Abstract
Background.
Individual measures and previous composite measures of subclinical vascular disease defined high risk for cardiovascular events, but did not detect low and modest risk. A different approach might better describe the spectrum from low to high risk.
Methods and Results.
In the Cardiovascular Health Study, 3,252 participants without history of clinical cardiovascular disease (M ± SD 74.3 years ± 5.1, 63% women, 17% African Americans) had noninvasive vascular assessments in 1992–1993. We assigned a score of 0, 1, or 2 (no, mild, or severe abnormalities) to ankle–arm index, electrocardiogram, and common carotid intima-media thickness, based on clinical cutoffs. A summary index (range 0–6, absent to severe disease) summed individual scores. Abdominal aortic ultrasound and brain magnetic resonance imaging were collected in a subsample. Mortality and incident cardiovascular events were identified through June 2008. Event and death rates increased across index grades. Comparing grades 1 to 5+ with absent disease, and adjusting for demographics, hazard ratios for cardiovascular events within 8 years ranged from 1.1 (95% confidence interval 0.8–1.6) to 4.7 (3.4–6.9) and, for mortality, from 1.5 (1.0–2.3) to 5.0 (3.3–7.7) (p for trend across grades <.001 for both outcomes). Adjustment for cardiovascular risk factors did not substantially change the associations. The index improved mortality risk classification over demographics and risk factors in participants who did not die during the follow-up. Including in the index the aortic ultrasound and the brain magnetic resonance imaging further improved risk classification.
Conclusions.
Older adults with minimal subclinical vascular disease had low cardiovascular events risk and mortality. This approach might more fully account for vascular burden.
Keywords: Epidemiology, Aging, Atherosclerosis, Cardiovascular disease, Mortality
MEASURES of vascular disease at different vascular beds, including carotid, coronary, and peripheral arteries, have been examined in the Cardiovascular Health Study (CHS) as predictors of cardiovascular disease (CVD) (1–3). Each of these measures appears to provide independent information about the risk for clinical CVD even when multiple measures are included in the same model (4). A previous composite indicator from the CHS (5) considered any severe subclinical disease at different anatomic sites to identify high-risk individuals. This indicator was not designed to capture low to modest levels of disease.
Lower levels of subclinical CVD increase the likelihood of maintaining intact health and function in the elderly population (6). Although most older adults will develop clinical or subclinical CVD by the age of 80 years, a minority remains remarkably disease free (7). Additionally, patterns of peripheral and coronary vascular disease are markedly heterogeneous (8). To better understand the full impact of subclinical CVD on health and function in late life, we investigated the risk for death and cardiovascular events associated with an indicator that resulted from a novel way of combining a number of noninvasive measures. This indicator was designed to reflect a gradient of subclinical CVD, ranging from absent to severe. We also investigated whether the associations with the outcomes were independent of traditional cardiovascular risk factors. As a secondary aim, we assessed if this indicator better described the spectrum of vascular disease, by capturing a wider range of risk, compared with the previous CHS indicator for any severe subclinical CVD (5).
METHODS
Population
The CHS (9) enrolled 5,888 men and women ≥65 years old from four U.S. communities, combining the original cohort enrolled in 1989–1990 and a minority cohort (13% of the total sample), included in 1992–1993. Exclusion criteria were being institutionalized, wheelchair-bound, or on treatment for cancer. Participants gave informed consent, and the study was approved by participating institutions’ institutional review boards. We selected 3,695 participants (450 from the minority cohort) without history of myocardial infarction, angina, coronary artery bypass surgery or percutaneous transluminal angioplasty, congestive heart failure, stroke or transient ischemic attack, carotid surgery, peripheral vascular bypass surgery, angioplasty, or intermittent claudication in 1992–1993.
Noninvasive Measures of Vascular Disease and Vascular Burden Index
To calculate the vascular burden index, we combined measures of ankle–arm blood pressure, electrocardiogram, and common carotid ultrasound, collected in 1992–1993. For each measure, a score of 0, 1, or 2 (no, mild, or severe abnormalities) was assigned based on clinical cutoffs for normal, intermediate, and high values (Table 1). The cutoff for the ankle–arm index (10) was based on mortality and cardiovascular events risk in the CHS cohort, and electrocardiogram abnormalities were previously defined in the same study (11). For common carotid artery intima-media thickness, the upper 20th percentile (1.19 mm, in our sample) was previously used in CHS to define risk (12). Consistently, we used the lower 20th percentile (0.89 mm) to differentiate low from moderate disease.
Table 1.
Cut-offs for the Noninvasive Measures of Vascular Disease and Attributed Scores Used to Obtain the Standard Three-Measures Vascular Burden Index (the first three upper measures) and the Extended Five-Measures Index (cut-offs choice rationales are detailed in the Methods section)
| Score | 0 | 1 | 2 |
| Ankle–arm index | (1, 1.4] | (0.9, 1.0] or >1.4 | ≤0.9 |
| Electrocardiogram | No abnormalities | Minor abnormalities* | Major abnormalities† |
| Common carotid intima-media thickness | Lower 20th percentile | 20th–80th percentile | Upper 20th percentile |
| AAA | No AAA | Infrarenal to suprarenal ratio ≥1.2 (men only) or 3 cm ≤ IRD < 3.5 cm | IRD ≥ 3.5 cm |
| Magnetic resonance imaging brain infarcts | No infarcts ≥3 mm | 1 infarct ≥3 mm | ≥2 infarcts ≥3 mm |
Notes: AAA = abdominal aortic aneurysms; IRD = infrarenal diameter. “(” indicates end point not included; “]” indicates end point is included.
Minor Q or QS waves, high R waves, minor isolated ST–T abnormalities, ST elevation, incomplete right bundle branch block, RR1, long QT interval, short PR, right axis deviation.
Ventricular conduction defects, major Q or QS abnormalities, minor Q or QS with ST–T wave abnormalities, left ventricular hypertrophy, isolated major ST–T wave changes, atrial fibrillation or first-degree atrioventricular block.
The index (range 0–6, absent to severe disease) summed the individual scores and could be calculated for 3,252 participants without a history of clinical CVD. The 443 participants excluded because of missing data, compared with the participants included, were older (M ± SD 76.9 years ± 6.1 vs 74.3 years ± 5.1 years), more likely women (68.8% vs 62.7%), and had slightly higher systolic blood pressure and lower low-density lipoprotein cholesterol (data not shown).
Additional Noninvasive Measures of Vascular Disease
Abdominal aortic ultrasound and brain magnetic resonance imaging, also available in 1992–1993, are generally more difficult to obtain in clinical and research settings, so we initially excluded them from the index. Their contribution to risk definition was explored in secondary analyses. Abdominal aortic ultrasound (13) cutoff was based on previous CHS events risk estimates. For brain magnetic resonance imaging, the presence of one or more infarcts ≥3 mm was associated with poor functional outcomes in the CHS (14), and in other studies predicted mortality in the elderly population (15). This extended five-measures index ranged from 0 to 10 points (absent to severe disease), and could be calculated for 2,393 participants.
Previous CHS Indicator for Any Severe Subclinical Disease
The previous indicator (5,12) detected the presence of any of the following in the 1992–1993 exam: major electrocardiogram abnormalities, increased common or internal carotid artery wall thickness (>80th percentile) or stenosis (>25%), low ankle–arm index (≤0.9), and positive Rose Questionnaire for angina or intermittent claudication.
Covariates
We considered demographics (age, gender, and race), baseline systolic and diastolic blood pressure (mm Hg), diabetes (fasting glucose ≥126 mg/dL or specific treatment), smoking history (never, vs former or current smoker), high- and low-density lipoprotein cholesterol cholesterol (mg/dL), and body mass index (kg/m2) (9).
Outcomes
All-cause mortality and incident nonfatal and fatal cardiovascular events (any myocardial infarction, angina, stroke, transient ischemic attack, congestive heart failure, peripheral arterial disease, or cardiovascular death) were identified, according to CHS ascertainment and adjudication methods (16,17), through June 2008.
Statistical Analysis
We used analysis of variance (continuous variables) and chi-square tests for trend (categorical variables) to test differences in baseline characteristics across index levels. After calculating crude event and death rates, we used Cox proportional hazards models to determine the association of the index with events and mortality. We tested for proportional hazards using Schoenfeld residuals and the assumption was not met, with p = .045 and p = .014 for cardiovascular events and mortality, respectively. Dividing the available follow-up time in half, at 8 years, allowed the proportional hazards assumption to be met within each half, resulting in different estimates of risk for early and late events. These models were first adjusted for demographics, then for cardiovascular risk factors. For each outcome, the statistical significance of the difference between the −2 log likelihood of two models including demographics and risk factors, one without and the other with the index, was tested against a chi-square distribution with 5 df (p < .05), to determine the improvement after the inclusion of the index. Because of the differences in cardiovascular outcomes often reported between sexes, we tested index–gender interaction and showed sex-stratified models. We also tested interactions between the index and age, race and cardiovascular risk factors. We finally repeated the analyses after including the abdominal ultrasound and the brain magnetic resonance imaging.
Secondary analyses.—
We assessed event and death rates for participants with and without any subclinical CVD according to the previous CHS indicator, and we plotted Kaplan–Meier curves for events-free and overall survival by our index and Kuller et al.’s one (5). Among participants with all noninvasive measures of subclinical disease, we compared the ability of the three subclinical indices (the previous indicator for any subclinical CVD, the three-measures index, and the extended five-measures index) to reclassify participants into risk categories using the net reclassification improvement (NRI) by Pencina and colleagues (18) The baseline risk categories were determined from a model using only the demographic and risk factor covariates for death or CVD within 8 years. Risk categories were quartiles of the distribution of predicted probabilities for each outcome and the NRI was calculated separately for those who did and did not experience the outcome of interest. The Hosmer–Lemeshow goodness-of-fit test was used to assess model calibration (19).
Analyses were performed using SPSS 14.0 (Chicago, IL) and STATA 11 (Stata Corp., College Station, TX).
RESULTS
Participants without history of clinical CVD and with data on the three basic vascular measures were 3,252 (M ± SD 74.3 years ± 5.1, 62.7% women, 17% African Americans). The composite index was skewed to low values in both sexes (Figure 1), with a median value of 2 and inter quartile range of 1–3. A higher proportion of men had severe disease. Age and the proportion of men and African Americans increased across index levels, as did different cardiovascular risk factors in men and women (Table 2). Event rates (Table 3) increased monotonically across index grades. Compared with absent subclinical CVD (index grade = 0) and adjusted for demographics, index grades >1 were associated with a significantly higher risk for both early and late events. Similar associations were observed for total mortality (Table 4). Adjustment for risk factors only slightly modified the results. When added to demographics and risk factors in Cox models, the index was significantly associated with both cardiovascular events and mortality (p value of the likelihood ratio test <.001 for both outcomes). None of the tested interactions (with age, race, or risk factors) were significant for either outcome. We found similar results in sex-stratified models. At each index grade, men had higher incidence rates than women (Tables 3 and 4).
Figure 1.
Distribution of the vascular burden index in participants without clinical cardiovascular disease.
Table 2.
Baseline Characteristics of the Sample
| Sex | 0 (n = 279) | 1 (n = 889) | 2 (n = 873) | 3 (n = 732) | 4 (n = 311) | 5+ (n = 168) | p Value | |
| Female, n (%) | 201 (72.0) | 590 (66.4) | 562 (64.4) | 430 (58.7) | 177 (56.9) | 78 (46.4) | <.001 | |
| Age, M ± SEM | Women | 72.2 ± 3.4 | 73.3 ± 4.3 | 74.1 ± 4.9 | 75.2 ± 5.4 | 75.6 ± 5.8 | 77.0 ± 5.5 | <.001 |
| Men | 72.7 ± 4.0 | 73.6 ± 4.5 | 74.3 ± 5.0 | 74.8 ± 5.4 | 76.2 ± 5.9 | 78.1 ± 6.3 | <.001 | |
| African American, n (%) | Women | 13 (6.5) | 61 (10.3) | 95 (16.9) | 100 (23.3) | 55 (31.1) | 27 (34.6) | <.001 |
| Men | 5 (6.4) | 37 (12.4) | 56 (18.0) | 50 (16.6) | 23 (17.2) | 33 (36.7) | <.001 | |
| Diabetes, n (%) | Women | 10 (5.1) | 39 (6.9) | 62 (11.4) | 65 (15.5) | 27 (16.1) | 21 (27.6) | <.001 |
| Men | 3 (4.0) | 32 (10.9) | 44 (14.3) | 57 (19.5) | 27 (20.8) | 21 (24.7) | <.001 | |
| Ever smoked, n (%) | Women | 83 (42.13) | 231 (40.2) | 222 (39.9) | 167 (39.7) | 92 (53.2) | 38 (49.4) | .044 |
| Men | 54 (71.1) | 198 (67.6) | 211 (68.5) | 201 (66.8) | 102 (77.3) | 63 (70.8) | .399 | |
| BMI, M ± SEM | Women | 26.0 ± 4.7 | 26.7 ± 4.9 | 27.3 ± 5.1 | 27.5 ± 5.7 | 27.1 ± 5.4 | 26.2 ± 5.1 | .004 |
| Men | 25.2 ± 3.1 | 26.3 ± 3.6 | 26.5 ± 3.8 | 27.0 ± 3.4 | 27.5 ± 4.3 | 25.7 ± 3.8 | <.001 | |
| SBP (mm Hg), M ± SEM | Women | 129.0 ± 17.1 | 131.7 ± 18.0 | 136.0 ± 21.3 | 139.6 ± 21.0 | 146.2 ± 23.5 | 151.3 ± 21.7 | <.001 |
| Men | 130.2 ± 16.5 | 130.9 ± 17.8 | 135.7 ± 20.7 | 137.4 ± 20.7 | 140.9 ± 24.9 | 145.0 ± 24.6 | <.001 | |
| DBP (mm Hg), M ± SEM | Women | 70.7 ± 10.3 | 70.3 ± 9.7 | 70.7 ± 11.1 | 71.1 ± 12.4 | 71.6 ± 11.8 | 72.9 ± 12.9 | .351 |
| Men | 73.9 ± 10.1 | 73.1 ± 10.7 | 74.610.6 | 74.7 ± 10.8 | 73.6 ± 11.9 | 72.6 ± 15.4 | .322 | |
| HDL, M ± SEM | Women | 60.6 ± 14.7 | 60.2 ± 15.0 | 57.7 ± 13.7 | 57.6 ± 15.5 | 56.3 ± 14.1 | 58.0 ± 13.6 | .001 |
| Men | 49.4 ± 10.8 | 48.0 ± 11.4 | 48.6 ± 11.9 | 48.2 ± 11.7 | 48.3 ± 12.7 | 46.4 ± 10.8 | .798 | |
| LDL, M ± SEM | Women | 127.9 ± 31.1 | 129.2 ± 31.7 | 130.3 ± 32.6 | 131.1 ± 35.1 | 138.1 ± 37.4 | 141.2 ± 37.5 | .001 |
| Men | 125.4 ± 32.3 | 119.5 ± 30.8 | 122.1 ± 31.2 | 121.0 ± 32.6 | 124.1 ± 32.0 | 114.9 ± 32.8 | .306 |
Notes: p value for linear trend. BMI = body mass index; DBP = diastolic blood pressure; HDL = high-density lipoprotein; LDL = low-density lipoprotein; SBP = systolic blood pressure.
Table 3.
Risk of Incident Early and Late Cardiovascular Events in the Whole Sample and by Sex
| Incident Fatal or Nonfatal Cardiovascular Disease Events | |||||||
| Index Grade | n | n Events | Rate/100 Person-y | Model 1 |
Model 2 |
||
| Hazard Ratio (95% confidence interval) (<8 y) | Hazard Ratio (95% confidence interval) (>8 y) | Hazard Ratio (95% confidence interval) (<8 y) | Hazard Ratio (95% confidence interval) (>8 y) | ||||
| Whole sample | |||||||
| 0 | 279 | 103 | 3.1 | 1.0 (referent) | 1.0 (referent) | 1.0 (referent) | 1.0 (referent) |
| 1 | 889 | 365 | 3.9 | 1.1 (0.8–1.6) | 1.3 (1.0–1.8) | 1.1 (0.8–1.6) | 1.2 (0.9–1.7) |
| 2 | 873 | 450 | 5.5 | 1.8 (1.4–2.5) | 1.5 (1.1–2.1) | 1.8 (1.3–2.5) | 1.5 (1.1–2.0) |
| 3 | 732 | 412 | 6.5 | 2.2 (1.6–3.0) | 1.6 (1.2–2.2) | 2.0 (1.5–2.8) | 1.4 (1.0–2.0) |
| 4 | 311 | 197 | 8.3 | 2.6 (1.9–3.6) | 2.2 (1.6–3.2) | 2.2 (1.6–3.2) | 1.8 (1.2–2.7) |
| 5+ | 168 | 127 | 15.6 | 4.7 (3.4–6.9) | 3.4 (2.1–5.8) | 3.9 (2.7–5.7) | 3.2 (1.9–5.4) |
| Test for trend p < .001 | Test for trend p < .001 | ||||||
| Women | |||||||
| 0 | 201 | 72 | 3.0 | 1.0 (referent) | 1.0 (referent) | 1.0 (referent) | 1.0 (referent) |
| 1 | 590 | 237 | 3.7 | 1.0 (0.7–1.5) | 1.3 (0.9–2.0) | 1.1 (0.7–1.6) | 1.4 (0.9–2.0) |
| 2 | 562 | 281 | 5.1 | 1.6 (1.1–2.2) | 1.6 (1.1–2.3) | 1.6 (1.1–2.4) | 1.7 (1.1–2.5) |
| 3 | 430 | 235 | 6.0 | 1.8 (1.2–2.6) | 1.7 (1.2–2.6) | 1.8 (1.2–2.6) | 1.6 (1.1–2.5) |
| 4 | 177 | 105 | 7.3 | 2.0 (1.4–3.0) | 2.2 (1.4–3.6) | 1.9 (1.2–3.0) | 2.1 (1.3–3.5) |
| 5+ | 78 | 57 | 13.5 | 4.0 (2.6–6.2) | 3.2 (1.6–6.3) | 3.4 (2.1–5.5) | 3.4 (1.7–6.9) |
| Test for trend p < .001 | Test for trend p < .001 | ||||||
| Men | |||||||
| 0 | 78 | 31 | 3.4 | 1.0 (referent) | 1.0 (referent) | 1.0 (referent) | 1.0 (referent) |
| 1 | 299 | 128 | 4.4 | 1.5 (0.8–2.7) | 1.2 (0.7–2.0) | 1.3 (0.7–2.4) | 1.0 (0.6–1.7) |
| 2 | 311 | 169 | 6.3 | 2.6 (1.4–4.8) | 1.4 (0.8–2.3) | 2.3 (1.2–4.1) | 1.2 (0.7–2.0) |
| 3 | 302 | 177 | 7.4 | 3.3 (1.8–6.0) | 1.4 (0.8–2.3) | 2.7 (1.5–4.9) | 1.1 (0.7–1.9) |
| 4 | 134 | 92 | 9.9 | 4.0 (2.2–7.5) | 2.2 (1.2–4.0) | 3.0 (1.6–5.7) | 1.7 (0.8–2.9) |
| 5+ | 90 | 70 | 17.8 | 7.5 (4.0–14.1) | 3.6 (1.7–7.9) | 5.2 (2.8–10.0) | 2.9 (1.3–6.1) |
| Test for trend p < .001 | Test for trend p < .001 | ||||||
Note: Model 1: adjusted for age (agey5), sex (gend01), and race (race01); Model 2: Model 1 + cardiovascular risk factors (body mass index, systolic and diastolic blood pressure, high-density lipoprotein and low-density lipoprotein cholesterol, diabetes, and smoking).
Table 4.
Total Early and Late Mortality in the Whole Sample and by Sex
| Total Mortality | |||||||
| Index Grade | n | n Events | Rate/100 Person-y | Model 1 |
Model 2 |
||
| Hazard Ratio (95% confidence interval) (<8 y) | Hazard Ratio (95% confidence interval) (>8 y) | Hazard Ratio (95% confidence interval) (<8 y) | Hazard Ratio (95% confidence interval) (>8 y) | ||||
| Whole Sample | |||||||
| 0 | 279 | 115 | 3.1 | 1.0 (referent) | 1.0 (referent) | 1.0 (referent) | 1.0 (referent) |
| 1 | 889 | 440 | 4.0 | 1.5 (1.0–2.3) | 1.1 (0.9–1.4) | 1.6 (1.1–2.5) | 1.1 (0.8–1.4) |
| 2 | 873 | 512 | 5.0 | 1.9 (1.3–2.8) | 1.3 (1.0–1.7) | 2.0 (1.3–3.0) | 1.4 (1.1–1.7) |
| 3 | 732 | 473 | 5.8 | 2.2 (1.5–3.2) | 1.4 (1.1–1.8) | 2.3 (1.5–3.5) | 1.3 (1.0–1.7) |
| 4 | 311 | 241 | 7.8 | 2.6 (1.7–4.0) | 2.0 (1.6–2.7) | 2.6 (1.7–4.1) | 1.9 (1.4–2.6) |
| 5+ | 168 | 147 | 11.7 | 5.0 (3.3–7.7) | 1.9 (1.3–2.7) | 4.8 (3.0–7.5) | 1.8 (1.2–2.6) |
| Test for trend p < .001 | Test for trend p < .001 | ||||||
| Women | |||||||
| 0 | 201 | 75 | 2.7 | 1.0 (referent) | 1.0 (referent) | 1.0 (referent) | 1.0 (referent) |
| 1 | 590 | 269 | 3.6 | 1.7 (1.0–2.8) | 1.1 (0.8–1.5) | 2.0 (1.1–3.5) | 1.1 (0.8–1.5) |
| 2 | 561 | 303 | 4.5 | 1.9 (1.2–3.2) | 1.4 (1.0–1.8) | 2.3 (1.3–4.1) | 1.5 (1.1–2.0) |
| 3 | 430 | 268 | 5.4 | 2.3 (1.3–3.8) | 1.5 (1.1–2.0) | 2.6 (1.5–4.7) | 1.4 (1.0–1.9) |
| 4 | 177 | 131 | 7.1 | 2.5 (1.4–4.4) | 2.4 (1.7–3.4) | 2.7 (1.5–5.1) | 2.3 (1.6–3.3) |
| 5+ | 78 | 66 | 9.9 | 4.9 (2.7–8.7) | 1.9 (1.2–3.0) | 4.7 (2.5–8.9) | 1.8 (1.1–2.9) |
| Test for trend p < .001 | Test for trend p < .001 | ||||||
| Men | |||||||
| 0 | 78 | 40 | 4.0 | 1.0 (referent) | 1.0 (referent) | 1.0 (referent) | 1.0 (referent) |
| 1 | 299 | 171 | 4.8 | 1.4 (0.7–2.6) | 1.1 (0.7–1.6) | 1.2 (0.6–2.3) | 1.0 (0.7–1.6) |
| 2 | 311 | 209 | 6.0 | 1.7 (0.9–3.2) | 1.3 (0.8–1.9) | 1.5 (0.8–2.8) | 1.2 (0.8–1.8) |
| 3 | 302 | 205 | 6.3 | 2.0 (1.1–3.7) | 1.2 (0.8–1.9) | 1.8 (0.9–3.3) | 1.2 (0.8–1.9) |
| 4 | 134 | 110 | 8.7 | 2.7 (1.4–5.1) | 1.6 (1.0–2.6) | 2.2 (1.2–4.3) | 1.5 (0.9–2.5) |
| 5+ | 90 | 81 | 13.8 | 4.9 (2.5–9.3) | 1.8 (1.0–3.3) | 4.2 (2.2–8.1) | 1.6 (0.9–3.0) |
| Test for trend p < .001 | Test for trend p < .001 | ||||||
Note: Model 1: adjusted for age (agey5), sex (gend01), and race (race01); Model 2: Model 1 + cardiovascular risk factors (body mass index, systolic and diastolic blood pressure, high-density lipoprotein and low-density lipoprotein cholesterol, diabetes, smoking).
After incorporating abdominal ultrasound and brain magnetic resonance imaging, the index could be recalculated for 2,393 participants (M ± SD 74.3 years ± 4.9, 62.1% women, 15.3% African Americans). From index grade 1 to 6+, compared with grade 0, cardiovascular event rates ranged from 2.6 to 16.3 per 100 person-years and hazard ratios from 1.6 (95% confidence interval 1.1–2.3) to 7.8 (5.1–12.1) for the first 8 years after adjustment for demographics. We observed similar trends for mortality.
Secondary Analyses
According to the previous CHS indicator for any subclinical disease (yes/no), 1,202 participants (37%) were free and 2,048 (63%) had subclinical vascular disease. Events rates per 100 person-years were 3.9 in participants without and 6.5 in participants with vascular disease, and deaths rates per 100 person-years 3.8 and 6.0, respectively. Vascular disease was associated with an increased risk for both CVD events (hazard ratio 1.6, 95% confidence interval 1.4–1.9) and mortality (hazard ratio 1.4, 95% confidence interval 1.3–1.5) for the first 8 years, after adjustment for demographics and risk factors.
In unadjusted models (Figure 2), low grades of subclinical disease with the new index were associated with longer CVD-free and overall survival, and participants with high grades with the new index showed reduced CVD-free and overall survival over time, compared with the previous indicator.
Figure 2.
Unadjusted Kaplan–Meier curves for survival free from cardiovascular events (A) and overall survival (B), by subclinical vascular disease burden according to the new index and the previous Cardiovascular Health Study indicator for any severe subclinical cardiovascular disease (5).
The NRI allowed testing if the new indicator provided a better classification of the participants into risk categories for both outcomes, compared with demographics, risk factors, and the previous CHS indicator. The NRI was calculated separately for those who did and did not experience the outcome of interest. Using a more reader-friendly illustration, we expected that the new index would classify a higher proportion of participants who did experience the outcome into higher risk categories, and a higher proportion of participants who did not experience the outcome into lower risk categories. Quartiles of the predicted probability of death within 8 years for this older cohort were <11%, 11%–17%, 17%–28%, and >28%. Adding the previous indicator for any subclinical disease to a model including risk factors and demographics did not improve risk classification for mortality. Tables 5 and 6 show the NRI models comparing the three-measures index and the previous CHS indicator for both outcomes. The indices based on three (Table 5) or five measures improved risk classification over the indicator variable in the subgroup who did not die (p = .006 for three measures and p < .001 for five measures), and the index based on five measures showed an improvement over the three-measures index in the same subgroup (p = .001). There were no improvements in risk classification for the subgroup of participants who died within 8 years (Table 5). For incident CVD within 8 years, quartile cut-points of the predicted probabilities were 23%, 29%, and 40%. Adding the indicator for any subclinical disease improved reclassification in the group with events (p = .002). The three-measures index did not improve classification over the previous indicator (Table 6), but the five-measures index did (p = .014). The five-measures index improved classification over the three-measures index among only those without a CVD event (p = .010). All models were well calibrated, with Hosmer–Lemeshow goodness-of-fit test p values >.50.
Table 5.
Prediction of 8-Year Mortality According to the New Index for Subclinical Vascular Disease, Compared With the Previous Cardiovascular Health Study Indicator for Any Subclinical Disease
| Predicted Probability (quartiles) | New Subclinical Vascular Burden Index |
Total | ||||
| <11% | 11%–17% | 17%–28% | >28% | |||
| Participants who survived (p = .006) | ||||||
| Previous indicator for any subclinical vascular disease | <11% | 403 (81.3) | 91 (18.4) | 2 (0.4) | 0 | 496 (100) |
| 11%–17% | 105 (23.3) | 277 (61.4) | 66 (14.6) | 3 (0.7) | 451 (100) | |
| 17%–28% | 13 (2.8) | 79 (17.3) | 318 (69.4) | 48 (10.5) | 458 (100) | |
| >28% | 0 | 4 (1.3) | 69 (21.6) | 247 (77.2) | 320 (100) | |
| Total | 521 (30.2) | 451 (26.1) | 455 (26.4) | 298 (17.3) | 1,725 (100) | |
| Participants who died (p = NS) | ||||||
| Previous indicator for any subclinical vascular disease | <11% | 29 (72.5) | 11 (27.5) | 0 | 0 | 40 (100) |
| 11%–17% | 16 (20.0) | 50 (62.5) | 14 (17.5) | 0 | 80 (100) | |
| 17%–28% | 4 (3.6) | 14 (12.6) | 73 (65.8) | 20 (18.0) | 111 (100) | |
| >28% | 0 | 0 | 24 (9.1) | 239 (90.9) | 263 (100) | |
| Total | 49 (9.9) | 75 (15.2) | 111 (22.5) | 259 (52.4) | 494 (100) | |
Notes: NS = nonsignificant. Values in the cells are expressed as n (%). Values in italics: participants who were reclassified toward a lower risk using the new subclinical vascular index, compared with the previous indicator for any subclinical vascular disease. Values in bold: participants who were reclassified toward a higher risk.
Table 6.
Prediction of 8-Year Cardiovascular Events According to the New Index for Subclinical Vascular Disease, Compared With the Previous Cardiovascular Health Study Indicator for Any Subclinical Disease
| Predicted Probability (quartiles) | New Subclinical Vascular Burden Index |
Total | ||||
| <23% | 23%–29% | 29%–40% | >40% | |||
| Participants who did not experience CVD events (p = NS) | ||||||
| Previous indicator for any subclinical vascular disease | <23% | 337 (80.0) | 67 (15.9) | 17 (4.0) | 1(0.2) | 422 (100) |
| 23%–29% | 64 (29.8) | 98 (45.6) | 46 (21.4) | 7 (3.3) | 215 (100) | |
| 29%–40% | 39 (15.5) | 31 (12.4) | 147 (58.6) | 34 (13.6) | 251 (100) | |
| >40% | 3 (1.5) | 15 (7.3) | 29 (14.2) | 158 (77.1) | 205 (100) | |
| Total | 443 (40.5) | 211 (19.3) | 239 (21.9) | 200 (18.3) | 1,093(100) | |
| Participants who experienced CVD events (p = NS) | ||||||
| Previous indicator for any subclinical vascular disease | <23% | 180 (73.5) | 45 (18.4) | 19 (7.8) | 1 (0.4) | 245 (100) |
| 23%–29% | 50 (29.2) | 65 (38.0) | 52 (30.4) | 4 (2.3) | 171 (100) | |
| 29%–40% | 34 (11.9) | 26 (9.1) | 158 (55.1) | 69 (24.0) | 287 (100) | |
| >40% | 2 (0.5) | 15 (3.6) | 61 (14.4) | 345 (81.6) | 423 (100) | |
| Total | 266 (23.6) | 151 (13.4) | 290 (25.8) | 419 (37.2) | 1,126 (100) | |
Notes: CVD = cardiovascular disease; NS = nonsignificant. Values in the cells are expressed as n (%). Values in italics: participants who were reclassified toward a lower risk using the new subclinical vascular index, compared with the previous indicator for any subclinical vascular disease. Values in bold: participants who were reclassified toward a higher risk.
DISCUSSION
This study illustrates not only the range of subclinical vascular disease in older adults without clinical CVD but also the strong association between this index of subclinical disease and the risk for clinical CVD and total mortality: we observed a “dose–response” effect across levels of subclinical CVD, independently of demographics and conventional cardiovascular risk factors. Mild to moderate disease was highly prevalent. For each index grade, men had higher event and death rates than women. The very low rate of events and mortality in participants without subclinical disease identify them as an exceptional group, with potentially unique characteristics in term of life histories or genetic profiles.
In CHS, noninvasive subclinical CVD measures were individually associated with higher event relative risks (1–3). When entered simultaneously in a unique model, all independently contributed to the prediction (4). An indicator for any severe subclinical CVD, previously developed in CHS, predicted high CVD risk (5). Compared with the new indicator, the previous indicator did not discriminate between mild and moderate disease. Our approach provides more discrimination from very low to extremely high risks, and illustrates the greater accuracy of combining measures to predict risk. In particular, the new index was able to improve the predictive validity in participants who did not die during the 10 years follow-up. These findings suggest that the new index could have a better chance of capturing the low end of the subclinical CVD spectrum. Noteworthy, low levels of subclinical CVD are associated with the ability of maintaining intact health and function in the elderly population (6).
Adding two other measures of subclinical CVD further increased the ability of discriminating among levels of disease and of predicting mortality and incident CVD. However, the simple combination of ankle–arm index, electrocardiogram, and carotid intima-media thickness, easy and relatively nonexpensive, already provides a well description of the spectrum.
Our composite index may be useful for more fully accounting for vascular burden in clinical studies by capturing the full spectrum of cardiovascular comorbidity (20). Although the term “subclinical” is commonly used to describe vascular disease in absence of clinical manifestations, individual subclinical CVD measures have been associated with functional abnormalities in multiple domains (21,22). Our index could be translated to other aspects of aging research, such as the evaluation of risk factors for progressive and chronic declines in specific systems’ performance and in global functioning, and the assessment of factors associated with exceptionally healthy aging.
While reaffirming the importance of combining organ disease markers for a more detailed estimation of cardiovascular risk, our findings, showing a high prevalence of subclinical CVD in this cohort and a strong association with cardiovascular morbidity and mortality, remark the need of maximizing pharmacological and nonpharmacological preventive efforts across the spectrum of CVD, and illustrate the potential for aging with minimal vascular disease.
Some limitations have to be acknowledged. The relatively few participants with incomplete vascular assessments were older and, in general, had a higher risk factor burden. Also, the study included only a small proportion of nonwhite participants, who might have more risk factors and subclinical disease, so that these findings might not be generalized to nonwhite populations, requiring confirmation by further evidence.
Strengths include the large population-based sample, the number of noninvasive tests to measure CVD, the standardized approach to define prevalent and incident events and death, and the ability to control for risk factors.
In summary, older adults with minimal subclinical CVD had exceptional disease-free survival and low mortality, compared with people with moderate to severe disease. The considered noninvasive measures of vascular disease have relatively low cost and are diffused in clinical practice and research, so that this index may be useful to characterize the spectrum of subclinical CVD and to estimate risks. This index could also be used for more fully accounting for vascular burden in clinical studies: Its ability to cover the spectrum of vascular disease should be used to elucidate possible pathways linking age-related vascular disease to other aspects of aging and progressive disability.
FUNDING
M.I. was a research scholar at the Pepper Older Americans Independence Center of the University of Pittsburgh (P30 AG024827) and was supported by an educational grant from the “Gianandrea Pugi” Foundation (Florence, Italy) and from the CHS All Stars Study (R-01 AG-023629). This study was also supported by contracts N01-HC-35129, N01-HC-45133, N01-HC-75150, N01-HC-85079 through N01-HC-85086, N01 HC-15103, N01 HC-55222, and U01 HL080295 from the National Heart, Lung, and Blood Institute, with additional contribution from the National Institute of Neurological Disorders and Stroke. A full list of participating investigators and institutions in the CHS is at http://chs3.chs.biostat.washington.edu/chs/. This research was supported in part by the Intramural Research Program of the National Institute on Aging, NIH (K.V.P.). The “Oristano Program for Cardiovascular Disease Research” supported part of this work. The National Heart, Lung, and Blood Institute approved the manuscript before submission. All the other sponsors had no role in the collection, analysis, or interpretation of the data or in the decision to submit the study for publication.
CONFLICT OF INTEREST
None of the authors report conflicts of interest.
References
- 1.Newman AB, Shemanski L, Manolio TA, et al. Ankle-arm index as a predictor of cardiovascular disease and mortality in the Cardiovascular Health Study. The Cardiovascular Health Study Group. Arterioscler Thromb Vasc Biol. 1999;19:538–545. doi: 10.1161/01.atv.19.3.538. [DOI] [PubMed] [Google Scholar]
- 2.O’Leary DH, Polak JF, Kronmal RA, Manolio TA, Burke GL, Wolfson SK., Jr. Carotid-artery intima and media thickness as a risk factor for myocardial infarction and stroke in older adults. Cardiovascular Health Study Collaborative Research Group. N Engl J Med. 1999;340:14–22. doi: 10.1056/NEJM199901073400103. [DOI] [PubMed] [Google Scholar]
- 3.Newman AB, Naydeck BL, Ives DG, et al. Coronary artery calcium, carotid artery wall thickness, and cardiovascular disease outcomes in adults 70 to 99 years old. Am J Cardiol. 2008;101:186–192. doi: 10.1016/j.amjcard.2007.07.075. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Psaty BM, Furberg CD, Kuller LH, et al. Traditional risk factors and subclinical disease measures as predictors of first myocardial infarction in older adults: the Cardiovascular Health Study. Arch Intern Med. 1999;159:1339–1347. doi: 10.1001/archinte.159.12.1339. [DOI] [PubMed] [Google Scholar]
- 5.Kuller LH, Shemanski L, Psaty BM, et al. Subclinical disease as an independent risk factor for cardiovascular disease. Circulation. 1995;92:720–726. doi: 10.1161/01.cir.92.4.720. [DOI] [PubMed] [Google Scholar]
- 6.Newman AB, Arnold AM, Naydeck BL, et al. Cardiovascular Health Study Research Group. “Successful aging”: effect of subclinical cardiovascular disease. Arch Intern Med. 2003;163:2315–2322. doi: 10.1001/archinte.163.19.2315. [DOI] [PubMed] [Google Scholar]
- 7.Newman AB, Naydeck BL, Sutton-Tyrrell K, Feldman A, Edmundowicz D, Kuller LH. Coronary artery calcification in older adults to age 99: prevalence and risk factors. Circulation. 2001;104:2679–2684. doi: 10.1161/hc4601.099464. [DOI] [PubMed] [Google Scholar]
- 8.Newman AB, Naydeck BL, Sutton-Tyrrell K, et al. Relationship between coronary artery calcification and other measures of subclinical cardiovascular disease in older adults. Arterioscler Thromb Vasc Biol. 2002;22:1674–1679. doi: 10.1161/01.atv.0000033540.89672.24. [DOI] [PubMed] [Google Scholar]
- 9.Fried LP, Borhani NO, Enright P, et al. The Cardiovascular Health Study: design and rationale. Ann Epidemiol. 1991;1:263–276. doi: 10.1016/1047-2797(91)90005-w. [DOI] [PubMed] [Google Scholar]
- 10.O’Hare AM, Katz R, Shlipak MG, Cushman M, Newman AB. Mortality and cardiovascular risk across the ankle-arm index spectrum: results from the Cardiovascular Health Study. Circulation. 2006;113:388–393. doi: 10.1161/CIRCULATIONAHA.105.570903. [DOI] [PubMed] [Google Scholar]
- 11.Furberg CD, Manolio TA, Psaty BM, et al. Major electrocardiographic abnormalities in persons aged 65 years and older (the Cardiovascular Health Study). Cardiovascular Health Study Collaborative Research Group. Am J Cardiol. 1992;69:1329–1335. doi: 10.1016/0002-9149(92)91231-r. [DOI] [PubMed] [Google Scholar]
- 12.Kuller L, Borhani N, Furberg C, et al. Prevalence of subclinical atherosclerosis and cardiovascular disease and association with risk factors in the Cardiovascular Health Study. Am J Epidemiol. 1994;139:1164–1179. doi: 10.1093/oxfordjournals.aje.a116963. [DOI] [PubMed] [Google Scholar]
- 13.Newman AB, Arnold AM, Burke GL, O’Leary DH, Manolio TA. Cardiovascular disease and mortality in older adults with small abdominal aortic aneurysms detected by ultrasonography: the Cardiovascular Health Study. Ann Intern Med. 2001;134:182–190. doi: 10.7326/0003-4819-134-3-200102060-00008. [DOI] [PubMed] [Google Scholar]
- 14.Longstreth WT, Jr., Dulberg C, Manolio TA, et al. Incidence, manifestations, and predictors of brain infarcts defined by serial cranial magnetic resonance imaging in the elderly: the Cardiovascular Health Study. Stroke. 2002;33:2376–2382. doi: 10.1161/01.str.0000032241.58727.49. [DOI] [PubMed] [Google Scholar]
- 15.Liebetrau M, Steen B, Hamann GF, Skoog I. Silent and symptomatic infarcts on cranial computerized tomography in relation to dementia and mortality: a population-based study in 85-year-old subjects. Stroke. 2004;35:1816–1820. doi: 10.1161/01.STR.0000131928.47478.44. [DOI] [PubMed] [Google Scholar]
- 16.Psaty BM, Kuller LH, Bild D, et al. Methods of assessing prevalent cardiovascular disease in the Cardiovascular Health Study. Ann Epidemiol. 1995;5:270–277. doi: 10.1016/1047-2797(94)00092-8. [DOI] [PubMed] [Google Scholar]
- 17.Ives DG, Fitzpatrick AL, et al. Surveillance and ascertainment of cardiovascular events. The Cardiovascular Health Study. Ann Epidemiol. 1995;5:278–285. doi: 10.1016/1047-2797(94)00093-9. [DOI] [PubMed] [Google Scholar]
- 18.Pencina MJ, D’Agostino RB, Vasan RS. Evaluating the added predictive ability of a new marker: from area under the ROC curve to reclassification and beyond. Stat Med. 2008;27:157–172. doi: 10.1002/sim.2929. [DOI] [PubMed] [Google Scholar]
- 19.Hosmer DW, Lemeshow S. Applied Logistic Regression. 2nd ed. New York, NY: Wiley; 2000. [Google Scholar]
- 20.Karlamangla A, Tinetti M, Guralnik J, Studenski S, Wetle T, Reuben D. Comorbidity in older adults: nosology of impairment, diseases, and conditions. J Gerontol A Biol Sci Med Sci. 2007;62:296–300. doi: 10.1093/gerona/62.3.296. [DOI] [PubMed] [Google Scholar]
- 21.Elbaz A, Ripert M, Tavernier B, et al. Common carotid artery intima-media thickness, carotid plaques, and walking speed. Stroke. 2005;36:2198–2202. doi: 10.1161/01.STR.0000181752.16915.5c. [DOI] [PubMed] [Google Scholar]
- 22.McDermott MM, Liu K, Greenland P, et al. Functional decline in peripheral arterial disease: associations with the ankle brachial index and leg symptoms. JAMA. 2004;292:453–461. doi: 10.1001/jama.292.4.453. [DOI] [PubMed] [Google Scholar]