The Relation of Fasting Insulin to Blood Pressure in a Multiethnic Population: The Miami Community Health Study

The Relation of Fasting Insulin to Blood Pressure in a Multiethnic Population: The Miami Community Health Study RICHARD P. DONAHUE, PhD, RONALD J. PRINEAS, MD, JUDY A. BEAN, PhD, ROSEMARY A. DECARLO DONAHUE, PhD, RONALD B. GOLDBERG, MD, JAY S. SKYLER, MD, AND NEIL SCHNEIDERMAN, PhD

PURPOSE: The aim of this study was to examine the associations among fasting insulin, adiposity, waist girth, and blood pressure among a nondiabetic multiethnic population. METHODS: A cross-sectional study was performed among 25–44-year-old African-Americans (n 5 159), Cuban-Americans (n 5 128), and non-Hispanic whites (n 5 207) selected from Date County, Florida. Fasting insulin levels were correlated with resting blood pressure level within each ethnic group. The separate effects of percentage body fat and waist girth on the association between blood pressure and insulin were analyzed in multiple linear regression and analysis of covariance. RESULTS: Fasting insulin was positively associated with systolic (r 5 0.26–0.39; P , 0.01) and diastolic blood pressure (r 5 0.19–0.30; P 5 0.10 to P , 0.001) among women of all ethnic groups and among non-Hispanic white men (r 5 0.27; P , 0.05). Stepwise linear regression analyses revealed statistically significant associations between systolic and diastolic blood pressure and fasting insulin level in non-Hispanic whites independent of other covariates, including sex and percentage body fat (P , 0.001). Fasting insulin was also independently and significantly related to systolic blood pressure among African-Americans (P 5 0.02). Among Cuban-Americans, sex and percentage body fat were the main correlates of blood pressure level. Analysis of covariance revealed a relationship between insulin and blood pressure that was independent of waist girth among men and women. CONCLUSIONS: Fasting insulin level and blood pressure were positively associated among AfricanAmericans and non-Hispanic whites. This association was not entirely due to the common association with percentage body fat or waist girth. Ann Epidemiol 1998;8:236–244.  1998 Elsevier Science Inc. KEY WORDS:

Insulin, Blood Pressure, African-American, Hispanic-American, Cross-Sectional Study.

INTRODUCTION High blood pressure is a major cause of disability and death in the United States, particularly among ethnic minority populations (1). In the general population, there is a continuous relationship between blood pressure and risk of coronary heart disease (CHD) (2). Indeed, the absolute risk for cardiovascular disease is greatest among those in the nonhypertensive portion of the blood pressure distribution (3). The increasing diversity of the ethnic composition in the United States has important public health implications

From the University of Miami School of Medicine, Department of Epidemiology and Public Health, Miami, FL (R.P.D., R.J.P., J.A.B.); Department of Social and Preventive Medicine, SUNY Buffalo, Buffalo, NY (R.P.D.); University of Miami School of Nursing, Miami, FL (R.A.D.D.); and Department of Medicine, University of Miami School of Medicine, Miami, FL (R.B.G., J.S.S.); University of Miami, Department of Psychology, Miami, FL (N.S.). Address reprint requests to: Richard P. Donahue, PhD, Department of Social and Preventive Medicine, SUNY Buffalo, 270 Farber Hall, Buffalo, NY 14214. Received April 3, 1997; accepted October 30, 1997.  1998 Elsevier Science Inc. All rights reserved. 655 Avenue of the Americas, New York, NY 10010

for the prevention and control of several chronic conditions, including high blood pressure and obesity. These two conditions may be linked through a common association with hyperinsulinemia/insulin resistance. It has been suggested, however, that the association between insulin level and blood pressure link is absent among African-Americans. Clearly, there is a need to better understand the physiologic determinants of blood pressure, especially among minority groups (4). Numerous reports from several centers (5–10) have generated inconsistent findings and have led to differences of opinion (11–13) regarding the role of hyperinsulinemia/ insulin resistance in regulation of blood pressure. Many of these studies have examined only one or at most two ethnic groups or included persons receiving pharmacologic treatment for hypertension or non-insulin-dependent diabetes mellitus, which confound the results. Inconsistent findings are also the result of several other factors: (i) the close correlation between insulin (or a resistance to its action) and obesity; (ii) the importance of obesity level as well as its anatomic location in affecting insulin and blood pressure 1047-2797/98/$19.00 PII S1047-2797(97)00208-1

AEP Vol. 8, No. 4 May 1998: 236–244

Selected Abbreviations and Acronyms CHD 5 coronary heart disease IGT 5 impaired glucose tolerance BMI 5 body mass index WHR 5 waist-to-hip ratio CT 5 computed tomography

levels; and (iii) whether fasting insulin concentrations can adequately reflect insulin resistance viz-a-viz its association with blood pressure. We have previously shown that a direct measure of insulin resistance is related to blood pressure level in 53 AfricanAmerican and non-Hispanic white persons who participated in the current study (14). Cuban-Americans were not included in that report. The present report addressed the above concerns in a triethnic population–based sample. Three specific issues were studies. First, are there ethnic differences in the association between blood pressure and fasting insulin level? Second, to what degree is the blood pressure/insulin association mediated by body fat? Third, is fasting insulin level associated with blood pressure in both men and women after consideration of waist girth as a measure of visceral fat?

MATERIALS AND METHODS Study Population The Miami Community Health Study is an epidemiologic study designed to assess the correlates of CHD risk factors among 24–44-year-old African-Americans, Cuban-Americans, and non-Hispanic whites living in Dade County, Florida. The study cohort was selected from census tracts (1990 U.S. Census) located within 10 miles of the University of Miami School of Medicine with populations composed of at least 80% of the targeted ethnic group. Many census tracts have a relatively homogeneous ethnic composition which makes this type of identification and recruitment feasible. After appropriate census tracts were selected, several blocks were defined within each census tract from which to recruit eligible participants. This list of addresses was supplemented with information from a marketing service (I Rent America, Dallas, Texas) to obtain up-to-date information on each residential area. Letters of invitation were sent to each residence, and a bilingual telephone recruiter conducted an enumeration interview within contacted households to obtain a list of all eligible persons. Persons without a telephone were contacted by letter with a tear-off form to indicate their level of interest. Interested people were contacted at a location where they had access to a phone. Once a potential study candidate was contacted, his or her eligibility was assessed by telephone interview. Eligibility

Donahue et al. BLOOD PRESSURE, INSULIN, AND ETHNICITY

237

was limited to those who were aged 25–44 years at the time of telephone contact; self-identified as either AfricanAmerican, Cuban-American, or non-Hispanic white; and who did not have a current or past history of diagnosed hypertension, diabetes mellitus, CHD or stroke; who were not receiving any medication known to affect carbohydrate metabolism or blood pressure level (e.g., antihypertensive, cholesterol-lowering or psychotropic agents); who had a permanent address in the target area; and who were able to read and understand the written informed consent form and to participate fully in all phases of the study protocol. Pregnant or lactating women were excluded until at least three months after giving birth or cessation of lactation. If more than one eligible person in a household was identified, the one whose next birthday was closest to the data of the telephone call was selected. If a person was excluded for medical reasons, another member of that household could have been recruited. Only one eligible person per household was recruited and examined. An attempt was made to interview by telephone all eligible study candidates whether or not they agreed to participate. In this manner we were able to compare eligible participants with nonparticipants according to a number of health and sociodemographic indices. Appointments for participants were scheduled 1–14 days in advance, with reminder telephone calls shortly before the examination date. Participants who postponed their clinic visit or failed to attend were rescheduled a minimum of three times before they were considered no longer likely to participate. The clinic examination consisted of two clinic visits of 3–4 hours each, separated by 1–2 weeks in most instances. This report deals with data collected during the first clinic examination: resting blood pressure, anthropometry, and several standardized questionnaires to assess lifestyle and health habits. All participants were required to fast (except water) for 12 hours and to refrain from smoking or vigorous physical activity before coming into the clinic. A standard 75-g oral glucose tolerance test was administered with determinations of glucose and insulin at 215, zero, 60, and 120 minutes. The first two sample results were averaged to provide a fasting value. Diabetes (fasting glucose, . 140 mg/ dL or a 2-hour glucose, . 200 mg/dL) and impaired glucose tolerance (IGT) (fasting glucose, , 140 mg/dL and 2-hour glucose, 140–199 mg/dL) were defined according to criteria of the World Health Organization (15). Persons testing positive for diabetes (n 5 20) were not included in this report. Glucose was assayed by the glucose oxidase method (Yellow Springs Instrument, Yellow Springs, OH). Insulin was assayed with standard radioimmunoassay techniques. The lower limit of detection was 3.0 mU/ml. All persons with fasting levels below the minimum were assigned a value of 3.0 mU/ml. The interassay coefficient of variation for fasting glucose was 2.7% and that for fasting insulin was

238

Donahue et al. BLOOD PRESSURE, INSULIN, AND ETHNICITY

15%. Intralaboratory coefficients were , 2% and 11%, respectively. Blood Pressure Seated blood pressure was measured three times using a standard mercury manometer by trained and certified technicians (16). The onset of the first phase (systolic) and onset of the fifth phase (diastolic) Korotkoff sounds were recorded. The mean of the second and third measures was used in analyses. Anthropometry All anthropometric measures were made with the participant wearing light clothes and no shoes. Weight was recorded on a balance-beam scale and recorded to the nearest 0.25 lb. Height was measured to the nearest 0.5 cm. Body mass index (BMI) was calculated as weight (kg)/height (m2). The waist girth was measured in the standing position by applying a linen tape measure horizontally midway between the iliac crest and the lowest lateral portion of the rib cage, and anteriorly midway between the umbilicus and the xiphoid process. This measure is the smallest circumference at waist level. The mean of two measurements (recorded to the nearest 0.5 cm) was used as a proxy for visceral fat. To estimate percentage body fat, we calculated body density from subcutaneous skinfold thickness measurements using age- and gender-specific prediction equations published by Durnin and Womersley (17). For example, the equation for men aged 30–39 years is: body density 5 1.1165 2 0.0484 log (triceps skinfold 1 subscapular skinfold). For non-Hispanic white and Cuban-American participants, percentage body fat was calculated from body density using the Siri equation (18): percentage body fat 5 (4.95/density 2 4.50) 3 100. For African-Americans (who have a greater density of lean body mass than whites), percentage body fat was calculated by the equation published by Schutte and colleagues (19): percentage body fat 5 (4.374/density 2 3.928) 3 100. The study was approved by the University of Miami School of Medicine Internal Review Board, and all participants gave written informed consent. Statistical Analysis Differences between mean values of continuous variables were assessed by one-way analysis of variance. Spearman or Pearson correlation coefficients were computed to examine the association between fasting insulin and other continuous variables. Fasting insulin level was not normally distributed so it was log transformed (natural log). Multiple stepwise linear regression analyses were performed to examine the relation of insulin to blood pressure according to ethnicity. In one model, age, sex, and fasting insulin were included as independent variables. In a second model percentage body fat was added to the above list of independent variables.

AEP Vol. 8, No. 4 May 1998: 236–244

These variables were chosen a priori to provide an unconfounded estimate of the association between blood pressure and insulin level. The change is the beta coefficient for insulin was evaluated to determine the extent to which its link with blood pressure is mediated by body fat. The appropriate cross-product term to examine a gender-by-insulin interaction was included and was found not to be statistically significant (P . 0.15). An insulin-by-ethnicity interaction term was also nonsignificant. The effects of the waist girth on blood pressure were evaluated after stratification at the sex-specific median value. A similar strategy was employed for fasting insulin. Analysis of covariance adjusting for ethnicity and age was used to estimate the association between insulin level (high or low) and waist girth (high or low) on blood pressure among men and women separately. All analyses were performed by use of the Statistical Analysis System (20). All statistical tests were two-sided.

RESULTS Of 21,599 people contacted, the eligibility status of 6440 was unknown due to their refusal to be queried over the telephone, leaving 15,159 (70.2%) who agreed to a telephone interview. Most of these people were ineligible (n 5 14,179), predominantly because they were outside the age criteria, leaving 980 who met all the eligibility criteria. Nineteen people (2%) were found to be medically ineligible after they had come into the clinic; they are not considered further in this report. One hundred twenty-five people (12.8%) repeatedly failed to keep their appointment by the end of the field operations, and 322 (32.9%) refused. The final sample size numbered 514 subjects (52.4%). Table 1 compares all participants (n 5 514) to eligible nonparticipants (n 5 447; 322 refusals plus 125 who failed to keep their appointment) according to a number of characteristics assessed during the telephone interview. Participants were significantly more likely to be men (51.0% vs. 39.7%; P , 0.001). There were no statistically significant differences in the distribution of the ethnic composition, mean age, frequency of reported family history of diabetes or CHD, perceived health, or the frequency of having smoked 100 cigarettes during one’s lifetime. Restriction of the comparisons to the 494 subjects in this report yielded similar results. Table 2 presents descriptive characteristics of the study population according to ethnicity and sex. As expected, the mean age of the study population was z35 years. Men had higher mean levels of blood pressure than women across all ethnic groups, although the magnitude of this difference was smallest among African-Americans. Fasting insulin levels were higher in men than in women, except among the non-Hispanic whites among whom there was no difference,

AEP Vol. 8, No. 4 May 1998: 236–244

TABLE 1. Comparison of participants and nonparticipants: the Miami Community Health Study, 1991–1995 Characteristic

239

Donahue et al. BLOOD PRESSURE, INSULIN, AND ETHNICITY

Participants (n 5 514)

Nonparticipants (n 5 447)

P

51.0

39.7

, 0.001

33.3 25.7 41.0 35.3

36.5 28.3 35.2 36.6

0.19

49.0 44.2 6.8

51.8 44.6 3.6

0.08

21.3 76.0 2.7

20.4 78.1 1.5

0.43

17.9 77.9 4.2

22.7 73.1 4.2

0.18

92.0

89.5

0.33

48.1 51.9

47.5 52.5

0.87

25.0 75.0

30.0 70.0

0.08

14.0

13.7

0.03

Gender, % male Ethnicity (%) African-American Cuban American Non-Hispanic white Mean age (years) Reported family history of: High blood pressure or stroke (%) Yes No Don’t know Diabetes (%) Yes No Don’t know Coronary heart disease (%) Yes No Don’t know Perceived health, excellent or good (%) Smoked 100 cigarettes in lifetime (%) Yes No Currently smoke cigarettes (%) Yes No Mean education level (years)

0.58

and fasting glucose was consistently higher in men than women. Measures of body fatness and fat distribution revealed the well-known pattern of a higher percentage of body fat and smaller waist circumferences in women than men. The frequency of IGT was highest in the African-

American (17.6%) participants and lowest among the Cuban-Americans (12.5%). The univariate associations between fasting insulin and selected variables are shown in Table 3. Fasting insulin was significantly and positively associated with systolic and diastolic blood pressure level among women in all ethnic groups, and among non-Hispanic white men. Fasting insulin was highly correlated with percentage body fat and waist girth in all sex-ethnic groups, with the correlation coefficients generally higher in women. In order to evaluate further the association among blood pressure level, insulin concentration, and percentage body fat, two regression models were analyzed. Among AfricanAmericans (Table 4), age and insulin concentration were independently associated with systolic blood pressure level in model 1. In model 2, percentage body fat replaced insulin while the coefficient for sex nearly tripled. For diastolic blood pressure in model 1, insulin, sex, and age were independent predictors and explained 8.3% of the variance. In model 2, percentage body fat entered the model, and neither insulin nor age remained statistically significant. The beta coefficient for gender virtually doubled, suggesting that once differences in percentage body fat were controlled, men had a predicted diastolic blood pressure 6.2 mmHg higher than women. The multiple R2 increased slightly (from 8.3 to 10.3%), perhaps due to the close association between percentage body fat and insulin. Among Cuban-Americans, fasting insulin was not associated with systolic blood pressure in either model 1 or 2. Fasting insulin and gender were positively correlated with diastolic blood pressure (model 1) and explained 12.1% of the variance. In model 2, percentage body fat displaced fasting insulin, and again the beta coefficient for gender increased. However, only an additional 2.1% of the variance was added, for a total of 14.2%. Among non-Hispanic whites, fasting insulin and gender were associated with systolic and disastolic blood pressure level in both models. The addition of percentage body fat as an explanatory variable in model 2 failed to displace either insulin or sex, though

TABLE 2. Selected descriptive characteristics of the study population African-American Characteristic Age (years) Systolic blood pressure (mmHg) Diastolic blood pressure (mmHg) Fasting insulin (mU/mL) Fasting glucose (mg/dL) Body fat (%) Waist (cm) IGT (%) BMI (kg/m2)

Women (n 5 91) 35.9 110.8 72.5 8.3 84.3 39.7 85.0 17.6 28.5

(5.8) (12.3) (10.3) (5.5) (7.7) (6.0) (14.5) (6.8)

Cuban-American

Men (n 5 68) 35.0 113.6 75.3 12.0 86.9 30.5 90.1 17.7 27.3

(6.0) (10.1) (9.1) (13.0) (9.8) (5.4) (12.6) (4.4)

Women (n 5 62) 34.9 103.8 69.2 7.2 85.1 36.8 78.1 14.5 25.3

(5.7) (10.1) (7.7) (4.0) (6.9) (5.8) (12.8) (5.1)

Non-Hispanic whites

Men (n 5 66) 32.5 112.4 74.5 10.1 88.8 28.5 95.5 10.6 28.8

(5.0) (8.7) (8.9) (7.8) (9.0) (5.2) (14.8) (5.8)

Women (n 5 94) 36.5 106.5 70.5 9.5 85.6 35.6 79.4 18.1 26.2

(5.1) (10.7) (8.9) (10.0) (8.6) (6.4) (17.7) (7.0)

Men (n 5 113) 36.1 112.6 75.5 9.6 90.6 28.2 91.7 14.2 26.5

(5.7) (10.2) (8.3) (13.3) (10.1) (5.8) (11.7) (4.3)

240

Donahue et al. BLOOD PRESSURE, INSULIN, AND ETHNICITY

AEP Vol. 8, No. 4 May 1998: 236–244

TABLE 3. Spearman rank order correlation coefficients between fasting insulin (mU/mL) and selected variables according to ethnicity and gender African-American Variable Systolic blood pressure (mmHg) Diastolic blood pressure (mmHg) Body fat (%) Waist girth (cm) a

P P c P d P b

, , , ,

Cuban-American

Non-Hispanic whites

Women (n 5 81)

Men (n 5 52)

Women (n 5 58)

Men (n 5 65)

Women (n 5 92)

Men (n 5 107)

0.26a 0.19d 0.34a 0.52a

0.13 0.11 0.44c 0.32a

0.27b 0.28b 0.46c 0.49c

0.03 0.05 0.28c 0.30a

0.39c 0.31c 0.36c 0.44c

0.27a 0.24a 0.22a 0.24a

0.01. 0.05. 0.001. 0.10.

the coefficient for insulin was reduced by 24% (e.g., for systolic blood pressure: 4.5–3.4/4.5 3 100% 5 24%). Percentage body fat added an additional 4–6% of the variance explained in each instance. The effects of visceral adiposity, as indexed by waist girth, on the blood pressure/insulin association was examined in sex-specific analyses. Among women, there were statistically significant main effects of insulin level and waist girth on systolic (P , 0.001 for each) and diastolic blood pressure level (P 5 0.04 and P , 0.001, respectively; data not shown). Subgroup comparisons within each level of waist girth adjusted for age and ethnicity (Figure 1) revealed that blood pressure increased significantly with increased insulin and that this association was most striking among those whose waist girth was above the median level. Among men (Figure 2), insulin and waist levels were also independently related to systolic (P 5 0.06 and P 5 0.009) and diastolic (P 5 0.04 and P , 0.001; data not shown) blood pressure levels. The differences within subgroups were not as evident as in women but did show a monotonic increase in blood

pressure with increased insulin. Tests for a waist-by-insulin interaction were not statistically significant in either gender (P . 0.20).

DISCUSSION These analyses examined the independent and joint relationships among body fat, its anatomic location, and fasting insulin with blood pressure level in a population-based, multiethnic sample of nondiabetic persons. In the present report, fasting insulin was related independently of percentage body fat to systolic blood pressure among AfricanAmericans and non-Hispanic whites, whereas sex and percentage body fat were the major correlates of blood pressure among Cuban-Americans. Insulin was associated with diastolic blood pressure only among non-Hispanic whites once the effects of percentage body fat were considered. An important issue is how or to what degree insulin is involved in blood pressure regulation and whether it is

TABLE 4. Multiple stepwise linear regression of selected variables on blood pressure according to ethnicity African-American Dependent variable Systolic blood pressure Insulin Sex Age Body fat (%) Multiple R2 (%) Diastolic blood pressure Insulin Sex Age Body fat (%) Multiple R2 (%) a

Cuban-American

Model 2b

Model 1

Model 2

Model 1

Model 2

3.0 (0.02)c 3.1 (0.10) 0.35 (0.03) — 6.1

1.18 (0.39) 8.9 (, 0.001) 0.35 (0.03) 0.59 (, 0.001) 13.6

1.8 (0.21) 8.6 (, 0.001) 0.06 (0.69) — 17.9

0.33 (0.82) 12.3 (, 0.001) 20.07 (0.64) 0.42 (0.006) 23.0

4.5 (, 0.001) 5.5 (, 0.001) 0.02 (0.83) — 15.7

3.4 (, 0.001) 9.0 (, 0.001) 20.18 (0.19) 0.45 (0.001) 20.1

2.1 (0.06) 3.2 (0.04) 0.30 (0.02) — 8.3

1.3 (0.27) 6.2 (0.007) 0.22 (0.12) 0.30 (0.07) 10.3

2.5 (0.04) 4.5 (0.003) 0.05 (0.67) — 12.1

1.4 (0.28) 8.2 (, 0.001) 20.03 (0.79) 0.35 (0.009) 14.2

3.3 (, 0.001) 4.5 (, 0.001) 0.15 (0.15) — 14.0

2.5 (, 0.001) 7.4 (0.002) 20.02 (0.85) 0.38 (, 0.001) 20.5

Model 1 contains insulin; sex (0, women; 1, men); and age (years) as independent variables. Model 2 contains insulin; sex (0, women; 1 men); age (years); and body fat (%) as independent variables. c Regression coefficient and (P value). b

Non-Hispanic whites

Model 1a

AEP Vol. 8, No. 4 May 1998: 236–244

Donahue et al. BLOOD PRESSURE, INSULIN, AND ETHNICITY

241

FIGURE 1. Mean level of systolic blood pressure (mmHg, adjusted for age and ethnicity) according to median level of fasting insulin and waist girth: women.

entirely appropriate in multivariate analyses to adjust for adiposity. Although variables such as BMI can be measured with little error, this is certainly not the case for fasting insulin concentration. Indeed, the ratio of the intraindividual variation to the interindividual variation for fasting insulin is relatively large, making it difficult to find an association between insulin and cardiovascular risk factors (21). Given the positive correlation between percentage body fat and insulin concentration, a portion of the association between insulin and blood pressure may be due to the higher concentrations of insulin in persons with high adiposity. Thus the inclusion of indices of body fat as potential confounders in multiple regression analyses may result in an overadjustment for that variable and an underestimate of the association with insulin level. Stratification by level of the waist girth (or other measures of visceral fat) as well as insulin level allows for a better understanding of the roles that each variable may play in blood pressure regulation. The findings in this report indicated that consideration of waist girth did not eliminate the association between insulin and blood pressure level. Indeed, the analyses indicated significant independent effects of insulin concentration on blood pressure that were additive to those of the waist girth, particularly among women. There has been considerable debate of whether hyperinsulinemia (or insulin resistance) is related to blood pressure among various ethnic groups. Saad et al. (22) found that fasting insulin was unrelated to mean blood pressure level in a select sample of Pima Indians and African-Americans.

These findings contrast with those of Faulkner et al. (23) and the CARDIA Study (24), which reported a positive association between fasting insulin (or insulin resistance) and blood pressure among African-Americans. The association with visceral fat was not reported. Results among thirdgeneration Japanese-Americans (7), Hispanics living in Colorado (8), and Mexican-Americans (9) have also demonstrated a significant association between blood pressure and insulin level. Nevertheless, hyperinsulinemia and insulin resistance are not interchangeable, although they are reasonably well correlated among nondiabetic persons (23). The present study is unique in that it includes men and women of three ethnic backgrounds, randomly drawn from a single county, who were unmedicated and nondiabetic at entry. The current results also show that control for the waist girth as a reflection of visceral fat did not eliminate the positive association between fasting insulin and blood pressure among African-American and non-Hispanic whites. Studies in older populations (25) or obese participants (26) have failed to support a link between blood pressure and insulin level. The metabolic interrelationships among blood pressure, body fat (and its location), and insulin levels are no doubt in part responsible for the conflicting findings of statistical independence. Many studies published to date have used BMI to assess obesity, while fewer studies have examined the effects of visceral obesity on the association between blood pressure and insulin. Boyko and colleagues (7) showed that computed tomography (CT)–measured intraabdominal fat did not eliminate the blood pressure–

242

Donahue et al. BLOOD PRESSURE, INSULIN, AND ETHNICITY

AEP Vol. 8, No. 4 May 1998: 236–244

FIGURE 2. Mean level of systolic blood pressure (mmHg, adjusted for age, and ethnicity) according to median level of fasting insulin and waist girth: men.

insulin relationship among third generation JapaneseAmericans. Similarly, Johnson and colleagues (27) reported that insulin was related to blood pressure level independent of CT-measured fat among 81 men. The strength of the insulin/blood pressure association was weaker with increased waist circumference—a finding also observed in the San Antonio Heart Study (28), but which we failed to note despite the wide range of adiposity and waist girth in this population. The small degree of overlap in the waist size distribution between the sexes (Table 2) prohibited the inclusion of sex and waist girth (along with insulin) in a single regression model. Waist girth was chosen rather than the waist-to-hip ratio (WHR) as the variable of greater relevance in these analyses because the waist circumference has been shown to correlate with metabolic factors more closely than the WHR, particularly among women (29) and likely reflects the amount of visceral adiposity rather than its distribution (reflected by the WHR). If women are less insulin-resistant than men as we have suggested (30), this may account for some of the sex differences in blood pressure level. Whether the waist girth is an equally good indicator of visceral body fat in both sexes and in different ethnic groups is an important unanswered question. Conway and colleagues (31) used CT to assess abdominal fat and reported that African-American women (n 5 8) had less visceral fat than non-Hispanic white women (n 5 10) at comparable WHR levels. To our knowledge, there are no comparative data for men or for

Cuban-Americans. This is an important area for future research. Several limitations and strengths of the current study deserve comment. Our response rate of z53% limits the generalizability of the results to other populations. As in many other epidemiologic studies, the participants in this study appeared healthier than the nonparticipants and perhaps were more health conscience, as evidenced by their lower frequency of cigarette smoking. They also reported a marginally lower frequency of a family history of high blood pressure or stroke, which may serve to attenuate relationships between blood pressure and other factors. The vast majority of the ineligible persons of both sexes were excluded because they were outside the age criteria, and few were excluded due to medical conditions associated with insulin resistance or blood pressure. Our decision to exclude known hypertensive or diabetic persons likely resulted in the omission of persons with the greatest degree of hyperinsulinemia/insulin resistance. This may partly explain the lack of a univariate association between insulin and blood pressure in African-American and Cuban-American men. Thus these results are conservatively biased towards the hypothesis of no association. Nevertheless, the study of a young to middle-aged, unmedicated population with little subclinical heart disease strengthens the internal validity of the study results. Although we used an insulin assay that cross-reacts to some extent with proinsulin, among nondiabetic persons

AEP Vol. 8, No. 4 May 1998: 236–244

the concentration of proinsulin is low relative to insulin and, as shown by Haffner and coworkers (32), is unlikely to have materially altered our findings. Moreover, the dichotomization of fasting insulin in the analysis of covariance makes a meaningful misclassification unlikely. Another consideration is that we utilized anthropometric indices of obesity and its location, rather than direct methods such as CT or magnetic resonance imaging which are expensive and, in the case of CT, require irradiation of the study participants. The equations to predict percentage body fat have been validated predominantly in white populations. More work is needed in nonwhite populations to validate the use of anthropometric measures (33). Because hyperinsulinemia/insulin resistance is involved in blood pressure regulation, a number of important public health and clinical issues emerge. The combined effects of weight reduction and increased insulin sensitivity on blood pressure lowering have not been adequately quantified, nor is it clear whether increased insulin sensitivity reduces elevated blood pressure equally in overweight and lean persons. Although the present study cannot address this issue directly, other studies have indicated that weight loss and physical activity, both known to increase insulin sensitivity, can reduce blood pressure (34, 35). Given the high (and increasing) prevalence of obesity, especially among minority women (36), nationwide efforts in preventing and controlling obesity and encouraging moderate exercise through favorable alterations in insulin resistance, may very well have favorable consequences on reducing blood pressure (37).

Donahue et al. BLOOD PRESSURE, INSULIN, AND ETHNICITY

243

7. Boyko EJ, Leonetti D, Bergstrom RW, Newell-Morris L, Fujimoto WY. Visceral adiposity, fasting plasma insulin, and blood pressure in Japanese-Americans. Diabetes Care. 1995;18:174–181. 8. Every NR, Boyko EJ, Keane EM, Marshall JA, Rewers M, Hamman RF. Blood pressure, insulin, and c-peptide levels in San Luis Valley, Colorado. Diabetes Care. 1993;16:1543–1550. 9. Haffner SM, Valdez RA, Hazuda HP, Mitchell BD, Morales PA, Stern MP. Prospective analysis of the insulin-resistance syndrome (Syndrome X). Diabetes. 1992;41:715–722. 10. Falkner B, Kushner H, Tulenko T, Sumner AE, Marsh JB. Insulin sensitivity, lipids, and blood pressure in young American blacks. Arteriosclerosis. 1995;15:1798–1804. 11. Donahue RP, Skyler JS, Schneiderman N, Prineas RJ. Hyperinsulinemia and elevated blood pressure: Cause, confounder, or coincidence? Am J Epidemiol. 1990;132:827–834. 12. Jarrett RJ. In defense of insulin: A critique of Syndrome X. Lancet. 1992;340:469–471. 13. Haffner SM. Editorial: Insulin and blood pressure: Fact or fantasy? J Clin Endocrin and Metab. 1993;76:541–543. 14. Donahue RP, DeCarlo Donahue RA, Prineas RJ, Bean JA, Gutt M, Skyler JS, et al. Insulin sensitivity and blood pressure in a biethnic sample: The Miami Community Health Study. J Clin Epidemiol. 1996;49:859–864. 15. World Health Organization. Diabetes mellitus—Report of a WHO study group. WHO Publication No. 727. Geneva: World Health Organization; 1985. 16. Prineas RJ, Gillum RF, Horibe H, Hannan PJ. The Minneapolis Children’s Blood Pressure Study. Part 1. Standards of measurements for children’s blood pressure. Hypertension. 1980;2(Suppl.I):I-8–I-24. 17. Durnin JUGA, Womersley J. Body fat assessed from total body density and its estimation from skinfold thicknesses: Measurements on 481 men and women aged from 16 to 72 years. Br J Nutr. 1974;32:77–97. 18. Siri WE. Body composition from fluid spaces and density: Analysis of methods. In: Techniques for Measuring Body Composition. Washington, D.C.: National Academy of Sciences, National Research Council; 1961:223–244.

This research was supported by grant no. R01 HL 44600 from the National Institutes of Health. We acknowledge the technical assistance of Clara Bonney, Killiam Lopez, and Delia Stephens in the preparation of this manuscript.

19. Schutte JE, Townsend EJ, Hugg J, Shoup RF, Malina RM, Blomqvist CG. Density of lean body mass is greater in blacks than in whites. J Appl Physiol. 1984;56:1647–1649.

REFERENCES

21. Donahue RP, Barrett-Connor E, Orchard TJ, Gutai J. Summary of the workshop on endogenous insulin and sex hormones in atherosclerosis and coronary heart disease. Arteriosclerosis. 1988;8:544–548.

20. SAS Institute Inc. SAS user’s guide: Statistics, version 5. Cary, NC: SAS Institute; 1985.

1. Epstein FH. The epidemiology of essential hypertension. In: Robertson JIS. Handbook of Hypertension. Vol 1. Clinical Aspects of Essential Hypertension. Amsterdam: Elsevier; 1983:1–20.

22. Saad MF, Lillioja MB, Nyomba BL, Castillo C, Ferraro R, DeGregorio M, et al. Racial differences in the relation between blood pressure and insulin resistance. N Engl J Med. 1991;324:733–739.

2. MacMahon SW, Cutler JA, Neaton JD, Furberg CD, Cohen JD, Kuller LH, et al. Relationship of blood pressure to coronary and stroke morbidity and mortality in clinical trials and epidemiological studies. J Hypertension. 1986;4(4):S14–S17.

23. Falkner BF, Hulman S, Jushner H. Insulin-stimulated glucose utilization and borderline hypertension in young adult blacks. Hypertension 1993;22:18–25.

3. Hypertension Detection and Follow-up Program Cooperative Group: Five-year findings of the Hypertension Detection and Follow-up Program. I. Reduction of mortality of persons with high blood pressure, including mild hypertension. JAMA. 1979;242:2562–2571. 4. The Fifth Report of the Joint National Committee on Detection, Evaluation, and Treatment of High Blood Pressure. NIH Publication No. 93-1088, Bethesda, MD: National Heart, Lung, and Blood Institute; January, 1993.

24. Manolio TA, Savage PJ, Burke GL, Liu K, Wagenknecht LE, Sidney S, et al. Association of fasting insulin with blood pressure and lipids in young adults. Arteriosclerosis. 1990;10:430–436. 25. Asch S, Wingard DL, Barrett-Connor EL. Are insulin and hypertension independently related? Ann Epidemiol. 1991;1:215–230. 26. Saad MF, Knowler WC, Pettitt DJ, Nelson RG, Mott DM, Bennett PH. Insulin and hypertension: Relationship to obesity and glucose intolerance in Pima Indians. Diabetes. 1990;39:1430–1435.

5. Ferrannini E, Buzzigoli G, Bonadonna R, Giorico MA, Oleggini M, Graziadei L, et al. Insulin resistance in essential hypertension. N Engl J Med. 1987;317:350–357.

27. Johnson D, Prud’homme D, De´spres JP, Nodeau A, Tremblay A, Bouchard C. Relation of abdominal obesity to hyperinsulinemia and high blood pressure in men. Int J Obesity. 1992;16:881–890.

6. Modan M, Halkin H, Almog S, Lusky A, Eshkol A, Shefi M, et al. Hyperinsulinemia. A link between hypertension, obesity and glucose intolerance. J Clin Invest. 1985;75:809–817.

28. Haffner SM, Fong P, Hazuda HP, Pugh JA, Patterson JK. Hyperinsulinemia, upper body adiposity, and cardiovascular risk factors in nondiabetics. Metabolism. 1988;37:338–345.

244

Donahue et al. BLOOD PRESSURE, INSULIN, AND ETHNICITY

AEP Vol. 8, No. 4 May 1998: 236–244

29. Pouliot MC, De´pres JP, Lemieux S, Moorjani S, Bouchard C, Tremblery A, et al. Waist circumference and abdominal sagittal diameter: Best simple anthropometric indexes of abdominal visceral adipose tissue accumulation and related cardiovascular risk in men and women. Am J Cardiol. 1994;73:460–468.

34.

30. Donahue RP, Prineas RJ, DeCarlo Donahue RA, Skyler J. The female “insulin advantage” and effects of body fat: The Miami Community Health Study. Int J Obesity. 1996;20:76–82.

35.

31. Conway J, Yanovski SZ, Avila NA, Hubbard VS. Visceral adipose tissue differences in black and white women. Am J Clin Nutr. 1995;61:765–771.

36.

32. Haffner SM, Bowsher RR, Mykka¨nen L, Hazuda HP, Mitchell BD, Valdez RA, et al. Proinsulin and specific insulin concentration in highand-low risk populations for NIDDM. Diabetes. 1994;43:1490–1493. 33. Schreiner PJ, Terry JG, Evans GW, Hinson WH, Crouse III JR, Heiss

37.

G. Sex-specific associations of magnetic resonance imaging-derived intraabdominal and subcutaneous fat areas with conventional anthropometric indices: The Atherosclerosis Risk in Communities Study. Am J Epidemiol. 1996;144:335–345. Rocchini AP, Katch V, Shork A, Kelch RP. Insulin and blood pressure during weight loss in obese adolescents. Hypertension. 1987;10: 262–273. Krotkiewski M, Mandroukas K, Sjostrom L, et al. Effects of longterm physical training on body fat, metabolism, and blood pressure. Metabolism. 1979;28:650–658. Prevalence of overweight for Hispanics—United States, 1982–1984. JAMA. 1990;263:631–632. Torjesen PA, Birkeland KI, Anderssen SA, Hjermann I, Home I, Irdal P. Lifestyle changes may reverse development of the insulin resistance syndrome. The Oslo Diet and Exercise Study: A randomized trial. Diabetes Care. 1997;20:26–31.