Arterial hypertension and mortality in the elderly

AJH

2002; 15:958 –966

Arterial Hypertension and Mortality in the Elderly Edoardo Casiglia, Alberto Mazza, Vale´ rie Tikhonoff, Roberta Scarpa, Francesco Guglielmi, and Achille C. Pessina Background: The aim of this study was to evaluate at a population level whether hypertension is a risk factor for cardiovascular mortality and to verify whether or not this is true for both genders at any age. Methods: This population-based, long-lasting, prospective study includes a 14-year mortality (institutional epidemiology in primary care). Unselected, unbiased subjects (5185) aged 22 to 95 years were recruited from the Italian general population, and divided into normotensive (⬍140 mm Hg systolic blood pressure [BP] and ⬍90 mm Hg diastolic BP and untreated) and hypertensive groups. The main aim was to identify the significant predictors of mortality due to stroke, coronary artery disease, heart failure, and pulmonary embolism, and to quantify the age-adjusted relative risk of hypertension in men and women, at different age classes (⬍70, 70 to 79, ⱖ80 years) for each mortality cause. The analysis was repeated among 1091 normotensive and 1091 hypertensive age-matched subjects to clean statistics from the effects of age.

A

Results: There were 846 cardiovascular deaths, 178 due to stroke, 273 to coronary disease, 351 to heart failure, and 44 to pulmonary embolism. Hypertension predicted stroke mortality, but not that due to other causes. This prediction was only significant in women, not in men. No prediction was possible after the age of 80 years. Agematching increased the significance level of stroke mortality prediction in women aged ⬍80 years; in these women, systolic BP predicted stroke mortality directly and diastolic inversely. Conclusions: In this population, hypertension predicted only stroke mortality in women aged ⬍80 years. High systolic and low diastolic BP were predictive of stroke mortality, confirming a prognostic role for high pulse pressure. Am J Hypertens 2002;15:958 –966 © 2002 American Journal of Hypertension, Ltd. Key Words: Hypertension, mortality, elderly, general population, women.

rterial hypertension is considered one of the most powerful risk factors, and the general belief is that it increases cardiovascular mortality both in men and women. The Framingham study gave the greatest contribution to this opinion.1–3 Although this concept is continuously re-echoed in the literature, the only way to demonstrate a negative prognostic effect of hypertension is represented by the studies based on general population that include hypertensive patients as well as normotensive all-age subjects. In fact, animal studies cannot be directly used for humans for evident reasons, and cohort studies cannot be generalized without introducing biases. And the number of populationbased studies dealing with this topic is limited. The present study aimed at evaluating 1) whether or not arterial hypertension predicts cardiovascular mortality at a population level; 2) if so, whether or not this is true for both genders and at all ages, and 3) which component of

cardiovascular mortality depends on blood pressure (BP) levels.

Received March 29, 2002. First decision May 9, 2002. Accepted June 20, 2002. From the Department of Clinical and Experimental Medicine, University of Padova, Padova, Italy.

Address correspondence and reprint requests to Dr. E. Casiglia, Department of Clinical & Experimental Medicine, University of Padova, Via Giustiniani, 2, 35128 Padova, Italy; e-mail: edoardo.casiglia@ unipd.it

0895-7061/02/$22.00 PII S0895-7061(02)03002-9

Methods Study Population A cohort of North Italian subjects aged 22 to 95 years were called in a population-based frame with a recruitment procedure only based on the Register Office files; 5185 subjects (ie, 73% of called ones) agreed with the study protocol, gave informed consent, and were recruited, whereas the remaining 27% did not answer or did not accept the study protocol. The subjects included and those not included in the study had similar age and gender distribution (data not shown). The study lasted 14 years and was carried out in accordance with the Declaration of Helsinki of the World Medical Association, and approved by the local Ethics Com-

© 2002 by the American Journal of Hypertension, Ltd. Published by Elsevier Science Inc.

AJH–November 2002–VOL. 15, NO. 11

STROKE MORTALITY IN HYPERTENSIVE WOMEN

959

FIG. 1. Plan of the study. Solid line) subjects who were untreated for hypertension at the initial screening; dashed line) treated subjects whose inclusion in the study was postponed (wash-out phase). The study lasted 168 months or until death in all subjects.

mittees. The procedures followed were in accordance with institutional guidelines. Data Collection A Hypertension Unit consisting of doctors expert in hypertension was activated for this study. Attention was given to assure that BP measurements were not biased by the residual effect of antihypertensive treatment (Fig. 1) . Those individuals who were being treated for hypertension (52%) underwent a 4-week wash-out period (representing the best compromise with a realistic development of the study). In such subjects, the beginning of the study was postponed, and the first BP values were taken at the end of this period. As shown in Fig. 1, the study lasted 168 months (or until death) in all cases. No drop-out was observed during the wash-out phase, due to assiduous observation by the Hypertension Unit and strict cooperation between the Hypertension Unit and the general practitioners. The schedule of BP measurements is shown in Fig. 1. Supine BP (Korotkoff phases 1 and 5) was taken eight times in 2 months (in triplicate at 15-min intervals at visit 1 and 2, in duplicate at 15-min intervals at visit 3). To reduce the white-coat effect and the alarm reaction, if any, only the average of the last two of the eight BP measurements was taken into consideration for the analysis of data and for the diagnosis of hypertension (discussed later). Only after visit 3, subjects labeled as hypertensive were free to receive antihypertensive treatment. This was necessary in 707 subjects. Because this is a long-lasting study, periodical controls were scheduled to determine how many subjects developed hypertension during the follow-up and to ascertain whether later antihypertensive treatment was necessary. Historical data were recorded. All subjects underwent physical, laboratory, and instrumental examination (electrocardiogram, echocardiogram, chest X-ray, spirometry, and myocardial scintigraphy when necessary).

as the cut-off point. Subjects were labeled as not old when aged less than 70 years; according to the literature, elderly subjects aged 70 years or more were further divided into old and very old (80 years and older).4 According to current guidelines,5 subjects were labeled as hypertensive (HT) when at the end of the wash-out phase the systolic BP was ⱖ140 mm Hg or the diastolic BP was ⱖ90 mm Hg, and as normotensive (NT) when both systolic and diastolic BP were below these values. Subjects with one of the following criteria: Minnesota code6 equal to 1.1, 1.2, or 1.3 (if absent 6.4.1) or 4.1 or 4.4 (if absent 6.4.1, 7.1.1, and 7.2.1) or 5.1, 5.2, 5.3, or 5.4 (if absent 6.4.1, 7.1.1, 7.2.1, and 7.4), positive myocardial scintigraphy, positive stress test, history of myocardial infarction confirmed by hospital files, history of angina pectoris confirmed by hospital or physician’s files, or appropriate antianginal chronic treatment were labeled as having coronary artery disease (CAD). Subjects having clear and permanent neurologic consequences or a positive history confirmed by hospital or physician’s files or a positive computed tomography were considered to have had a stroke. Atrial fibrillation required the Minnesota code 8.3.1 or 8.3.3; atrioventricular block, code 6.2; left bundle branch block, code 7.1.1 or 7.1.2; and left anterior hemiblock, code 7.7. Subjects having enlarged heart at chest X-ray, dyspnea, or peripheral edema were considered to have congestive heart failure (CHF). Body mass index (BMI) was calculated as the ratio of weight (in kilograms) to squared height (in meters). Subjects with repeated fasting blood glucose of ⬎7 mmol/L (126 mg/dL), a history of diabetes, or treatment with antidiabetic drugs were considered to be diabetic.7 Subjects with total cholesterol of ⬎5.2 mmol/L (200 mg/dL) were hypercholesterolemic,8 those with serum triglycerides ⬎1.7 mmol/L (150 mg/dL) were hypertriglyceridemic.9 A diagnosis of proteinuria10 required a urine level of 2 g/L.

Definitions and Cut-Off Values

Mortality Data

Because the definition of “elderly” is uncertain, the median age of the study population (70 years) was considered

Mortality monitoring was based on the Register Office (which in Italy is very accurate when death is concerned)

960

STROKE MORTALITY IN HYPERTENSIVE WOMEN

AJH–November 2002–VOL. 15, NO. 11

Table 1. General characteristics of the entire population and of the age- and gender-matched groups

Age (y) Males (%) Body mass index (kg/m2) Systolic blood pressure (mm Hg) Diastolic blood pressure (mm Hg) Arterial hypertension (%) Heart rate (beats/min) Blood glucose (mmol/L) Serum cholesterol (mmol/L) Serum triglycerides (mmol/L) Serum uric acid (mmol/L) Proteinuria ⬎2 g/L (%) Diabetes (%) History of stroke (%) History of CAD (%) Current smokers (%)

Entire Population (n ⴝ 5185)

Age-Matched Groups (n ⴝ 2182)

61.3 ⫾ 17.9 [61.1–62.1] 57.0 25.8 ⫾ 4.2 [25.7–25.9] 153.2 ⫾ 26.6 [152.5–153.9] 87.4 ⫾ 12.2 [87.1–87.8] 71.0 78.1 ⫾ 12.6 [77.7–78.2] 5.7 ⫾ 1.6 [5.6–5.9] 5.6 ⫾ 1.2 [5.5–5.7] 1.38 ⫾ 1.0 [1.3–1.4] 0.31 ⫾ 0.08 [0.30–0.32] 7.1 24.8 7.4 15.0 22.0

62.8 ⫾ 14.7 [60.1–61.5] 50.0 25.6 ⫾ 4.0 [25.4–25.8] 143.1 ⫾ 24.0 [142.0–144.2]* 85.2 ⫾ 11.3 [84.7–85.7]* 50.0* 78.0 ⫾ 12.9 [77.4–78.6] 5.7 ⫾ 1.5 [5.6–5.8] 5.6 ⫾ 1.1 [5.5–5.7] 1.3 ⫾ 0.9 [1.2–1.4] 0.30 ⫾ 0.08 [0.29–0.31] 6.1 23.2 8.5 15.0 26.2

CAD ⫽ coronary artery disease. 95% confidence intervals in brackets. * P ⬍ .05 v entire population.

and double-checked for causes of death by referring to hospitals, retirement homes, or physicians’ files. Deaths due to CAD (myocardial infarction or fatal angina pectoris), stroke, CHF, or pulmonary embolism were classified as cardiovascular, the other deaths were labeled as noncardiovascular. Statistical Analysis Descriptive Statistics Values of continuous variables were expressed as mean ⫾ standard deviation; 95% confidence intervals (95% CIs) were indicated when appropriate. Analysis of variance with the Bonferroni’s correction was used to compare grouped continuous variables, and the Pearson’s ␹2 test to compare the prevalence of categoric variables. Because age, number of men, BMI, blood glucose, serum cholesterol, triglycerides, and uric acid, as well as the prevalence of proteinuria, diabetes, smoking, and history of CAD and stroke, were higher in HT than in NT subjects, grouped BP values were always adjusted for these confounders. Predictive Role of Hypertension as a Categoric Item Annual mortality rates were compared with the Kaplan-Mayer approach after generating cumulative survival curves. Multivariate stepwise proportional hazard Cox regression was used to identify the variables having a prognostic role. The categoric variable hypertension was included as a covariate in such analyses. The analysis was first performed in the entire population, then for each gender separately. All variables were used in multivariate analysis, rather than their interaction; the covariates used to adjust for confounders are specified

time to time in the Results section (treatment of hypertension, present or absent, was always included as a covariate). Relative risk (RR) of hypertension was calculated for each mortality. Predictive Role of BP as a Continuous Item A further Cox analysis was performed using the continuous variables systolic BP and diastolic BP as covariates, instead of hypertension. The aim of this procedure was to evaluate whether, independent of the label of hypertension, mortality increased with increasing BP. This analysis was performed separately in NT subjects and then in HT patients, to clarify whether the quality of NT guaranteed a protection for any level of BP. Because systolic and diastolic BP were not independent to each other, their natural logarithm was used instead of original values. Adjustment for Age The role of hypertension was distinguished from that of age, which was achieved by age adjustment and age matching. Multivariate adjustment for age was obtained by including age as a covariate in all proportional hazard regressions of mortality. Age matching was obtained with the case-by-case method, searching in the entire population for two groups of subjects (NT and HT); each subject of the first group corresponded to someone in the second group with the same age. This way 1091 NT and 1091 age-matched HT subjects were identified. The 2182 matched subjects were obviously obtained with a nonrandom procedure; the experimental cohort created by this operation was nevertheless perfectly acceptable for the aims of the present study, having general characteristics comparable to those of the entire population apart from BP (Table 1).

AJH–November 2002–VOL. 15, NO. 11

STROKE MORTALITY IN HYPERTENSIVE WOMEN

961

Table 2. General characteristics of normotensive and hypertensive subjects Normotensive (n ⴝ 1487) Age (y) Body mass index (kg/m2) Systolic blood pressure (mm Hg) Diastolic blood pressure (mm Hg) Arterial hypertension (%) Heart rate (beats/min) Blood glucose (mmol/L) Serum cholesterol (mmol/L) Serum triglycerides (mmol/L) Serum uric acid (mmol/L) Proteinuria ⬎2 g/L (%) Diabetes (%) History of stroke (%) History of CAD (%) Current smokers (%)

48.7 24.0 124.4 76.5

⫾ ⫾ ⫾ ⫾

77.9 5.44 5.24 1.32 4.96

⫾ ⫾ ⫾ ⫾ ⫾

19.5 [47.1–9.5] 3.4 [23.8–24.1] 9.3 [123.9–124.9] 7.6 [76.1–76.9] 28.7 12.8 [77.3–78.6] 1.30 [5.38–5.53] 1.1 [5.19–5.3] 0.8 [1.30–1.44] 1.32 [4.8–5.0] 3.8 12.8 1.3 16.5 35.0

Hypertensive (n ⴝ 3698) 66.8 26.5 164.8 91.8

⫾ ⫾ ⫾ ⫾

78.1 6.02 5.73 1.69 5.3

⫾ ⫾ ⫾ ⫾ ⫾

14.2 [66.3–67.2] 4.2 [26.4–26.7] 22.2 [164.1–165.5] 10.8 [91.4–92.1] 71.3* 12.5 [77.7–78.5] 1.64 [5.97–6.08] 1.7 [5.7–5.8] 1.05 [1.66–1.73] 1.4 [5.2–5.4] 8.5* 29.8* 2.9* 19.0* 16.7*

Abbreviation as in Table 1. 95% confidence intervals in brackets. * P ⬍ .05 v normotensive subjects.

Results General Characteristics of the Population The general characteristics of the study population are summarized in Table 1. Mean age was 61.6 ⫾ 17.9 years (range, 22 to 95 years), median age 70 years; 2932 subjects (1541 men and 1391 women) were not old, 1859 (1135 and 724, respectively) old, and 394 (274 and 120, respectively) very old; 1487 were NT and 3698 HT. The general characteristics of the NT and HT subjects are shown in Table 2. The BP values by age classes are shown in Fig. 2. Prevalence of hypertension was 57.2% in the not old, 86.5% in the old (P ⬍ .0001 v not old), and 85% in the very old subjects (P ⬍ .0001 v not old, insignificant difference versus old). Mortality Rates: Univariate Unadjusted Analysis During the 14 years of follow-up there were 1724 deaths (overall mortality rate, 33.2%); 846 were cardiovascular

FIG. 2. Systolic and diastolic blood pressure values by age. * Indicates P ⬍ .01 v age class ⬍30 years.

(178 due to stroke [67 men, 111 women], 273 to CAD [105 men, 168 women], 351 to CHF [144 men, 207 women], and 44 to pulmonary embolism [16 men, 28 women], and 878 noncardiovascular deaths. Rates are summarized in Table 3. At univariate analysis, mortality due to stroke, CAD, and CHF was apparently higher, both in men and women, in HT than in NT subjects (P ⬍ .01). The same trend was observed for noncardiovascular mortality, suggesting a nonspecific (noncardiovascular) trend probably attributable to the difference in age of the two categories of subjects (48.7 ⫾ 15.6 years [95% CI 47.7– 49.7] in the NT group, 66.3 ⫾ 14.2 years [95% CI 66.3– 67.3] in the HT group, P ⬍ .0001). Analysis of Mortality Corrected for the Effects of Age The correction for the effects of age was performed by age adjustment and age matching. These two methods gave concordant results. Age Adjustment The multivariate RR of hypertension, calculated after adjusting for age, are shown in Fig. 3. When the entire population was taken into account (top panel), arterial hypertension had a prognostic role only for stroke mortality (RR 1.40, 95% CI 1.01–2.3), whereas CAD and CHF mortality were not predicted by hypertension. A subsequent gender-specific analysis (middle and bottom panels) showed that the effect of arterial hypertension on stroke mortality was only present in women (RR 1.90, 95% CI 1.02– 4.12); in men no prediction of mortality by the label of hypertension was possible. Age Matching To confirm once more that the results of univariate analysis were influenced by age, a cohort of 2182 subjects case-to-case matched for age and gender

962

Causes of Death Stroke

Coronary Disease

Entire population (1487 normotensives and 3698 hypertensives) 2.5% 4.0% 0.6% 10.2% 6.4%* 7.9%* 0.9% 19.6%*

Women (n ⫽ 2950)

{ 694 NT (age 49.0 ⫾ 19.2 y) { 1541 HT (age 66.8 ⫾ 15.7 y)* 793 NT (age 48.4 ⫾ 19.9 y) { 2157 HT (age 69.0 ⫾ 12.5 y)*

1.2% 4.3%*

All (n ⫽ 2182)

1091 NT (age 55.2 ⫾ 18.2 y) 1091 HT (age 55.2 ⫾ 18.1 y)

All (n ⫽ 5185) Men (n ⫽ 2235)

Men (n ⫽ 1154) Women (n ⫽ 1028)

1487 NT (age 48.7 ⫾ 19.6 y) 3968 HT (age 66.8 ⫾ 14.2 y)*

{ { {

Heart Failure

Pulmonary Embolism

Other Causes

Overall Mortality 18.5% 39.2%*

1.6% 3.6%*

1.7% 6.0%*

4.3% 7.4%*

0.4% 0.8%

12.1% 23.9%*

20.2% 41.9%*

0.9% 4.8%*

3.1% 6.6%*

3.7% 8.2%*

0.8% 1.0%

8.6% 16.6%*

17.0% 37.3%*

1.6% 3.1%*

1091 normotensives and 1091 age-matched hypertensives 3.4% 5.2% 0.8% 13.7% 3.6% 4.9% 0.4% 14.2%

24.7% 26.2%

577 NT (age 57.2 ⫾ 18.5 y) 577 HT (age 57.2 ⫾ 18.5 y)

1.9% 2.6%

2.1% 2.4%

5.0% 4.2%

0.5% 0.7%

14.4% 17.0%

23.9% 27.0%

514 NT (age 58.1 ⫾ 17.4 y) 514 HT (age 58.1 ⫾ 17.4 y)

1.4% 3.7%*

4.9% 4.9%

5.4% 5.8%

1.2% 0%

12.8% 10.9%

25.7% 25.3% AJH–November 2002–VOL. 15, NO. 11

NT ⫽ normotensive; HT ⫽ hypertensive. * P ⬍ .01 v NT.

STROKE MORTALITY IN HYPERTENSIVE WOMEN

Table 3. Fourteen-year mortality rates (%) in the entire population and in a cohort of 1091 normotensive (NT, systolic blood pressure ⬍140 mm Hg and diastolic ⬍90 mm Hg) and 1091 age-matched hypertensive (HT, systolic blood pressure ⱖ140 mm Hg or diastolic ⱖ90 mm Hg) untreated subjects

AJH–November 2002–VOL. 15, NO. 11

STROKE MORTALITY IN HYPERTENSIVE WOMEN

963

FIG. 3. Stroke, coronary artery disease (CAD), and coronary heart failure (CHF) mortality in relation to the label of hypertension. Age-adjusted relative risk (RR) of hypertension among 1487 normotensive (E) and 3698 hypertensive subjects (●) are shown. *P ⬍ .001 v normotensive.

was selected (among the matched subjects, prevalence of hypertension was by definition 50%, ie, lower than in the entire study population, and BP values were consequently lower, but as previously mentioned, the other characteristics were comparable; see Table 1). In this experimental matched cohort there were 556 deaths (14-year rate 25.5%); 252 of them were cardiovascular (52 due to stroke, 76 to CAD, 111 to CHF, and 13 to pulmonary embolism). Rates are shown in Table 3. Only stroke mortality was significantly higher in HT than in NT matched subjects, and only in women (RR of hypertension 2.57 95% CI 1.04 – 6.36). Also in this cohort, no prediction of mortality was possible for men, either based on the label of hypertension or on BP values.

ous variables, separately, among NT women and then among HT women, they had a role only in HT women. Among HT women, the Cox coefficient was positive for systolic (⫹6.4 ⫾ 2.1, indicating a worsening prognostic role of high systolic BP) and negative for diastolic BP (⫺5.6 ⫾ 2.2, indicating a protective effect of a high diastolic BP). After dividing women into classes of age, arterial hypertension predicted stroke mortality in not old and old, but not in very old women (Fig. 4).

Mortality Model in Women

On the basis of univariate analysis, arterial hypertension appeared to be a predictor of stroke, CAD, and CHF mortality (Table 3). Apart from stroke, this was only a bias due to age; HT patients were, in fact, older than the NT and therefore more prone to die of cardiovascular disease (as well as of any other cause) than the NT group. After the effect of age was removed with age correction or even better with age matching, only stroke mortality remained significantly and multivariately influenced by hypertension. In the experimental conditions represented by two age-matched groups (in which the between-group effects of age were completely abolished), the risk of stroke mortality was double in HT patients than in NT subjects, whereas those of CAD and CHF were not statistically different in the two groups. Therefore, the present data create the impression that stroke is a direct consequence of hypertension, at least in women, whereas the relationship

Apart from hypertension, four other items were multivariate predictors of stroke mortality in women, namely atrial fibrillation (RR 3.35, 95% CI 1.93–5.80), diabetes (RR 1.86, 95% CI 1.28 –2.70), history of stroke (RR 2.61, 95% CI 1.54 – 4.43), and history of CAD (RR 1.46, 95% CI 1.01–2.25). Other independent variables such as hypercholesterolemia, hypertriglyceridemia, smoking habit, BMI, proteinuria, heart rate, intermittent claudication, coffee and alcohol consumption, atrioventricular block, left bundle branch block, left anterior hemiblock, and antihypertensive treatment did not reach statistical significance (ie, did not play any prognostic role). Values of multivariate ␹2 and probability of each significant covariate are summarized in Table 4. When systolic and diastolic BP were used as continu-

Discussion Hypertension Predicts Only Stroke Mortality

964

STROKE MORTALITY IN HYPERTENSIVE WOMEN

AJH–November 2002–VOL. 15, NO. 11

Table 4. Results of the multivariate analysis for the dependent variable “stroke mortality” among women Covariates

All (n ⫽ 2950) NT (n ⫽ 793) HT (n ⫽ 2157)

2

{ ␹P ⬍ { ␹P ⬍ { ␹P ⬍ 2

2

SBP

DBP

Age

AF

Diabetes

History of Stroke

History of CAD

7.15* .007 NS NS 3.53‡ .05

5.35† .02 NS NS 6.28§ .01

117.60 ⬍.0001 32.43 ⬍.0001 58.12 ⬍.0001

16.34 ⬍.0001 NS NS 15.29 ⬍.0001

11.23 .001 NS NS 8.26 .005

10.88 .001 3.99 .005 9.17 .005

8.21 .004 NS NS 7.98 .005

SBP ⫽ systolic blood pressure; DBP ⫽ diastolic blood pressure; AF ⫽ atrial fibrillation (Minnesota code 8.3.1 or 8.3.3); NS ⫽ not significant; other abbreviations as in Table 1–3. ␹2 and P values of the accepted covariates are indicated. For SBP and DBP, the Cox coefficients [with standard errors in brackets] are also indicated; negative coefficient indicates protective effect. * Coefficient 6.8 [SE 1.9]; † coefficient ⫺5.0 [SE 2.1]; ‡ coefficient 6.4 [SE 2.1]; § coefficient ⫺5.6 [SE 2.2].

between hypertension and CAD or between hypertension and CHF, if any, is largely influenced by age. This is not the first notation in this respect, as not all the studies demonstrated a clear contribution of hypertension to development of cardiovascular risk, particularly in the elderly. This is pointed out both by epidemiologic studies and clinical trials. The Framingham researchers, who notoriously sustain the importance of BP in cardiovascular risk, in a recent reanalysis had to conclude that CAD mortality does not depend on BP in a linearly increasing manner.11 In the National Health and Nutrition Examination Survey, diastolic and mean BP were not predictive of cardiovascular

mortality,12 and Glynn13 and Langer14 –16 and their colleagues even found—in separate studies—a paradoxically higher survival at higher BP, particularly in old and very old subjects. Pekkanen et al,17 studying 1619 men aged 40 to 59 years at baseline, found that BP was no longer associated with coronary risk after 20 years of follow-up, when the subjects had become old. Finally, Staessen et al18 found an absence of a relation between CAD events and systolic BP in untreated patients. Finally, although the intervention trials (obviously including hypertensive patients only) are not a good model for determining the role of hypertension as a predictor, meta-analysis of randomized large-scale trials interestingly indicated that reduction of CAD events in treated hypertensive patients could be lower than expected.19 A recent meta-analysis of 4233 persons with hypertension even showed an increased risk of events in those in whom the lowest BP was achieved.20 The Julius’ opinion is that “ . . . BP lowering does not equally affect all complications of hypertension. Particularly bothersome is the fact that treatment decreases stroke but fails to substantially reduce CAD events. These variations in treatment outcomes may reflect the multifactorial pathophysiology of hypertension: essential hypertension is frequently associated with BP-independent coronary risk factors.”21 Is this sufficient to conclude that the some subjects are immune from hypertension and even protected by high BP values? We do not take a position in this debate. Role of Gender

FIG. 4. Fourteen-year stroke mortality rate (%) in relation to the label of hypertension after age stratification in women. In the age class ⬍70 years there are 626 normotensives (■) and 915 hypertensives (䊐), in the age class 70 to 79 years 122 (■) and 943 (䊐), in the age class ⬎80 years 45 (■) and 299 (䊐). RR ⫽ multivariate relative risk of hypertension. * Multivariate P ⬍ .0001 between (■) and (䊐). NS ⫽ nonsignificant difference.

Our data clearly show that hypertension had no influence on stroke mortality in men. The reason for this can only be the object of speculation. Women seem to be more prone to cerebral ischemia after nocturnal BP falls22 and less responsive to stroke prevention with acetylsalicylic acid.23 Furthermore, women’s arrival to the hospital after a stroke is usually delayed in comparison to men,24 and finally, women probably have a greater incidence of stroke com-

AJH–November 2002–VOL. 15, NO. 11

plications.25 Whether or not such characteristics are sufficient to explain the higher stroke mortality observed in women in response to hypertension remains to be clarified. Role of Advanced Age Another piece of evidence coming from the present study is that, among women, the weight of high BP on stroke mortality was declining with age (RR of hypertension was 4.78 before 70 years, 3.69 between 70 and 79 years, and 0.37 [insignificant] in women aged 80 years or older). One possible explanation is that the relative risk of hypertension decreases with increasing age simply because the absolute risk of age increases. The covariate hypertension is overwhelmed (both statistically and biologically) by the more powerful covariate of age. This is probably why many investigators found that the importance of arterial hypertension as a cardiovascular risk factor is less pronounced20 –26 or even inverted14,15 with the advanced age. Another possible explanation is that elderly persons are the survivors of a population where a significant mortality due to hypertension-related complications has already made its mark, eliminating those having the worst risk pattern or more subjected to experiment lethal consequences of high BP. BP as a Continuous Variable As pointed out, the present study demonstrates that the label of HT was predictive of stroke mortality in women. When BP was considered as a continuous variable, systolic BP had a prognostic role on stroke mortality among HT women, with a Cox coefficient of 6.4 ⫾ 2.1 (the higher systolic BP, the greater the stroke mortality). This was not the case in NT women, where no systolic value below 140 mm Hg was dangerous, thus indirectly validating the reliability of this upper limit of BP normalcy. At variance with systolic BP, the Cox coefficient of diastolic BP was negative in HT women, indicating a worse survival with decreasing diastolic values. Accordingly, many researchers demonstrated that low diastolic BP values, especially when associated with a high systolic BP (high pulse pressure) can increase the risk of stroke27,28 with a higher incidence of carotid stenosis29,30 or a reduction in cerebral flow.31,32 The studies of Langer et al14 and of Benetos et al,33 showing a higher mortality in treated HT having a prevalent decrease in diastolic BP, are in keeping with this view. In conclusion, in a population such as that reported in this study arterial hypertension predicts stroke mortality in women but not in men. Coronary artery disease, CHF, or pulmonary embolism mortality is not predicted by hypertension in either men or women. More in details, arterial hypertension predicts stroke mortality only in HT women aged less than 80 years. Stroke mortality increases with increasing BP values in HT women; this does not happen in NT women, where stroke mortality does not increase at

STROKE MORTALITY IN HYPERTENSIVE WOMEN

965

any age with increasing BP until the cut-off of 140/90 mm Hg is reached.

References 1. 2. 3.

4.

5.

6.

7. 8. 9.

10.

11. 12.

13.

14.

15.

16.

17.

18.

19. 20.

21.

Kannel WB: Fifty years of Framingham Study contribution to understanding hypertension. J Hum Hypert 2000;14:83–90. Kannel WB: Elevated systolic blood pressure as a cardiovascular risk factor. Am J Cardiol 2000;15:251–255. Franklin SS: Cardiovascular risks related to increased diastolic, systolic and pulse pressure. An epidemiologist’s point of view. Pathol Bil 1999;47:594 –603. Dontas AS, Toupadaki N, Tzonou A, Kasviki Charvati P: Survival in the oldest old: Death risk factors in old and very old subjects. J Aging Health 1996;8:220 –237. Joint National Committee on Prevention Detection, Evaluation and Treatment of High Blood Pressure: The sixth report of the Joint National Committee on prevention, detection, evaluation and treatment of high blood pressure. Arch Intern Med 1997;157:2413–2446. Kors JA, Crow RS, Hannan PJ, Rautaharju PM, Folsom AR: Comparison of computer-assigned Minnesota codes with the visual standard method for new coronary heart disease events. Am J Epidemiol 2000;151:790 –797. Casiglia E, Palatini P: Cardiovascular risk factors in the elderly. J Hum Hypertens 1998;12:575–581. Thelle DS: Epidemiology of hypercholesterolemia and European management guidelines. Cardiology 1990;77(4 Suppl):42–47. Mantero Atienza E, Beach RS, Sotomayor MG, Christakis G, Baum MK: Nutritional status of institutionalized elderly in south Florida. Arch Latinoam Nutr 1992;42:242–249. Olivarius ND, Mogensen CE: Danish general practitioners’ estimation of urinary albumin concentration in the detection of proteinuria and microalbuminuria. Br J Gen Pract 1995;45:71–73. Port S, Demer L, Jennrich R, Walter D, Garfinkel A: Systolic blood pressure and mortality. Lancet 2000;355:175–180. Fang J, Madhavan S, Alderman MH: Pulse pressure: A predictor of cardiovascular mortality among young normotensive subjects. Blood Press 2000;9:260 –266. Glynn RJ, Che CU, Guralnik JM, Taylor JO, Hennekens CH: Pulse pressure and mortality in older people. Arch Intern Med 2000;160: 2765–2772. Langer RD, Ganiats TG, Barrett-Connor E: Paradoxical survival of elderly men with high blood pressure. Br Med J 1989;298:1356 – 1357. Langer RD, Ganiats TG, Barrett-Connor E: Factors associated with paradoxical survival at higher blood pressures in the very old. Am J Epidemiol 1991;134:29 –38. Langer RD, Criqui MH, Barrett Connor EL, Klauber MR, Ganiats TG: Blood pressure change and survival after age 75. Hypertension 1993;22:551–559. Pekkanen J, Tervahauta M, Nissinen A, Karvonen MJ: Does the predictive value of baseline coronary risk factors change over a 30-year follow-up? Cardiology 1993;82:181–190. Staessen JA, Gasowski J, Wang JG, Thijs L, Den-Hond E, Boissel JP, Coope J, Ekbom T, Gueyffier F, Liu L, Kerlikowske K, Pocock S, Fagard RH: Risks of untreated and treated isolated systolic hypertension in the elderly: Meta-analysis of outcome trials. Lancet 2000;385:865–872. McVeigh GE, Flack J, Grimm R: Goals of antihypertensive therapy. Drugs 1995;49:161–175. Boutitie F, Gueyffier F, Pocock S, Fagard R, Boissel JP, for the INDANA Project Steering Committee: J-shaped relationship between blood pressure and mortality in hypertensive patients: New insights from the meta-analysis of individual patients date. Ann Intern Med 2002;136:438 –448. Julius S: Five decades of antihypertensive treatment: The unresolved issues. J Hypertens 2000;18(Suppl):3–7.

966

STROKE MORTALITY IN HYPERTENSIVE WOMEN

22. Imai Y, Tsuji I, Nagai K, Watanabe N, Ohkubo T, Sakuma M, Hashimoto J, Itoh O, Satoh H, Hisamichi S, Abe K: Circadian blood pressure variation related to morbidity and mortality from cerebrovascular and cardiovascular diseases. Ann NY Acad Sci 1996;783: 172–185. 23. Barnett NJ: Stroke in women. Can J Cardiol 1990;6:(Suppl B):11– 17. 24. Weaver WD, White HD, Wilcox RG, Aylward PE, Morris D, Guerci A, Ohman EM, Barbash GI, Betriu A, Sadowski Z, Topol EJ, Califf RM: Comparisons of characteristics and outcomes among women and men with acute myocardial infarction treated with thrombolytic therapy. JAMA 1996;275:777–782. 25. Abramov D, Tamariz MG, Sever JY, Christakis GT, Bhatnagar G, Heenan AL, Goldman BS, Fremes SE: The influence of gender on the outcome of coronary artery bypass surgery. Ann Thorac Surg 2000;70:800 –805. 26. Kannel WB: Hypertension as a risk factor for cardiac events— epidemiologic results of long-term studies. J Cardiovasc Pharmacol 1993;21(Suppl 2):27–37. 27. Voko Z, Bots ML, Hofman A, Koudstaal PJ, Witteman JC, Breteler MM: J-shaped relation between blood pressure and stroke in treated hypertensives. Hypertension 1999;34:1181–1185.

AJH–November 2002–VOL. 15, NO. 11

28. Selmer R: Blood pressure and twenty-year mortality in the city of Bergen, Norway. Am J Epidemiol 1992;136:428 –440. 29. Sutton-Tyrrell K, Alcorn HG, Wolfson SK, Kelsey SF, Kuller LH: Predictors of carotid stenosis in older adults with and without isolated systolic hypertension. Stroke 1993;24:355–361. 30. Suurkula M, Agewall S, Fagerberg B, Wendelhag I, Widgren B, Wikstrand J: Ultrasound evaluation of atherosclerotic manifestations in the carotid artery in high-risk hypertensive patients. Risk Intervention Study (RIS) Group. Arterioscler Thromb 1994;14:1297–1304. 31. Millar JA, Lever AF, Burke V: Pulse pressure as a risk factor for cardiovascular events in the MRC Mild Hypertension trial. J Hypertens 1999;17:1065–1072. 32. Lee ML, Rosner BA, Weiss ST: Relationship of blood pressure to cardiovascular death: the effects of pulse pressure in the elderly. Ann Epidemiol 1999;9:101–107. 33. Benetos A, Zureik M, Morcet J, Thomas F, Bean K, Safar M, Ducimetie`re P, Guize L: Decrease in diastolic blood pressure combined with an increase in systolic blood pressure is associated with a higher cardiovascular mortality in men. J Am Coll Cardiol 2000; 35:673–680.