Regional Disparities in the Incidence of Elevated Blood Pressure Among Young Adults
The CARDIA Study
Background Within the United States, little is known about regional disparities in blood pressure (BP), their changes over time, or explanations for their existence.
Methods and Results A population-based cohort of 5115 black and white men and women, 18 to 30 years old in 1985-1986 (balanced on age, race, sex, and education), was followed up for 7 years in four centers: Birmingham, Ala; Chicago, Ill; Minneapolis, Minn; and Oakland, Calif. Differences in elevated BP (EBP) prevalence among centers at years 0, 2, 5, and 7 and in 7-year incidence of EBP were assessed. Sociodemographic and dietary variables, physical activity, weight, smoking, and alcohol were considered. At year 0, no regional differences were seen. Seven years later, there was marked variability in prevalence of EBP overall and for both black and white men, from a low in Chicago (9% for black men and 5% for white men) to a high in Birmingham (25% for black men and 14% for white men). Birmingham also had the highest 7-year incidence (11%) and overall prevalence at year 7 (14%). The adjusted odds ratios, with Birmingham as referent (95% CIs), for 7-year incidence of EBP overall were 0.38 (0.24, 0.60) for Chicago, 0.37 (0.24, 0.57) for Minneapolis, and 0.74 (0.52, 1.07) for Oakland.
Conclusions Regional disparities are absent at baseline but become apparent as the cohort ages. These differences are not fully explained by the available behavioral and sociodemographic characteristics.
Sizable regional differences in morbidity and mortality from CVD both among and within countries have been reported.1 2 3 4 5 6 7 8 9 10 11 12 13 14 Multiple possible contributing factors ranging from geographic bias in classifications on death certificates1 through physical properties of the environment, such as water hardness and climate,2 3 have been investigated, but none of these have been shown to account fully for these observed differences.
The Stroke Belt concept, first proposed in the mid-1960s,4 highlights a concentration of high stroke mortality for the southeastern United States.5 6 7 8 Because hypertension is a well-established risk factor for stroke, its geographic variability is a plausible contributor to the observed variability in end-organ damage, such as ischemic heart disease and stroke. Indeed, although they are scarce, there are reports of regional differences in the prevalence of hypertension. For example, data from the 1976-1980 National Health and Nutrition Examination Survey (NHANES II) showed increased prevalence of hypertension for black women in the southeastern United States.9 Also, marked geographic variability in the relationship between SBP and age was first reported almost three decades ago.10 Recent data from INTERSALT, a cross-sectional study in 32 countries, support this concept.11 12 13 In striking contrast, little is known about the amount and possible causes of regional variability in the incidence of hypertension in the United States.
In this paper, we consider differences in prevalence and in 7-year incidence of EBP for black and white young adults among the four sites of the CARDIA Study. We do so before and after adjustment for pertinent sociodemographic, behavioral, dietary, and other correlates of BP. An important issue is whether the observed regional differences can be attributed to factors known to be associated with BP levels or whether other factors play important roles.
The CARDIA study is a prospective, multicenter, epidemiological study of the evolution of CVD risk factors in young adults. In 1985-1986, 5115 black and white men and women 18 to 30 years old were recruited and examined at Birmingham, Ala; Chicago, Ill; Minneapolis, Minn; and Oakland, Calif. For the Birmingham, Minneapolis, and Chicago centers, participants were randomly selected from communities or from specific census tracts. For the Oakland center, participants were randomly selected from enrollees in the Kaiser Permanente Medical Care Program. Recruitment was stratified to achieve nearly equal numbers at each center by race (black, white), sex, education (high school or less, more than high school), and age (18 to 24 years, 25 to 30 years). More detailed descriptions of the sampling plan and initial cohort characteristics have been published.14 15 Participants were reexamined in year 2 (1987-1988, n=4624), year 5 (1990-1991, n=4352), and year 7 (1992-1993, n=4084); 43 participants had died by year 7. The 7-year follow-up rate was 80%. Follow-up was >75% among blacks and those with less education or income.
Data Collection and Quality Control
Examinations were performed on fasting participants in the morning. Caffeine, food, heavy physical activity, smoking (30-minute proscription), and alcohol intake were proscribed for ≥2 hours before BP measurement. Early in the examination, after a 5-minute rest, BP was measured at three 1-minute intervals on the right arm of seated subjects with a random-zero sphygmomanometer. First- and fifth-phase Korotkoff sounds were recorded for SBP and DBP, respectively. Means of the second and third BP measurements were used in analyses.
Age, sex, race, years of education, alcohol intake, cigarette use, antihypertensive medication use, oral contraceptive use, and family history of hypertension were ascertained by questionnaire at each examination. Questionnaires assessed physical activity over the preceding year at each examination16 and dietary history (1-month recall period) at baseline and year 7.17 Dietary variables were total daily caloric intake, total protein, total fat, saturated fatty acids, monounsaturated fatty acids, dietary cholesterol (in % kcal), sodium, potassium, magnesium, and calcium (in mg/1000 kcal), amount and frequency of salt added in cooking, and frequency of salt added at the table. Height and weight were measured without shoes and outer garments, and BMI (weight [kg]/height [m]2) was calculated.
At each examination, all data collection personnel were trained and certified centrally and subsequently recertified.18 In particular, BP measurement training and certification included pretraining and posttraining tests, audiocassette tapes, and Y-tube stethoscope testing. Duplicate BP measurements for quality control were conducted, and terminal digit preferences were addressed when necessary. We inspected technician-specific single-point histograms depicting the distributions of SBP, DBP, and pulse pressure (SBP−DBP). Although there was variability in the BP distributions from technician to technician, as expected, there were no apparent differences between centers in this variability. Potential technician outliers were identified for further analyses.
Because of the age of our participants, the prevalence of hypertension was low. Therefore, we focused on EBP, ie, high-normal and greater levels based on the Fifth Report of the Joint National Committee on Detection, Evaluation, and Treatment of Hypertension (JNC V).19 Subjects were defined as having EBP if the SBP was ≥130 mm Hg or DBP was ≥85 mm Hg or if they reported the use of antihypertensive medication at the time of the examination.
Prevalence of EBP was calculated for each examination, with the number of individuals at that examination as the denominator. The 7-year incidence of EBP was calculated for those seen at both year 0 and year 7 examinations by the ratio of the number of participants seen at year 7 who had EBP at year 7 but not at year 0 divided by the number of participants seen at year 7 who did not have EBP at year 0. We compared the 7-year incidences and prevalences of EBP among the four centers for each year and each race/sex group by use of χ2 statistics.
We performed multivariable adjustment of center-specific differences in EBP incidence with MLR, with EBP at year 7 as the dependent variable.20 Other variables were site (center); age (years); sex; race; years of education (four categories); alcohol intake (mL/wk); cigarette use (current versus former or never smoker); current oral contraceptive use; family history of hypertension in mother, father, or sibling (collected at years 0 and 5); physical activity (continuous score); BMI; and the dietary variables described above. Seven-year changes in BMI, physical activity, and total caloric intake were used as additional covariates. Backward stepwise regression, with P<.20 as the criterion for keeping variables in the final model, was used. Age and indicator variables for each site and race/sex group (when appropriate) were retained in all models.
Participants who changed centers between examinations were assigned to the center at which they underwent the year 7 examination for the year 7 prevalence and incidence analyses; 1.8% of participants were examined at year 7 at a center different from year 0. All analyses were repeated, restricted to those living within 300 miles of their year 7 center (88% of year 7 participants). This produced no significant changes.
To further analyze whether observed center differences might be due to technician effects, we created an indicator variable for each technician and reran our MLR models. Technicians who were potential outliers on the basis of bivariate analyses were initially kept in the models. Center-specific differences in 7-year prevalence or incidence of EBP were not affected; hence, the final models do not include these covariates. Further, we used MLR models (without covariates for technicians) to compute, for each technician, the prevalence of EBP expected from the final model for this technician’s subjects. We then compared this expected prevalence with the prevalence of EBP observed for the same technician’s subjects. Only for one technician, who had measured BP on 309 subjects at year 7, was there a significant difference between observed and expected. We entered an indicator variable for this technician into the final models and reran all analyses. The results did not change.
Although the main dependent variable was EBP, center differences for SBP and DBP in years 0, 2, 5, and 7 were compared by one-way ANOVA. All analyses were performed overall and for race/sex groups. Results were consistent with the previous outcomes. To account for possible regional differences in hypertension management, we repeated all analyses after excluding subjects on antihypertensive medications at year 0 and/or year 7. This did not change our results.
There were no baseline differences in EBP, mean SBP, or DBP between subjects seen and not seen at year 7, either overall or at any of the four centers. Because hypertension in the United States becomes increasingly manifest at older ages, we repeated the above analyses restricted to participants ≥25 years old at baseline to see whether geographic variability patterns were accentuated. Analyses were originally performed with STATA Statistical Software, version 4.0,21 and then verified with SAS, Version 6.10.22
Mean age of the cohort at year 7 was 32 years, with no differences among centers. Overall, 63% of the cohort reported a family history of hypertension, with the highest proportion being from Birmingham (67%), especially black women (80%) (Table 1⇓). Only 2% of the cohort reported active antihypertensive medication use, with the highest overall in Birmingham, especially for black men. There were regional differences in behavioral factors, such as smoking and physical activity. Of special note, Birmingham showed the highest sodium intake and lowest potassium and magnesium intakes, both overall and in all race/sex groups. Cohort characteristics at the other examination times were similar to year 7 patterns (data not shown).
At baseline (year 0), unadjusted overall EBP prevalence for the entire cohort ranged from 6.3% in Chicago to 9.3% in Oakland (Fig 1⇓). At years 2, 5, and 7, EBP prevalence was always highest in Birmingham. By years 5 and 7, Chicago and Minneapolis exhibited low EBP prevalences. By year 7 (Fig 2⇓), there were marked differences among sites for both black men (Birmingham was highest at 25%, followed by Oakland at 20%) and white men (Birmingham and Oakland very close at 14% and markedly higher than Chicago at 5% and Minneapolis at 7%). Despite the very low prevalence of EBP in white women at year 0, it appears that prevalence of EBP for white women increased most for Birmingham.
Similarly, 7-year EBP incidence was highest overall in Birmingham, especially for black men (Fig 3⇓). For white men, the rates in Birmingham and Oakland were comparable and higher than for Chicago or Minneapolis. Although not statistically significant, EBP incidence was highest in Birmingham for both black and white women. We repeated bivariate prevalence and incidence analyses restricted to those 25 to 30 years old at baseline (data not shown). Seven-year incidence patterns were accentuated for the entire cohort and also for each race/sex group: for example, for older black men, 7-year EBP incidence was 25% in Birmingham and 3% in Chicago.
We adjusted EBP incidence for age (and race/sex when appropriate) only and also developed fully adjusted models (Table 2⇓). Using Birmingham for reference, the adjusted ORs for 7-year incidence of EBP were significantly <1 in Chicago (age/sex/race only adjusted OR=0.36; fully adjusted OR=0.38) and Minneapolis (ORs=0.42 and 0.37, respectively). The ORs for Oakland were 0.68 and 0.74 (P=NS). Within each race/sex group, findings were similar in black men and white men (Table 2⇓). In black women, the fully adjusted ORs were 0.75, 0.65, and 0.55 in Chicago (P=NS), Minneapolis (P=NS), and Oakland (P=.04), respectively. In white women, these ORs were 0.74, 0.60, and 1.00, all P≥.05. Of note, the coefficients for sodium intake were not significant in any of the models at P<.20, and therefore, sodium was eliminated by backward stepping. Because of the clinical importance of sodium intake, we also forced it (in mg/1000 kcal) into the models: the fully adjusted ORs for the cities did not change, and therefore we display the final models without sodium. Consideration of salt added in cooking or at the table also did not change these results.
Our data show striking differences in 7-year incidence of EBP among the four study sites. The differences in EBP prevalence among centers at year 0 could be due to chance; by year 7, however, these differences were statistically significant overall, as well as for black and white men separately. Multivariable analyses suggest that known dietary, behavioral, and familial correlates of hypertension as measured in CARDIA do not fully explain these regional disparities.
Differences among the sites for women were not statistically significant and were less pronounced than for men, albeit qualitatively similar. The lower prevalence and incidence of EBP in women are consistent with hypertension developing later in US women than in men.23 This “age lag” may explain why differences are only suggested for women in our study. Sex differences in geographic variability of BP have also been noted in Britain.24 25
Our reanalyses focusing on those 25 to 30 years old at baseline are consistent with regional disparities in EBP being more apparent for older ages. We hypothesize that these differences (1) result from the geographic variability exhibited by the combination of genetic and environmental factors that causes hypertension and (2) represent the effect of aging on those members of the population who will ultimately become hypertensive. In fact, we see marked similarity between Chicago and Minneapolis, the two Midwestern sites, and thus interpret our data as showing differences in EBP rates between three regions (Southeast, Midwest, and Pacific Coast) rather than four sites. Perhaps differential migratory patterns among blacks from the South to the Midwest, rather than the Pacific Coast, might explain part of the differences. However, although some studies have addressed the relationship between migration and hypertension,26 we are unaware of data substantiating this conjecture.
Surprisingly, the role of differences in BP levels and the incidence of hypertension among regions in explaining the geographic variability of CVD has received little attention in the United States. The Cardiovascular Health Study, another current US four-center study, albeit of a population significantly older than that of CARDIA, has also noted considerable differences in SBP and DBP by study site that are unexplained by other predictors of hypertension.27 In particular, residence in the easternmost site (Washington County, Maryland) was associated with consistently lower BP. Also, the Strong Heart Study has shown a lower prevalence of hypertension among American Indians in North and South Dakota than in Arizona and Oklahoma.28 In addition, vital statistics show regional variability in death rates from hypertensive disease, with highest rates in the South Atlantic and Pacific regions, especially for nonwhites (mainly blacks) in the South Atlantic region.5 6 Similarly, data from the Department of Veterans Affairs show increased all-cause mortality among hypertensives followed up in clinics in the southeastern United States.29 Conversely, attempts to explain the increased stroke mortality in the Stroke Belt have shown higher prevalence of hypertension in the Southeast than in the rest of the country for black women only,9 even though severe hypertension was higher for both male and female blacks.
In Britain3 25 30 31 32 and France,31 geographic variation in BP has been studied more extensively and found to be established in the population by age 25 to 29 years.24 Also, data from seven European countries suggest a possible dependence of the association between SBP and ischemic heart disease on geographic area, with a stronger impact of high SBP on ischemic heart disease mortality in southern Europe than in northern Europe.8 Furthermore, worldwide differences in the distribution of BP have been studied since at least the 1920s and were summarized by Epstein and Eckoff in 1967.10 At that time, the concept that the relationship between SBP and age varies across populations was first proposed. This concept has recently been strengthened by the INTERSALT study11 12 13 of 52 centers in 32 countries, which included US sites in three states: Illinois, Mississippi, and Hawaii. For these sites, BPs were lower in Chicago (Illinois) and higher in the Mississippi and Hawaii populations.12 INTERSALT demonstrated that BMI, heavy alcohol intake, and potassium excretion have significant independent associations with BP in individual subjects, but cross-center comparisons were inconsistent.13 For sodium excretion, there were consistent within- and cross-population associations with BP. These regression coefficients were at least three times higher for the older (40 to 59 years) than the younger (20 to 39) population.11 To the best of our knowledge, INTERSALT data have not been analyzed directly to show to what extent among-center differences in BP or hypertension are explained by correlates of hypertension.
The Stroke Belt concept and the known geographic variability in CVD suggest that geographic variation in hypertension (and other CVD risk factors) contributes to the geographic variability in its end-organ damage. The Stroke Belt may be beginning to “shift,” and temporal changes in the location of areas with the highest stroke mortality rates are being seen.6 7 8 This suggests that there are environmental factors apart from geography per se that contribute to regional disparities. In fact, analyses performed on British populations show that lifestyle and environmental factors contribute to but do not fully explain geographic variability in BP.3 25
Our study has several limitations. First, we have studied four urban areas only. Nonetheless, our data show differences among four sites (or three regions), consistent with the increased incidence of stroke in the Stroke Belt and observed patterns of hypertension-related CVD mortality.5 6 Second, despite our efforts in technician training and quality control, technician effects may have contributed to the observed differences. Adjustment for individual technicians, however, did not change the outcome. Still, intraindividual variability and measurement error, together with other unknown or unmeasured potential explanatory variables, may add to the inability of the models to explain all observed differences. Third, cohort retention was high but not complete; however, analyses of appropriate subsets make it unlikely that loss to follow-up biased our results. Finally, because incidence of EBP is low for young women, we may not have had enough power to detect center differences among women. Similarly, some variables known to be associated with hypertension, such as sodium or alcohol intake, may not have reached statistical significance in our models because of lack of power.
Geographic differences in EBP persisted after adjustment for correlates of hypertension: BMI and weight gain, physical activity and its change over 7 years, dietary intake, alcohol and tobacco use, education, oral contraceptive use, and family history of hypertension. In fact, adjustment for all these factors had only limited impact on geographic differences, as shown in Table 2⇑. What, then, could account for the differences among sites that we have found? Hypertension is a multifactorial disorder, resulting from a combination of genetic and environmental factors.33 34 There are, of course, many potential environmental factors that we could not include,3 35 and clearly, we used only a coarse genetic marker, namely, family history of hypertension. It is possible that there are differences in gene polymorphisms among US urban areas; however, given the high degree of mobility of the US population, this is an unlikely explanation for our findings. Also, a trait as complex as hypertension is likely to be caused by multiple genes and gene-environment interactions.
In summary, the prevalences of EBP among young adults at year 7 as well as 7-year incidence were different among our four study sites and were highest in our one site in the Stroke Belt (Birmingham), most markedly so for black men, and consistent for all race/sex groups. These geographic differences were not apparent at baseline but developed over a 7-year period. Further study should investigate whether these differences become more marked as the cohort ages, and continued follow-up should contribute to a better understanding of the known geographic variability in CVD morbidity and mortality.
Selected Abbreviations and Acronyms
|BMI||=||body mass index|
|CARDIA||=||Coronary Artery Risk Development in Young Adults|
|DBP||=||diastolic blood pressure|
|EBP||=||elevated blood pressure|
|MLR||=||multiple logistic regression|
|SBP||=||systolic blood pressure|
This work was supported in part by contracts N01-HC-48047, N01-HC-48048, N01-HC-48049, N01-HC-48050, and N01-HC-95095 from the National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, Md, and by contract U18-HS09446-01 from the Agency for Health Care Policy and Research.
Reprint requests to Catarina I. Kiefe, PhD, MD, University of Alabama at Birmingham, 1717 Eleventh Ave S, MT 700, Birmingham, AL 35205-4785.
- Received January 15, 1997.
- Revision received February 28, 1997.
- Accepted March 5, 1997.
- Copyright © 1997 by American Heart Association
Stolley PD, Kuller LH, Nefzeger MD, Tonascia S, Lilienfeld AM, Miller GD, Diamond EL. Three-area epidemiological study of geographic differences in stroke mortality, II: results. Stroke. 1977;8:551-557.
Borhani NO. Changes and geographic distribution of mortality from cerebrovascular disease. Am J Public Health. 1965;55:673-681.
Ingster LM, Feinleib M. Geographic patterns of CVD mortality in the U.S.A. Cardiovasc Risk Factors. 1996;6:186-192.
Gillum RF, Ingram DD. Relation between residence in the southeast region of the United States and stroke incidence: the NHANES I Epidemiologic Follow-up Study. Am J Epidemiol. 1996;144:665-672.
Casper ML, Wing S, Anda RF, Knowles M, Pollard RA. The shifting stroke belt: changes in the geographic pattern of stroke mortality in the United States, 1962 to 1988. Stroke. 1995;26:755-760.
Howard G, Evans GW, Pearce K, Howard VJ, Bell RA, Mayer EJ, Burke GL. Is the stroke belt disappearing? An analysis of racial, temporal and age effects. Stroke. 1995;26:1153-1158.
Roccella EJ, Lenfant C. Regional and racial differences among stroke victims in the United States. Clin Cardiol. 1989;12:IV-18-IV-22.
Epstein FH, Eckoff RD. The epidemiology of high blood pressure: geographic distributions and etiological factors. In: Stamler J, Stamler R, Pullman TN, eds. The Epidemiology of Hypertension. New York, NY: Grune & Stratton, Inc; 1967.
Elliott P, Stamler J, Nichols R, Dyer AR, Stamler R, Kesteloot H, Marmot M, for the INTERSALT Cooperative Research Group. Intersalt revisited: further analyses of 24 hour sodium excretion and blood pressure within and across populations. BMJ. 1996;312:1249-1253.
Rodriguez BL, Labarthe DR, Huang B, Lopez-Gomez J. Rise of blood pressure with age: new evidence of population differences. Hypertension. 1994;24:779-785.
Intersalt Cooperative Research Group. Intersalt: an international study of electrolyte excretion and blood pressure: results for 24 hour urinary sodium and potassium excretion. BMJ. 1988;297:319-328.
Mariotti S, Capocaccia R, Farchi G, Menotti A, Verdecchia A, Keys A. Age, period, cohort and geographical area effects on the relationship between risk factors and coronary heart disease mortality: 15-year follow-up of the European cohorts of the Seven Countries Study. J Chronic Dis. 1986;39:229-242.
Cutter GR, Burke GL, Dyer AR, Friedman GD, Hilner JE, Hughes GH, Hulley SB, Jacobs DR Jr, Liu K, Manolio TA, Oberman A, Perkins LL, Savage PJ, Serwitz JR, Sidney S, Wagenknecht LE. Cardiovascular risk factors in young adults: the CARDIA baseline monograph. Control Clin Trials. 1991;12:1S-78S.
The CARDIA Coordinating Center. Coronary Artery Risk Development in (Young) Adults (CARDIA): year 7 Exam Manual of Operations. Birmingham, Ala: the CARDIA Study Coordinating Center, University of Alabama at Birmingham; April 1992.
Hosmer DW, Lemeshow S. Applied Logistic Regression. New York, NY:: John Wiley & Sons; 1989.
StataCorp. Stata Statistical Software: Release 4.0. College Station, Tex: Stata Corp; 1995.
SAS Institute Inc. SAS Software Ver 6.10 for OS/2. Cary, NC: 1993.
Burt VL, Whelton P, Roccella EJ, Brown C, Cutler JA, Higgins M, Horan MJ, Labarthe D. Prevalence of hypertension in the US adult population: results from the Third National Health and Nutrition Examination Survey, 1988-1991. Hypertension. 1995;25:305-313.
Prior IAM. Migration, hypertension, and Pacific perspectives for prevention. In: Lovenburg W, Yamori Y, eds. Nutritional Prevention of Cardiovascular Disease. Orlando, Fla: Academic Press, Inc; 1984:137-153.
Tell GS, Rutan GH, Kronmal RA, Bild DE, Polak JF, Wong ND, Borhani NO, for the Cardiovascular Health Study (CHS) Collaborative Research Group. Correlates of blood pressure in community-dwelling older adults: the Cardiovascular Health Study. Hypertension. 1994;23:59-67.
Howard BV, Lee ET, Yeh JL, Go O, Fabsitz RR, Devereux RB, Welty TK. Hypertension in adult American Indians: the Strong Heart Study. Hypertension. 1996;28:256-264.
Miller JP, Perry HM Jr, Rossiter JE, Baty JD, Carmody SE, Sambhi MP. Regional differences in mortality during 15-year follow-up of 11 936 hypertensive veterans. Hypertension. 1994;23:431-438.
Marmot MG. Geography of blood pressure and hypertension. Br Med Bull. 1984;40:380-386.
Lindpaintner K. Blood pressure and heredity: is it all in the genes, or not? Hypertension. 1993;22:147-149.
Hong Y, de Faire U, Heller DA, McClearn GE, Pedersen N. Genetic and environmental influences on blood pressure in elderly twins. Hypertension. 1994;24:663-670.