler, C. A., & Sandefur, G. D. (2002). Gender differences in the exchange of social support with friends, neighbors, and co-workers at midlife. Social Science Research, 31, 365–391. Link, B. G., & Phelan, J. C. (1995). Social conditions as fundamental causes of disease. Journal of Health and Social Behavior (extra issue), 80–94. Markides, K. S., & Eschbach, K. (2005). Aging, migration, and mortality: Current status of research on the Hispanic paradox. Journals of Gerontology: Social Sciences, 60B, S68–S75.
123
Hypertension Among Older Adults
911
Matthews, K. A., Kiefe, C. I., Lewis, C. E., Liu, K., Sidney, S., & Yunis, C. (2002). Socioeconomic trajectories and incident hypertension in a biracial cohort of young adults. Hypertension, 39, 772–776. Mendes de Leon, C. F., & Glass, T. A. (2004). The role of social and personal resources in ethnic disparities in late-life health. In N. B. Anderson, R. A. Bulatao, & B. Cohen (Eds.), Critical perspectives on racial and ethnic differences in health in later life (pp. 353–405). Washington, DC: The National Academies Press. Meyer, C. M., Armenian, H. K., Eaton, W. W., & Ford, D. E. (2004). Incident hypertension associated with depression in the baltimore epidemiological catchment area follow-up study. Journal of Affective Disorders, 83, 127–133. Myers, H. F., & Hwang, W.-C. (2004). Cumulative psychosocial risks and resilience: A conceptual perspective on ethnic health disparities in late life. In N. B. Anderson, R. A. Bulatao, & B. Cohen (Eds.), Critical perspectives on racial and ethnic differences in health in later life (pp. 492–539). Washington, DC: The National Academies Press. National Center for Health Statistics. (2007). Health, United States, 2007, with chartbook on trends in the health of Americans. Hyattsville, MD. Ostchega, Y., Dillon, C. F., Hughes, J. P., Carroll, M., & Yoon, S. (2007). Trends in hypertension prevalence, awareness, treatment, and control in older US adults: Data from the national health and nutrition examination survey 1988 to 2004. Journal of the American Geriatric Society, 55(7), 1056–1065. Pavalko, E. K., & Woodbury, S. (2000). Social roles as process: Caregiving careers and women’s health. Journal of Health and Social Behavior, 41(1), 91–105. Pearlin, L. I., Schieman, S., Fazio, E. M., & Meersman, S. (2005). Stress, health, and the life course: Some conceptual perspectives. Journal of Health and Social Behavior, 46(2), 205–219. Pugliesi, K., & Shook, S. L. (1998). Gender, ethnicity, and network characteristics: Variation in social support resources. Sex Roles, 38(3/4), 215–238. Read, J. G., & Gorman, B. K. (2006). Gender inequalities in U.S. adult health: The interplay of race and ethnicity. Social Science and Medicine, 62, 1045–1065. Read, J. G., & Gorman, B. K. (2007). Racial/ethnic differences in hypertension and depression among U.S. adult women. Ethnicity and Disease, 17(2), 383–390. Ross, C. E., & Mirowsky, J. (2010). Gender and the health benefits of education. The Sociological Quarterly, 51, 1–19. Sevick, M. A., Rolih, C., & Pahor, M. (2000). Gender differences in morbidity and mortality related to depression: A review of the literature. Aging: Clinical and Experimental Research, 12(6), 407–416. Shih, R. A., Fernandes, M. M., & Bird, C. E. (2010). The application of biomarker data to the study of social determinants of health. In C. E. Bird, P. Conrad, A. M. Freemont, & S. Timmermans (Eds.), Handbook of medical sociology (6th ed., pp. 395–417). Nashville: Vanderbilt University Press. Smedley, B. D., Stith, A. Y., & Nelson, A. R. (2003). In B. D. Smedley, A. Y. Stith, & A. R. Nelson (Eds.), Unequal treatment: Confronting racial and ethnic disparities in health care. Washington, DC: National Academies Press. StataCorp. (2007). Stata statistical software: Release 10.0. College Station, TX: Stata Corp. Stewart, D., Johnson, W., & Saunders, E. (2006). Hypertension in Black Americans as a special population: Why so special? Current Cardiology Reports, 8(6), 405–410. Taylor, R. J., & Chatters, L. M. (1986). Patterns of informal support to elderly black adults: Family, friends, and church members. National Association of Social Workers, pp. 432–438. Uchino, B. N., Cacioppo, J. T., & Kiecolt-Glaser, J. K. (1996). The relationship between social support and physiological processes: A review with emphasis on underlying mechanisms and implications for health. Psychological Bulletin, 119, 488–531. Vega, W. A., & Amaro, H. (1994). Latino outlook: Good health, uncertain prognosis. Annual Review of Public Health, 15, 39–67. Winkleby, M. A., & Cubbin, C. (2004). Racial/ethnic disparities in health behaviors: A challenge to current assumptions. In N. B. Anderson, R. A. Bulatao, & B. Cohen (Eds.), Critical perspectives on racial and ethnic differences in health in later life (pp. 450–491). Washington, DC: The National Academies Press.
123