Predictors of New-Onset Diastolic and Systolic Hypertension
The Framingham Heart Study
Background— Factors leading differentially to the development of isolated diastolic (IDH), systolic-diastolic (SDH), and isolated systolic (ISH) hypertension are poorly understood. We examined the relations of blood pressure (BP) and clinical risk factors to the new onset of the 3 forms of hypertension.
Methods and Results— Participants in the Framingham Heart Study were included if they had undergone 2 biennial examinations between 1953 and 1957 and were free of antihypertensive therapy and cardiovascular disease. Compared with optimal BP (SBP <120 and DBP <80 mm Hg), the adjusted hazard ratios (HRs) for developing new-onset IDH over the ensuing 10 years were 2.75 for normal BP, 3.29 for high-normal BP (both P<0.0001), 1.31 (P=0.40) for SDH, and 0.61 (P=0.36) for ISH. The HRs of developing new-onset SDH were 3.32, 7.96, 7.10, and 23.12 for the normal BP, high-normal BP, ISH, and IDH groups, respectively (all P<0.0001). The HRs of developing ISH were 3.26 for normal and 4.82 for high-normal BP (both P<0.0001), 1.39 (P=0.24) for IDH, and 1.69 (P<0.01) for SDH. Increased body mass index (BMI) during follow-up predicted new-onset IDH and SDH. Other predictors of IDH were younger age, male sex, and BMI at baseline. Predictors of ISH included older age, female sex, and increased BMI during follow-up.
Conclusion— Given the propensity for increased baseline BMI and weight gain to predict new-onset IDH and the high probability of IDH to transition to SDH, it is likely that IDH is not a benign condition. ISH arises more commonly from normal and high-normal BP than from “burned-out” diastolic hypertension.
Received July 17, 2004; revision received October 22, 2004; accepted December 20, 2004.
Diastolic hypertension, defined as a diastolic blood pressure (DBP) of 90 mm Hg or higher, often occurs in conjunction with a systolic blood pressure (SBP) ≥140 mm Hg: so-called systolic-diastolic hypertension (SDH).1 When diastolic hypertension occurs with an SBP <140 mm Hg, it is called isolated diastolic hypertension (IDH).2 Isolated systolic hypertension (ISH) is defined as an SBP ≥140 and a DBP of <90 mm Hg.1 We previously described the age-related longitudinal tracking of DBP and SBP in the original Framingham Heart cohort,3 but the pathways leading to and the risk factors that predispose to the development of the new-onset hypertension subtypes have not been fully characterized.
See p 1094
In younger adults, IDH is more common than SDH.4 Indeed, the Third National Health and Nutrition Examination Survey (NHANES III, 1988 to 1991)5 showed that IDH was the most frequent form of diastolic hypertension in young adults <40 years old and comparable to SDH in prevalence from age 40 to 49 years. Together, IDH and SDH accounted for more than 75% of younger adult individuals with untreated hypertension.5 Surprisingly, despite its frequency in young hypertensive individuals, IDH has been considered by some to be an artifact of measurement6 or to be of no clinical importance.2
In contrast, ISH becomes the dominant form of hypertension from the sixth decade of life and beyond.5,7 NHANES III data also showed that 3 out of 4 adults with hypertension were 50 years of age or older, and approximately 80% of untreated and inadequately treated individuals with hypertension age 50 years and older have ISH.5 It is unclear, however, whether the majority of ISH in older people arises from younger people with “burned-out” diastolic hypertension rather than de novo in persons with nonhypertensive BP.
The first goal of the present study was to determine, by use of longitudinal follow-up in the original Framingham Heart Study cohort from 1963 to 1967, the risk for the development of IDH, SDH, and ISH among nonhypertensive subjects and from hypertension subtypes other than the specific outcome analyzed. For example, when the outcome analyzed was IDH, subjects with IDH were excluded from the baseline population. The second goal was to identify clinical risk factors that predispose differentially to the development of new-onset IDH, SDH, and ISH. The Framingham Heart Study is uniquely suited to explore these questions because it provides an opportunity to characterize the natural history of largely untreated BP during the early years of the study, thus minimizing treatment bias.
The Framingham Heart Study began in 1948 with the enrollment of 5209 men and women, 28 to 62 years of age at entry, with all subjects undergoing an extensive cardiovascular history, physical examination, 12-lead ECG, and various blood chemistry tests. Detailed descriptions of study design, the method of determining BP measurements, and the method of classifying cardiovascular end points have been published elsewhere.3,8
Participants in the original Framingham Heart Study cohort were included in this investigation if they (1) had undergone both biennial examinations 3 and 4 (1953 to 1957); (2) were not receiving antihypertensive treatment; (3) were free of previous cardiovascular disease, such as myocardial infarction, stroke, or heart failure (the advent of a cardiovascular event may interrupt the natural history of the normal transition from one BP category to another); and (4) were not classified as having the hypertension outcome of interest at baseline. Values from the first and second biennial examinations were not used because both SBP and DBP on these examinations were slightly higher than those recorded on subsequent examinations. This finding was consistent with an early-visit “white-coat” hypertensive effect. Follow-up was through biennial examination 9 (1965–1967). We used 1953 to 1967 as our follow-up period because a more contemporary one would have resulted in a substantial increase in subjects receiving antihypertensive therapy and/or experiencing cardiovascular events (because of older age) and thus limiting our ability to draw conclusions about the natural history of new-onset IDH, SDH, and ISH. The term “new-onset” is a category-specific term that refers to the first time that the specific hypertension subtype of interest develops in someone who did not previously have it; it does not include subsequent additional transitions to other hypertension subtypes.
Blood Pressure Classification
Those qualifying for the study were classified according to the Sixth Report of the Joint National Committee on Detection, Evaluation, and Treatment of High Blood Pressure (JNC-VI).9 Subjects were divided into 6 categories (Table 1) on the basis of similar BP groups on both examinations 3 and 4 or on the average of BP from examinations 3 and 4, as follows: group 1, optimal BP (SBP <120 and DBP <80 mm Hg); group 2, normal BP (SBP 120 to 129 or DBP 80 to 84 mm Hg); group 3, high-normal BP (SBP 130 to 139 or DBP 85 to 89 mm Hg); group 4, IDH (SBP <140 and DBP ≥90 mm Hg); group 5, SDH (SBP ≥140 and DBP ≥90 mm Hg); and group 6, ISH (SBP ≥140 and DBP <90 mm Hg). Those who met the above qualifications were followed up for a mean of 10 years through biennial examination 9 (1965 to 1967) for the development of new-onset IDH, SDH, and ISH, defined as occurring on 2 consecutive biennial examinations or the average of BP from 2 consecutive biennial examinations. Thus, criteria for defining (1) baseline BP groups, (2) baseline BP exclusions, and (3) new-onset BP end points were identical.
Incidence rates for new-onset IDH, SDH, and ISH (per 1000 person-years) were determined from baseline through examination 9, both unadjusted and after adjustment for age and sex. Cox proportional-hazards regression was performed, providing hazard ratios10 for developing IDH, SDH, or ISH for each baseline BP category, using those with optimal BP as a reference group. We adjusted for covariates that were related to the development of hypertension in a previous Framingham heart Study investigation11 as follows: age, sex, body mass index (BMI in kg/m2) at baseline and change in BMI (as a time-dependent covariate), total cholesterol (mg/dL), blood glucose (mg/dL), heart rate (bpm), and current smoking. Similarly, Cox regression analysis was performed to provide standardized hazard ratios for 1-SD increments in risk factors at baseline in the prediction of incident hypertension subtypes. The development of new-onset IDH, SDH, and ISH was determined at the follow-up examinations occurring at any time during the 3, 5, 7, 9, and 11 years of follow-up after the baseline examinations. Individuals were censored if they started antihypertensive medication or if they sustained a myocardial infarction, stroke, or heart failure before developing the new-onset hypertension subtype of interest. SAS statistical software (SAS Institute)12 was used. A probability value of P<0.05 was considered significant.
Demographic and Clinical Characteristics
Of the original 5209 Framingham Heart Study participants, 1027 did not attend both examinations 3 and 4, 155 were receiving antihypertensive therapy, and 112 had experienced a previous cardiovascular event. There remained 3915 subjects who met the inclusion criteria and were eligible for this study, comprising 1726 men and 2189 women; the mean age was 48.5 years. Table 1 shows the mean values of selected risk factors by BP groupings at baseline. There was a predominance of women in all BP groupings except for IDH, in which younger men accounted for 65% of the subjects, and high-normal BP, in which there were equal numbers of men and women.
Of the subjects who met the inclusion criteria, 193 were classified as having IDH at baseline (54 on both biennial examination 3 or 4 and 139 from the average of examinations 3 and 4). Among the 3722 participants without IDH at baseline, 240 developed new-onset IDH during follow-up. Of these, 48 were censored (23 for starting antihypertensive therapy, 18 for sustaining cardiovascular events, and 7 for meeting both exclusion criteria), leaving 192 individuals (111 men and 82 women, 5.2% of the baseline sample) who qualified as new onset of IDH (Table 2). After adjustment for age and sex, incidence rates of IDH (per 1000 person-years) were 2.1, 7.5, 9.2, 3.8, and 2.1 from optimal BP, normal BP, high-normal BP, SDH, and ISH, respectively. At the time of diagnosis of IDH, the mean age was 50.3 years and the mean BP was 131.7/93.4 mm Hg.
Of the subjects who met the inclusion criteria, 618 were classified as having SDH at baseline (416 on both biennial examination 3 and 4 and 202 from the average of examinations 3 and 4). Among the remaining 3297 participants without SDH at baseline, 685 developed new-onset SDH during follow-up. Of these, 132 were censored (58 for starting antihypertensive therapy, 67 for sustaining cardiovascular events previously, and 7 for fulfilling both exclusions), leaving 553 individuals (254 men and 299 women, 16.8% of the baseline sample) who qualified as having new onset of SDH (Table 2). After adjustment for age and sex, incidence rates of SDH (per 1000 person-years) were 4.3, 14.7, 31.7, 32.3, and 79.5 from optimal BP, normal BP, high-normal BP, IDH, and ISH, respectively. At the time of diagnosis of SDH, the mean age was 53.4 years and the mean BP was 152.5/96.2 mm Hg.
Of the subjects who met the inclusion criteria, 349 were classified as having ISH at baseline (137 on the basis of BP on both biennial examination 3 and 4 and 212 by the average of BP on examination 3 and 4). Of the remaining 3566 subjects without ISH at baseline, 808 developed new-onset ISH during follow-up. Of these, 203 individuals were censored (127 for starting antihypertensive therapy, 51 for sustaining a cardiovascular event, and 25 for having both exclusion criteria before developing ISH), leaving 605 individuals (238 men and 367 women, 17.0% of the baseline population) who developed ISH (Table 2). After adjustment for age and sex, incidence rates of ISH (per 1000 person-years) were 8.1, 24.3, 36.1, 12.1, and 13.6 for optimal BP, normal BP, high-normal BP, IDH, and SDH, respectively. The mean age at diagnosis of ISH was 58.6 years, and the mean BP was 149.4/82.6 mm Hg.
Hazard Ratios for Developing Outcome Variables
With persons with optimal BP at the baseline visit used as a reference group, hazard ratios (HRs) and 95% CIs for developing IDH, SDH, and ISH are shown in Table 3. After adjustment for age, sex, baseline BMI, change in BMI during follow-up, total cholesterol, blood glucose, heart rate, and smoking, the HRs for developing IDH were 2.75 (95% CI, 1.79 to 4.23) for normal BP, 3.29 (95% CI, 2.09 to 5.19) for high-normal BP (both probability values <0.0001), and 1.31 (95% CI, 0.70 to 2.43, P=0.40) for SDH and 0.61 (95% CI, 0.21 to 1.76, P=0.36) for ISH.
The HRs for developing SDH, after adjustment for risk factors as above, were 3.32 (95% CI, 2.38 to 4.62) for normal BP, 7.96 (95% CI, 5.73 to 11.04) for high-normal BP, 7.10 (95% CI, 4.80 to 10.50) for ISH, and 23.12 (95% CI, 15.85 to 33.71) for IDH (all probability values <0.0001).
The HRs for developing ISH, after adjustment for risk factors as above, were 3.26 (95% CI, 2.48 to 4.29) for normal BP, 4.82 (95% CI, 3.65 to 6.37) for high-normal BP (both probability values <0.0001), 1.39 (95% CI, 0.80 to 2.43, P=0.24) for IDH, and 1.69 (95% CI, 1.21 to 2.37, P<0.01) for SDH.
By use of dummy variables for the prediction of new-onset IDH, all categories were significantly different from each other except for the comparison between normal and high-normal BP and between normal BP and ISH. For the prediction of new-onset SDH, all categories were significantly different from each other except for the comparison between high-normal BP and SDH and between high-normal BP and ISH. For the prediction of new-onset ISH, all categories were significantly different from each other except for the comparison between optimal BP and IDH and between IDH and SDH (data not shown for IDH, SDH, or ISH subtypes).
Risk Factors for Development of Hypertension
HRs of risk factors for the development of IDH, SDH, and ISH are shown in Table 4. After adjustment for baseline BP groups, HRs for developing IDH were 0.62 (95% CI, 0.52 to 0.74) for each 1 SD of age, 0.54 (95% CI, 0.39 to 0.74) for female sex, 1.24 (95% CI, 1.06 to 1.44) for each 1 SD of baseline BMI, and 1.28 (95% CI, 1.06 to 1.55) for each 1-SD increase in BMI over time. Baseline heart rate, total cholesterol, blood glucose, and smoking were not predictive of IDH incidence.
After adjustment for baseline BP groups, the HR for developing SDH was 1.92 (95% CI, 1.72 to 2.15) for each 1 SD of increase in BMI during follow-up. Baseline age, sex, baseline BMI, heart rate, total cholesterol, blood glucose, and smoking were not predictive of SDH incidence.
After adjustment for baseline BP groups, HRs for developing ISH were 1.82 (95% CI, 1.66 to 1.99) for 1 SD of age, 1.40 (95% CI, 1.16 to 1.68) for female sex, and 1.15 (95% CI, 1.04 to 1.28) for an increase in BMI during follow-up. BMI at baseline was not associated with an increased likelihood of ISH. In separate models, BMI at baseline did not predict new-onset ISH in younger (age <50 years, P=0.22) or older subjects (age ≥50 years, P=0.31), in men (P=0.71) or women (P=0.91). Baseline heart rate, total cholesterol, blood glucose, and smoking were not predictive of ISH incidence.
Average Maximum DBP Values Reached Before the Development of ISH
Of subjects who developed ISH, 59% did not have antecedent diastolic hypertension either at baseline or at any examination before ISH onset (average maximum DBP of 80.8 mm Hg), 23% had a maximum DBP of 90 to 94 mm Hg (average maximum DBP of 91.6 mm Hg), and 18% had a maximum DBP of 95 mm Hg or higher (average maximum DBP of 99.4 mm Hg).
When the new-onset BP category was based on 1 rather than 2 consecutive biennial examinations, the results were essentially unchanged (data not shown).
To estimate the potential bias of censoring subjects at the time of starting antihypertensive therapy or on sustaining a cardiovascular complication, HRs for developing IDH, SDH, and ISH were recalculated (1) after inclusion of subjects who started therapy and/or events subsequent to baseline and (2) after excluding all subjects from baseline who subsequently started therapy and/or had events. The overall results of the study for new-onset IDH and SDH were essentially unchanged, or there was a modest strengthening of HRs across all BP categories when either the inclusion or exclusion methods were substituted for censoring. In contrast, there was a notable increase in the HR for new-onset ISH from 1.69 to 3.79 (95% CI, 2.81 to 5.09) after substitution of the inclusion method for censoring and an increase in the HR to 3.83 (95% CI, 2.64 to 5.55) after substitution of the exclusion method for censoring.
The major findings of this study were that normal and high-normal BP had the highest HRs for new onset of IDH and ISH, whereas IDH had by far the highest HR for new-onset SDH over a 10-year follow-up period. The majority (59%) of those who developed ISH did not have antecedent diastolic hypertension; indeed, their average maximum DBP before developing ISH was 81 mm Hg. Only 18% of those who developed new-onset ISH had a previous DBP of 95 mm Hg or higher. Baseline BMI was a predictor of IDH, and an increase in BMI during follow-up was a predictor of IDH, SDH, and ISH. Other predictors of hypertensive subtypes were young age and male sex for IDH and old age and female sex for ISH.
Note that what was previously called normal and high-normal BP in the Sixth Joint Report of the National Committee on Detection, Evaluation, and Treatment of High Blood Pressure9 (JNC-VI) has now been renamed prehypertension in the JNC 71 report to indicate a likelihood of increased cardiovascular risk. We have chosen to report our data using these former JNC-VI categories to examine any differences in risk of developing hypertensive subtypes between these categories.
The frequent progression from prehypertension to IDH in young adults is consistent with underlying increased peripheral resistance.13 Brachial DBP and SBP rise with increases in peripheral resistance, but the rise in peripheral SBP, unlike DBP, is partially attenuated by the reduction in peripheral amplification that occurs with the development of hypertension in young adult men and to a lesser extent in young women.13,14 There are 2 lines of evidence supporting this concept. Wilkinson and colleagues,15 using pulse-wave analysis with applanation tonometry, showed that a rise in peripheral DBP was accompanied by an attenuated rise in peripheral SBP as a result of decreased peripheral amplification in subjects <50 years old but not in those ≥50 years old. Similarly, Millasseau and colleagues16 showed that carotid-femoral pulse-wave velocity, an indicator of aortic stiffness, was more closely correlated with peripheral DBP than SBP in subjects <50 years old; this was not found in those ≥50 years old.
In older people, ISH results from increased elastic artery stiffness, which increases pulse-wave velocity and wave reflection amplitude.13 In the present study, alteration in hemodynamics, in the absence of direct measurements, can be inferred from longitudinal changes in BP variables. Because the majority of persons who developed new-onset ISH had no previous diastolic hypertension, we infer that large-artery stiffness and not vascular resistance was the predominant mechanism for the rise in SBP.3,17 In contrast, a minority of persons with new-onset ISH had antecedent diastolic hypertension. This small subset of individuals may have developed ISH from large-artery stiffness superimposed on antecedent high vascular resistance, consistent with so-called “burned-out” diastolic hypertension.18
Significance of IDH
The results of the present study indicate that there is a significantly increased likelihood of IDH developing in persons with antecedent prehypertension. Furthermore, participants with IDH at baseline were 23 times as likely to develop SDH as those with optimal blood pressure, and 82.5% of persons with baseline IDH developed new-onset SDH during the ensuing 10 years of follow-up. IDH was characterized as occurring in overweight or obese younger men, who in turn had a high risk for the future development of SDH. These findings, therefore, question the validity of earlier studies that concluded that IDH was artifactual6 or was without increased cardiovascular risk.2,19–22 The present study did not examine cardiovascular outcomes, so we can only hypothesize that IDH is associated with increased risk for clinical events.
Overweight and Obesity as Predisposing Factors to Hypertensive Subtypes
Strong evidence23 implicates weight gain and obesity as prime factors in the future development of diastolic hypertension. Indeed, both baseline BMI and subsequent weight gain were strong determinants of new-onset IDH and SDH in the present study. Surprisingly, BMI at baseline was not a predictor of ISH; however, an increase in BMI during follow-up was a significant but weak predictor. The lack of association of baseline BMI with the incidence of ISH in our study is supportive of a cross-sectional Mexican survey,24 which failed to show a significant association of excess body weight with ISH.
Strengths and Limitations of the Study
Because of the narrow diagnostic thresholds that define and separate nonhypertensive status and various hypertensive subtypes from each other, there is the probability of random variation of BP measurements from one biennial examination to the next, and hence, the possibility of misclassification of baseline and follow-up BP categories. To minimize this problem, we defined our 6 baseline BP categories on the basis of BP recorded at biennial examinations 3 and 4 and by the average of examinations 3 and 4; these criteria were used to remove persons at baseline with the specific hypertensive subtype outcome variable. Similarly, we required that the new-onset hypertension be present at 2 consecutive or the average of 2 consecutive examinations during follow-up. These requirements may have resulted in a survival bias because the presence of 2 consecutive examinations required patient survival and attendance at both visits; however, because the results from the use of 2 consecutive examinations for the development of new-onset outcome variables were ostensibly the same as or even stronger than noted with a single examination, this potential bias does not appear to be important.
Furthermore, we favored censoring persons rather than including them in the study after they had started antihypertensive therapy during follow-up. Substitution of the inclusion method for censoring resulted in a greater likelihood of SDH converting to ISH, because with treatment, DBP usually decreased to <90 mm Hg, whereas SBP often remained >140 mm Hg. We also favored censoring persons rather than excluding them from the entire study if they were started on antihypertensive therapy during follow-up. Substitution of the exclusion method for censoring also resulted in a greater likelihood of SDH converting to ISH. This may have occurred because the exclusion of many subjects with severely elevated DBP who began therapy resulted in a larger percentage of subjects with mildly elevated DBP that would more easily transition to ISH over the 10-year follow-up period. We believe that censoring is the least problematic and the most conservative of the 3 approaches, in that it preserves the natural history of age-related BP changes until the initiation of therapy or onset of cardiovascular events: In this way, selection bias is minimized. Finally, because the subjects of the original Framingham sample were largely middle-class and white, our findings may not be generalizable to other ethnic groups.
Given the propensity for increased BMI and weight gain in the development of new-onset of IDH and the high probability of IDH to transition to SDH, it is likely that IDH is not a benign condition. With the predicted aging of the US population over the next 20 years, the number of elderly persons with ISH should continue to grow5 unless there is a dramatic improvement in primary prevention. The majority of individuals with new-onset of ISH arise from those with normal or high-normal BP. Our findings support the close monitoring of persons who are currently prehypertensive and the use of lifestyle measures to prevent the progression from prehypertension to diastolic and/or systolic hypertension.
The Framingham Heart Study is funded by National Institutes of Health contract NO1-HC-25195.
Guest Editor for this article was Roberto Bolli, MD.
Pickering TG. Isolated diastolic hypertension. J Clin Hypertension. 2003; 5: 411–413.
Franklin SS, Guston W IV, Wong ND, Larson MG, Weber MA, Kannel WB, Levy D. Hemodynamic patterns of age-related changes in blood pressure. The Framingham Heart Study. Circulation. 1997; 96: 308–315.
Franklin SS, Milagros JJ, Wong ND, L’Italien GJ, Lapuerta P. Predominance of isolated systolic hypertension among middle-aged and elderly US hypertensives: analysis based on National Health and Nutrition Examination Survey (NHANES III). Hypertension. 2001; 37: 869–874.
Blank SG, Mann SJ, James GD, West JE, Pickering TG. Isolated elevation of diastolic blood pressure: real or artifactual? Hypertension. 1995; 26: 383–389.
Burt VL, Whelton P, Roccella EJ, Brown C, Cutler JA, Higgins M, Goran MJ, Labarthe D. Prevalence of hypertension in the US adult population: results from the third National Health and Nutrition Survey (1988–1991). Hypertension. 1995; 25: 305–313.
Cox DR. Regression models and life tables. J R Stat Soc B. 1972; 34: 187–220.
SAS Institute Inc. SAS/STAT software: changes and enhancements through release 6.11. Cary, NC: SAA Institute Inc; 1996: 807–884.
Nichols WW, O’Rourke MF. McDonald’s Blood Flow in Arteries. 4th ed. London, UK: Arnold, Hodder Headline Group; 1998.
Wilkinson IB, Franklin SS, Hall IR, Tyrell S, Cockcroft JR. Pressure amplification explains why pulse pressure is unrelated to risk in young subjects Hypertension. 2001; 38: 1461–1466.
Millasseau S, Ritter JM, Chowienczyk P. Relationship between blood pressure and aortic pulse wave velocity in young and old subjects with essential hypertension. J Hypertens. 2003; 21 (suppl 4): S253. Abstract.
Nichols WW, Nicolini FA, Pepine CJ. Determinants of isolated systolic hypertension in the elderly. J Hypertens. 1991; 10 (suppl 6): S73–S77.
Fang J, Madhavan S, Cohen H, Alderman MH. Isolated diastolic hypertension: a favorable finding among young and middle-aged hypertensive subjects. Hypertension. 1995; 25: 377–382.
Petrovitch H, Curb JD, Bloom-Marcus E. Isolated systolic hypertension and risk of stroke in Japanese-American men. Stroke. 1995; 26: 25–29.
Nielsen WB, Lindenstrom E, Vestbo J, Jensen GB. Is diastolic hypertension an independent risk factor for stroke in the presence of normal systolic blood pressure in the middle-aged and elderly? Am J Hypertens. 1997; 10: 634–639.
Arroyo P, Fernandez V, Avila-Rosas H. Overweight and hypertension: data from the 1992–1993 Mexican Survey. Hypertension. 1997; 30 [pt 2]: 646–649.