Author + information
- Received December 4, 2000
- Revision received March 19, 2001
- Accepted April 5, 2001
- Published online July 1, 2001.
- Robert H Fagard, MD, PhDa,* (, )
- Karel Pardaens, PhDa,
- Jan A Staessen, MD, PhDa and
- Lutgarde Thijs, MSca
- ↵*Reprint requests and correspondence:
Dr. Robert H. Fagard, U. Z. Gasthuisberg-Hypertensie, Herestraat 49, B-3000 Leuven, Belgium
The goal of this study was to assess the prognostic power of the pulse pressure-to-stroke index (PP-to-SVi) ratio for cardiovascular events and mortality in patients with uncomplicated hypertension.
The prognostic significance of pulse pressure (PP) has been studied repeatedly, but few data are available on the PP-to-SVi ratio.
Invasive hemodynamic measurements, including brachial intra-arterial pressure and stroke index by the direct oxygen Fick method, were performed in the period 1972 to 1982 in 192 patients with uncomplicated hypertension; their outcome was ascertained in 1994.
Age at baseline averaged 37 ± 12 years; brachial artery pressure was 165 mm Hg ± 30/89 ± 17 mm Hg; PP averaged 76 mm Hg ± 18 mm Hg, and the PP-to-SVi ratio was 1.67 mm Hg/(ml/m2) ± 0.73 mm Hg/(ml/m2). During 3,057 patient years of follow-up, 19 patients died, and 44 experienced at least one fatal or nonfatal cardiovascular event. Cox regression analysis revealed that the PP-to-SVi ratio was a significant predictor of fatal and nonfatal cardiovascular events and of all-cause mortality after control for age and gender (p < 0.01). Its predictive power persisted after additional adjustment for mean arterial pressure and heart rate. Each 0.75-mm Hg/(ml/m2) increase in the PP-to-SVi ratio was independently associated with a 79% increase in the risk of a cardiovascular event (p = 0.01) and a 2.05-fold greater risk of all-cause mortality (p = 0.01).
The PP-to-SVi ratio is a significant and independent predictor of cardiovascular events and mortality in selected patients with uncomplicated hypertension.
In the last decade, a number of studies have provided strong evidence that pulse pressure (PP), measured as the difference between systolic and diastolic blood pressure (BP) at the level of the brachial artery, is an independent predictor of future left ventricular (LV) mass, vascular hypertrophy and cardiovascular morbidity and mortality (1–15). A high PP results, at least partly, from stiffening of the arterial tree, which leads to an increase of systolic BP and a decrease of diastolic BP at any given level of mean BP. Thus, PP has been used as a crude index of arterial stiffness, or, conversely, of arterial compliance in these prognostic studies. However, arterial stiffness is defined as the relationship between changes in pressure and volume—and vice versa for arterial compliance—so that consideration of volume changes could improve the estimate of the vascular wall properties. Volume changes of the arterial system are difficult to measure in man, but stroke volume (SV) has been used as a substitute for these changes. Therefore, the ratio of SV to PP has been proposed as an estimate of arterial compliance (16). In spite of a number of theoretical objections to the use of this ratio or its reciprocal, Randall et al. (17)and Chemla et al. (18)have shown that it correlates well with more standard estimates of arterial compliance. The purpose of the present study was: 1) to identify the prognostic value of the pulse pressure-to-stroke index (PP-to-SVi) ratio in patients with uncomplicated hypertension, and 2) to investigate whether this ratio would be better than PP alone with regard to the prediction of mortality and cardiovascular (CV) events. We applied invasive gold standard techniques for the measurement of brachial artery PP and stroke index (SVi) and used the PP-to-SVi ratio rather than the inverse because the development of pressure results from the ejection of blood in the arterial tree by the LV. The age-dependent difference between peripheral and central PP was accounted for by inclusion of age in a multivariable regression analysis.
This analysis includes patients who were referred to the Hypertension Unit of the University Hospitals of Leuven between 1972 and 1982 and who underwent a hemodynamic study for research purposes. Routine investigations included history, clinical examination, appropriate laboratory tests, eye-fundus examination, chest X-ray, electrocardiography at rest, exercise testing, pulmonary function tests, intravenous urography and renal arteriography when indicated. Patients were excluded when World Health Organization (WHO) stage III organ damage was present at the time of the investigation (19). Patients with evidence of ischemic heart disease, heart failure, stroke, claudication, retinal exudates or hemorrhages or renal insufficiency were excluded. No patient had valvular heart disease. All patients were in sinus rhythm, and none had evidence of pulmonary disease or diabetes. Patients had never been treated for high BP or had their antihypertensive medication stopped for two weeks or more. The Ethics Committee of the Faculty of Medicine approved the hemodynamic studies, and participants gave informed consent after the nature of the procedures had been explained.
All hemodynamic measurements were performed in the morning in the same laboratory, where room temperature was 18°C to 22°C, a few days after hospital admission. The brachial artery was cannulated to measure systolic and diastolic intra-arterial pressure and to sample arterial blood; PP was the difference between systolic and diastolic BP, and mean BP was obtained from electrical damping. A venous catheter (Swan-Ganz, 93.110.5F) was introduced in the antecubital vein and positioned in the pulmonary artery to sample mixed venous blood. Pressures were registered on a Mingograph 81 recorder (Siemens Medical, Germany). Cardiac output (l/min) was determined by the direct oxygen Fick method as described (20)and normalized for body size as cardiac output divided by body surface area (cardiac index; [l/min]/m2). Heart rate ([HR] beats/min) was recorded from the electrocardiogram. Stroke index (ml/m2) was determined from cardiac index and HR. The PP-to-SVi ratio (mm Hg/[ml/m2]) was then calculated as the ratio of PP and SVi. The hemodynamic measurements were obtained with patients in supine rest 30 min after the technical procedures.
After the baseline examination, the patients were referred to their usual source of care. Follow-up with regard to vital status, causes of death and occurrence of cardiovascular events was performed in 1989 and in 1994, as described in detail (20,21). Events were coded according to the Ninth Revision of the International Classification of Diseases. The following CV events, known to be associated with hypertension (22), were considered: sudden death, myocardial infarction, cerebrovascular accident, heart failure, new-onset angina pectoris, transient ischemic attack and peripheral vascular disease. Objective evidence was required for acceptance as described (21). The vital status of all patients could be determined in 1994 (20). However, two living patients could not be traced in 1994, so their data from the 1989 survey (21)were used in this analysis.
Database management and statistical analysis were performed with the SAS software (SAS Institute Inc., Cary, North Carolina). Data are reported as mean ± SD or median and range. The unpaired Student ttest and the chi-square statistic were used for the comparison of groups. Multivariable regression analysis was applied to identify the determinants of PP and the PP-to-SVi ratio. The Cox proportional hazards regression model was used for survival analysis (23,24). Two categories of end points were considered: 1) all-cause mortality, and 2) fatal and nonfatal CV events combined. In patients with more than one CV event, only the first event was considered. The prognostic value of a variable is given by its relative hazard rate (RHR). We estimated how much the incidence of the events would change for approximately one standard deviation increase of PP, SVi and of the PP-to-SVi ratio, respectively. The principal covariates considered in multivariable regression analysis were: age, age squared, gender, mean BP and HR. Two-sided p values of ≤0.05 were considered statistically significant.
Characteristics at baseline
Age of the 192 subjects averaged 37 ± 12 years, body mass index 25.9 kg/m2± 3.6 kg/m2and body surface area 1.85 m2± 1.18 m2; 68% were men. Conventional BP on admission to the hospital was 180 mm Hg ± 31/111 ± 20 mm Hg. Of the patients, 29% had never received antihypertensive drugs, and treatment was interrupted at least two weeks before the baseline investigations in the others. The majority of the patients had essential hypertension (n = 165); 10 had a renal artery stenosis, which was considered significant in three, and 17 had evidence of renal parenchymal disease but without renal insufficiency. Serum creatinine averaged 93.8 μmol/l ± 17.7 μmol/l. Serum cholesterol averaged 5.59 mmol/l ± 1.11 mmol/l, and the sum of the electrocardiographic voltages SV1+ RV5averaged 3.08 mV ± 1.04 mV; 39% were current smokers. Heart rate and intra-arterial BP averaged 80 beats/min ± 14 beats/min and 165 mm Hg ± 30/89 ± 17 mm Hg, respectively, at the hemodynamic investigation. Pulse pressure was 76 mm Hg ± 18 mm Hg, SVi 49.7 ml/m2± 13.8 ml/m2and the PP-to-SVi ratio 1.67 mm Hg/(ml/m2) ± 0.73 mm Hg/(ml/m2). Table 1shows that age, age squared, gender and mean BP explained 51% of the variance of PP, whereas 60% of the variance of the PP-to-SVi ratio was explained by age, age squared, HR and mean BP; there was no significant effect of etiology of hypertension.
Total follow-up time, that is, time until death or until the last available information, amounted to 3,057 patient years, with a median follow-up of 16.5 (range 0.7 to 21.1) years. Fifteen men and four women died at an average age of 53.3 (range 32.5 to 78.1) years or 6.9 (range 0.7 to 20.1) years after the baseline examination. Thirty-two men and 12 women suffered at least one fatal or nonfatal CV event, the first event occurring at the age of 56.6 (range 32.5 to 78.1) years or a median of 8.3 (range 0.7 to 2.2) years after the baseline studies. Table 2summarizes the causes of death and the first occurring CV events.
Baseline characteristics of patients with and patients without events during follow-up
Table 3shows that, compared with survivors without a CV event, patients who suffered a CV event or died were older and had lower SVi and higher BP, PP and a higher PP-to-SVi ratio.
Prognostic value of PP and of the PP-to-SVi ratio
Both age and its quadratic term were required to describe the relation of CV events with age, but age squared was not significant for all-cause mortality. Etiology of hypertension was not a significant predictor of outcome. As shown in Table 4, PP significantly predicted CV events (p = 0.03) and death (p = 0.02) after control for age and gender, but its prognostic power did not persist after additional adjustment for mean BP (p ≥ 0.13). By contrast, the PP-to-SVi ratio was a strong predictor of both CV events (p = 0.007) and mortality (p < 0.001), which remained significant after adjustment for mean BP and HR; the adjusted RHR associated with a 0.75 mm Hg/(ml/m2) higher PP-to-SVi ratio was 1.79 (p = 0.01) for CV events and 2.05 (p = 0.01) for all-cause mortality. Adding electrocardiographic voltages to these models did not affect the results.
When the PP-to-SVi ratio, age, age squared, gender, mean BP and HR were offered to the models, multivariable regression analysis retained the PP-to-SVi ratio (p = 0.007), gender (p < 0.001), age (p = 0.01) and age squared (p = 0.03) as significant predictors of CV events and the PP-to-SVi ratio (p < 0.001), gender (p = 0.001) and age (p = 0.08) for the prediction of all-cause mortality.
Stroke index alone did not predict the incidence of CV events after control for age and gender (p = 0.40). The RHR was, however, significant for mortality (p = 0.02) and amounted to 0.42 (95% confidence interval: 0.20 to 0.87) per 15 ml/m2increase of SVi; the RHR was 0.49 (p = 0.07) after additional adjustment for mean BP. When PP and SVi were offered separately to the models, together with age, age squared, gender, mean BP and HR, PP (RHR: 1.50; p = 0.03), but not SVi, predicted the incidence of CV events in addition to age, age squared and gender. By contrast, both PP (RHR: 1.80; p = 0.04) and SVi (RHR: 0.47; p = 0.04) predicted death in addition to age and gender.
The main finding of this study is that the PP-to-SVi ratio is a significant and independent predictor of CV events and all-cause mortality in hypertensive patients with limited target organ damage (WHO stage ≤II) and no associated cardiovascular diseases or other clinical conditions (19,25)at baseline. The results, furthermore, suggest that the PP-to-SVi ratio may be a more robust predictor of outcome than PP alone in this study population.
Prognostic value of PP and of the PP-to-SVi ratio
The prognostic value of PP has been examined in the population (1,4,8,9,12,14), in hypertensive patients (2,3,5,7,10,13,15), in survivors of myocardial infarction (6)and in patients with LV dysfunction (11). Pulse pressure was found to be predictive for future LV mass and vascular hypertrophy (5)and of cardiovascular morbidity and mortality (1–4,6–15). Overall, its prognostic value remained significant after adjustment for variable sets of risk factors, which usually included systolic, diastolic or mean BP. Pulse pressure was derived from conventional upper arm BP measurements in most studies. Two studies applied ambulatory BP monitoring (5,7), and the results suggested that the prognostic value of PP can only be partially attributed to the alerting reaction evoked by the clinic visit. Recently, de Simone et al. (26)assessed SVi noninvasively by use of echocardiography and derived PP from conventionally measured BP in 294 men and women with arterial hypertension. Through a mean 10-year follow-up period, 50 subjects experienced at least one CV complication, and 14 subjects died. The stroke index-to-pulse pressure ratio (SVi-to-PP) significantly predicted the incidence of all CV events but not of fatal events. In this study we assessed the PP-to-SVi ratio by invasive techniques, that is, intra-arterial measurement of BP and determination of SVi by the direct oxygen Fick method, which is considered the gold standard for hemodynamic evaluations. We found that the invasive PP-to-SVi ratio was a significant predictor of CV events and of all-cause mortality, which was independent of age, gender, mean BP and HR. Moreover, the PP-to-SVi ratio appeared to be a more robust predictor than PP alone, which lost its predictive power when mean BP was included in the model. This could result from the fact that the PP-to-SVi ratio better reflects vascular wall properties because it accounts for the dependence of PP on SVi. Stiffening of the arterial tree is due to loss of elasticity of the vascular wall and to atherosclerosis, which is likely to precede the occurrence of clinical CV events and death. On the other hand, the independent prognostic power of a low SVi, at least for mortality, may have contributed to the prognostic power of the PP-to-SVi ratio. Although none of our patients had a history, signs or symptoms of heart failure, a low SVi may reflect impaired LV function.
A number of theoretical objections have been raised against the use of the SVi-to-PP ratio, or its reciprocal, as a measure of arterial compliance or stiffness (17,27). First, only part of the SV contributes to the increase in arterial volume within the time that peak pressure is reached. This portion of SV is, in fact, not known, but it is assumed that SV is closely related to the relevant change in volume. Second, SV depends on LV function and a number of other factors, but it can be argued that this objection is of lesser importance because PP results from the actual increase in volume at the time of the peak pressure. In spite of these objections, it has been shown that the SVi-to-PP ratio is strongly related to more standard assessments of arterial compliance in studies in animals (17)and in man (18). As in all studies on the prognostic power of PP in man, we have calculated PP as the difference between systolic and diastolic BP measured at the level of the brachial artery, whereas aortic pressure would have been more appropriate. Peripheral PP is indeed higher than PP derived from central measurements, and the difference decreases with advancing age. It is likely that the inclusion of age in the multivariable models accounted, at least partly, for this effect. Whereas the prognostic value of the PP-to-SVi ratio appeared to be more robust than that of PP alone, the lack of an independent effect of PP after adjustment for mean BP may be due to the smaller number of subjects in this invasive hemodynamic study than in other studies. On the other hand, our study population was younger than in those most previous reports, and the results are, in fact, compatible with those of Sesso et al. (28)who recently reported that PP lost its prognostic significance for CV disease in men aged ≤60 years when mean BP was included in the model. Finally, the use of invasive hemodynamic measurements is undoubtedly a limitation with regard to its applicability in the assessment of the hypertensive patient. On the other hand, the use of intra-arterial BP measurements and of the direct oxygen Fick method for the determination of SVi can be regarded as a strength of the study because of their greater accuracy and validity. The predictive power of the invasive PP-to-SVi ratio may be stronger than that of the ratio obtained by noninvasive means. In view of these limitations, a more direct and clinically applicable assessment of vascular stiffness would be needed to answer the question whether the predictive power of PP and the PP-to-SVi ratio is truly a reflection of increased vascular stiffness.
The authors gratefully acknowledge the assistance of N. Ausseloos, J. Delsupehe and J. Romont.
☆ Supported, in part, by the Belgian National Research Council, N.F.W.O., Brussels, Belgium. R. Fagard is holder of the Prof. A. Amery Chair in Hypertension Research, founded by Merck, Sharp and Dohme (Belgium).
- blood pressure
- heart rate
- pulse pressure
- PP-to-SVi ratio
- pulse pressure-to-stroke index ratio
- relative hazard rate
- stroke volume
- stroke index
- SVi-to-PP ratio
- stroke index-to-pulse pressure ratio
- World Health Organization
- Received December 4, 2000.
- Revision received March 19, 2001.
- Accepted April 5, 2001.
- American College of Cardiology
- Mitchell G.F,
- Moyè L.A,
- Braunwald E,
- et al.
- Domanski M.J,
- Mitchell G.F,
- Norman J.E,
- Exner D.V,
- Pitt B,
- Pfeffer M.A
- Franklin S.S,
- Khan S.A,
- Wong N.D,
- Larson M.G,
- Levy D
- Remington J.W,
- Noback C.B,
- Hamilton W.F,
- Gold J.J
- Randall O.S,
- Westerhof N,
- van den Bos G.C,
- Alexander B
- Chemla D,
- Hébert J.-L,
- Coirault C,
- et al.
- Report of a WHO Expert Committee
- Fagard R,
- Staessen J,
- Thijs L,
- Amery A
- Kannel W.B,
- Stokes J III.
- Cox D.R
- Mattheus D.E,
- Farewell V