Diastolic Blood Pressure and Mortality in the Elderly With Cardiovascular Disease
Isolated systolic hypertension is predominantly observed in the elderly because of increased arterial stiffness. Aggressive treatment leads to excessive lowering of diastolic blood pressure and favors the presence of a J-shaped curve association with mortality. We investigated whether, in the elderly, this pattern of association is a simple epiphenomenon of increased arterial stiffness and impaired cardiac function. In a cohort of 331 hospitalized subjects >70 years old (mean age±SD: 85±7 years), aortic pulse wave velocity and pressure wave reflections, by pulse wave analysis, and cardiac function, by ultrasound, were assessed. During a 2-year follow-up period, 110 subjects died. No association of prognosis with systolic pressure, pulse pressure, or pulse wave velocity was observed. A J-shaped association between diastolic pressure and overall and cardiovascular mortality was observed. Unadjusted Cox regression analysis showed that patients in the first tertile of diastolic pressure (≤60 mm Hg) had higher mortality. In Cox regression analysis, diastolic pressure ≤60 mm Hg was a predictor of mortality independently from cardiac–vascular properties, cardiovascular risk factors, and drug treatment. Multivariate regression analysis showed that increased age and low total peripheral resistance, but not left ventricular function, were the cardinal determinants of low diastolic pressure. An “optimal” diastolic pressure of 70 mm Hg in subjects with isolated systolic hypertension was found. We showed that, in the frail elderly, a value of diastolic blood pressure ≤60 mm Hg is associated with reduced survival, independent from large artery stiffness and left ventricular function, suggesting that more rational antihypertensive therapy, not only based on systolic pressure level, is needed.
- diastolic blood pressure
- arterial stiffness
- pressure wave reflections
- total peripheral resistance
The goal of antihypertensive treatment is to prevent cardiovascular (CV) complications through the reduction of systolic (SBP) and diastolic blood pressure (DBP). However, since the primary work of Cruickshank et al,1 several reports, but not all, have shown that, in hypertensive subjects treated with drugs, low DBP is frequently associated with increased mortality (reviewed in Reference 2). This finding was constantly difficult to evaluate. First, it is difficult in epidemiological studies to assess a J- or U-shaped association with mortality, and it is often easier, using a semilogarithmic scale, to show a linear relation between DBP and mortality. Second, in humans, the decrease of DBP is the consequence of both the aging process3 and the result of drug treatment, making the net drug effect quite difficult to define. Finally it should be noted that isolated systolic hypertension is difficult to treat and, therefore, aggressive treatment may lead to excessive lowering of diastolic blood pressure and that, in the oldest old, treating high SBP is not always related to reduced overall mortality.3–6
In the recent years, in subjects >50 years of age with advanced renal failure, Blacher et al7 showed that increased aortic stiffness and low DBP were independent predictors of CV risk. The distinction between these 2 pathophysiological mechanisms is challenging, because DBP is a component of pulse pressure (PP), increased PP is the principal hemodynamic consequence of increased aortic stiffness, and no data are available until now. In 1998, in a population of 16 913 subjects followed for 13 years, Tuomilehto et al8 indicated that low DBP alone was a significant predictor of CV and non-CV mortality among persons aged >50 years (most >70 years). However, the evaluation of the underlying pathophysiological mechanisms was limited, particularly regarding hemodynamic parameters.
In this study, a cohort of very old frail subjects was investigated prospectively (mean age±SD: 85±7 years). We tried to delineate for the first time the pathophysiological role of DBP on total and CV mortality in relation to large artery stiffness and pressure wave reflections, as well as to total peripheral resistance and cardiac function.
From May 2000 to November 2001, 331 consecutive patients entering the geriatric departments of Charles Foix and Emile Roux Hospitals, Ile de France, were included in the PRonostic cardiovasculaire Optimisation Therapeutique En GERiatric Study with respect to the following inclusion criteria: age >70 years old; past story of CV disease involving coronary heart disease, cerebrovascular disease, hypertension, or any other CV events of the upper or lower limbs, thoracic or abdominal aorta, or renal arteries; Mini Mental Status Examination >15 of 30; absence of fatal disease with life expectancy <1 month; and willingness to give a written informed consent to participate in this study. Patients with cachexia (body mass index: <17 kg/m2) and/or evolutive cancer and/or advanced renal failure (plasma creatinine: >250 μmol/L) were not included in the study.
The study cohort was then composed of 331 subjects (86 men and 245 women) with mean age±SD of 85±7 years. The PRonostic cardiovasculaire Optimisation Therapeutique En GERiatric Study was approved by the Committee for the Protection of Human Subjects in Biomedical Research of Saint Germain Hospital (Ile de France). Written informed consent was obtained from all participants after relevant information was provided to them and to their relatives. Only the parameters that are relevant to the present analysis are presented here.
Social, Anthropometric, and Clinical Parameters
Information compiled from the questionnaire filled out at inclusion included gender, age, weight, height, personal history of CV event, the presence of diabetes mellitus, dyslipidemia, hypertension, smoking habits, and previous diseases. The reason for hospitalization and the level of education (1 indicates primary school; 2, college degree; 3, bachelor degree; and 4, university degree) were also registered. In all of the subjects, such information agreed with that given by relatives and/or recorded from the most recent previous hospitalization.
Antihypertensive drugs included diuretics (38.0%), calcium channel antagonists (27.9%), angiotensin-converting enzyme inhibitors (26.1%), β-blockers (12.3%), α-blockers (4.0%), and central-acting agents (3.1%), either alone or in combination. Three percent of patients were medically treated for dyslipidemia (drugs including statins or fibrates). Fourteen percent of patients were medically treated for diabetes mellitus (drugs including sulfonamides and/or biguanids or insulin).
Assessment of BP, Arterial Stiffness, and Pressure Wave Reflections
The measurements were performed in the morning, after an overnight fast, with each patient in the supine position. Brachial BP was measured after 15 minutes of rest using the semiautomatic oscillometric device Dynamap (Kontron). Five measurements 2 minutes apart were averaged. Data on the validity of the oscillometric devices in the elderly and especially in the presence of increased levels of arterial stiffness are lacking; therefore, our results should be viewed under this limitation.
The relative enhancement of carotid SBP because of reflected pressure waves (augmentation index [AI]%) was assessed by means of applanation tonometry and application of pulse wave analysis at the level of the carotid artery; the carotid pressure waveform was calculated as described previously.9 It was available in 296 subjects. Aortic hemodynamics were also estimated by the use of generalized transfer function from radial pressure wave (Sphygmocor AtCor). The time of the arrival of the reflected wave (reflected wave time transit) and the timing at the level of the central arteries (reflected wave time transit/left ventricular ejection time) were measured. The ratio of diastolic pressure time index versus tension time index (ie, the integral of pressure and time during diastole and systole, respectively), has been shown to correlate well with the ratio of subepicardial to subendocardial blood flow, and, therefore, it represents an index of subendocardial viability, defined as subendocardial viability ratio (Buckberg index).10,11 It was automatically obtained from the aortic pressure waveform by the Sphygmocor apparatus. Because the validity of the generalized transfer function in such an elderly population is not known, the data that are presented on central AI in this study are derived from direct carotid artery tonometry, which is a very good surrogate of invasively acquired aortic AI.9
Aortic pulse wave velocity (PWV) was determined using the foot-to-foot method as described previously12 (Complior, Colson); it was available in 283 subjects. The superficial distance covered by the pulse wave was measured directly from the carotid to the femoral artery. This method for distance assessment may overestimate PWV by ≈2 m/s on average.13
Measurement of Carotid and Cardiac Ultrasound Parameters
The common carotid artery intima–media thickness and wall motion were measured by a high-resolution B-mode (7.5 MHz transducer, Kontron 440; n=291). Measurements were done on the right and left common carotid artery, 2 cm proximal to the bifurcation, always performed in plaque-free arterial segments. It was automatically determined from changes of density on the section perpendicular to the vessel wall using specific software.
Echocardiograms were recorded with an ultrasound system (Kontron 440) using a 2.5-MHz phase-array transducer (n=297). Cardiac measurements were performed according to the American Society of Echocardiography by M-mode measurements. It was possible to evaluate left ventricular volumes (v) only with left ventricular diameters (D) assuming that the geometric shape of the ventricle was a prolate ellipse.14 Then the volume of this ellipse was expressed as follows: V=(4π/3) (2D/2) (D/2) (D/2)=πD3/3≈D3.
Cardiac output (Q) was calculated with the formula: SV×heart rate, where SV is stroke volume. Total peripheral vascular resistance (TPR), as: TPR=MBP/Q, where MBP indicates mean blood pressure. Echocardiograms were also used to evaluate the diastolic index: E wave deceleration slope time (DT).
Measurement of Biological Parameters
Venous blood samples were obtained in subjects after an overnight fast and after determination of routine biochemistry and lipid profile by standard methods was performed.
Follow-up started from the baseline examination and lasted until April 2004. Of all 331 participants in the present study, 3 (1%) were lost to follow-up. Information was obtained from the patient himself, from relatives, or from general practitioners. Interim telephone and clinic contacts were used to assess all of the hospitalizations, outpatient CV diagnoses, and overall mortality. In the case of hospitalization, discharge reports from medical specialists were obtained. Fatal and nonfatal CV events and all-cause mortality were reported. Follow-up time was defined by the time from the baseline visit until the first event date (for those who had an event) or was censored at the last contact date (for those who did not have any event or for the 3 patients who were lost to follow-up).
In this exploratory analysis, the proportions of subjects were pooled by 10-mm Hg strata of DBP, SBP, and PP, and the distribution of events (%) was evaluated to determine whether statistical relations were linear.
Survival analysis based on Kaplan–Meier curves and log-rank tests was used to assess the unadjusted association between tertiles of DBP (first tertile [n=114]: ≤60 mm Hg; second tertile [n=110]: 61 to 70 mm Hg; third tertile [n=102]: >70 mm Hg). To test the effect of other peripheral and central hemodynamics (focusing mainly on the cardiac and vascular properties), we performed a similar analysis according to tertiles for SBP, PP, MBP, TPR, heart rate, large artery stiffness (PWV), pressure wave reflections (AI), and left ventricular systolic and diastolic function (ejection factor [EF] and DT, respectively).
Multivariate linear regression analysis was applied to find the determinants of DBP. All of the CV risk factors, as well as functional and structural vascular and cardiac parameters, were evaluated by means of bivariate correlation with DBP. Then multivariate linear regression analysis was applied to find the independent predictors of DBP. The final model was verified by the enter, backward, and stepwise methods (final results represent stepwise analysis).
In addition, the validity of the association between DBP and all-cause mortality, as well as CV mortality, was tested using extended adjustments by various Cox regression models. In these models, all of the potential confounding factors (hospital of inclusion, socioeconomic parameters, classical CV risk factors, drugs, cardiac parameters, biochemical indices, and especially vascular parameters) were entered step by step. In the Cox models, DBP was used either as a dichotomized variable (first tertile [≤60 mm Hg] versus second and third tertiles [>60 mm Hg]), because no significant difference regarding survival was observed between the second and third tertiles of DBP. Finally, subgroup analysis of those subjects with uncontrolled systolic hypertension (SBP ≥140 mm Hg) was performed regarding the effect of DBP and SBP on overall mortality.
T test for continuous variables and χ2 test for qualitative parameters were applied to investigate for differences between subjects with DBP ≤60 mm Hg and DBP >60 mm Hg. Statistical analysis was performed on an SPSS 11.5. P≤0.05 was considered statistically significant.
Percentage of All-Cause Death and CV Death by 10 mm Hg of BP Strata
The percentage of all-cause death and CV death (Figure 1a and 1b) was related to DBP in a J-shaped pattern. On the contrary, a flat relation among SBP, PP, and overall mortality was found (Figures 2a and 3⇓a), as well as an inverse linear relation between SBP and CV mortality and a J-shaped relation between PP and CV mortality.
Unadjusted Kaplan–Meier Curves: BP and Arterial Stiffness
In Figure 4, a clear association of the DBP tertile (brachial or carotid) with (Figure 4a) the total mortality-free survival (P=0.004; Figure 4b) and CV mortality-free survival (P=0.008) is described. Corresponding P values for other brachial BP, PWV, AI, TPR, left ventricular systolic and diastolic function (EF and DT, respectively), and heart rate are shown in Table 1. Note that none of these factors were related to overall and/or CV mortality except for EF.
Unadjusted Kaplan–Meier curves also showed that the presence of diabetes mellitus, the lower tertile of hematocrit and plasma albumin, and the higher tertile of plasma creatinine were significantly associated with reduced survival (data not shown).
Determinants of DBP
In Table 2, the independent predictors of DBP are described. Age, TPR, AI, PWV, heart rate, and the educational status were independent predictors of DBP.
Cox Regression Models
In Table 3, the predicting effect of DBP as a dichotomous variable (first DBP tertile versus the [second and third added] DBP tertile) on total mortality-free survival and on CV mortality-free survival was adjusted by various Cox regression models. DBP ≤60 mm Hg was an independent predictor of mortality even after adjustment for age, gender, and hospital of inclusion (model 1), as well as for additional adjustment for mental status (model 2); classical CV risk factors and previous coronary heart disease and stroke (model 3); medication (model 4); cardiac function and structure (model 5); PWV (model 6a); TPR (model 6b); and AI (model 6c). Only after adjustment for the combined effect of vascular properties (PWV, AI, and TPR) did DBP ≤60 mm Hg lose its predicting value.
Similar results were found concerning DBP and CV mortality-free survival (Table 2). The independent effect of DBP was lost after adjustment for the effect of vascular properties. Similarly, after adjustment for biochemical factors (hematocrit, plasma albumin, and plasma creatinine) or weight, low DBP was an independent predictor of overall and CV mortality (data not shown).
Comparison of First DBP Tertile (≤60 mm Hg) versus Combined Second and Third DBP Tertile (>60 mm Hg)
Subjects with DBP ≤60 mm Hg were older and had smaller weight, hematocrit, plasma albumin, total cholesterol, low-density lipoprotein cholesterol, and triglycerides (Table 4). No differences were found regarding other classical CV risk factors, biochemical and social parameters (education level and living habits), study design parameters (center effect; data not shown), and the reason of hospitalization (data not shown). Subjects with DBP ≤60 mm Hg tended to be treated with more drugs (P=0.079) and a higher percentage of diuretics (P=0.071).
Subjects with DBP ≤60 mm Hg (Table 5) had lower SBP, PP, and MBP. Carotid–femoral PWV and carotid intima–media thickness did not differ significantly between the 2 groups. AI was lower in subjects with DBP ≤60 mm Hg, but this difference was abolished after adjustment for MBP. On the contrary, the Buckberg index was lower in subjects with low DBP after adjustment for MBP. Ejection fraction, LV mass, and DT were similar between the 2 groups. Finally, subjects with low DBP had lower TPR (Table 5).
In subjects with uncontrolled SBP (≥140 mm Hg), subgroup analysis verified the lack of association between tertiles of SBP and total mortality (first: 13 events/39 subjects=33.3%; second: 11 events/36 subjects=30.5%; third: 10 events/36 subjects=27.7%; log rank by Kaplan–Meier P=0.863). On the contrary, subjects in both the lowest (n=37, mean DBP±SD: 62.9±5.8 mm Hg) and the highest (n=40, mean DBP±SD: 83.6±7.5 mm Hg) tertiles of DBP had higher mortality events (14 events/37 subjects=37.4% and 14 events/40 subjects=35%, respectively) than the middle tertile (n=34, mean DBP±SD: 72.7±1.7 mm Hg, 6 events/34 subjects=17.6%). Kaplan–Meier analysis of the second versus the combined first and third tertiles of DBP showed a marginally significant difference (P=0.056).
This study was the first prospective investigation in an elderly population in which pressure wave reflections, arterial stiffness, cardiac function, and TPR were measured to investigate the potential pathophysiological association of low DBP and mortality. We showed that, in this very aged population, a J-curved association between DBP and mortality (all-cause or CV) was present. DBP was modulated by age, TPR, pressure wave reflections (AI), large artery stiffness (PWV), heart rate, and educational level. The lower survival in subjects with DBP ≤60 mm Hg was independent from the hospital of inclusion, mini mental status examination, classical CV risk factors, previous health state, coronary heart disease and stroke, and biochemical parameters, as well as drug treatment. Moreover, no cardiac or vascular parameter could solely explain this association.
Considerations on the Population
The population of the present study carries many particularities because of the high prevalence of atherosclerotic disease (coronary, cerebral, and peripheral vascular disease reaching ≈62%), which must be carefully considered and may potentially limit the extrapolation of our results to other elderly populations. Only 80 subjects were <80 years of age, and 131 were >90 years old (mean age: 85.1 years; range: 70 to 103 years). This major trait of the population may be responsible for a number of CV particularities. First, carotid femoral PWV was consistently augmented, passing the 20 m/s in ≈10% of the population (mean PWV: 14.4 m/s; range: 7.2 to 28.9 m/s). Nevertheless, large arterial wall properties (assessed by PWV) were identical between DBP groups. We have shown in the past that, at >70 years of age, PWV no longer correlated with age.15 Second, and in relation to the first, only 10% of the subjects had seriously impaired left ventricular function (EF: <45%). Third, one third of the population had an extreme decrease of DBP, that is, <60 mm Hg, and only 8 subjects had uncontrolled DBP >90 mm Hg; 111 subjects (one third of the population) had uncontrolled SBP ≥140 mm Hg. Taken together, these findings suggest that the overall population was composed mainly of “survivors.”16
Our negative results concerning prediction of mortality by hemodynamic parameters are important to consider. SBP and PP were not associated with prognosis, all-cause mortality, or CV mortality. Previous study in the oldest old reported this absence of relation between SBP and overall mortality.17 One could considerer that the patients with the more severe hypertension had probably died before having the “age opportunity” of entering this study. The remaining poorly controlled subjects with hypertension in this survey could benefit from a survival effect. Similar explanation could be given for PP and PWV.
Considerations on the Pathophysiology of the J Curve
Four potential “pathophysiological” mechanisms have been proposed to explain the existence of a J curve. First, the J curve may be an epiphenomenon of more severe underlying chronic illness, which thereby increases mortality.4 Second, low DBP could also be a marker of cardiac function. Indeed, in the population of North Karelia,8 especially in patients >70 years of age, the DBP–mortality relation was considered as a direct main result of cardiac failure, and there was an age dependence regarding the effect of low DBP on mortality. Third, the J curve may represent an epiphenomenon of increased arterial stiffness, a well-known independent marker of advance vascular disease and of increased mortality, leading to high PP and low DBP.18–20 Finally, low DBP may compromise coronary perfusion during the diastolic phase of the cardiac cycle, especially in subjects with coronary heart disease.1,2,21
In the current study we found some differences concerning indices of chronic illness between DBP groups (weight, plasma albumin, lipids, and hematocrit), but none of them was sufficient, from a statistical point of view, to explain the association of low DBP with mortality (adjusted data not shown). Moreover, we excluded from the study all of the patients with cachexia, evolutive cancer, and/or advanced renal failure. Concerning the presence of pre-existing CV disease and/or risk factors, no significant differences were found between the 2 groups. Finally, no significant differences regarding the reason for admission in the hospital, before inclusion in the study, were found.
In the hospitalized population presented in this study, a low prevalence of heart failure was observed, suggesting other pathophysiological mechanisms underneath the J-shaped curve. Moreover, adjustment for both structural and functional cardiac status did not modify the results. Finally, intima–media thickness of the carotid artery, as well as PWV, that is, 2 classical markers of the CV health state, were also similar between the 2 groups.
We also showed that, although low DBP is classically related to higher PWV, in this study the effect of DBP on mortality was independent from arterial stiffness and/or pressure wave reflections and was not associated with an increase in PP. Age and low TPR (consequently lower pressure wave reflections), but not EF, account for 13% of DBP variation and may be responsible for the low level of DBP in the first tertile. Because the peripheral resistance and the pressure wave reflections are closely interrelated and both affect DBP, it is difficult to define their individual effects on mortality. Yet, our results imply, for the first time, that low AI may have a deleterious effect, independent from TPR, exclusively on CV mortality.
Although this study did not provide direct proof for the deleterious effect of extreme low DBP on coronary perfusion during diastole, epidemiological data1,2,21 support this pathophysiological approach. Even in the absence of evident contractile dysfunction in older subjects, subendocardial myocardial dysfunction may exist.22 Indirect evidence from this study support this hypothesis, suggesting that, for the same level of MBP, the subendocardial viability index (Buckberg index) is reduced in subjects with DBP ≤60 mm Hg. Recent epidemiological data have shown that for the same cutoff value (≤60 mm Hg), DBP had a deleterious effect on the survival of patients with coronary artery disease.2 We suggest that, in the frail elderly with a high burden of CV disease, even in the absence of evident severe coronary heart disease, in a nonhypertrophied fibrotic heart, oxygen delivery may be impaired in the case of DBP ≤60 mm Hg and, further, could impair subendocardial contractility. Finally, it should be noted that, although every possible effort to identify the cause of death was done, CV mortality was probably underestimated.
Our subgroup analysis in subjects with uncontrolled systolic hypertension showed that low DBP at ≈60 mm Hg was as harmful as a value of 80 mm Hg and that the optimal DBP level was 70 mm Hg. Therefore, aggressive treatment of isolated systolic hypertension, in a “fragile” population with low systemic TPR, as the one included in this study, may counterbalance the potential favorable effect from SBP decrease. Taking into consideration that in these subjects their limited life expectancy may restrict the actual impact of treatment, this seems to be a real-life scenario.
In conclusion, we showed that in the frail oldest old with a high burden of CV disease, DBP is not linearly, but in a J-shaped curve, associated with mortality, with a cutoff level at ≤60 mm Hg. We also showed that this association was not a simple epiphenomenon because of concomitant chronic illness, cardiac failure, or increased arterial stiffness but was associated with reduced peripheral resistance/pressure wave reflections and potentially aggressive blood pressure reduction, possibly jeopardizing coronary perfusion. More data are needed on the necessity to hold back antihypertensive medications in elderly with low TPR and DBP <70 mm Hg.
We are deeply indebted to the PROTEGER patients and their relatives, who made this study possible.
Sources of Funding
This work was supported by the Société Française d’Hypertension Artérielle and the Fondation de France.
- Received February 20, 2007.
- Revision received March 12, 2007.
- Accepted April 25, 2007.
Franklin SS, Larson MG, Khan SA, Wong ND, Leip EP, Kannel WB, Levy D. Does the relation of blood pressure to coronary heart disease risk change with aging? The Framingham Heart Study. Circulation. 2001; 103: 1245–1249.
Boshuizen HC, Izaks G, van Buuren S, Ligthart GJ. Blood pressure and mortality in elderly people aged 85 and older: community based study. BMJ. 1998; 316: 1780–1784.
Amery A, Birkenhager W, Brixko R, Bulpitt C, Clement D, Deruyttere M, De Schaepdryver A, Dollery C, Fagard R, Forette F. Efficacy of antihypertensive drug treatment according to age, sex, blood pressure and previous cardiovascular disease in patients over the age of 60. Lancet. 1986; 2: 586–592.
Blacher J, Guerin A, Pannier B, Marchais S, Safar M, London G. Impact of aortic stiffness on survival in end-stage renal disease. Circulation. 1999; 99: 2434–2439.
Asmar R, Benetos A, Topouchian J, Laurent P, Pannier B, Brisac AM, Target R, Levy BI. Assessment of arterial distensibility by automatic pulse wave velocity measurement: validation and clinical application studies. Hypertension. 1995; 26: 485–490.
Popp RL, Harrison DC. Ultrasonic cardiac echocardiography for determining stroke volume and valvular regurgitation. Circulation. 1970; 16: 493–502.
Langer RD, Ganiats TG, Barret-Connor F. Factors associated with paradoxical survival at higher blood pressures in the very old. Am J Epidemiol. 1991; 134: 29–38.
Meaume S, Benetos A, Henry OF, Rudnichi A, Safar ME. Aortic pulse wave velocity predicts cardiovascular mortality in subjects >70 years of age. Arterioscler Thromb Vasc Biol. 2001; 21: 2046–2050.
Benetos A, Rudnichi A, Safar M, Guize L. Pulse pressure and cardiovascular mortality in normotensive and hypertensive subjects. Hypertension. 1998; 32: 560–564.
Blacher J, Protogerou AD, Safar ME. Cardiovascular risk and the macrocirculation. In: Safar ME, ed. Macro- and Microcirculation in Hypertension. London, UK: Lippincott Williams & Wilkins; 2005: 83–97.
Cruickshank JM. Coronary blood flow reserve and the J curve relation between diastolic blood pressure and myocardial infarction. BMJ. 1988; 297: 1227–1230.
Lumens J, Delhaas T, Arts T, Cowan BR, Young AA. Impaired subendocardial contractile myofiber function in asymptomatic aged humans, as detected using MRI. Am J Physiol. 2006; 291: H1573–H1579.