Geographic Patterns in Overall and Specific Cardiovascular Disease Incidence in Apparently Healthy Men in the United States
Background and Purpose— Residence in the Southeastern United States (US) has been linked to increased stroke incidence and mortality. However, data on regional variability in overall cardiovascular disease (CVD) and specific coronary heart disease incidence are sparse.
Methods— We assessed the risk of major CVD (nonfatal stroke, nonfatal myocardial infarction, or death from CVD) and specific CVD associated with region of residence (Northeast, Southeast, Midwest, and West) in 17 927 apparently healthy male participants of the Physicians’ Health Study. Subjects were aged 40 to 84, most were white (93%), and had no previous CVD at baseline. We used residence in the Northeast as the reference group and proportional hazards models to adjust for potential confounding.
Results— We found no difference in risk of major CVD between regions of residence. Further, we found no consistent association between myocardial infarction and CVD death and region of residence. In contrast, we found a significantly increased risk of total stroke (HR, 1.22; 95% CI, 1.02 to 1.47) associated with residence in the Southeast compared with the Northeast. This relative risk was further increased for ischemic stroke (HR, 1.30; 95% CI, 1.06 to 1.58). We saw no difference in risk of any outcome when categorizing state of residence into tertiles based on mean winter temperature, mean summer temperature, or into 2 groups based on latitude.
Conclusions— In this homogenous and well-characterized cohort of US male physicians, we found greater incidence of ischemic stroke, but not other vascular events among those living in the Southeastern US, compared with other regions.
In 1980, the National Heart Lung and Blood Institute classified the 11 states (Indiana, Kentucky, Tennessee, Mississippi, Alabama, Georgia, South Carolina, North Carolina, Virginia, Arkansas, and Louisiana) with stroke mortality rates 10% higher than the national average (ie, ≥65 deaths/100 000 persons) as the “Stroke Belt,” and has since funded research to reduce the risk of stroke in this region.1 Others have also identified the Stroke Belt, reporting similar disparities in stroke mortality rates.2–4 Howard et al5 documented a national decline in stroke mortality, predicting regional variability to persist with other regions emerging with high stroke rates. The National Health and Nutrition Examination Survey I Epidemiologic Follow-up Study found the highest stroke incidence rates in the Southeast United States (US) compared with other regions.6 In white males 45 to 74 years of age, risk associated with the Southeastern US was not attenuated after adjusting for conventional stroke risk factors. Further, there was no difference in incidence rates between subjects residing inside and outside standard metropolitan statistical areas.6 However, stroke subtype was not assessed and the only control for socioeconomic status was education level.
The Southeast has also been associated with higher mortality rates from congestive heart failure,7 higher prevalence of self-reported heart disease,8 and greater prevalence of hypertension among black females,9 white men, as well as black men and women,10 compared with other regions. In contrast, the West has been associated with increased incidence of ischemic heart disease, compared with the Southeast, Northeast, and Midwest regions.11
Increased stroke incidence has been reported in several studies in the colder months and in other studies during the warmer months. Others have found no seasonal differences in stroke rates.12 Seasonal variation in ischemic stroke incidence, if real, may reflect the influence of acute changes in temperature on known stroke risk factors (eg, blood pressure, blood clotting time, fibrinogen levels, etc).12 Other potential explanations for this increased stroke risk associated with residence in the Southeast include residual confounding caused by lifestyle factors and factors correlated with weather conditions. We therefore aimed to prospectively evaluate the association between geographic region of residence and incidence of overall and specific cardiovascular disease (CVD), using data from a well-characterized cohort of mostly white, apparently healthy, US male physicians, aged 40 to 84 years, and free of CVD at baseline.
Materials and Methods
The Physicians Health Study is a completed randomized trial designed to test the benefits and risks of aspirin (325 mg every other day) or beta-carotene (50 mg every other day) in the primary prevention of cardiovascular disease and cancer among 22 071 apparently healthy men aged 40 to 84 at baseline in 1982. The methods and results of the Physicians Health Study have been described in detail previously.13,14 Post-trial follow-up is continuing.15 All participants gave informed consent. The Institutional Review Board of Brigham and Women’s Hospital approved the Physicians Health Study. For this analysis, we included follow-up information through February 28, 2004 (maximum of 22.5 years of follow-up). As of that date, follow-up was 97% complete.
At baseline, subjects completed a questionnaire detailing information on cardiovascular risk factors and lifestyle habits. Of the original trial cohort, we excluded 702 participants who did not reside within the 50 US states or in Washington DC. Of these 21 369 subjects, we retained only those subjects with complete information on age at baseline, frequency of smoking, exercise, and alcohol intake, height, weight, previous diagnoses of hypertension (diagnosis of hypertension or systolic blood pressure ≥140 mm Hg or diastolic blood pressure ≥90 mm Hg), elevated cholesterol (diagnosis of high cholesterol or cholesterol ≥240 mg/100 mL), diabetes, and family history of myocardial infarction (MI) before age 60, and those without a coronary revascularization procedure before receiving the baseline questionnaire procedure. This left 17 927 subjects for analysis.
Using the state listed in subjects’ baseline mailing address, we categorized subjects into the 4 geographic regions Northeast, Southeast, Midwest, and West (Table 1) according to the 2000 US Census.16
Every 6 months in the first year and annually thereafter, participants self-reported incidence of study outcomes, including CVD events. Medical records were obtained for all reported cardiovascular end points and reviewed by an Endpoints Committee of physicians, except coronary revascularization procedures. The occurrence of MI was confirmed if symptoms met World Health Organization criteria and if the event was associated with abnormal levels of cardiac enzymes or diagnostic electrocardiograms. Nonfatal stroke was confirmed if the participant had a new focal–neurologic deficit of vascular origin and sudden onset that persisted for >24 hours. Strokes were classified as ischemic, hemorrhagic, or unknown after review of all available information including diagnostic testing, such as brain scans. This classification had a high interrater agreement (κ=0.82).17 Cardiovascular deaths were confirmed by autopsy reports, death certificates, medical records, and information obtained from next of kin or family members. We evaluated the following outcome events: major CVD (a combined end point composed of nonfatal stroke, nonfatal MI, or death from CVD), total MI, total stroke, ischemic stroke, death from CVD, and reports of coronary revascularization procedures (bypass surgery or percutaneous coronary angioplasty). There were too few hemorrhagic strokes to create a separate category for this analysis.
Within each region, we calculated crude CVD incidence rates. Next, we calculated age standardized CVD incidence rates by region, using the age distribution of the Northeast as the standard.
We used Cox proportional hazards models to estimate the risk of major CVD, specific major CVD, as well as ischemic stroke and coronary revascularization procedures associated with geographic region, adjusting for potential confounding factors. Indicator variables were generated for age at baseline (40 to 49, 50 to 69, and ≥70), frequency of alcohol intake (once per day or more, once per month through 6 times per week, rarely/never), smoking (never, past, current 1 to 19 cigarettes per day, or current ≥20 cigarettes per day), exercise frequency (once per day or more, 1 to 6 per week, 1 to 3 per month, rarely/never), body mass index (weight in kg divided by height in meters squared; <25, 25 to 29, ≥30), history of cholesterol ≥240 mg/dL, history of hypertension (defined as before), and parental history of MI before age 60.
In a first sensitivity analysis, we re-categorized geographic region into 2 groups (Stroke Belt versus non-Stroke Belt). Additionally, we re-categorized geographic region into 2 groups based on latitude. Second, we gathered the monthly mean ambient temperature measurements from all monitoring stations in the US that were in operation every year from 1982 to 2002 (Earth Info, Boulder, Colo). We then computed winter mean (December to February) and summer mean (June to August) temperatures for each state. We then re-categorized geographic region into 3 groups based on winter mean temperature, and then again into 3 groups based on summer mean temperature (Table 1).
Study subjects were predominantly white (93%). At baseline, the 17 927 subjects were middle aged (mean±SD, 53.5±9.0), predominantly never smokers (51%), and past smokers (39%), with a mean body mass index of 24.8±2.8 kg/m2. A large proportion was classified as overweight (38%) or obese (4%). Subjects generally exercised to sweat at least once per week (73%), and 25% drank daily whereas 15% drank rarely or never. Some subjects were previously diagnosed with elevated cholesterol (12%), hypertension (24%), diabetes (2%), or had a family history of MI before age 60 (10%). There appeared to be little difference in the prevalence of these risk factors and cardiovascular events by geographic region (Table 2). During a mean follow-up time (±SD) of 20.4±4.2 years, subjects experienced 2607 major CVD events, 1018 strokes of which 846 were classified as ischemic stroke, 1247 MIs, 2479 coronary revascularization procedures, and 995 CVD deaths.
Incidence Rates by Region
The Southeast and Midwest had higher crude and age-standardized major CVD, total stroke, ischemic stroke, coronary revascularization, and CVD death incidence rates compared with the Northeast (Table 3). MI age-standardized incidence rates were lowest in the Southeast, whereas major CVD, stroke, ischemic stroke, and coronary revascularization rates were highest in that region (Table 3).
In unadjusted analyses, the Southeast was associated with significantly decreased risk of MI, and marginally significant (P<0.10) increased risk of coronary revascularization, and ischemic stroke, compared with the Northeast. The Midwest was also associated with significantly increased risk of coronary revascularization procedures compared with the Northeast. There were no other significant associations between geographic region and other CVD outcomes (Table 4). After adjustment for potential confounding factors, the Southeast was no longer associated with significantly decreased risk of MI compared with the Northeast. However, the Southeast was associated with significantly increased risks of total stroke (HR, 1.22; 95% CI, 1.02 to 1.47) and ischemic stroke (HR, 1.30; 95% CI, 1.06 to 1.58) compared with the Northeast (Table 4). Furthermore, we found significantly increased risk of coronary revascularization with residence in the Southeast and the Midwest.
Compared with the rest of the US, residence in the Southeast was associated with marginally significantly increased risk of total stroke (HR, 1.16; 95% CI, 1.00 to 1.34) and significantly increased risk of ischemic stroke (HR, 1.20; 95% CI, 1.02 to 1.41). The Southeast was also associated with significantly decreased risk of MI (HR, 0.86; 95% CI, 0.75 to 0.99), but not coronary revascularization (HR, 1.08; 95% CI, 0.98 to 1.19), compared with the rest of the US. Residence in the Southeast was not associated with risk of major CVD or CVD death (data not shown). Furthermore, compared with the rest of the US, residence in the Midwest was not associated with any of the evaluated outcomes (data not shown).
Because we found significantly increased risk of ischemic stroke but not for MI and coronary revascularization, we limited further sensitivity analyses to the ischemic stroke outcome only. First, residence in the Stroke Belt was associated with marginally significantly increased risk of ischemic stroke (HR, 1.17; 95% CI, 0.98 to 1.40) compared with the rest of the US. When we categorized region of residence based approximately on latitude, the South was not associated with increased risk of ischemic stroke (HR, 1.00; 95% CI, 0.87 to 1.14) compared with the North.
Next, we ranked states into tertiles based on each state’s mean winter temperature. Compared with the region with the lowest mean winter temperatures (region 1: mean, 4.6°C; state’s mean winter temperature ranged from −9.0°C to −0.9°C), neither the region with the next highest mean winter temperatures (region 2: 1.3°C; range, −0.6°C to 3.5°C; HR, 1.00; 95% CI, 0.85 to 1.17) nor the region with the highest mean winter temperatures (region 3: 9.0°C; range, 4.3°C to 22.6°C; HR, 0.88; 95% CI, 0.74 to 1.06) was associated with increased risk of ischemic stroke.
Last, we ranked states into tertiles based on each state’s mean summer temperature. Compared with the region with the lowest mean summer temperatures (region 1: mean, 19.7°C; state’s mean summer temperature ranged from 11.4°C to 21.6°C), neither the region with the next highest mean summer temperatures (region 2: 23.0°C; range, 21.7°C to 24.5°C; HR, 0.97; 95% CI, 0.81 to 1.15) nor the region with the highest mean summer temperatures (region 3: 26.1°C; range, 24.6°C to 27.6°C; HR, 0.90; 95% CI, 0.77 to 1.06) was associated with increased risk of ischemic stroke.
In this well-characterized cohort of mostly white, apparently healthy, US male physicians, aged 40 to 84, with no CVD or cancer at baseline, we did not find increased risk of major CVD associated with region of residence in any analysis. When we examined individual CVD events, we found significantly increased risk of total stroke and ischemic stroke associated with the Southeast that was not attenuated after adjusting for conventional CVD risk factors. We were also able to replicate the pattern of increased stroke in the Southeastern US (ie, Stroke Belt) seen in previous mortality and incidence studies, as well as when using an alternative definition of Stroke Belt (North Carolina, South Carolina, Georgia, Alabama, Tennessee, Mississippi, Louisiana, Arkansas; data not shown), but not with any region when classified by latitude or mean seasonal temperature. In contrast, we found no consistent association between geographic region and risk of MI or CVD death. Our data suggested an association between residence in the Southeast and Midwest with coronary revascularization procedures, which were, however, attenuated and no longer statistically significant in our sensitivity analyses. The increased risk of ischemic stroke in the Southeast region may be attributable to differences in other regional environmental conditions or lifestyle factors not included in this analysis.
Our stroke risk estimates for the Southeast region for predominantly white, male, physicians 40 to 84 years of age (HR, 1.22; 95% CI, 1.02 to 1.47), were slightly lower than those reported by Gillum and Ingram6 for white males 45 to 74 years of age (HR, 1.41; 95% CI, 1.02 to 1.92). Unadjusted stroke incidence rates were markedly lower in our study, compared with Gillum and Ingram.6 This likely reflects that our cohort only included physicians. On average, physicians are considered to be of higher socioeconomic status, to have healthier lifestyles, better diets, better access to health care services, and more frequent use of preventive health care, compared with nonphysicians. Another explanation why our incidence rates are lower is that our study was conducted from 1982 to 2004, whereas the study period in the study by Gillum and Ingram6 was much earlier (1971 to 1987). Because stroke mortality rates have declined over time,5 and assuming incidence rates followed a similar pattern, we would expect to see lower incidence rates in our study.
A possible reason for the observed association between residence in the Southeastern US and incidence of ischemic stroke is diet. In the third National Health and Nutrition Examination Survey, the Southeast region had the highest blood pressures values, the highest intakes of monounsaturated and polyunsaturated fatty acids, cholesterol, and sodium, and the lowest intake of fiber and several vitamins and minerals18 compared with the other three regions. We had no detailed information on dietary intake at baseline in the Physicians Health Study. However, statistical adjustment using available information on vegetable and milk intake made little difference (data not shown).
The National Health and Nutrition Examination Survey III reported lower mean serum selenium levels among people residing in the Southern US compared with the other 3 US regions.19 Selenium levels have previously been inversely associated with risk of nonfatal MI and fatal CHD,20 total, cardiovascular, and stroke mortality,21 and directly associated with coronary revascularization procedures.20
Studies of temperature and stroke incidence have been inconsistent, with some finding no association, others reporting associations with cold months, and others with warm months.12 One study found no difference in stroke incidence rates between days with abrupt changes in weather conditions and those without such changes.22 We ranked states based on mean winter temperature, mean summer temperature and latitude. Although we found no association with geographic region based on either latitude or seasonal mean temperature, other characterizations of the regional variability in temperature or other weather conditions may be associated with increased risk of ischemic stroke.
Our study had several strengths, including the large number of outcome events and participants, prospective method of data collection, and the homogenous nature of our cohort, which reduced confounding by many sociodemographic factors. Furthermore, with the exception of coronary revascularization, all outcome events were confirmed after medical record review.
Several limitations should be considered when interpreting out results. The mailing address given by each subject at baseline was used to classify each subject into US regions. This mailing address could have been the subject’s home or the subject’s work address. Some subjects may work and live in different cities and states, which could be important if we were examining local variability in CVD incidence. However, it is unlikely that subjects worked and lived in different census regions; therefore, this likely did not result in substantial misclassification of census region. If a few cases of misclassification did in fact occur, this was likely nondifferential with respect to outcome, resulting in underestimation of risk.
If subjects moved during follow-up from other regions (cold weather states in the Northeast and Midwest) to the Southeast (eg, Florida, South Carolina, etc), this would have made the regions’ incidence rates more similar. Therefore, this also would have led to an underestimate of the risk of ischemic stroke associated with residence in the Southeast.
These findings are directly generalizable to middle-aged to older, white, US males of higher socioeconomic status. However, gender and racial differences in stroke mortality3 and stroke incidence6,23 have been demonstrated. Therefore, it is likely that different racial and gender groups may have markedly different ischemic stroke incidence rates in the same region of residence. Thus, these analyses should be replicated in women and nonwhites.
In summary, in this large, prospective cohort of apparently healthy, mostly white, US male physicians, we found no evidence of regional differences in risk of major CVD. In further analyses, we found consistent significantly increased risk of total and ischemic stroke, but not other vascular events associated with living in the Southeast compared with the Northeast or the rest of the US, after adjustment for traditional CVD risk factors.
The authors are indebted to the participants of the Physician’s Health Study for their outstanding commitment and cooperation, and to the entire Physicians’ Health Study staff for their expert and unfailing assistance.
Sources of Funding
The study was supported by NCI grants CA 34944 and CA 40360, and NHLBI grants HL 26490 and HL 34595.
- Received January 25, 2007.
- Revision received March 29, 2007.
- Accepted April 4, 2007.
United States National Heart, Lung, and Blood Institute. Stroke Belt Initiative. Project accomplishments and lessons learned. Available at: http://www.nhlbi.nih.gov/health/prof/heart/other/sb_spec.pdf. Accessed March 26, 2007.
Lanska DJ. Geographic distribution of stroke mortality in the United States: 1939–1941 to 1979–1981. Neurology. 1993; 43: 1839–1851.
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.
Perry HM, Roccella EJ. Conference report on stroke mortality in the southeastern United States. Hypertension. 1998; 31: 1206–1215.
Howard G, Howard VJ, Katholi C, Oli MK, Huston S. Decline in US stroke mortality: an analysis of temporal patterns by sex, race, and geographic region. Stroke. 2001; 32: 2213–2220.
Gillum RF, Ingram DD. Relation between residence in the southeast region of the United States and stroke incidence: The NHANES I Epidemiologic Followup Study. Am J Epidemiol. 1996; 144: 665–673.
Obisesan TO, Vargas CM, Gillum RF. Geographic variation in stroke risk in the United States. Region, urbanization, and hypertension in the Third National Health and Nutrition Examination Survey. Stroke. 2000; 31: 19–25.
Hennekens CH, Buring JE, Manson JE, Stampfer M, Rosner B, Cook NR, Belanger C, LaMotte F, Gaziano JM, Ridker PM, Willett W, Peto R. Lack of effect of long-term supplementation with beta carotene on the incidence of malignant neoplasms and cardiovascular disease. N Engl J Med. 1996; 334: 1145–1149.
Christen WG, Gaziano JM, Hennekens CH. Design of Physicians’ Health Study II–a randomized trial of beta-carotene, vitamins E and C, and multivitamins, in prevention of cancer, cardiovascular disease, and eye disease, and review of results of completed trials. Ann Epidemiol. 2000; 10: 125–134.
US Bureau of the Census. Summary population and housing characteristics. 2000 census of population and housing. Part 1. 2002. Available at: http://www.census.gov/prod/cen2000/phc-1-1-pt1.pdf. Accessed March 26, 2007.
Berger K, Kase CS, Buring JE. Interobserver agreement in the classification of stroke in the physicians’ health study. Stroke. 1996; 27: 238–242.
Hajjar I, Kotchen T. Regional variations of blood pressure in the United States are associated with regional variations in dietary intakes: the NHANES-III data. J Nutr. 2003; 133: 211–214.
Yoshizawa K, Ascherio A, Morris JS, Stampfer MJ, Giovannucci E, Baskett CK, Willett WC, Rimm EB. Prospective study of selenium levels in toenails and risk of coronary heart disease in men. Am J Epidemiol. 2003; 158: 852–860.
Virtamo J, Valkeila E, Alfthan G, Punsar S, Huttunen JK, Karvonen MJ. Serum selenium and the risk of coronary heart disease and stroke. Am J Epidemiol. 1985; 122: 276–282.
Field TS, Hill MD. Weather, Chinook, and stroke occurrence. Stroke. 2002; 33: 1751–1757.
Sacco RL, Boden-Albala B, Gan R, Chen X, Kargman DE, Shea S, Paik MC, Hauser WA. Stroke incidence among white, black, and Hispanic residents of an urban community: the Northern Manhattan Stroke Study. Am J Epidemiol. 1998; 147: 259–268.