Ten-year incidence of elevated blood pressure and its ... - Nature

Journal of Human Hypertension (1999) 13, 13–21  1999 Stockton Press. All rights reserved 0950-9240/99 $12.00 http://www.stockton-press.co.uk/jhh

ORIGINAL ARTICLE

Ten-year incidence of elevated blood pressure and its predictors: The CARDIA Study AR Dyer1, K Liu1, M Walsh2, C Kiefe3, DR Jacobs, Jr4 and DE Bild5 1

Department of Preventive Medicine, Northwestern University Medical School, Chicago, IL; 2College of American Pathologists, Skokie, IL; 3CARDIA Coordinating Center, University of Alabama at Birmingham, Birmingham, AL; 4Division of Epidemiology, School of Public Health, University of Minnesota, Minneapolis, MN; 5National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, MD, USA

Few prospective studies have examined associations of lifestyle factors or variables in the insulin resistance syndrome (syndrome X) with incidence of elevated blood pressure (BP) in black subjects and women. This report estimates the 10-year incidence of high blood pressure (HBP) and high normal blood pressure (HNBP) in the biracial cohort of the Coronary Artery Risk Development in (Young) Adults Study (CARDIA), and examines lifestyle factors and four syndrome X variables, measured at baseline, as predictors. CARDIA examined 5115 black and white men and women aged 18–30 years in 1985–1986, and re-examined them at 2, 5, 7, and 10 years. The 10-year incidence of HBP was 16.4% in black men, 7.8% in white men, 13.1% in black women, and 3.2% in white women, while the 10-year incidence of HBP or HNBP was 29.5%, 16.2%, 19.2%, and 6.3%, respectively, in the four sex–race subgroups. Predictors included body mass index, waist circumference, physi-

cal activity, alcohol intake, pulse rate, cigarette smoking, education, fasting insulin, triglycerides, uric acid, and high-density lipoprotein cholesterol, as well as age and systolic BP. In univariate analyses, each of these variables was significantly related to incidence in at least one of the four sex–race groups. In multivariate analyses that included control for age and systolic pressure, independent predictors included fasting insulin in white men and women, triglycerides in white men, uric acid and pulse rate in black men, waist circumference in white men and black women, and education (inverse) in white men and black and white women. These results suggest that lower socioeconomic status, as assessed by education level, and one or more syndrome X variables, ie, fasting insulin, triglycerides, uric acid, may be associated with development of elevated BP in young adults.

Keywords: blood pressure; insulin; triglycerides; uric acid; obesity; prospective study

Introduction A number of lifestyle factors are considered risk factors for hypertension, including excessive consumption of calories leading to overweight and obesity, physical inactivity, and excessive alcohol consumption.1–15 Heart rate and cigarette smoking have also been positively related to incidence of hypertension,3,5–8,12,14,16–18 although cigarette smoking has generally been inversely related to blood pressure (BP) in cross-sectional studies.19–27 Education and social status have been inversely related to BP and hypertension in both cross-sectional and prospective studies.12,16,23,28,29 Only four prospective studies examined one or more of these variables as possible risk factors for hypertension separately in black subjects9,12,16 or women.8,9,12 Hence, even though there is a consensus that overweight and obesity, physical Correspondence: Dr Alan R Dyer, Department of Preventive Medicine, Northwestern University Medical School, 680 N. Lake Shore Dr., Suite #1102, Chicago, IL 60611– 4402, USA Received 25 June 1998; revised and accepted 28 September 1998

inactivity, and excessive alcohol intake are risk factors for hypertension, there is a paucity of data on these lifestyle factors, and on heart rate, cigarette smoking, and education level, as predictors in black subjects and women. The role of insulin and insulin resistance in the development of hypertension has been the subject of intense interest and debate in recent years,30–40 and Reaven has defined a cluster of abnormalities in carbohydrate and lipid metabolism in association with hypertension called syndrome X.31,34 Syndrome X is characterised by resistance to insulinstimulated glucose uptake, glucose intolerance, hyperinsulinaemia, increased triglycerides, decreased high-density lipoprotein cholesterol (HDL-C), hyperuricaemia, and hypertension. The common feature of syndrome X is insulin resistance.31 Several prospective studies have examined one or more syndrome X variables as predictors of hypertension, including fasting insulin,13,14,41–44 triglycerHDL-C,13,14,42,44,45 and uric ides,8,13,14,42,44,45 6,18,45,46 acid, with independent positive associations for fasting insulin,41,43,44 triglycerides,8,13,42,44 and

Elevated blood pressure and its predictors AR Dyer et al

14

uric acid18,45 in some studies, but not all, and an independent inverse association for HDL-C in one study.13 However, of these variables only uric acid has been examined for an association with hypertension incidence separately in black and white subjects,9 and only three studies have examined associations in women.8,9,43 Hence, there is also a paucity of data on associations of these syndrome X variables with hypertension incidence in black subjects and women. The objectives of this report are: (1) to estimate and compare the 10-year incidence of high BP and high normal BP in black and white men and women from the Coronary Artery Risk Development in (Young) Adults Study (CARDIA), and (2) to examine lifestyle factors and four syndrome X variables, assessed at baseline, as predictors.

Subjects and methods Subjects CARDIA is a longitudinal investigation of the impact of lifestyle and other factors on evolution of coronary heart disease risk factors during young adulthood. Study design, recruitment of participants, and methods have been described elsewhere.47,48 In 1985–1986, 5115 black and white men and women aged 18–30 years were recruited and examined at four clinical sites in the USA: Birmingham, Alabama; Chicago, Illinois; Minneapolis, Minnesota; and Oakland, California. Participants were re-examined at 2, 5, 7, and 10 years post-baseline, with re-examination rates among surviving cohort members of 90.4%, 85.7%, 80.6%, and 78.5%, respectively. Examination procedures and laboratory methods Blood pressure was measured at each exam on the right arm using a Hawksley random zero sphygmomanometer (WA Baum Company, Copaigue, NY, USA) with the participant seated and following a 5min rest. Three measurements were taken at 1-min intervals. Systolic and diastolic pressures were recorded as Phase I and Phase V Korotkoff sounds. The average of the second and third measurements was taken as the pressure of record. A 30-sec pulse rate was recorded prior to the first measurement. Standardised questionnaires were used to collect self-report data on diagnosis and treatment of hypertension, diabetes, and other conditions. At baseline, participants were asked to fast for at least 12 h and not to smoke or engage in heavy physical activity for at least 2 h prior to the exam. Following BP measurement, venipuncture was performed with the participant seated, and blood was collected with minimal stasis for assessment of lipids, insulin, and uric acid. After plasma or serum separation, aliquots were stored at −70°C until shipped on dry ice to a central laboratory. Serum insulin was measured using an immunoassay technique at the University of Pittsburgh Diabetes Research Center.49 Plasma lipids and lipoproteins were measured by the University of Washington

Northwest Lipid Research Clinic Laboratory. Total triglycerides and HDL-C were determined using enzymatic procedures,50 and HDL-C was measured after dextran sulfate-magnesium precipitation.51 Uric acid was measured by American Bioscience Laboratories (Van Nuys, CA, USA) by the uricase method using a standard assay. Body weight (light clothing) was measured to the nearest 0.5 pound using a beam balance scale. Height without shoes was measured to the nearest 0.5 cm using a vertically mounted centimetre ruler and a metal carpenter’s square. Body mass index (BMI) was calculated as weight (kg) divided by height squared (m2). Waist circumference was measured at the minimal abdominal girth, in duplicate, to the nearest 0.5 cm. Interviewers using standardised questionnaires obtained data on years of education, number of cigarettes smoked per day, physical activity level (amount of moderate and heavy activity in the past year),52 and alcohol intake. For these analyses, education was dichotomized as 12 or fewer years of education versus more than 12, based on the greatest number of years reported at any of the five exams. Values from follow-up exams were included in the definition of the variable for education, because education was not complete at baseline for some cohort members due to their young ages. Baseline smoking was also dichotomized into current cigarette smoker versus not a current smoker. The quantity and type of alcohol containing drinks reported were used to calculate alcohol consumption as milliliters of ethanol per day.53 Statistical analysis End-point definitions: Ten-year incidence was determined for two end-points: (1) High BP, defined as first occurrence at any follow-up exam of systolic BP ⭓140 mm Hg, or diastolic BP ⭓90 mm Hg, or taking antihypertensive medication; and (2) either high BP or high normal BP, defined as systolic BP ⭓130 mm Hg, or diastolic BP ⭓85 mm Hg, or taking antihypertensive medication. In what follows, we refer to this latter end-point as elevated BP or EBP. The high BP end-point is based on BP cutpoints used in the fifth report of the Joint National Committee on Detection, Evaluation, and Treatment of High Blood Pressure (JNC V) to define hypertension based on BP readings taken on multiple occasions, while EBP is based on cutpoints used in JNC V to define high normal BP or hypertension.54 To be included in analyses of incidence rates, CARDIA participants had to have at least one follow-up exam at which both systolic and diastolic blood pressures were obtained and to be free of high BP or EBP at baseline. Analyses to estimate incidence of high BP were based on 4747 men and women, after excluding one person with incomplete BP measurement at baseline, 226 due to not having BP measured post-baseline, and 141 for high BP at baseline. Analyses to estimate incidence of EBP were based on 4494 men and women, after excluding also 253 persons for high normal BP at baseline. Life-table techniques were used to estimate 10-


year incidence for each end-point, using the four time intervals between exams, ie, 0–2 years, 2–5 years, 5–7 years, and 7–10 years.55 In these analyses, participants were censored at the exam at which high BP or EBP first occurred, or at their last followup exam. Participants were included in any followup interval in which BP was measured at the exam at the end of the interval, whether or not BP was measured at the previous follow-up exam. Participants who missed one or more follow-up exams, but had a subsequent exam, were excluded only from the time intervals in which the missed exam was at the end of the interval. Hence, participants could ‘drop-in’ or ‘drop-out’ in these life-table analyses, and ‘drop-ins’ were assumed not to have EBP at missed exams. Predictors of elevated blood pressure: Because the number of persons with high BP at follow-up was relatively small for some sex–race subgroups, analyses to determine predictors used EBP as the endpoint. For these analyses, persons were excluded from the cohort of 4494 used to estimate incidence of EBP, if they fasted fewer than 8 h at baseline (n = 131), were missing data on any of the variables examined as predictors (n = 140), had a fasting baseline glucose of 7.8 mmol/L (140 mg/dL) or greater (n = 8), or reported a previous diagnosis of diabetes at baseline (n = 20). Thus, 4195 black and white men and women were included. Exclusions for high fasting glucose values and self-reported diabetes reflect our interest in examining syndrome X variables as predictors. Multiple logistic regression was used for these analyses. Individuals were included in each 2–3 year follow-up interval in which they had a BP measurement at the end of the interval, if they had not yet met the definition of EBP in any prior interval. Hence, a person who had BP measured at baseline and at each follow-up exam and did not have EBP prior to the year 10 exam appeared four times in the data set. To account for the fact that an individual could appear more than once in the data set, separate constant terms were included in the logistic model for each interval of follow-up. Use of logistic regression in this manner provides results that are essentially equivalent to a Cox proportional hazards regression in which individuals with EBP at any exam are assumed to have the same length of followup, ie, are tied observations, and the logistic approach is used to handle these tied observations.56,57 Variables examined as predictors included age, systolic BP, BMI, waist circumference, physical activity, alcohol intake, pulse, cigarette smoking, education, fasting insulin, triglycerides, uric acid, and HDL-C. All variables were based on baseline values, except education which was based on the greatest number of years reported at any exam. Odds ratios and 95% confidence intervals were calculated for each variable in a univariate logistic model and then in a multivariate model that included all variables, except BMI. To provide further insight into associations of these variables with EBP and interrelationships among variables, odds ratios and con-

fidence intervals for five additional multivariate models were also obtained for each sex–race group. The first model included age, systolic pressure, waist circumference, physical activity, alcohol intake, pulse, cigarette smoking, and education, but not syndrome X variables. This model was fit because adjustment of associations of waist circumference and alcohol intake with EBP for syndrome X variables could represent overadjustment if these two variables contribute to increased levels of triglycerides, uric acid, HDL cholesterol, or fasting insulin.33,58–60 The four remaining models included each syndrome X variable alone, plus the eight variables listed above. These models were fit because the strong interrelationships among the syndrome X variables could obscure associations of these variables with EBP in the full multivariate model. All multivariate models included waist circumference, rather than BMI, because waist circumference was generally more strongly related to EBP than BMI. For continuous variables, odds ratios and confidence intervals were calculated for a one standard deviation increase in each variable, with standard deviations determined separately for each sex–race group. In accord with previous CARDIA reports, analyses were also conducted in which both insulin and triglycerides were log-transformed to reduce skewness.58,59 Because physical activity and alcohol intake also had very skewed distributions in each sex–race group, these variables were also log transformed after adding one unit and 1 ml/day, respectively, to each participant’s values. Because results of these analyses were very similar to those for untransformed data, and odds ratios were easier to interpret when the data were not transformed, only results without log-transformation of these four variables are reported here. In addition, because an association between alcohol intake and incidence of EBP might not be linear, as suggested by other studies,61 analyses were also conducted based on dichotomization of intake at 35 ml per day in men and 15 ml per day in women. These cutpoints were the approximate 85th percentiles for all men and all women, respectively.

Results Incidence of high blood pressure and elevated blood pressure Ten-year incidence of high BP was greater in black compared to white subjects and in men compared to women, and ranged from 3.2% in white women to 16.3% in black men (Figure 1). Similarly, 10-year incidence of EBP was also greater in blacks compared to whites, and in men compared to women, ranging from 6.3% in white women to 29.5% in black men (Figure 2). Predictors of elevated blood pressure Descriptive statistics for measurements and characteristics examined as predictors of EBP in the 4195 men and women are given by sex and race in Table 1.

15


16

Figure 1 Incidence at any follow-up exam of SBP ⭓140 mm Hg, DBP ⭓90 mm Hg, or of self-report of antihypertensive treatment.

Figure 2 Incidence at any follow-up exam of SBP ⭓130 mm Hg, DBP ⭓85 mm Hg, or of self-report of antihypertensive treatment.

In univariate analyses, among lifestyle and other variables, BMI, waist circumference, and cigarette smoking were positively related to 10-year incidence of EBP, and education and physical activity

inversely related, in each sex–race group (Table 2). Pulse and alcohol intake were positively related in three groups. These associations were statistically significant for BMI and waist circumference in all four groups, for cigarette smoking, education, and pulse in three, and for physical activity and alcohol intake in one. Among the four syndrome X variables, fasting insulin, triglycerides, and uric acid were positively related to EBP in each sex–race group, with the association being statistically significant in all four for fasting insulin, and in three for triglycerides and uric acid. HDL-C was also inversely and significantly related to incidence of EBP in three of the groups. In the full multivariate model (Table 3), waist circumference and cigarette smoking remained positively related to EBP in each sex–race group, with statistically significant associations for white men and black women for waist circumference. In the multivariate model that excluded the four syndrome X variables (Table 4), waist circumference was significantly related to EBP in all four groups, with susbtantially larger odds ratios than those for the full model (Table 3). Although associations of physical activity with EBP remained inverse in each group in the multivariate models, none of the associations was statistically significant. Education continued to have a significant inverse association with EBP in white men and in black and white women. Alcohol intake was significantly related to EBP in black men in the model that excluded the syndrome X variables, but not in the full model. In the other three sex–race groups, alcohol intake had a weak inverse relationship with EBP. Pulse rate was significantly related to EBP in black men in the full model, and in white men, if the syndrome X variables were excluded. In models that included BMI, rather than waist circumference, BMI was positively and significantly related to EBP in each sex–race group in the models that excluded the syndrome X variables. However, in the full model, BMI was not

Table 1 Physiologic measurements and characteristics by sex and race in participants without elevated blood pressurea (EBP) at baseline in 1985–1986: The CARDIA Study Variableb

Black men (n = 856)

Age (years) Systolic pressure (mm Hg) Diastolic pressure (mm Hg) Pulse (beats/30 sec) Physical activity score Alcohol intake (ml/day) Body mass index (kg/m2) Waist circumference (cm) Fasting insulin (pmol/L) Triglycerides (mmol/L) Serum uric acid (␮mol/L) HDL cholesterol (mmol/L) Education ⬎12 years (%) Cigarette smoker (%)

24.0 ± 3.8c 113.2 ± 8.6 69.0 ± 8.8 31.9 ± 4.7 529.3 ± 339.2 16.5 ± 28.5 24.2 ± 4.0 79.7 ± 9.0 72.1 ± 53.0 0.77 ± 0.47 354.8 ± 69.4 1.38 ± 0.34 64.0 36.7

White men (n = 950) 25.3 ± 3.4 112.2 ± 8.7 69.2 ± 8.2 33.4 ± 5.1 513.5 ± 305.2 18.6 ± 25.2 24.0 ± 3.3 82.1 ± 8.2 64.1 ± 37.2 0.97 ± 0.69 369.0 ± 67.8 1.22 ± 0.28 83.9 27.9

Black women (n = 1203) 24.3 ± 3.8 107.0 ± 8.6 66.2 ± 8.4 35.8 ± 5.3 277.3 ± 229.2 5.1 ± 12.1 25.4 ± 6.1 75.5 ± 12.1 92.5 ± 62.6 0.70 ± 0.34 259.6 ± 60.0 1.43 ± 0.33 74.0 31.0

Systolic pressure ⭓130 mm Hg or diastolic pressure ⭓85 mm Hg or antihypertensive medication. Baseline, except education, which is based on the greatest number of years reported at any of the five exams. c Mean ± standard deviation. a

b

White women (n = 1186) 25.4 ± 3.4 104.2 ± 8.4 65.6 ± 7.7 36.1 ± 5.5 400.2 ± 261.3 8.9 ± 15.6 23.0 ± 4.2 71.5 ± 8.8 65.2 ± 47.0 0.78 ± 0.43 268.9 ± 55.7 1.45 ± 0.34 85.3 26.1


Table 2 Univariate odds ratios (OR) and 95% confidence intervals (CI) for predictors of 10-year incidence of elevated blood pressure (EBP): The CARDIA Study Variable

ORa

95% CI

Black menb Age Systolic pressure Pulse Physical activity Alcohol intake Body mass index Waist circumference Insulin Triglycerides Uric acid HDL cholesterol Education ⬎12 yrs Cigarette smoker

1.15 2.38*** 1.25** 0.91 1.22*** 1.33*** 1.36*** 1.24*** 1.11 1.31*** 1.01 0.98 1.35*

(0.99, (1.99, (1.08, (0.79, (1.09, (1.17, (1.19, (1.12, (0.97, (1.14, (0.88, (0.72, (1.01,

1.32) 2.86) 1.43) 1.05) 1.36) 1.51) 1.54) 1.39) 1.26) 1.50) 1.16) 1.31) 1.79)

Black women Age Systolic pressure Pulse Physical activity Alcohol intake Body mass index Waist circumference Insulin Triglycerides Uric acid HDL cholesterol Education ⬎12 yrs Cigarette smoker

1.43*** 2.45*** 0.96 0.85 1.06 1.49*** 1.55*** 1.27*** 1.34*** 1.40*** 0.82* 0.65** 1.40*

(1.23, (2.07, (0.83, (0.73, (0.94, (1.31, (1.37, (1.13, (1.18, (1.23, (0.70, (0.47, (1.03,

1.67) 2.89) 1.12) 1.00) 1.20) 1.69) 1.75) 1.43) 1.52) 1.61) 0.96) 0.89) 1.90)

OR

95% CI

Table 3 Multivariate odds ratios (OR) and 95% confidence intervals (CI) for predictors of 10-year incidence of elevated blood pressure (EBP): The CARDIA Study Variablea

White men 1.15 2.49*** 1.34*** 0.84 1.03 1.66*** 1.76*** 1.50*** 1.35*** 1.50*** 0.68*** 0.57** 1.45*

(0.96, (2.00, (1.15, (0.69, (0.87, (1.43, (1.50, (1.32, (1.21, (1.26, (0.56, (0.37, (1.00,

1.37) 3.11) 1.57) 1.01) 1.22) 1.94) 2.06) 1.70) 1.50) 1.79) 0.83) 0.86) 2.09)

White women 1.12 2.75*** 1.53*** 0.55*** 0.94 1.62*** 1.67*** 1.37*** 1.28*** 1.23 0.70* 0.25*** 1.44

(0.86, (2.12, (1.22, (0.39, (0.70, (1.37, (1.42, (1.22, (1.12, (0.96, (0.53, (0.15, (0.84,

1.45) 3.56) 1.92) 0.77) 1.26) 1.92) 1.97) 1.54) 1.45) 1.57) 0.93) 0.42) 2.47)

a

Based on increase of approximately one standard deviation for continuous variables (see Table 1 for values), and 1.0 for dichotomous variables. b Number of persons with EBP: black men, 211; white men, 135; black women, 188; white women, 62. *P ⱕ 0.05; **P ⱕ 0.01; ***P ⱕ 0.001.

significantly related to EBP in any group (data not shown). Among the syndrome X variables, uric acid remained positively related to EBP in each sex–race group, with a significant association in black men in the full multivariate model, and in white men and black women, if uric acid was the only syndrome X variable included. Fasting insulin and triglycerides continued to have positive associations with EBP in three groups in the full model, with the association statistically significant for insulin for white men and women and for triglycerides in white men. Triglycerides was also significantly related to EBP in black women, if it was the only syndrome X variable included. While HDL-C generally continued to have inverse associations with EBP in these analyses, none of the associations was statistically significant. Analyses that included alcohol intake as a dichotomous variable generally provided results comparable to those in which alcohol intake was entered as a continuous variable. In black and white men, the odds ratios (95% confidence interval) associated with alcohol consumption of 35 ml or more per day were 2.19 (1.53, 3.15) and 1.14 (0.72, 1.80) in the univariate analyses, respectively, and 1.77 (1.19, 2.66) and 1.01 (0.61, 1.67) in the full multivariate

ORb

95% CI

OR

95% CI

Age Systolic pressure Pulse Physical activity Alcohol intake Waist circumference Insulin Triglycerides Uric acid HDL cholesterol Education ⬎12 yrs Cigarette smoker

Black men 1.12 (0.95, 1.32) 2.28*** (1.89, 2.75) 1.18* (1.01, 1.37) 0.92 (0.80, 1.07) 1.13 (1.00, 1.28) 1.08 (0.92, 1.28) 1.09 (0.95, 1.26) 0.88 (0.73, 1.05) 1.21* (1.03, 1.41) 0.99 (0.85, 1.15) 1.01 (0.72, 1.41) 1.35 (0.97, 1.87)

White men 1.14 (0.94, 1.39) 2.41*** (1.91, 3.04) 1.18 (0.98, 1.42) 0.89 (0.73, 1.09) 0.94 (0.77, 1.14) 1.23* (1.00, 1.51) 1.20* (1.03, 1.40) 1.18** (1.04, 1.34) 1.16 (0.96, 1.40) 0.93 (0.74, 1.15) 0.48** (0.30, 0.76) 1.51 (0.99, 2.29)

Age Systolic pressure Pulse Physical activity Alcohol intake Waist circumference Insulin Triglycerides Uric acid HDL cholesterol Education ⬎12 yrs Cigarette smoker

Black women 1.39*** (1.18, 1.64) 2.35*** (1.97, 2.81) 0.89 (0.76, 1.05) 0.94 (0.80, 1.11) 0.97 (0.83, 1.13) 1.19* (1.00, 1.40) 0.99 (0.83, 1.18) 1.14 (0.98, 1.32) 1.14 (0.97, 1.33) 0.93 (0.78, 1.11) 0.67* (0.48, 0.94) 1.29 (0.92, 1.81)

White women 1.17 (0.89, 1.54) 2.22*** (1.68, 2.93) 1.27 (0.97, 1.66) 0.79 (0.56, 1.12) 0.96 (0.73, 1.26) 1.11 (0.88, 1.41) 1.21* (1.03, 1.42) 1.11 (0.92, 1.33) 1.04 (0.79, 1.37) 1.04 (0.75, 1.43) 0.42** (0.23, 0.78) 1.21 (0.65, 2.24)

a

The multivariate model included all listed variables. Based on increase of approximately one standard deviation for continuous variables (see Table 1 for values), and 1.0 for dichotomous variables. *P ⱕ 0.05; **P ⱕ 0.01; ***P ⱕ 0.001.

b

model. In black and white women, odds ratios associated with consumption of 15 ml or more per day were 1.25 (0.77, 2.02) and 0.73 (0.36, 1.49) in univariate analyses, respectively, and 0.93 (0.55, 1.55) and 0.75 (0.34, 1.63) in the full model. Odds ratios and confidence intervals for other variables were generally little affected by whether alcohol intake was modelled as a continuous variable or as a dichotomous categorical variable.

Discussion Ten-year incidence rates of high BP and EBP were greater in blacks than whites and in men compared to women. Lifestyle factors as well as syndrome X variables were found to be predictors of EBP, with generally stronger associations for both sets of variables for white men and women compared to black men and women. While several lifestyle factors are considered risk factors for hypertension, including overweight and obesity, physical inactivity, and excessive alcohol consumption,1–15 and heart rate and cigarette smoking have been positively related to incidence of hypertension,3,5–8,12,14,16–18 and education inversely related,12 only four prospective studies have examined one or more of these variables as predictors of hypertension in black subjects9,12,16 or women.8,9,12 Similarly, while several prospective studies have examined one or more syndrome X variables as predictors of hypertension,6,8,13,14,18,41–46 only one prior

17


18

Table 4 Odds ratios (OR) and 95% confidence intervals (CI) for predictors of 10-year incidence of elevated blood pressure (EBP) in five multivariate models: The CARDIA Study Variablea

ORb

95% CI

Black men 1.05 (0.90, 1.23) 2.29*** (1.90, 2.75) 1.18* (1.02, 1.36) 0.92 (0.80, 1.07) 1.15* (1.02, 1.29) 1.16* (1.01, 1.34) 1.05 (0.76, 1.46) 1.27 (0.92, 1.77) 1.10 (0.96, 1.26) 0.95 (0.81, 1.12) 1.19* (1.03, 1.39) 1.00 (0.86, 1.16) Black women Age 1.40*** (1.19, 1.65) Systolic pressure 2.37*** (1.99, 2.81) Pulse 0.91 (0.77, 1.06) Physical activity 0.92 (0.78, 1.08) Alcohol intake 0.99 (0.85, 1.15) Waist circumference 1.29*** (1.13, 1.48) Education ⬎12 yrs 0.66* (0.48, 0.93) Cigarette smoker 1.32 (0.95, 1.84) Insulin 1.07 (0.91, 1.26) Triglycerides 1.18* (1.02, 1.35) Uric acid 1.17* (1.00, 1.36) HDL cholesterol 0.89 (0.75, 1.05) Age Systolic pressure Pulse Physical activity Alcohol intake Waist circumference Education ⬎12 yrs Cigarette smoker Insulin Triglycerides Uric acid HDL cholesterol

OR

95% CI

White men 1.13 (0.93, 1.37) 2.40*** (1.91, 3.02) 1.24* (1.04, 1.48) 0.90 (0.74, 1.09) 0.91 (0.75, 1.11) 1.55*** (1.31, 1.83) 0.47** (0.30, 0.75) 1.42 (0.95, 2.14) 1.24** (1.07, 1.44) 1.22*** (1.09, 1.37) 1.25* (1.04, 1.50) 0.81 (0.66, 1.00) White women 1.09 (0.84, 1.42) 2.27*** (1.73, 2.98) 1.24 (0.96, 1.61) 0.79 (0.56, 1.11) 0.96 (0.74, 1.24) 1.26* (1.04, 1.51) 0.41** (0.22, 0.75) 1.16 (0.64, 2.11) 1.21* (1.03, 1.42) 1.12 (0.94, 1.33) 1.08 (0.83, 1.41) 0.98 (0.71, 1.34)

a

Results for age through cigarette smoker are from models that included only these eight variables; results for insulin, triglycerides, uric acid, and HDL cholesterol are from models that included the first eight variables, plus the individual syndrome X variable. b Based on increase of approximately one standard deviation for continuous variables (see Table 1 for values), and 1.0 for dichotomous variables. *P ⱕ 0.05; **P ⱕ 0.01; ***P ⱕ 0.001.

prospective study examined associations separately in blacks and whites, and only three examined associations in women.8,9,43 Comparisons of results of these analyses with those of other longitudinal studies of hypertension incidence are difficult, since investigators have generally used different definitions of hypertension or EBP, and have controlled for different variables in multivariate analyses. In addition, some investigators have adjusted for baseline BP, while others have not. For these analyses, we used a definition of EBP that included high normal BP in addition to hypertension,54 and adjusted for baseline BP. We adjusted for baseline pressure even though this could represent an overadjustment, eg, if a higher baseline pressure was due to one or more of the variables studied here, because if not, prospective findings might simply reflect the baseline associations of these variables with BP. In these analyses, BMI and waist circumference were significantly related to EBP in each sex–race group in univariate analyses, and in multivariate analyses if syndrome X variables were excluded from the model. However, in the full model, while BMI and waist circumference were positively related in each group, none of the associations was statistically significant for BMI, and waist circumference was significantly related only in black women

and white men. In both univariate and multivariate analyses, associations with EBP were weakest in black men. Because of the known associations of obesity with syndrome X variables, and the view of some investigators that obesity is itself part of syndrome X,33 adjustment of the association of waist circumference with EBP for syndrome X variables could represent overadjustment, particularly if obesity contributes to higher values of insulin, triglycerides, and uric acid and to lower values of HDL-C. In the only other study to report the association of obesity with incidence of hypertension separately for black and white men and women, Ford and Cooper12 found BMI to be related to incidence of hypertension in each sex–race group with control for age, income, education, physical activity, alcohol intake, heart rate, total cholesterol, and serum magnesium, using data from the NHANES I Epidemiologic Follow-up Study.12 However, neither baseline BP nor syndrome X variables were included as covariates in these analyses. In a Kaiser Permanente study,9 BMI was positively related to incidence of hypertension in both blacks and whites, with a stronger association in whites. In the Meharry Cohort Study,17 a study of black male physicians, BMI was not significantly related to incidence of hypertension with control for baseline BP and other covariates. In CARDIA, while physical activity was inversely related to EBP in each sex–race group in both univariate and multivariate analyses, only the association in white women in univariate analyses was statistically significant. In the NHANES I Study,12 leisure time physical activity was significantly and inversely related to hypertension incidence in black men, white men, and black women in univariate analyses, but not multivariate analyses. In a previous report fron CARDIA,62 change in physical activity was also not significantly related to change in BP. The lack of strong associations between physical activity and BP in CARDIA could be due to attenuation caused by inaccuracy of self-reported physical activity data. Alcohol intake was most strongly related to EBP in black men, with significant associations in the univariate analysis and the multivariate analysis that did not adjust for syndrome X variables. Alcohol intake of 35+ ml per day was also significantly related to EBP in black men in both univariate and multivariate analyses. Alcohol intake was not related to incidence of EBP in white men or black and white women in univariate or multivariate analyses, whether treated as a categorical or continuous variable. In the NHANES I Study,12 alcohol consumption of three or more drinks per day was independently related to hypertension incidence only in white men, and in the Framingham Offspring Study,8 alcohol intake was a risk factor for hypertension in women, but not men. In the Kaiser Permanente study,9 alcohol intake of three or more drinks per day was significantly related to incidence of hypertension in whites, including women, but not blacks. The generally weak associations of alcohol intake with EBP incidence reported here are consistent with the previously reported weak cross-


sectional associations of alcohol intake with BP in CARDIA,53,62 and studies that suggest that the alcohol–BP association is weaker in young adults than middle-aged and older adults.53,61 In CARDIA, heart rate was positively and significantly related to EBP in black men, white men, and white women in univariate analyses. The association remained positive in these same three groups in multivariate analyses, with similar values for the odds ratio, even though the association was significant only in black men. In the NHANES I Study,12 heart rate was a risk factor for hypertension only in white men, and in the Meharry Cohort Study,17 heart rate was significantly related to hypertension incidence only with adjustment for age. In the Framingham Offspring Study,8 heart rate was an independent risk factor for hypertension in both men and women in analyses that did not include control for baseline BP. In another Kaiser Permanente study,18 a positive association between heart rate and incidence of hypertension could be completely explained by baseline BP. While cigarette smoking has generally been inversely related to BP in cross-sectional studies,19–27 it was positively related to hypertension incidence in the Israeli Ischemic Heart Disease Study6 and the Meharry Cohort Study17 without control for other risk factors. In CARDIA, cigarette smoking was positively related to EBP in each sex–race group in both univariate and multivariate analyses, with significant associations in univariate analyses in black men and women and white men. In CARDIA, the most consistent findings were those for education and EBP. Education of greater than 12 years, reported at baseline or any follow-up exam, was significantly and inversely related to EBP in white men and black and white women in both univariate and multivariate analyses. In the NHANES I Study,12 education was significantly and inversely related to hypertension incidence in these same sex–race groups in univariate analyses. In multivariate analyses with control for age, heart rate, BMI, alcohol intake, physical activity, serum cholesterol, serum magnesium, and household income, the association remained significant in white women and was of borderline significance in the other two sex–race groups. It is unclear why education was related to EBP in white men and women and black women, but not black men, in both CARDIA and NHANES I, although it could be that education is a poorer marker for socioeconomic status in black men compared to the other sex–race groups, or that socioeconomic status is simply not related to BP in black men. In the NHANES I Study,12 income was not a significant predictor of hypertension in black men or any other sex–race group. However, in the Charleston Heart Study,16 the incidence of hypertension was greater in black men categorised as being of low social class compared to men of higher social class. Among the syndrome X variables considered here, insulin, triglycerides, and uric acid were all positively related to EBP in each sex–race group in univariate analyses, with significant associations for insulin in all four groups and for triglycerides and

uric acid in three. HDL-C had a significant inverse association in three sex–race groups. In the full multivariate analyses, associations of insulin and triglycerides with EBP remained positive in three groups, with significant associations in white men and women for insulin and for triglycerides in white men. The associations of uric acid with EBP remained positive in each sex–race group, although the association was significant only in black men. HDL-C had little or no association with EBP in the multivariate analyses. Because the strong interrelationships among the four syndrome X variables could obscure associations of this group of variables with EBP, associations of these four variables with EBP were examined individually with adjustment only for the lifestyle and other variables. These analyses also indicated significant associations of uric acid with EBP in white men and black women and triglycerides with EBP in black women. Thus, in these analyses fasting insulin was related to EBP in white men and women, uric acid in black men and women and white men, and triglycerides in white men and black women. Other studies that have examined the associations of one or more of these syndrome X variables with hypertension incidence have been inconsistent,6,8,13,14,41–46 and no previous study examined associations in black and white men and women separately, although a Kaiser Permanente study18 reported that the association of uric acid with incidence of hypertension was stronger in women than men and in whites than blacks. A number of possible mechanisms have been proposed for an association of insulin and insulin resistance with BP and hypertension, including effects on the sympathetic nervous system, proliferation of vascular smooth muscle cells, ion transport, and increased renal sodium reabsorption.31–33,36,38 However, the role of insulin and insulin resistance in development of hypertension has been a subject of intense debate.30–40 Fasting insulin has not been consistently related to BP and hypertension in crosssectional studies,44,63,64 and experimental studies have failed to reveal a BP response to acute increases in plasma insulin levels.39,40 Studies also suggest that insulin and insulin resistance are positively related to BP only in persons with a positive family history of hypertension,65,66 ie, that insulin or insulin resistance may be a genetic marker for hypertension risk. In a study by Grandi et al,65 insulin resistance appeared to be related to a genetic pattern, independently of the presence of hypertension, ie, insulin was significantly higher and insulin sensitivity significantly lower in both normotensives and hypertensives with a positive family history of hypertension. Further, Weisser et al66 found that fasting insulin was positively correlated with BP only in persons with a family history of hypertension and diabetes mellitus, and in a study of twins, Mayer et al67 found that log fasting insulin was not related to BP in a matched analysis of monozygous pairs after adjustment for age and BMI. In this latter study, the partial correlation of log fasting insulin with BP was 0.00 for systolic BP and −0.02 for diastolic BP. Other studies have also found insulin

19


20

levels to be higher in normotensives with a positive family history of hypertension, compared to those without a positive history, even when mean BP levels were similar.68–70 In conclusion, the results from this study suggest that lower socioeconomic status as assessed by education level and one or more variables from the insulin resistance syndrome, ie, fasting insulin, triglycerides, and uric acid, may be associated with the development of elevated BP in young adult black and white men and women, with stronger associations in whites than blacks.

Acknowledgements This study was supported by Contracts NO1-HC48047, NO1-HC-48048, NO1-HC-48049, and NO1HC-48050 from the National Heart, Lung, and Blood Institute.

References 1 National High Blood Pressure Education Program Working Group Report on Primary Prevention of Hypertension. Arch Intern Med 1993; 153: 186–208. 2 Levine DM et al. Behavior changes and the prevention of high blood pressure. Workshop II. Circulation 1993; 88: 1387–1390. 3 Levy RL, White PD, Stroud WD, Hillman CC. Sustained hypertension: predisposing factors and causes of disability and death. JAMA 1947; 135: 77–80. 4 Kannel WB et al. The relation of adiposity to blood pressure and development of hypertension: The Framingham Study. Ann Intern Med 1967; 67: 48–59. 5 Paffenbarger RS Jr, Thorne MC, Wing AL. Chronic disease in former college students: VIII. Characteristics in youth predisposing to hypertension in later years. Am J Epidemiol 1968; 88: 25–32. 6 Kahn HA et al. The incidence of hypertension and associated factors: the Israel ischemic heart disease study. Am Heart J 1972; 84: 171–182. 7 Paffenbarger RS Jr, Wing AL, Hyde RT, Jung DL. Physical activity and incidence of hypertension in college alumni. Am J Epidemiol 1983; 117: 245–257. 8 Garrison RJ, Kannel WB, Stokes J III, Castelli WP. Incidence and precursors of hypertension in young adults: The Framingham Offspring Study. Prev Med 1987; 16: 235–251. 9 Friedman GD et al. Precursors of essential hypertension: body weight, alcohol and salt use, and parental history of hypertension. Prev Med 1988; 17: 387– 402. 10 Selby SV, Friedman GD, Quesenberry CP Jr. Precursors of essential hypertension: the role of body fat distribution pattern. Am J Epidemiol 1989; 129: 43–53. 11 Paffenbarger RS Jr, Jung DL, Leung RW, Hyde RT. Physical activity and hypertension: An epidemiological view. Ann Med 1991; 23: 319–327. 12 Ford E, Cooper RS. Risk factors for hypertension in a national cohort study. Hypertension 1991; 18: 598– 606. 13 Haffner SM, Ferrannini E, Hazuda HP, Stern MP. Clustering of cardiovascular risk factors in confirmed prehypertensive individuals. Hypertension 1992; 20: 38– 45. 14 Shetterly SM, Rewers M, Hamman RF, Marshall JA. Patterns and predictors of hypertension incidence among Hispanics and non-Hispanic whites: the San Luis Valley Diabetes Study. J Hypertens 1994; 12: 1095–1102.

15 Blair SN, Goodyear NN, Gibbons LW, Cooper KH. Physical fitness and incidence of hypertension in healthy normotensive men and women. JAMA 1984; 252: 487– 490. 16 Keil JE, Tyroler HA, Sandifer SH, Boyle E Jr. Hypertension: effects of social class and racial admixture. The results of a cohort study in the Black population of Charleston, South Carolina. Am J Public Health 1977; 67: 634 –639. 17 Thomas J et al. Risk factors and the incidence of hypertension in black physicians: The Meharry Cohort Study. Am Heart J 1985; 110: 637–645. 18 Selby JV, Friedman GD, Quesenberry CP Jr. Precursors of essential hypertension: pulmonary function, heart rate, uric acid, serum cholesterol, and other serum chemistries. Am J Epidemiol 1990; 131: 1017–1027. 19 Sive PH et al. Distribution and multiple regression analysis of blood pressure in 10,000 Isreali men. Am J Epidemiol 1971; 93: 317–327. 20 Stamler J et al. Multivariate analysis of the relationship of seven variables to blood pressure: findings of the Chicago Heart Association Detection Project in Industry, 1967–1972. J Chron Dis 1975; 28: 527–548. 21 Havlik RJ et al. Evidence for additional blood pressure correlates in adults 20–56 years old. Circulation 1980; 61: 710–715. 22 Reed D, McGee D, Yano K. Biological and social correlates of blood pressure among Japanese men in Hawaii. Hypertension 1982; 4: 406– 414. 23 Abernathy JR et al, for the USA-USSR Steering Committee for Cardiovascular Area I: Pathogenesis of Atherosclerosis. Correlates of systolic and diastolic blood pressure in men 40 to 59 years of age sampled from United States of America and Union of Soviet Socialist Republics Lipid Research Clinics Populations. Am J Cardiol 1988; 61: 1071–1075. 24 Bonaa KH, Thelle DS. Associations between blood pressure and serum lipids in a population. Circulation 1991; 83: 1305–1314. 25 Hughes K et al. Relationships between cigarette smoking, blood pressure, and serum lipids in the Singapore general population. Int J Epidemiol 1993; 22: 637–643. 26 Huang Z et al, for the PRC-USA Cardiovascular and Cardiopulmomary Epidemiology Research Group. A north-south comparison of blood pressure and factors related to blood pressure in the People’s Republic of China: a report from the PRC-USA Collaborative Study of Cardiovascular Epidemiology. J Hypertens 1994; 12: 1103–1112. 27 Masana L, Farinaro E, De Henauw S, Nicaud V. Blood pressure in young adults with and without a parental history of premature coronary heart disease in Europe: the EARS study. J Hum Hypertens 1996; 10: 207–213. 28 Dyer AR, Stamler J, Shekelle RB, Schoenberger JA. The relationship of education to blood pressure: findings on 40,000 employed Chicagoans. Circulation 1976; 54: 987–992. 29 Stamler R et al, in behalf of the INTERSALT Cooperative Research Group. Higher blood pressure in adults with less education: some explanations from INTERSALT. Findings of the INTERSALT Study. Hypertension 1992; 19: 237–241. 30 Ferrannini E et al. Insulin resistance in essential hypertension. N Engl J Med 1987; 317: 350–357. 31 Reaven GM. Role of insulin resistance in human disease. Diabetes 1988; 37: 1595–607. 32 Reaven GM, Hoffman BB. Hypertension as a disease of carbohydrate and lipoprotein metabolism. Am J Med 1989; 87 (Suppl 6A): 2S–5S. 33 DeFronzo RA, Ferrannini E. Insulin resistance: a multifaceted syndrome responsible for NIDDM, obes-


34 35

36

37 38 39 40 41

42

43 44 45

46 47 48 49 50 51

52

ity, hypertension, dislipidemia, and atherosclerotic cardiovascular disease. Diabetes Care 1991; 14: 173– 194. Reaven GM. Role of insulin resistance in human disease (syndrome X): an expanded definition. Annu Rev Med 1993; 44: 121–131. Schmidt MI et al, for the Atherosclerosis Risk in Communities Study Investigators. Clustering of dyslipidemia, diabetes, and hypertension and its association with fasting insulin and central and overall obesity in a general population. Metabolism 1996; 45: 699–706. Donahue RP, Skyler JS, Schneiderman N, Prineas RJ. Hyperinsulinemia and elevated blood pressure: cause, confounder, or coincidence? Am J Epidemiol 1990; 132: 827–836. Jarrett RJ. In defence of insulin: a critique of syndrome X. Lancet 1992; 340: 469– 471. Haffner SM. Editorial: Insulin and blood pressure: fact or fantasy? J Clin Endocrinol Metab 1993; 76: 541–543. Anderson EA, Mark AL. The vasodilator action of insulin: Implications for the insulin hypothesis of hypertension. Hypertension 1993; 21: 136–141. Hall JE. Renal and cardiovascular mechanisms of hypertension in obesity. Hypertension 1994; 23: 381– 394. Skarfors ET, Lithell HO, Selinus I. Risk factors for the development of hypertension: a 10-year longitudinal study in middle-aged men. J Hypertens 1991; 9: 217– 223. Niskanen LK, Uusitupa MI, Pyorala K. The relationship of hyperinsulinemia to the development of hypertension in Type 2 diabetic patients and in non-diabetic subjects. J Hum Hypertens 1991; 5: 155–159. Lissner L et al. Fasting insulin in relation to subsequent blood pressure changes and hypertension in women. Hypertension 1994; 20: 797–801. Haffner SM, Miettinen H, Gaskill SP, Stern MP. Metabolic precursors of hypertension: The San Antonio Heart Study. Arch Int Med 1996; 156: 1994 –2000. Hunt SC, Stephenson SH, Hopkins PN, Williams RR. Predictors of an increased risk of future hypertension in Utah. A screening analysis. Hypertension 1991; 17: 969–976. Medalie JH, Goldbourt U. Precursors of hypertension in a general population. Circulation 1977; 56: III-43 (Abstract). Friedman GD et al. CARDIA: study design, recruitment, and some characteristics of the examined subjects. J Clin Epidemiol 1988; 41: 1105–1116. Cutter GR et al. Cardiovascular risk factors in young adults: the CARDIA baseline monograph. Control Clin Trials 1991; 12 (Suppl 1): 1S–77S. Herbert V, Lau KS, Gottlieb CW. Coated charcoal immunoassay of insulin. J Clin Endocrinol 1965; 25: 1375–1384. Warnick GR. Enzymatic methods for quantification of lipoprotein lipids. Methods Enzymol 1986; 129: 101– 123. Warnick GR, Benderson J, Albers JJ. Dextran sulphateMg2+ precipitation procedure for quantification of high-density lipoprotein cholesterol. Clin Chem 1982; 28: 1379–1388. Taylor HL et al. A questionnaire for the assessment of

53 54

55 56 57 58 59

60

61 62

63 64 65 66

67

68 69

70

leisure time physical activities. J Chron Dis 1978; 31: 741–755. Dyer AR et al. Alcohol intake and blood pressure in young adults: The CARDIA Study. J Clin Epidemiol 1990; 43: 1–13. Joint National Committee on Detection, Evaluation, and Treatment of High Blood Pressure. The fifth report of the Joint National Committee on Detection, Evaluation, and Treatment of High Blood Pressure. Arch Intern Med 1993; 153: 154 –183. Kahn HA, Sempos CT. Statistical methods in epidemiology Oxford University Press: New York (NY), 1989. Kalbfleisch JD, Prentice RL. The statistical analysis of failure time data. John Wiley and Sons: New York (NY), 1980. SAS Institute Inc. SAS/STAT user’s guide, release 6.03 edition. SAS Institute Inc.: Cary (NC), 1988. Manolio TA et al. Association of fasting insulin with blood pressure and lipids in young adults: The CARDIA Study. Arteriosclerosis 1990; 10: 430– 436. Folsom AR et al. Relation of body fatness and its distribution of cardiovascular risk factors in young blacks and whites: the role of insulin. Am J Epidemiol 1989; 130: 911–924. National Research Council, Committee on Diet and Health, Food and Nutrition Board, Committee on Life Sciences. Diet and Health: Implications for Reducing Chronic Disease. National Academy Press: Washington, DC, 1989, pp 431– 464. MacMahon S. Alcohol consumption and hypertension. Hypertension 1987; 9: 111–121. Liu K et al. Blood pressure in young blacks and whites: relevance of obesity and lifestyle factors in determining differences. The CARDIA Study. Circulation 1996; 93: 60–66. Muller DC et al. An epidemiological test of the hyperinsulinemia-hypertension hypothesis. J Clin Endocrinol Metab 1993; 76: 544 –548. Burchfiel CM et al. Distribution and correlates of insulin in elderly men: The Honolulu Heart Program. Arterioscler Thromb Vasc Biol 1995; 15: 2213–2221. Grandi AM et al. Hyperinsulinemia, family history of hypertension, and essential hypertension. Am J Hypertens 1996; 9: 732–738. Weisser B et al. Plasma insulin is correlated with blood pressure only in subjects with a family history of hypertension or diabetes mellitus: results from 11 001 participants in the Heureka Study. J Hypertens 1993; 11 (Suppl 5): S308–S309. Mayer EJ et al. Genetic and environmental influences on insulin levels and the insulin resistance syndrome: an analysis of women twins. Am J Epidemiol 1996; 143: 323–332. Ferrari P et al. Altered insulin sensitivity, hyperinsulinema, and dislipidemia in individuals with a hypertensive parent. Am J Med 1991; 91: 589–596. Facchini F et al. Insulin resistance, hyperinsulinemia, and dislipidemia in nonobese individuals with a family history of hypertension. Am J Hypertens 1992; 5: 694 –699. Neutel JM et al. Heredity and hypertension: impact on metabolic characteristics. Am Heart J 1992; 124: 435– 440.

21