The study sample consisted of individuals who were participants of the Framingham Offspring cohort and the details and design have been described.7
Offspring (and their spouses) of the Framingham Original cohort were recruited beginning in 1971. Participants have undergone follow-up examinations at the Heart Study clinic approximately every 4 to 6 years, including standardized questionnaires, a physical examination, and assessment of standard cardiovascular risk factors. The examining physician measured the blood pressure of seated participants using a mercury column sphygmomanometer, an appropriately sized cuff, and a standardized protocol.8
The average of 2 blood pressure readings taken by a physician served as the examination blood pressure, the primary exposure, and end point for subsequent analyses. Pulse pressure was calculated as systolic blood pressure minus diastolic blood pressure. Mean arterial pressure was calculated as diastolic blood pressure plus pulse pressure divided by 3. Presence of hypertension was defined as a systolic blood pressure of 140 mm Hg or greater, diastolic blood pressure of 90 mm Hg or greater, or the use of anti-hypertensive medication.
The present investigation is based on examination cycles 7 (1998–2001) and 8 (2005–2008), with examination cycle 7 representing the first cycle to include tonometry measurements. The study protocol was approved by the Boston University Medical Center institutional review board; written informed consent was obtained from all participants.
We evaluated 3 measures of arterial stiffness and pressure pulsatility derived from arterial tonometry, namely carotid-femoral pulse wave velocity (CFPWV), central forward pressure wave amplitude (FWA), and augmentation index. The CFPWV is the criterion standard for assessing aortic stiffness.9
The FWA depends on peak systolic blood flow and characteristic impedance of the aorta, which is the resistance of the aorta to the pulsatile component of blood flow.
The FWA and CFPWV are both dependent on aortic wall stiffness and lumen diameter. However, compared with CFPWV, FWA has a markedly (5-fold) greater dependence on aortic diameter and can be conceptualized as a measure of matching between peak systolic blood flow and diameter of the aorta. Because the central pressure wave propagates distally, it is partly reflected at the interface with muscular arteries due to impedance mismatch, creating a backward traveling reflected wave. Augmentation index is defined as the proportion of central pulse pressure that is attributable to a late systolic increase in pressure due to overlap between the forward and reflected pressure wave, and hence is a measure of peripheral wave reflection.
Arterial tonometry measures were acquired as previously described.10,11
After more than 5 minutes of rest, supine brachial systolic and diastolic blood pressures were obtained. For tonometry calibration only, we used an oscillometric blood pressure device for examination cycle 7 and an auscultatory device for examination cycle 8.10,11
Arterial tonometry with a simultaneously acquired electrocardiogram was obtained for the brachial, radial, femoral, and carotid arteries.10,11
All recordings were performed on the right side of the body. Transit distances were assessed by body surface measurements from the suprasternal notch to the pulse recording site. Details of signal analyses and data processing have been published elsewhere and are summarized in eFigure 1 and eFigure 2 at http://www.jama.com
Brachial Artery Flow and Flow-Mediated Dilation
Brachial artery flow velocity and flow-mediated dilation (FMD) measurements were performed using a Toshiba SSH-140A ultrasound system as previously described in detail.12
In brief, after acquiring baseline brachial artery flow velocity and brachial artery diameter, a cuff was inflated on the right forearm to at least 50 mm Hg above the participant’s systolic blood pressure, interrupting blood flow for 5 minutes. Brachial artery flow was measured again during the first 15 seconds after deflating the cuff and brachial artery diameter was reassessed 60 seconds after cuff release. Flow-mediated dilation was defined as the percentage change in brachial diameter between baseline and hyperemia.
Carotid-femoral PWV was inverse-transformed to reduce heteroscedasticity and was then multiplied by −1000 to restore directionality and convert the units to milliseconds per meter. Correlates of blood pressure and tonometry traits were explored using multivariable regression analysis. The primary outcomes (dependent variables) assessed were continuous systolic blood pressure, diastolic blood pressure, mean arterial pressure, and pulse pressure during examination cycle 8 as well as incident hypertension in those participants free of hypertension during examination cycle 7. The secondary outcomes were CFPWV, FWA, and augmentation index measured during examination cycle 8.
In a first modeling stage, we investigated clinical and biochemical variables during examination cycle 7 (not including blood pressure, tonometry measures, and brachial ultrasound traits) that were associated (P < .05) with at least 1 of the assessed continuous outcomes in a multivariable model. The following independent variables were eligible for entry into the model: age, sex, body mass index (calculated as weight in kilograms divided by height in meters squared), height, heart rate, diabetes, total cholesterol, high-density lipoprotein cholesterol, triglycerides, lipid-lowering treatment, fasting glucose, prevalent cardiovascular disease, current smoking, and time between examination cycles 7 and 8. All variables except lipid-lowering treatment and prevalent cardiovascular disease were identified as correlates of future blood pressure or tonometry traits and were included in later linear regression models. Models with tonometry traits as the dependent variable were additionally adjusted for antihypertensive treatment.
In a second stage, we used a stepwise procedure to identify blood pressure and tonometry measures collected during examination cycle 7 (systolic blood pressure, diastolic blood pressure, mean arterial pressure, pulse pressure, CFPWV, FWA, and augmentation index) that were associated with the assessed continuous outcome in multivariable models. To avoid colinearity effects among blood pressure components, models for systolic and diastolic blood pressure during examination cycle 8 considered only systolic and diastolic blood pressure during examination cycle 7, whereas models for pulse pressure and mean arterial pressure during examination cycle 8 considered only pulse pressure and mean arterial pressure during examination cycle 7; tonometry measures were included in all models. In a third step, we explored whether brachial flow or FMD, which were available in a subset of participants, further improved the regression models.
Correlates of incident hypertension were assessed in an analogous 3-stage design using multivariable logistic regression modeling. From the list of potential clinical covariates described above, modeling stage 1 identified only age, sex, body mass index, height, and triglycerides as significant correlates of incident hypertension in our study sample. Hence, for reasons of model parsimony, no other covariates were used in stages 2 and 3 of the logistic regression modeling for incident hypertension.
In all analyses, a 2-sided P value of less than .05 was regarded as significant. All analyses were performed using SAS software version 9.2 (SAS Institute Inc).