The Relationship Between Systolic Blood Pressure and Cardiovascular Risk—Results of the Brisighella Heart Study

Abstract

We examined the relationship of systolic (SBP) and diastolic (DBP) blood pressure, and pulse pressure to coronary heart disease and cerebrovascular disease risk in a prospective population‐based European cohort. The Brisighella Heart Study included 2939 men and women between the ages of 14–84 without prior coronary heart disease or cerebrovascular disease and not taking antihypertensive therapy at baseline. Cox regression was used to obtain hazard ratios (HRs) for coronary heart disease and cerebrovascular disease as a function of baseline blood pressure parameters over a 23‐year follow‐up. Higher combined coronary heart disease and cerebrovascular disease risk was evident in comparison to the referent of <120 mm Hg, with a 44% increased risk at SBP 120–139 mm Hg (HR, 1.44; 95% confidence interval [CI], 1.00–2.09; p=0.052), 76% increased risk at SBP 140–159 mm Hg (HR, 1.76; 95% CI, 1.16–2.69; p=0.009), and 109% increased risk at SBP ≥160 mm Hg (HR, 2.09; 95% CI, 1.31–3.35; p=0.0021). Trends of increasing risk with increasing levels of blood pressure were significant for SBP and pulse pressure, (p<0.0001) but not for DBP (p=0.058). In this European cohort, SBP was a stronger predictor of coronary heart disease and cerebrovascular disease events than DBP, and an increase in risk was already evident with highnormal SBP (120–139 mm Hg). The prognostic significance of pulse pressure was also demonstrated. The importance of SBP as seen in the Framingham Heart Study may be generalized to a European population with differences in diet and other risk factors.

In clinical practice, treatment decisions have often been based on diastolic blood pressure (DBP) instead of systolic blood pressure (SBP). ¹ Nevertheless, results of the Framingham Heart Study have suggested that SBP is an equal or better predictor of cardiovascular events than DBP in middle‐aged and elderly individuals, and these age groups represent the vast majority of individuals with hypertension. ² , ³

Epidemiologic studies had previously implied that the cardiovascular risk status of persons occupying certain regions of Europe (e.g., the Mediterranean area) was lower due to diet and wine consumption. ⁴ , ⁵ This poses a question about how results of the Framingham Heart Study should be extrapolated to other populations with differing characteristics. In clinical practice, some physicians may discount information provided by SBP if they focus on ethnic or environmental differences that seem to be protective of cardiovascular disease.

The recent Seven Countries Study ⁶ shed some light on this issue. The relationship between blood pressure and mortality was examined in several different regions of the world. The relative increase in long‐term mortality due to coronary heart disease (CHD) for a given increase in blood pressure was similar among different regions, even though the absolute risk varied substantially.

However, the results of the Seven Countries Study did not address the relative roles of SBP, DBP, and pulse pressure (PP). Furthermore, it did not separate CHD mortality or CHD events from total mortality. Information was not collected on diabetes at baseline. ⁷

The Brisighella Heart Study ⁸ is more comparable to the Framingham Heart Study in terms of design and data collection. It provides a unique opportunity to investigate the long‐term prognostic significance of blood pressure parameters in a non‐US population.

SUBJECTS AND METHODS

The design of the Brisighella Heart Study ⁸ has been described in detail elsewhere. Briefly, the Brisighella Heart Study is a prospective, population‐based cohort study of cardiovascular disease incidence and risk factors that began in 1972 with the enrollment of 2939 randomly selected men and women residents of Brisighella, Italy. The participants, aged 14–84 years and free of cardiovascular disease at enrollment, undergo examinations every four years, which include an extensive cardiovascular history and physical examination, 12‐lead electrocardiography, and various blood chemistries. Morbidity and mortality are monitored by clinic examinations and by review of interim hospitalizations.

Blood pressure was measured by a trained physician with a calibrated mercury sphygmomanometer on the dominant arm with the subject in the seated position at the end of the physical examination. Readings were taken to the nearest 2 mm Hg and using the fifth Koroktoff sound as a measure of DBP. The mean of three consecutive measurements 1 minute apart was computed both for SBP and DBP.

For the current analysis, the period of follow‐up was from 1973–1996. The long‐term prognostic significance of baseline SBP and DBP was determined before and after adjustment for the presence of well known and established cardiovascular risk factors such as age, gender, smoking, history of diabetes mellitus, and total serum cholesterol. The end points of interest were CHD indicated by angina pectoris (International Classification of Disease, Ninth revision [ICD‐9] code, 413.xx), fatal and nonfatal myocardial infarction (ICD‐9 code, 410.xx), and cerebrovascular disease indicated by transient ischemic attack or stroke (ICD‐9 code, 435.xx, 436.xx). All cardiovascular and cerebrovascular events were adjudicated by an expert panel review of medical records and death certificates and only the certified events have been included for the final evaluation. The reviewers who coded the cause of death were blinded with respect to the subject's cardiovascular risk factors.

Cox proportional hazards regression was used to calculate hazard ratios for categorical levels of SBP, DBP, and PP, after adjusting for age, sex, smoking, diabetes mellitus, and total serum cholesterol. These categories were <120 mm Hg, 120–139 mm Hg, 140–159 mm Hg, and >159 mm Hg for SBP, <70 mm Hg, 70–79 mm Hg, 80–89 mm Hg, >89 mm Hg for DBP, and <44 mm Hg, 44–53 mm Hg, 54–69 mm Hg, and >69 mm Hg for PP. The Maentel‐Haenzel test for trend was used to assess changes in the hazard ratio across blood pressure categories. First, three separate Cox models, (SBP, DBP, and PP) were evaluated as single continuous variables. Second, dual models (SBP and DBP) were evaluated by three descrete categories, as denoted above. Lastly, models included all three blood pressure components (SBP, DBP, and PP) and were again evaluated by three descrete categories.

Patient Demographics

Table I details the distribution of the baseline (at entry) study variables according to gender and hypertension status. There were 153 events in 2169 normotensive subjects and 197 events in 862 hypertensives. Among normotensive subjects (Table I [A]), the prevalence and mean value of most risk factors did not differ meaningfully according to gender. However, smoking was twice as prevalent in males vs. females, and mean triglyceride levels, mean arterial pressure, PP, SBP, and DBP were higher in males. Smoking rates were generally higher than those seen in the United States.

Table I.

Baseline Clinical Characteristics (Mean, Number, or Percent of Population), Stratified by Gender and Hypertension Status

A. N ormotensive
Characteristics	Malesm (N=1118 )	Females (N=1051 )	p Value
Cardiovascular events
Myocardial infarctions*	72	26
Strokes**	29	26
Age (years)	42.6±14.5	40.2±14.0	<0.0001
MAP (mm Hg)	89.4±8.1	87.5±8.7	<0.0001
SBP (mm Hg)	121.6±11.3	119.1±12.0	<0.0001
DBP (mm Hg)	73.4±8.2	71.8±8.6	<0.0001
PP (mm Hg)	48.3±9.8	47.3±10.0	0.0230
TC (mg/dL)	218.4±40.8	218.0±45.3	0.807
TG (mg/dL)	147.8±87.0	118.0±57.3	<0.0001
GL (mg/dL)	84.2±13.9	81.3±13.9	<0.0001
UR (mg/dL)	5.3±1.4	4.6±1.2	<0.0001
Diabetes mellitus	2.4	2.0	0.804
Smoker (%)	53.8	28.9	<0.0001
B. Hypertensive
Characteristics	Males (N=423 )	Females (N=439 )	p Value
Cardiovascular events
Myocardial infarctions*	65	50
Strokes**	47	35
Age (years)	58.3±13.4	58.7±10.6	0.638
MAP (mm Hg)	113.6±12.3	114.8±12.4	0.153
SBP (mm Hg)	162.2±18.8	164.9±19.2	0.037
DBP (mm Hg)	89.3±12.4	89.7±11.6	0.584
PP (mm Hg)	73.0±17.9	75.2±16.2	0.053
TC (mg/dL)	234.9±46.8	247.3±45.8	<0.0001
TG (mg/dL)	168.6±110.6	157.1±78.9	0.08
GL (mg/dL)	87.7±17.1	88.6±21.6	0.520
UR (mg/dL)	5.5±1.6	5.1±1.5	<0.0001
Diabetes mellitus	5.2	8.7	0.083
Smoker (%)	47.3	13.2	<0.0001
ISH (%)	64.3	64.9	0.850
MAP=mean arterial pressure; SBP=systolic blood pressure; DBP=diastolic blood pressure; PP=pulse pressure; TC=total cholesterol; TG=triglycerides; GL=glucose; UR=uric acid; ISH=isolated systolic hypertension defined as SBP >140 and DBP <90 mm Hg;*ICD‐9 code 410.xx (acute myocardial infarction), 413.xx (angina pectoris); **ICD‐9 code 435.xx (transient ischemic attack), 436.xx (stroke)

Among hypertensive subjects (Table I [B]), the risk factor distributions were very similar across gender except for a higher total cholesterol in females and a higher uric acid level and smoking prevalence in males. Also, more female hypertensives tended to have diabetes compared to males. Mean values of SBP exceeded 160 mm Hg, indicating that a majority of these subjects experienced stage 2 hypertension at baseline. ⁹ Sixty‐four percent of individuals with hypertension had isolated systolic hypertension, consistent with mean DBP values below 90 mm Hg.

For single blood pressure component models (Table II), the hazards ratios associated with a 10 mm Hg increment change in blood pressure and after adjusting for other risk factors, were as follows: PP, 1.13 (confidence interval [CI],1.12–1.14); SBP, 1.11 (CI, 1.10–1.12); DBP, 1.03 (CI, 1.02–1.05).

Table II.

Adjusted* Blood Pressure Parameters and Cardiovascular** Risk Models With SBP, DBP, and PP as Separate, Continuous Variables

Variable	Hazard Ratio	95% CI
SBP (per 10 mm Hg)	1.11	1.10–1.12
DBP (per 10 mm Hg)	1.03	1.02–1.05
PP (per 10 mm Hg)	1.13	1.12–1.14
SBP=systolic blood pressure; DBP=diastolic blood pressure; PP=pulse pressure; CI=confidence interval; *adjusted for age, smoking status, diabetes, gender, and total cholesterol; **stroke, fatal/nonfatal myocardial infarction

For dual blood pressure component models (Table III), the hazard ratios (in reference to a baseline of 1.00 for SBP <120 mm Hg) after adjusting for other risk factors were as follows: SBP 120–139 mm Hg, 1.44 (CI, 1.00–2.09; p=0.052); SBP 140–159 mm Hg, 1.76 (CI, 1.16–2.69; p=0.009); and SBP >159 mm Hg, 2.09 (CI, 1.31–3.35; p=0.0021). For DBP (in reference to a baseline of 1.00 for DBP <70 mm Hg), the hazard ratios were for 70–79 mm Hg, 0.81 (CI, 0.52–1.27; p=0.36), for DBP 80–89 mm Hg, 1.01 (CI, 0.65–1.58; p=0.96); and for DBP >89 mm Hg, 1.10 (CI, 0.65–1.86; p=0.73). Trends of increasing hazard ratios with increasing BP were significant for SBP (p=0.003), but not for DBP (p=0.865).

Table III.

Adjusted* Blood Pressure Parameters and Cardiovascular** Risk Model With SBP and DBP

Variable	Hazard Ratio	95% CI	p Value
SBP <120	1	Referent
SBP 120–139	1.44	1.00–2.09	0.0528
SBP 140–159	1.76	1.16–2.69	0.0085
SBP >159	2.09	1.31–3.35	0.0021
p value for trend			0.003
DBP <70	1	Referent
DBP 70–79	0.81	0.52–1.27	0.3583
DBP 80–89	1.01	0.65–1.58	0.9609
DBP >89	1.10	0.65–1.86	0.7309
p value for trend			0.865
SBP=systolic blood pressure; DBP=diastolic blood pressure; PP=pulse pressure; CI=confidence interval;*adjusted for age, smoking status, diabetes, gender, and total cholesterol; **myocardial infarction (fatal and nonfatal) and stroke

Table IV provides the results of a regression model containing all three blood pressure components. SBP and DBP demonstrated similar results to the dual models, except that the hazard ratio for DBP >89 mm Hg was significant at 1.59 (CI, 1.02–2.49; p=0.042). Adjusted hazard ratios for the model containing PP were as follows: PP, 44–53 mm Hg, 1.27 (CI, 0.78–2.05; p=0.34); PP 54–69 mm Hg, 1.75 (CI, 1.13–2.73; p=0.013); and PP >69 mm Hg, 1.73 (CI,1.10–2.72; p=0.017).

Table IV.

Adjusted* Blood Pressure Parameters and Cardiovascular** Risk Model With SBP, DBP, and PP

Variable	Hazard Ratio	95% CI	p Value
SBP <120	1	Referent
SBP 120–139	1.48	1.04–2.10	0.0313
SBP 140–159	1.92	1.32–2.80	0.0006
SBP >159	2.38	1.61–3.50	<0.0001
p value for trend			0.002
DBP <70	1	Referent
DBP 70–79	0.87	0.56–1.35	0.5391
DBP 80–89	1.27	0.85–1.89	0.2493
DBP >89	1.59	1.02–2.49	0.0418
p value for trend			0.0476
PP <44	1	Referent
PP 44–53	1.27	0.78–2.05	0.3405
PP 54–69	1.75	1.13–2.73	0.0130
PP >69	1.73	1.10–2.72	0.0169
p value for trend			0.021
SBP=systolic blood pressure; DBP=diastolic blood pressure; PP=pulse pressure; CI=confidence interval; *adjusted for age, smoking status, diabetes, gender, and total cholesterol; **fatal and nonfatal myocardial infarction and stroke

DISCUSSION

This analysis in a predominantly middle‐aged cohort demonstrated a consistent, strong, graded association between SBP and cardiovascular events, whereas this association was lacking with DBP. The Brisighella Heart Study, similar in design to the Framingham Heart Study, showed these findings in a rural Italian population with up to 23 years of follow‐up.

These results are generally consistent with other literature pointing to the importance of SBP. Kannel et al. ¹⁰ first suggested a stronger relation of SBP over DBP in predicting CHD risk in the Framingham Heart Study, and results of the Multiple Risk Factor Intervention Trial (MRFIT) ¹¹ pointed to SBP as a key determinant of mortality.

In a Framingham cohort with a mean age of 61 years, SBP and DBP were simultaneously entered into a Cox proportional model; the results were that SBP was directly related and DBP inversely related to CHD risk, thereby yielding PP as the strongest predictor of the three blood pressure components. ² Others have confirmed these observations in elderly populations. ¹² , ¹³ In a younger cohort, like that of Glynn and colleagues ¹⁴ and in the present study, SBP remains more important than DBP but PP is not the strongest predictor of events.

Of note in the current study was the observation that an increase in combined CHD and cerebrovascular disease risk was already evident with high‐normal SBP. This finding is consistent with other reports. ¹⁴ , ¹⁵ , ¹⁶ The Seven Countries Study ⁶ did not examine individual cardiovascular or cerebrovascular events, but higher mortality rates were already evident with high‐normal blood pressures. In a review of this study, MacMahon ¹⁷ concluded that the clinical benefits of blood pressure reduction are likely to extend to high‐normal blood pressure as well as to those individuals with hypertension.

Differences between countries in behaviors and environment may substantially affect cardiovascular disease risk. However, even in a European population known for a Mediterranean diet, high‐normal and elevated SBP is a common problem and it should be a relevant issue to the clinician.

The Framingham Heart Study represents only one population that is somewhat homogeneous, but our experience with a comparable European cohort suggests that the same principles about the relationship to CHD and CVD risk are evident. Clinicians should pay special attention to SBP for their middle‐aged and elderly patients. SBP should not be discounted, even among patients whose diet and evironment are commonly associated with lower cardiovascular disease risk.

Appendix A. Brisighella Heart Study, University of Bologna Brisighella Study Founder: Professor Gian Carlo Descovich; Members: Prof. A. Gaddi MD, PhD, Scientific Coordinator, Council Chairman of Bologna Medicine and Surgery School; E. Pelliconi, Brisghella Coordinator, Past Mayor; C. Sangiorgi, Current Mayor. Bologna Medical Staff: C. Borghi, MD; S. D'Addato, MD PhD; A. Fiorito; C. Galetti, MD; M.C. Grippo, MD; V. Immordino, MD; M. L. Malkowski, MD; Brisighella Team Project Coordinator: C. Mussoni, MD, PhD; A. Moretti, MD; S. Nascetti, MD; A. R. Reggiani, MD; S. Rimondi, MD. Data Handling and Biometrics: A. Dormi, Math D. Laboratory Staff: Z. Sangiorgi, BD; R. Mambelli. Brisighella General Practitioner Staff: L. Bagnaresi, MD; E. Belletti, MD; A. Gamberi, MD; A. Naldi, MD; C. Samorè, MD; P. Viozzi, MD. Anagraphic Staff: M. Nanni. Brisighella Heart Price Committee: E. Descovich; A. Gaddi; C. Sangiorgi.