Pulse Pressure and Mortality in Older People

ORIGINAL INVESTIGATION Pulse Pressure and Mortality in Older People

Robert J. Glynn, ScD; Claudia U. Chae, MD; Jack M. Guralnik, MD, PhD; James O. Taylor, MD; Charles H. Hennekens, MD, DrPH

Background: In older people, observational data are un- sure 80 to 89 mm Hg; relative to this group, the highest clear concerning the relationships of systolic and dias- death rate occurred in those with systolic pressure of tolic blood pressure with cardiovascular and total mor- 160 mm Hg or more and diastolic pressure less than 70 tality. We examined which combinations of systolic, mm Hg (relative risk, 1.90; 95% confidence interval, diastolic, pulse, and mean arterial pressure best predict 1.47-2.46). Both low diastolic pressure and elevated sys- total and cardiovascular mortality in older adults. tolic pressure independently predicted increases in cardiovascular (PϽ.001) and total (PϽ.001) mortality. Pulse Methods: In 1981, the National Institute on Aging ini- pressure correlated strongly with systolic pressure tiated its population-based Established Populations for (R=0.82) but was a slightly stronger predictor of both Epidemiologic Studies of the Elderly in 3 communities. cardiovascular and total mortality. In a model contain- At baseline, 9431 participants, aged 65 to 102 years, had ing pulse pressure and other potentially confounding vari- blood pressure measurements, along with measures of ables, diastolic pressure (P=.88) and mean arterial pres- medical history, use of medications, disability, and physi- sure (P=.11) had no significant association with mortality. cal function. During an average follow-up of 10.6 years among survivors, 4528 participants died, 2304 of car- Conclusions: Pulse pressure appears to be the best single diovascular causes. measure of blood pressure in predicting mortality in older people and helps explain apparently discrepant results Results: In age- and sex-adjusted survival analyses, the for low diastolic blood pressure. lowest overall death rate occurred among those with systolic pressure less than 130 mm Hg and diastolic pres- Arch Intern Med. 2000;160:2765-2772

N MIDDLE-AGED populations, both pulse pressure and a diminished associa- From the Division of Preventive systolic and diastolic blood pres- tion between systolic and diastolic pres- 21 Medicine, Department of sure have strong, linear relation- sure. This may lead to differing relation- Medicine, Brigham and ships with cardiovascular and to- ships of systolic and diastolic pressure with Women’s Hospital and Harvard tal mortality.1-5 Because of the mortality in older people. Medical School, Boston, Mass highI correlation between systolic and di- It is also possible that pulse pressure (Drs Glynn and Chae); astolic pressure, studies examining car- is the measure of blood pressure most Department of Biostatistics, diovascular risk in this age group com- strongly related to cardiovascular risk in Harvard School of Public monly find that diastolic pressure provides older people, and that consideration of Health, Boston (Dr Glynn); little additional prognostic information af- pulse pressure may explain the apparent Cardiology Division, 1-4 Department of Medicine, ter consideration of systolic pressure. In increased risk associated with low dias- Massachusetts General older people, however, observational stud- tolic pressure. Several prospective stud- 6-13 Hospital, Boston (Dr Chae); ies have been less consistent and have ies have found that elevated pulse pres- Epidemiology, Demography, commonly found U- or J-shaped relation- sure, which reflects increased arterial and Biometry Program, ships of blood pressure with mortality, es- stiffness with age, is associated with risk National Institute on Aging, pecially for diastolic pressure. In some of myocardial infarction, congestive heart Bethesda, Md (Dr Guralnik); studies, individuals with the lowest blood failure, and cardiovascular and total mor- East Boston Neighborhood pressure had the highest mortality.14,15 tality.22-27 However, pulse pressure is Health Center, East Boston, Treatment implications of these relation- strongly correlated with systolic pres- Mass (Dr Taylor); and 16-20 Department of Medicine, ships remain controversial. With the sure, and it remains unclear whether it pro- Epidemiology, and Public loss of aortic compliance, systolic pres- vides independent prognostic informa- 28 Health, University of Miami sure rises with age in industrialized coun- tion or is useful in clarifying J-curves. School of Medicine, Miami, Fla tries, while diastolic pressure declines af- Using data from 3 population- (Dr Hennekens). ter about age 60 years, leading to increased based cohorts of the Established Popu-

(REPRINTED) ARCH INTERN MED/ VOL 160, OCT 9, 2000 WWW.ARCHINTERNMED.COM 2765

Downloaded From: https://jamanetwork.com/ on 09/29/2021 SUBJECTS AND METHODS to 1992, giving an average follow-up of 10.6 years (range, 9.1-11.1 years) among survivors. Of the 9431 partici- SUBJECTS AND MEASURES pants, 4528 died, and death certificates were obtained for 4494. A single trained nosologist coded the underlying cause In 1981, the National Institute on Aging initiated its Es- of death according to the International Classification of Dis- tablished Populations for Epidemiologic Studies of the El- eases, Ninth Revision.36 We used these codes to classify deaths derly studies of community-dwelling persons aged 65 years as being caused by cardiovascular disease including stroke and older in 3 locations: East Boston, Mass; Washington (codes 401-459) or other causes. and Iowa counties, Iowa; and New Haven, Conn. During 1982 and 1983, surveys were conducted in the entire popu- STATISTICAL ANALYSIS lations of persons aged 65 years and older in East Boston and Iowa and in a stratified sample of residents of New Ha- We first determined correlations among the measures of ven. To maximize participation, trained interviewers vis- blood pressure to quantify their interrelationships. To ited the homes of all eligible participants. Participation rates examine the joint association of systolic and diastolic ranged from 80% to 85%, with 3809 participants in East pressure with mortality, we classified participants Boston, 3673 participants in Iowa, and 2812 participants according to categories of both variables. Categories in New Haven, for a total population of 10294 community- used were the same as in a previous study of blood pres- dwelling elderly subjects. Because some individuals par- sure in one of these populations,12 except that a priori ticipated through a proxy, 9431 participants had baseline we grouped individuals with systolic pressure between blood pressure measurements, and they constitute the co- 140 and 159 mm Hg and also formed a single group hort for the current study. among those with diastolic pressure between 70 and 79 In East Boston and New Haven, the trained inter- mm Hg. Thus, we partitioned the population into 16 viewer took 3 blood pressure measurements at 30-second groups according to category of systolic pressure (Ͻ130, intervals by means of a standard mercury sphygmoma- 130-139, 140-159, or Ն160 mm Hg) and category of nometer, after the participant had been seated for at least diastolic pressure (Ͻ70, 70-79, 80-89, or Ն90 mm Hg). 5 minutes, according to the protocol used in the Hyper- We used proportional hazards analyses to compare age- tension Detection and Follow-up Program.29 Two mea- and sex-adjusted total and cardiovascular death rates surements were taken in Iowa. For this study, systolic across these 16 categories. pressure was the average of all systolic measures; dias- To examine whether simpler models might summa- tolic pressure was the average of all diastolic measures; rize these relationships, we compared the ability of both pulse pressure was systolic minus diastolic pressure; single measures and pairs of measures of blood pressure and mean arterial pressure was [systolic+(2ϫdiastolic to predict total and cardiovascular mortality. We catego- pressure)]/3. rized pulse pressure and mean arterial pressure accord- The interviewer also collected information about ing to approximate quartiles in the population. We com- other characteristics potentially related to both blood pared the ability of alternative models to predict pressure and mortality. Participants reported their height mortality by means of the R2 statistic for survival analy- and weight, present and past use of cigarettes and alcohol, sis and the likelihood ratio–based discrimination index and whether they were ever told by a physician that they D (defined as the model likelihood ratio ␹2−1 divided by had myocardial infarction, stroke, or cancer. Angina was the −2 log likelihood of the null model) described by identified by the Rose questionnaire.30 All medications Harrell and colleagues.37 We used likelihood ratio tests used in the 2 weeks before the interview were identified to determine whether adding variables to a model sig- by direct inspection. Disability was identified through nificantly improved the fit. Additional models included reports of problems with activities of daily living31 and other variables that may affect both blood pressure and problems with physical function by a 3-item scale.32 We risk of death, but were restricted to the 4054 deaths in classified participants as low in physical activity when 8715 participants with complete data on these potential they reported not exercising vigorously at least once a confounding variables. week, rarely or never taking walks, and not working fre- We also examined whether relationships of blood pres- quently around the house or garden. Further information sure with mortality differed in the following subgroups: about demographic characteristics of the population33 and women aged 65 to 74 years, women aged 75 years or more, quality control of blood pressure measurements34,35 has men aged 65 to 74 years, men aged 75 years or more, us- been published elsewhere. ers and nonusers of antihypertensive drugs, people with Participants were contacted yearly during the first 6 chronic disease (angina; history of myocardial infarction, years after baseline, with in-home interviews at the 3- and stroke, or cancer; use of digoxin, loop diuretics, or hypo- 6-year follow-up and telephone interviews in the interim glycemic drugs), and those without these diseases. Be- years. Thereafter, mortality follow-up continued through cause undetected diseases may affect blood pressure in those the end of 1992 through local surveillance supplemented near death, we also examined deaths separately in the first by linkage to the National Death Index. These sources 3 years and thereafter. Separate analyses in each site found yielded complete follow-up of the cohort for mortality up comparable results.

lations for Epidemiologic Studies of the Elderly, pulse, and mean arterial pressure, best predicted car- including more than 9000 older individuals followed diovascular and total mortality. We also examined up for more than 10 years, we considered which mea- whether consideration of pairs of these measures sures of blood pressure, including systolic, diastolic, would enhance prediction.

(REPRINTED) ARCH INTERN MED/ VOL 160, OCT 9, 2000 WWW.ARCHINTERNMED.COM 2766

Downloaded From: https://jamanetwork.com/ on 09/29/2021 RESULTS 250 Observed Upper Limit INTERRELATIONSHIPS AMONG 200 Predicted BLOOD PRESSURE MEASURES Lower Limit 150 Among the 9431 elderly subjects in these 3 communities, systolic blood pressure had a correlation of 0.50 with 100 diastolic blood pressure. Although this correlation was large and highly significant (PϽ.001), it was substan- 50 tially weaker than the correlation of nearly 0.80 be- Pulse Pressure, mm Hg tween these measures found in studies of middle-aged 0 populations.1,22 –50 Overall, pulse pressure had a weak but slightly 5075 100 125 150 175 200 225 250 275 U-shaped relationship with diastolic pressure. On aver- Systolic Blood Pressure, mm Hg age, the lowest pulse pressures occurred among those Figure 1. Relationship of systolic pressure with pulse pressure in 9431 with diastolic pressures between 70 and 89 mm Hg. elderly individuals, with associated regression line and 95% prediction Among those with diastolic pressure less than 80 interval. mm Hg, the correlation with pulse pressure was −0.21 (PϽ.001), whereas above 80 mm Hg the correlation cant elevations in death rates. Conversely, the highest was 0.10 (PϽ.001). Conversely, pulse pressure and sys- death rates occurred among groups with the highest pulse tolic pressure had a strong, almost linear association pressures (top right of Table 1). The 1130 individuals in (Figure 1). The correlation of 0.82 between these the 3 groups with the highest death rates all had pulse measures further suggested that a straight line fits this pressures greater than 70 mm Hg, and their mean pulse relationship fairly well. Mean arterial pressure had pressure was 92.4 mm Hg (interquartile range, 84-99 strong correlations of 0.85 with systolic pressure and mm Hg). 0.89 with diastolic pressure but a weaker, although still We observed a similar relationship between catego- highly significant (PϽ.001), correlation of 0.39 with ries of systolic and diastolic blood pressure and cardio- pulse pressure. vascular mortality (Figure 2), suggesting that pulse pressure was an important determinant of cardiovascular JOINT RELATIONSHIPS WITH MORTALITY mortality. The same 3 groups with the highest total mortality also had the highest cardiovascular death rates, and The lowest risk of death occurred among the 321 par- these elevations were significant (PϽ.01) for each of these ticipants with diastolic pressures between 80 and 89 groups compared with the referent. mm Hg and systolic pressures less than 130 mm Hg, and we took this group as the referent (Table 1). The high- BEST PREDICTIVE MODELS est death rate occurred among the 173 participants with elevated systolic pressure (Ն160 mm Hg) and low dias- For both total (Table 2) and cardiovascular (Table 3) tolic pressure (Ͻ70 mm Hg), who had a 90% higher death mortality, an age- and sex-adjusted model including only rate than the referent group (PϽ.001). The next highest pulse pressure had slightly better predictive ability than death rates occurred among the individuals in the adja- a model including only systolic pressure or only dias- cent categories, with elevated systolic pressure and di- tolic pressure, as indicated by higher R2 and discrimina- astolic pressure of 70 to 79 mm Hg (relative risk [RR], tion index statistics. Those in the third quartile of pulse 1.59; 95% confidence interval [CI], 1.28-1.99) or bor- pressure (63-76 mm Hg) had a 19% increased risk of car- derline high systolic pressure (140-159 mm Hg) and low diovascular death, and those in the highest quartile (Ն77 diastolic pressure (RR, 1.56; 95% CI, 1.26-1.93). More mm Hg) had a 57% increased risk of cardiovascular death, than 100 deaths occurred in 14 of the 16 blood pressure compared with those in the lowest quartile. Adding di- groups, indicating the ability to estimate reliably the joint astolic pressure to the model including pulse pressure relationship of systolic and diastolic pressure with mor- alone left these effects of pulse pressure virtually un- tality in all groups except those with elevated diastolic changed. Within each category of diastolic pressure, per- pressure (Ն90 mm Hg) and systolic pressure less than sons with higher levels of pulse pressure had higher rates 140 mm Hg. of cardiovascular death (Figure 3). Mean arterial pres- The data in Table 1 strongly support the impor- sure had a weak association with both total (Table 2) and tance of pulse pressure as a determinant of the risk of cardiovascular (Table 3) mortality. In analyses includ- death in older people. The lowest death rates occurred ing both pulse pressure and mean arterial pressure, pulse among normotensive subjects with low pulse pressure pressure continued to have the same strong relation- (ie, systolic pressure Ͻ130 mm Hg and diastolic pres- ship with mortality, whereas mean arterial pressure had sure 70-89 mm Hg, or systolic pressure 130-139 mm Hg a nonsignificant relationship with these end points (data and diastolic pressure 80-89 mm Hg). The 1880 indi- not shown). viduals in these groups all had pulse pressure less than Alternative models that included both systolic and 60 mm Hg and their mean pulse pressure was 46.7 diastolic pressure found independent and highly signifi- mm Hg (interquartile range, 42-52 mm Hg). None of the cant effects of both measures and were significantly bet- groups in the lower left portion of Table 1 had signifi- ter than models including only one of these measures

(REPRINTED) ARCH INTERN MED/ VOL 160, OCT 9, 2000 WWW.ARCHINTERNMED.COM 2767

Downloaded From: https://jamanetwork.com/ on 09/29/2021 Table 1. Number of Participants (Number of Deaths) and Relative Risk (RR) of Death (95% CI) by Categories of Diastolic and Systolic Pressure*

Systolic Pressure, mm Hg

Diastolic Pressure, mm Hg Ͻ130 130-139 140-159 Ն160 Ͻ70 No. of participants (No. of deaths) 1184 (547) 461 (224) 559 (330) 173 (123) RR (95% CI) 1.35 (1.10-1.65) 1.31 (1.05-1.64) 1.56 (1.26-1.93) 1.90 (1.47-2.46) 70-79 No. of participants (No. of deaths) 1011 (399) 890 (409) 968 (494) 398 (245) RR (95% CI) 1.16 (0.94-1.43) 1.28 (1.04-1.58) 1.33 (1.09-1.63) 1.59 (1.28-1.99) 80-89 No. of participants {No. of deaths} 321 (116) 548 (204) 1088 (489) 592 (342) RR (95% CI) 1.00 (Referent) 1.05 (0.84-1.32) 1.24 (1.01-1.52) 1.49 (1.21-1.84) Ն90 No. of participants (No. of deaths) 36 (18) 97 (40) 448 (191) 657 (357) RR (95% CI) 1.55 (0.94-2.55) 1.24 (0.86-1.77) 1.21 (0.96-1.53) 1.47 (1.19-1.81)

*Relative risk is based on a proportional hazards model adjusted for age and sex and stratified by site. Groups with significantly elevated risk are shown in boldface. CI indicates confidence interval.

tive ability, compared with the models that also in-

2.2 cluded interactions between these variables summa- 2.2 2.0 rized in Table 1 and Figure 2. However, the predictive

2.0 Relative Risk 1.8 ability of the simpler model including only pulse pres- 1.8 1.6 sure was nearly equal to that of the model including both 1.6 1.4 1.4 systolic and diastolic pressure. Relative Risk 1.2 Further control for potential confounding vari- 1.2 1.0 1.0 ables had little effect on relationships of systolic pres- ≥160 sure or pulse pressure with either cardiovascular mor- ≥90 Figure 4 140-159 tality ( ) or total mortality (data not shown). Diastolic Pressure,80-89 mm Hg However, control for confounding variables attenuated 130-139 70-79 the apparent increased risk associated with low diastolic

< pressure. Adjusting for systolic pressure and confound- <70 130 Systolic Pressure, mm Hg ers, the risk associated with low diastolic pressure remained statistically significant (P=.02 for both total Figure 2. Relative risk of cardiovascular death by categories of diastolic and systolic pressure, adjusted for age and sex and stratified by site. and cardiovascular death). With adjustment for pulse pressure as well as confounding variables, diastolic (PϽ.001). After adjusting for age, sex, and diastolic pres- pressure had no significant association with mortality sure, higher levels of systolic pressure were associated (P=.88 for total mortality and P=.33 for cardiovascular with increased risk of cardiovascular death, with a sig- mortality). This suggests that pulse pressure and nificant 25% higher cardiovascular death rate (PϽ.001) comorbidity explain the apparent relationships between among those with borderline high systolic pressure (140- low diastolic pressure and mortality in aging adults. 159 mm Hg), relative to those with systolic pressure less With control for potential confounding variables and than 130 mm Hg, and a 59% higher cardiovascular death pulse pressure, mean arterial presure also had no asso- rate (PϽ.001) among those with high systolic pressure ciation with mortality (P=.11 for total mortality and (Ն160 mm Hg). After controlling for systolic pressure, P=.21 for cardiovascular mortality). low diastolic pressure (Ͻ70 mm Hg) was independently associated with a 38% increase in risk of cardio- SUBGROUP ANALYSES vascular death (PϽ.001), relative to those with diastolic pressure 80 to 89 mm Hg, and diastolic pressure of 70 We found little evidence that either use of antihyperten- to 79 mm Hg was associated with a 17% higher cardio- sive drugs or presence of important chronic diseases modi- vascular death rate (P=.006). If diastolic pressure was fied the relationships of pulse pressure and diastolic pres- not included in the model, then the effects of systolic pressure with cardiovascular mortality, summarized for the sure on risk were substantially reduced, and, similarly, entire population in Table 3. We did observe some ap- if the model did not include systolic pressure, then the parent differences in relationships by age. In analyses of effects of diastolic pressure were also reduced. This fur- women aged 65 to 74 years and in men aged 65 to 74 ther indicates the importance of including both systolic years, we found somewhat stronger relationships of pulse and diastolic pressure in predictive models in older people. pressure with mortality, compared with the effects in the For both total mortality and cardiovascular mortal- total population: relative to women in the lowest quar- ity, these models including independent effects of both tile of pulse pressure, those in the highest quartile had a systolic and diastolic pressure had comparable predic- 2.13 higher risk of cardiovascular death (95% CI, 1.58-

(REPRINTED) ARCH INTERN MED/ VOL 160, OCT 9, 2000 WWW.ARCHINTERNMED.COM 2768

Downloaded From: https://jamanetwork.com/ on 09/29/2021 Table 2. Comparison of Alternative Survival Analyses Predicting Total Mortality Based on 4528 Deaths in 9431 People

Alternative Models, RR (95% CI)* Blood Pressure, mm Hg Pulse Only Systolic Only Diastolic Only Mean Arterial Only Diastolic and Pulse Diastolic and Systolic Pulse Ͻ53 1.00 (Referent) 1.00 (Referent) 53-62 1.07 (0.98-1.17) 1.07 (0.98-1.17) 63-76 1.12 (1.02-1.22) 1.11 (1.02-1.21) Ն77 1.34 (1.23-1.46) 1.33 (1.22-1.45) Systolic Ͻ130 1.00 (Referent) 1.00 (Referent) 130-139 1.00 (0.91-1.09) 1.04 (0.95-1.14) 140-159 1.08 (1.00-1.17) 1.15 (1.06-1.25) Ն160 1.25 (1.15-1.37) 1.39 (1.26-1.53) Diastolic Ͻ70 1.16 (1.07-1.26) 1.13 (1.04-1.23) 1.27 (1.16-1.38) 70-79 1.05 (0.98-1.14) 1.06 (0.98-1.14) 1.11 (1.03-1.20) 80-89 1.00 (Referent) 1.00 (Referent) 1.00 (Referent) Ն90 1.11 (1.00-1.22) 1.08 (0.98-1.20) 1.02 (0.92-1.13) Mean arterial Ͻ90 1.00 (Referent) 90-97 0.93 (0.85-1.01) 98-105 0.97 (0.89-1.06) Ն106 1.02 (0.94-1.11) Model characteristics R 2 0.161 0.159 0.157 0.157 0.162 0.162 Discrimination index 0.0236 0.0233 0.0230 0.0229 0.0237 0.0237

*All models are adjusted for age and sex and stratified by site. RR indicates relative risk; CI, confidence interval.

Table 3. Comparison of Alternative Survival Analyses Predicting Cardiovascular Mortality Based on 2304 Cardiovascular Deaths in 9431 People

Alternative Models, RR (95% CI)* Blood Pressure, mm Hg Pulse Only Systolic Only Diastolic Only Mean Arterial Only Diastolic and Pulse Diastolic and Systolic Pulse Ͻ53 1.00 (Referent) 1.00 (Referent) 53-62 1.09 (0.96-1.25) 1.10 (0.96-1.25) 63-76 1.19 (1.05-1.35) 1.19 (1.05-1.35) Ն77 1.57 (1.39-1.77) 1.54 (1.37-1.74) Systolic Ͻ130 1.00 (Referent) 1.00 (Referent) 130-139 0.97 (0.85-1.11) 1.03 (0.90-1.17) 140-159 1.15 (1.03-1.29) 1.25 (1.12-1.41) Ն160 1.41 (1.25-1.59) 1.59 (1.39-1.81) Diastolic Ͻ70 1.20 (1.07-1.35) 1.16 (1.03-1.31) 1.38 (1.22-1.55) 70-79 1.08 (0.97-1.20) 1.08 (0.97-1.20) 1.17 (1.05-1.31) 80-89 1.00 (Referent) 1.00 (Referent) 1.00 (Referent) Ն90 1.24 (1.08-1.42) 1.20 (1.04-1.38) 1.10 (0.96-1.27) Mean arterial Ͻ90 1.00 (Referent) 90-97 0.97 (0.86-1.09) 98-105 0.98 (0.87-1.11) Ն106 1.14 (1.01-1.28) Model characteristics R 2 0.106 0.104 0.101 0.101 0.107 0.107 Discrimination index 0.0289 0.0283 0.0274 0.0273 0.0291 0.0290

*All models are adjusted for age and sex and stratified by site. RR indicates relative risk; CI, confidence interval.

2.88) and men in the highest quartile had a 1.87 in- at baseline. Older women in the highest quartile of pulse creased risk (95% CI, 1.43-2.44), relative to those with pressure had a 1.34 increased risk of cardiovascular death pulse pressure in the lowest quartile. Conversely, rela- (95% CI, 1.07-1.68) relative to older women in the low- tionships were weaker among those aged 75 years or older est quartile; in older men the comparable RR was 1.05

(REPRINTED) ARCH INTERN MED/ VOL 160, OCT 9, 2000 WWW.ARCHINTERNMED.COM 2769

Downloaded From: https://jamanetwork.com/ on 09/29/2021 3 years after baseline. In separate analysis of deaths oc-

2.2 curring after this time, pulse pressure had a strong graded 2.2 2.0 relationship with cardiovascular mortality, with an RR

2.0 Relative Risk 1.8 of 1.61 (95% CI, 1.38-1.88) comparing those in the high- 1.8 1.6 est and lowest quartiles of pulse pressure. 1.6 1.4 1.4

Relative Risk 1.2 1.2 1.0 COMMENT 1.0 ≥77 These prospective data indicate that, in older people, both ≥90 63-76 systolic and diastolic blood pressure provide important Diastolic Pressure,80-89 mm Hg and independent prognostic information about the risk 53-62 70-79 of cardiovascular and total mortality. As in middle-aged

<53 populations, higher systolic pressure predicted linear in- <70 Pulse Pressure, mm Hg creases in cardiovascular and total mortality, with little evidence of confounding by other comorbid conditions Figure 3. Relative risk of cardiovascular death by categories of diastolic and pulse pressure, adjusted for age and sex and stratified by site. and a clear increase in risk among those with borderline high (140-159 mm Hg) systolic pressure. Low diastolic pressure is also a marker of increased risk of death, in- 2.00 A B dependent of systolic pressure, although this associa-

1.75 tion is largely explained by the confounding effects of frailty and comorbid conditions. Pulse pressure, al- 1.50 though highly correlated with systolic pressure, has the advantage of incorporating the effects of both high sys- 1.25 tolic and low diastolic pressure. Prognostic models in-

1.00 cluding pulse pressure are simpler and have nearly the same predictive ability as models including both sys- 0.75 tolic and diastolic pressure. Furthermore, in these data, <130 130-139 140-159 ≥160 <53 53-62 63-76 ≥77 Systolic Pressure, mm Hg Pulse Pressure, mm Hg pulse pressure appears to be the best single blood pressure measure to predict mortality risk in the elderly. 1.75 C D In healthy, middle-aged populations, pulse pres- Relative Risk of Death sure is not a consistent and independent risk factor for 1.50 cardiovascular disease. In the Framingham,1 Western Col- laborative Group,3 and 4 Chicago, Ill–area prospective 1.25 studies,28 level of systolic and/or diastolic blood pressure had stronger relationships with cardiovascular risk 1.00 than pulse pressure. Results in this area are not entirely consistent, though, as some studies have found in- 0.75 creases in cardiovascular risk associated with higher sys- <70 70-79 80-89 ≥90 <70 70-79 80-89 ≥90 Diastolic Pressure, mm Hg tolic pressure after controlling for level of diastolic pressure.39,40 The Multiple Risk Factor Intervention Trial found Figure 4. Joint relationships of systolic and diastolic pressure with high rates of death from coronary heart disease in men cardiovascular mortality (A and C) and of pulse and diastolic pressure with cardiovascular mortality (B and D), based on proportional hazards models with elevated systolic pressure and low diastolic pres- including categories of 2 blood pressure measurements and adjusting for sure, but in other categories of systolic pressure, those age in years; sex; history of cancer, stroke, myocardial infarction, and with lower diastolic pressure had decreased death rates.41 angina; use of antihypertensive drugs, loop diuretics, digoxin, and hypoglycemic drugs; need for help with activities of daily living; problems Several studies have indicated that the prognostic with physical function; low activity level; overweight; current and past significance of systolic pressure increases with increas- cigarette smoking; and current use of alcohol. Bars indicate 95% confidence ing age, while that of diastolic pressure and mean arte- intervals. rial pressure decreases.1,4,42 After age 60 years, stiffening of the large arteries leads to decreased diastolic pressure (95% CI, 0.81-1.39). It is unclear whether these results and increased pulse pressure, and this changes the rela- reflect the natural variability of subgroup estimates or a tionship between low diastolic pressure and cardiovas- true modification of the impact of blood pressure in re- cular disease. Recent data from long-term follow-up of lation to age, which is supported by several previous stud- those aged 50 years or older in the Framingham Study ies.38 In none of these subgroup analyses were there any support an important role of pulse pressure in predict- significant relationships of diastolic blood pressure with ing incident coronary heart disease.27 In particular, that cardiovascular mortality after controlling for pulse pres- study found that both higher levels of systolic pressure sure and other potential confounding variables. and lower levels of diastolic pressure independently pre- We also observed some variability in results of sepa- dicted risk of coronary heart disease. However, because rate analyses examining cardiovascular deaths in the first this Framingham analysis examined a younger, health- 3 years of follow-up and thereafter. Neither pulse pres- screened population, it had limited ability to estimate presure nor diastolic pressure had any significant relation- cisely the joint relationship of systolic and diastolic blood ship with cardiovascular deaths observed during the first pressure with cardiovascular risk. Specifically, the

(REPRINTED) ARCH INTERN MED/ VOL 160, OCT 9, 2000 WWW.ARCHINTERNMED.COM 2770

Downloaded From: https://jamanetwork.com/ on 09/29/2021 Framingham analysis had only 9 subjects and observed of potential value in improving arterial compliance in- only 1 event in the group of persons with the highest clude angiotensin-converting enzyme inhibitors,68 ni- pulse pressure (systolic pressure Ն160 mm Hg and trates,69 and a low-salt diet.70 However, evidence from ran- diastolic pressure Ͻ70 mm Hg), compared with the 123 domized trials with morbidity and mortality end points deaths among 173 persons in this high-risk group in is needed to determine the relative value of these thera- our study. pies and whether reduction of pulse pressure should be Studies in high-risk and diseased populations also a specific treatment target. In evaluating cardiovascular support an important prognostic role for pulse pres- risk in older persons, both systolic and diastolic blood sure. In patients with newly diagnosed hypertension and pressure contribute independent information, but the best in patients with left ventricular dysfunction after myo- single measure of blood pressure to predict mortality cardial infarction, elevated pulse pressure is a major risk appears to be pulse pressure. factor for myocardial infarction and death.23,24 Pulse pressure appears to be the best measure of blood pressure to Accepted for publication May 3, 2000. predict congestive heart failure in the elderly.26 This study was supported by contract AG02107 from Several plausible mechanisms may explain the as- the National Institute on Aging, Bethesda, Md, and a grant sociation between elevated pulse pressure and cardio- from Bristol-Myers Squibb, Princeton, NJ vascular disease. Arterial stiffness increases after- Corresponding author: Robert J. Glynn, ScD, Divi- load43,44 and myocardial work,45 impairs ventricular sion of Preventive Medicine, Brigham and Women’s Hospi- relaxation,46,47 and causes ischemia.48-50 It is strongly cor- tal, 900 Commonwealth Ave E, Boston, MA 02215 (e-mail: related with left ventricular hypertrophy,51,52 a known risk [email protected]). factor for cardiovascular events.53 Arterial stiffness is also 54,55 correlated with the presence of atherosclerosis. The REFERENCES associated increase in shear stress and pulsatile strain may 56,57 promote primary atheroma development and may con- 1. Kannel WB, Gordon T, Schwartz MJ. Systolic versus diastolic blood pressure 58 tribute to rupture of vulnerable plaques. and risk of coronary heart disease: the Framingham Study. Am J Cardiol. 1971; Studies in middle-aged populations have generally 27:335-346. found higher RRs associated with elevated blood pres- 2. Wilhelmsen L, Wedel H, Tiblin G. Multivariate analysis of risk factors for coronary heart disease. Circulation. 1973;48:950-958. sure, compared with the magnitude of the observed ef- 3. Rosenman RH, Sholtz RI, Brand RJ. A study of comparative blood pressure mea- 5 fects noted in our data. Factors that could explain these sures in predicting coronary heart disease. Circulation. 1976;54:51-58. reduced effects include greater variability of blood pres- 4. Rabkin SW, Mathewson FAL, Tate RB. Predicting risk of ischemic heart disease sure in older persons and greater comorbidity, which is and cerebrovascular disease from systolic and diastolic blood pressure. Ann In- difficult to measure precisely. Correction of risk esti- tern Med. 1978;88:342-345. 5. MacMahon S, Peto R, Cutler J, et al. Blood pressure, stroke, and coronary heart mates for the variability of blood pressure measures iden- disease, I: effects of prolonged differences in blood pressure. Lancet. 1990;335: 5 tifies a much greater impact of blood pressure on risk, 765-774. and such corrections have a greater effect on risk esti- 6. Harris T, Cook EF, Kannel W, Schatzkin A, Goldman L. Blood pressure experi- mates in older adults.12 Similarly, our predictive models ence and risk of cardiovascular disease in the elderly. Hypertension. 1985;7: 118-124. for 10-year mortality in the elderly, based on baseline char- 7. Lindholm L, Lanke J, Bengtsson B. U-shaped association between mortality and acteristics, have only modest performance. However, this blood pressure in a thirteen year prospective study. Fam Pract. 1986;3:3-8. is also the case in younger populations,59 in whom the 8. Siegel D, Kuller L, Lazarus NB, et al. Predictors of cardiovascular events and mor- search for additional risk factors for cardiovascular events tality in the Systolic Hypertension in the Elderly Program pilot project. Am J Epi- continues. Even if RRs associated with elevated blood pres- demiol. 1987;126:385-399. 9. Coope J, Warrender TS, McPherson K. The prognostic significance of blood pressure are lower in the older population, high blood pressure in the elderly. J Hum Hypertens. 1988;2:79-88. sure remains of great public health importance because 10. Taylor JO, Cornoni-Huntley J, Curb JD, et al. Blood pressure and mortality risk absolute risks of cardiovascular death are far greater in in the elderly. Am J Epidemiol. 1991;134:489-501. older adults.12 11. Benfante R, Reed D, Frank J. Do coronary heart disease risk factors measured in the elderly have the same predictive roles as in the middle aged? Ann Epidemiol. These findings have implications for treatment as 1992;2:273-282. well as prognosis. They provide population-based evi- 12. Glynn RJ, Field TS, Rosner B, Hebert PR, Taylor JO, Hennekens CH. Evidence dence for the generalizability of the efficacy of treat- for a positive linear relation between blood pressure and mortality in elderly people. ment for hypertension in older patients, which has been Lancet. 1995;345:825-829. unequivocally demonstrated in randomized trials.60-65 13. Boshuizen HC, Izaks GJ, van Buuren S, Ligthart GJ. Blood pressure and mortality in elderly people aged 85 and older: community based study. BMJ. 1998;147: Higher levels of both systolic and pulse pressure had 1780-1784. strong linear relationships with cardiovascular and total 14. Mattila K, Haavisto M, Rajala S, Heikinheimo R. Blood pressure and five year sur- mortality. Borderline elevations in systolic pressure (140- vival in the very old. BMJ. 1988;296:887-889. 159 mm Hg) were associated with significant elevations 15. Langer RD, Ganiats TG, Barrett-Connor E. Paradoxical survival of elderly men with high blood pressure. BMJ. 1989;298:1356-1358. in risk, particularly among individuals who also had low 16. Cruickshank JM, Thorp IM, Zacharias FJ. Benefits and potential harm of lower- diastolic pressures. Our data also shed light on the ap- ing high blood pressure. Lancet. 1987;1:581-584. parent J-shaped relationship between diastolic pressure 17. Alderman MH, Ooi WL, Madhavan S, Cohen H. Treatment-induced blood pres- and mortality. In older adults, low diastolic pressure com- sure reduction and the risk of myocardial infarction. JAMA. 1989;262:920-924. monly occurs in individuals with high pulse pressure and 18. Farnett L, Mulrow CD, Linn WD, Lucey CR, Tuley MR. The J-curve phenomenon 12,66,67 and the treatment of hypertension: is there a point beyond which pressure re- among those with substantial comorbidity. Con- duction is dangerous? JAMA. 1991;265:489-495. trol for these variables explains the apparent elevations 19. Hansson L, Zanchetti A, Carruthers SG, et al. Effects of intensive blood- in risk associated with low diastolic pressure. Therapies pressure lowering and low-dose aspirin in patients with hypertension: principal

(REPRINTED) ARCH INTERN MED/ VOL 160, OCT 9, 2000 WWW.ARCHINTERNMED.COM 2771

Downloaded From: https://jamanetwork.com/ on 09/29/2021 results of the Hypertension Optimal Treatment (HOT) randomised trial. Lancet. afterload-dependent relaxation in intact canine heart. Am J Physiol. 1985;249: 1998;351:1755-1762. H747-H754. 20. Kaplan N. J-curve not burned off by HOT study. Lancet. 1998;351:1748-1749. 47. Kohno F, Kumada T, Kamabayashi M, Hayashida W, Ishikawa N, Sasayama S. 21. Franklin SS, Gustin WG IV, Wong ND, et al. Hemodynamic patterns of age- Change in aortic end-systolic pressure by alterations in loading sequence and related changes in blood pressure: the Framingham Study. Circulation. 1997; its relation to left ventricular isovolumic relaxation. Circulation. 1996;93:2080- 96:308-315. 2087. 22. Darne B, Girerd X, Safar M, Cambien F, Guize L. Pulsatile versus steady com- 48. Watanabe H, Ohtsuka S, Kakihana M, Sugishita Y. Coronary circulation in dogs ponent of blood pressure: a cross-sectional analysis and a prospective analysis with an experimental decrease in aortic compliance. J Am Coll Cardiol. 1993;21: on cardiovascular mortality. Hypertension. 1989;13:392-400. 1497-1506. 23. Fang J, Madhavan S, Cohen H, Alderman MH. Measures of blood pressure and 49. Kass DA, Saeki A, Tunin RS, Recchia FA. Adverse influence of systemic vascular myocardial infarction in treated hypertensive patients. J Hypertens. 1995;13: stiffening on cardiac dysfunction and adaptation to coronary occlusion. Circu- 413-419. lation. 1996;93:1533-1541. 24. Mitchell GF, Moye´ LA, Braunwald E, et al. Sphygmomanometrically determined 50. Buckberg GD, Fixler DE, Archie JP, Hoffman JIE. Experimental subendocardial pulse pressure is a powerful independent predictor of recurrent events after myo- ischemia in dogs with normal coronary arteries. Circ Res. 1972;30:67-81. cardial infarction in patients with impaired left ventricular function. Circulation. 51. Girerd X, Laurent S, Pannier B, Asmar R, Safar M. Arterial distensibility and left 1997;96:4254-4260. ventricular hypertrophy in patients with sustained essential hypertension. Am 25. Lee ML, Rosner BA, Weiss ST. Relationship of blood pressure to cardiovascular Heart J. 1991;122:1210-1214. death: the effects of pulse pressure in the elderly. Ann Epidemiol. 1999;9:101- 52. Saba PS, Roman MJ, Pini R, Spitzer M, Ganau A, Devereux RB. Relation of ar- 107. terial pressure waveform to left ventricular and carotid anatomy in normoten- 26. Chae CU, Pfeffer MA, Glynn RJ, Mitchell GF, Taylor JO, Hennekens CH. Pulse sive subjects. J Am Coll Cardiol. 1993;22:1873-1880. pressure predicts congestive heart failure in the elderly. JAMA. 1999;281:634- 53. Levy D, Garrison RJ, Savage DD, Kannel WB, Castelli WP. Prognostic implica- 639. tions of echocardiographically determined left ventricular mass in the Framing- 27. Franklin SS, Khan SA, Wong ND, Larson MG, Levy D. Is pulse pressure useful in ham Heart Study. N Engl J Med. 1990;322:1561-1566. predicting risk for coronary heart disease? the Framingham Heart Study. Circu- 54. Farrar DJ, Bond MG, Riley WA, Sawyer JK. Anatomic correlates of aortic pulse lation. 1999;100:354-360. wave velocity and carotid artery elasticity during atherosclerosis progression and 28. Dyer AR, Stamler J, Shekelle RB, et al. Pulse pressure, III: prognostic signifi- regression in monkeys. Circulation. 1991;83:1754-1763. cance in four Chicago epidemiologic studies. J Chronic Dis. 1982;35:283-294. 55. Hirai T, Sasayama S, Kawasaki T, Yagi S. Stiffness of systemic arteries in pa- 29. Hypertension Detection and Follow-up Program Cooperative Group. Variability tients with myocardial infarction: a noninvasive method to predict severity of coro- of blood pressure and the results of screening in the HDFP. J Chronic Dis. 1978; nary atherosclerosis. Circulation. 1989;80:78-86. 31:651-657. 56. Witteman JCM, Grobbee DE, Valkenberg HA, et al. J-shaped relation between 30. Rose GA, Blackburn H, Gillum RF, Prineas RJ. Cardiovascular Survey Methods. change in diastolic blood pressure and progression of aortic atherosclerosis. Lan- Geneva, Switzerland: World Health Organization; 1982:162-165. cet. 1994;343:504-507. 31. Branch LG, Katz S, Kniepmann K, et al. A prospective study of functional status 57. Lyon RT, Runyon-Hass A, Davis HR, Glagov S, Zarins CK. Protection from ath- among community elderly. Am J Public Health. 1984;74:266-268. erosclerotic lesion formation by reduction of artery wall motion. J Vasc Surg. 32. Rosow I, Breslau N. A Guttman health scale for the aged. J Gerontol. 1966;21: 1987;5:59-67. 556-559. 58. Cheng GC, Loree HM, Kamm RD, Fishbein MC, Lee RT. Distribution of circum- 33. Cornoni-Huntley J, Brock DB, Ostfeld AM, et al, eds. Established Populations for ferential stress in ruptured and stable atherosclerotic lesions: a structural analy- Epidemiologic Studies of the Elderly: Resource Data Book. Bethesda, Md: Na- sis with histopathological correlation. Circulation. 1993;87:1179-1187. tional Institutes of Health; 1986. NIH publication 86-2443. 59. Wilson PWF, D’Agostino RB, Levy D, Belanger AM, Silbershatz H, Kannel WB. 34. Glynn RJ, Field TS, Satterfield S, et al. Modification of increasing systolic blood Prediction of coronary heart disease using risk factor categories. Circulation. 1998; pressure in the elderly during the 1980s. Am J Epidemiol. 1993;138:365-379. 97:1837-1847. 35. Glynn RJ, Brock DB, Harris T, et al. Use of antihypertensive drugs and trends in 60. Amery A, Birkenha¨ger W, Brixko P, et al. Mortality and morbidity results from blood pressure in the elderly. Arch Intern Med. 1995;155:1855-1860. the European Working Party on High Blood Pressure in the Elderly trial. Lancet. 36. World Health Organization. International Classification of Diseases, Ninth Revi- 1985;1:1349-1354. sion (ICD-9). Geneva, Switzerland: World Health Organization; 1977. 61. Coope J, Warrender TS. Randomised trial of treatment of hypertension in el- 37. Harrell FE Jr, Lee KL, Mark DB. Multivariable prognostic models: issues in de- derly patients in primary care. BMJ. 1986;293:1145-1151. veloping models, evaluating assumptions and adequacy, and measuring and re- 62. SHEP Cooperative Research Group. Prevention of stroke by antihypertensive drug ducing errors. Stat Med. 1996;15:361-387. treatment in older persons with isolated systolic hypertension: final results of 38. Bulpitt CJ, Fletcher AE. Aging, blood pressure and mortality. J Hypertens. 1992; the Systolic Hypertension in the Elderly Program (SHEP). JAMA. 1991;265:3255- 10(suppl 7):S45-S49. 3264. 39. Build and Blood Pressure Study, 1959. Vol 1. Chicago, Ill: Society of Actuaries; 63. Dahlo¨f B, Lindholm LH, Hannson L, Scherste´n B, Ekbom T, Wester P-O. Mor- 1959. bidity and mortality in the Swedish Trial in Old Patients With Hypertension 40. Benetos A, Safar M, Rudnichi A, et al. Pulse pressure: a predictor of long-term (STOP-Hypertension). Lancet. 1991;338:1281-1285. cardiovascular mortality in a French male population. Hypertension. 1997;30: 64. MRC Working Party. Medical Research Council trial of treatment of hyperten- 1410-1415. sion in older adults: principal results. BMJ. 1992;304:405-412. 41. Neaton JD, Wentworth D. Serum cholesterol, blood pressure, cigarette smok- 65. Staessen JA, Fagard R, Thijs L, et al. Randomised double-blind comparison of ing, and death from coronary heart disease: overall findings and differences by placebo and active treatment for older patients with isolated systolic hyperten- age for 316099 white men. Arch Intern Med. 1992;152:56-64. sion. Lancet. 1997;350:757-764. 42. The Pooling Project Research Group. Relationship of blood pressure, serum cho- 66. Bots ML, Witteman JC, Hofman A, de Jong PT, Grobbee DE. Low diastolic blood lesterol, smoking habit, relative weight and ECG abnormalities to incidence of pressure and atherosclerosis in elderly subjects: the Rotterdam Study. Arch In- major coronary events: final report of the Pooling Project. J Chronic Dis. 1978; tern Med. 1996;156:843-848. 31:201-306. 67. D’Agostino RB, Belanger AJ, Kannel WB, Cruickshank JM. Relation of low dias- 43. Milnor WR. Arterial impedance as ventricular afterload. Circ Res. 1975;36:565- tolic blood pressure to coronary heart disease death in presence of myocardial 570. infarction: the Framingham Study. BMJ. 1991;303:385-389. 44. Urschel CW, Covel JW, Sonnenblick EH, Ross JR Jr, Braunwald E. Effects of de- 68. Mitchell GF, Pfeffer MA, Finn PV, Pfeffer JM. Equipotent antihypertensive agents creased aortic compliance on performance of the left ventricle. Am J Physiol. variously affect pulsatile hemodynamics and regression of cardiac hypertrophy 1968;214:298-304. in spontaneously hypertensive rats. Circulation. 1996;94:2923-2929. 45. Kelly RP, Tunin R, Kass DA. Effect of reduced aortic compliance on cardiac ef- 69. Latson TW, Hunter WC, Katoh N, Sagawa K. Effect of nitroglycerin on aortic im- ficiency and contractile function of in situ canine left ventricle. Circ Res. 1992; pedance, diameter and pulse-wave velocity. Circ Res. 1988;62:884-890. 71:490-502. 70. Avolio AP, Clyde KM, Beard TC, et al. Improved arterial distensibility in normo- 46. Hori M, Inoue M, Kitakaze M, et al. Loading sequence is a major determinant of tensive subjects on a low salt diet. Arteriosclerosis. 1986;6:166-169.

(REPRINTED) ARCH INTERN MED/ VOL 160, OCT 9, 2000 WWW.ARCHINTERNMED.COM 2772

Downloaded From: https://jamanetwork.com/ on 09/29/2021