Abstract
Background
Chronic inflammation, measured by interleukin-6, predicts incident disability among elderly people. However, little is known about the relation of C-reactive protein (CRP) to disability.
Method
Participants (>60 years old, N = 1680) were from the National Health and Nutrition Examination Survey 1999–2002. Disability in activities of daily living (ADL), instrumental activities of daily living (IADL), leisure and social activities (LSA), lower extremity mobility (LEM), and general physical activities (GPA) was obtained by self-report. Peak muscle power was the product of isokinetic peak leg torque and peak force velocity. Functional limitations were evaluated via habitual walking speed, which was obtained from a 20-foot timed walk. CRP levels were quantified by using latex-enhanced nephelometry.
Results
Elevated CRP levels were associated with disability in IADL, LSA, LEM, and GPA, independent of basic demographics, chronic medical diseases, health behaviors, as well as nutritional markers. The corresponding odds ratios of disability for each standard-deviation increase in natural-log-transformed CRP were 1.18 (95% confidence interval [CI], 1.02–1.35), 1.18 (95% CI, 1.00–1.39), 1.17 (95% CI, 1.03–1.33), and 1.17 (95% CI, 1.05–1.31), respectively. The relationship diminished after additional adjustment of leg power and/or walking speed, meaning that impairment in leg power and limitations in gait speed likely mediate the association between CRP and disability. CRP had an inverse relationship to leg power and walking speed. Likewise, additional adjustment for leg power substantially diminished the association between CRP and walking speed, suggesting a mediating effect of power between CRP and gait speed.
Conclusions
Independent of chronic diseases, elevated CRP is associated with multiple domains of disability through mediation of muscle power, habitual gait speed, or both. Future research is needed to understand CRP as a risk factor for disability in older populations.
C-Reactive protein (CRP), composed of five 23-kd subunits, is largely regulated by cytokines, especially interleukin-6 (1). CRP, an acute phase reactant and a marker of chronic inflammation, has been widely used for the evaluation of various inflammatory states (2). CRP is present mainly in serum and is also detectable on the surface of about 4% of normal peripheral blood lymphocytes (3). Acute phase reactant CRP is produced in the liver; CRP detectable on lymphocytes is produced by those cells (4). Chronic inflammation causes an accelerated protein catabolism and may thus relate to muscle wasting and sarcopenia (5,6). Reduction in muscle strength and mass are important consequences of muscle wasting and sarcopenia and are associated with the development of functional decline and disability (7,8). Ferrucci and colleagues (9) suggested that higher circulating level of inflammatory marker, specifically interleukin-6, was predictive of the development of disability in activities of daily living as well as mobility disability in older persons. They further reported that decline in muscle strength mediated the causal relationship from high interleukin-6 to development of new disability (10). Although CRP is also an important marker of inflammation, data examining the relation of CRP to disability, although existent, are relatively sparse (11).
The notion that peak muscle power, the product of force and velocity, may play an important role in functional independence in elderly people has recently attracted significant research interest. It has been suggested that muscle power may be more directly related to impaired physical performance than to strength in elderly persons (12,13). Moreover, power declines to an even greater degree than strength (14,15), suggesting that power is a major attribute in age-related functional decline. However, the roles of muscle power as well as walking function in the association between CRP and disability are essentially unknown and have not been investigated. Therefore, the aim of this cross-sectional study was to test the hypotheses that higher circulating levels of CRP are associated with disability in older adults and that muscle power and gait speed may mediate the association between elevated CRP and disability. We sought to test the hypotheses by analyzing data from the National Health and Nutrition Examination Survey (NHANES) 1999–2002.
Methods
Study Design and Population
The NHANES, a population-based survey designed to collect information on the health and nutrition of the U.S. population, used a stratified, multistage, and cluster sampling design to obtain a representative sample of the noninstitutionalized civilian U.S. population. The NHANES consists of a detailed home interview and a health examination conducted in a mobile examination center (MEC). Beginning in 1999, the NHANES became a continuous, annual survey rather than the periodic survey that it had been in the past. Detailed Survey Operations Manuals, Consent Documents, and Brochures of the NHANES 1999–2002 are available on the NHANES Web site (16,17).
The Physical Functioning Questionnaire was administered to 3703 individuals aged 60 years or older to assess the individual's level of difficulty in performing various tasks without using any special equipment. Of these, 3232 attended the MEC for examination, which included an assessment of the right isokinetic quadriceps muscle strength and a 20-foot timed walk test. Of the 3232 persons who attended the MEC, 457 were excluded from muscle strength examination because of the following safety reasons: chest or abdominal surgery in the past 3 weeks [20]; heart attack in the past 6 weeks [10]; brain aneurysm or stroke [173]; current neck or back pain [90]; difficulty in bending or straightening right knee [76]; or right knee or right hip replacement [88]. We further excluded 672 persons with missing data in muscle strength test and/or timed walk test because of person's refusal, limited time to do the examinations, person's coming late or leaving early, examinations being interrupted, equipment or data capture failure, technician/software/supply error, communication problems, or other reasons. Compared to persons with missing data, those with nonmissing values tended to be younger (70.6 years vs 73.4 years) and non-Hispanic white (58% vs 45.5%).
The NHANES isokinetic muscle testing was measured at a fixed angular velocity of 60 degrees/s. Participants (n = 350) with peak force velocity (PFVel) which varied >5 degrees/s from the chosen testing velocity were further excluded, leaving 1753 participants with reliable measures of knee extensor peak torque. Compared to participants with safety exclusion (457 participants), those excluded because of administrative reasons, communications, equipment error, or other reasons (672 participants), or those with inappropriate PFVel on muscle strength test (350 participants), those with reliable measures of muscle strength (1753 participants) tended to be younger (mean age 70.2 years vs 73.3 years, 73.4 years, and 72.4 years, respectively; p < .001) and less disabled (p < .001) in all disability domains (see below), and fewer were women (46.2% vs 48.8%, 54.7%, and 65.7%, respectively; p < .001). Of the 1753 participants, 73 were further excluded because of missing values in CRP, leaving 1680 participants as the final analytic sample.
Disability
Participants aged 60 years or older were asked 19 questions designed to measure their functional status. These questions were phrased to assess the individual's level of difficulty in performing the task without using any special equipment. The 19 questions, detailed in Table 1, were classified into five major domains: activities of daily living (ADL), instrumental activities of daily living (IADL), leisure and social activities (LSA), lower extremity mobility (LEM), and general physical activities (GPA).
Table 1.
Self-Report Functional Status
| Domains | Components (Difficulty in …) |
|---|---|
| Activities of daily living | Eating: using fork, knife, and drinking from cup |
| Dressing yourself | |
| Walking between rooms on same floor | |
| Getting in and out of bed | |
| Instrumental activities of daily living | Managing money |
| Doing household chores | |
| Preparing meals | |
| Leisure and social activities | Going out to movies and events |
| Attending social events | |
| Performing leisure activities at home | |
| Lower extremity mobility | Walking a quarter mile |
| Walking up 10 steps | |
| General physical activities | Stooping, crouching, kneeling |
| Lifting or carrying | |
| Standing up from armless chair | |
| Standing for long periods | |
| Sitting for long periods | |
| Reaching up over head | |
| Grasp/holding small objects |
Note: Participant was asked about the ability to perform a series of activities without using any special equipment.
A participant's answer to a given question was coded as “no difficulty,” “some difficulty,” “much difficulty,” or “unable to do.” The responses to the questions were totally based on a participant's subjective self-assessment without further explanations. Disability was defined as any difficulty in performing one or more activities within a given domain.
Knee Extensor Power and Habitual Walking Speed
Leg power was chosen as the surrogate measure of impairment for our investigation because of its previously stated relevance to more distal disablement outcomes (12,13,18). Right knee extensor force production was assessed using a Kinetic Communicator isokinetic dynamometer (Kin-Com MP, Chattecx Corp., Chattanooga, TN). Maximal voluntary concentric muscle force was measured in Newtons in the right quadriceps at an angular velocity of 60 degrees/s. Ideally, each participant would have a total of 6 trials during the strength test: 3 practice warm-ups and 3 trials for maximal voluntary effort. Highest PF (in Newtons) was obtained according to the following algorithm: for an examinee who had 4 or more trials, one highest PF was selected from trials 4–6 (trials for maximal voluntary effort); for an examinee with fewer than 4 trials, a highest PF was selected from the completed trials (warm-up trials). Most participants had a PFVel ≈60 degrees/s. Participants with extreme values of PFVel, i.e., >65 degrees/s or <55 degrees/s, were excluded. Knee extensor power was obtained from the following formula (19):
Habitual gait speed served as our surrogate measure of functional limitations because of its predictive relationship to subsequent adverse outcomes including disability (20). The 20-foot timed walk test was performed at the participant's usual pace. Participants were allowed to use a walker or cane during the timed walk test if needed. Habitual walking speed (m/s) was calculated as walking distance (20 feet = 6.15 meter) divided by time in seconds.
Measurement of CRP
CRP was measured as part of the NHANES 1999–2002 physical and laboratory examination. Standard phlebotomy techniques were used to obtain specimens, which were frozen to −20°C until used for laboratory analysis. CRP was analyzed using a highly sensitive assay technique. CRP was quantified by using latex-enhanced nephelometry with a Behring Nephelometer Analyzer System (Deerfield, IL). Detailed specimen collection and processing instructions are discussed in the NHANES Laboratory Procedures Manual and are available on the NHANES Web site (16,17).
Covariates
Age, gender, race/ethnicity, and smoking status were obtained by self-report. Diabetes was defined by self-report of a physician's diagnosis, the presence of a random plasma glucose level greater than 200 mg/dL, or the use of diabetic medications (including insulin injection and/or oral hypoglycemic agents). Three and sometimes four blood pressure (BP) determinations were taken using a mercury sphygmomanometer by an NHANES physician. BP was measured in the right arm unless specific conditions prohibited the use of the right arm. Averaged systolic and diastolic BPs were obtained. The presence of hypertension was defined by a self-report of doctor's diagnosis, the use of antihypertensive medications, or averaged BP >140/90 mmHg. Body mass index (BMI), calculated as weight in kilograms divided by the square of height in meters, was categorized according to the National Institutes of Health obesity standards: <18.5 = underweight, 18.5–24.9 = normal weight, 25.0–29.9 = overweight, and >30 = obese (21). Chronic medical conditions including myocardial infarction (>6 weeks), coronary heart disease, congestive heart failure, angina, chronic bronchitis, emphysema, and arthritis were ascertained by self-report questionnaires. Heart disease was defined if participants had myocardial infarction, coronary heart disease, congestive heart failure, or angina; chronic obstructive pulmonary disease (COPD) was defined if participants had chronic bronchitis or emphysema. A 2-minute timed Digit Symbol Substitution test was administered to determine the cognitive function of the NHANES participants. Cognitive impairment was defined if the scores of the Digit Symbol Substitution test were below the median (= 44) of the study population. Alcohol intake was determined by the questionnaire “In any one year, have you had at least 12 drinks of any type of alcohol beverage?” and was dichotomized. Anemia was defined according to the World Health Organization as hemoglobin, obtained by the Beckman Coulter autoanalyzer (Fullerton, CA), <12 g/dL in women and <13 g/dL in men (22). Serum vitamin B12 and folate levels were measured by using the Bio-Rad Laboratories “Quantaphase II Folate/vitamin B12” radioassay kit (Hercules, CA). Plasma homocysteine was measured by the Abbott homocysteine assay (Abbott Park, IL) (an automated fluorescence polarization immunoassay [FPIA]), a method equivalent to high performance liquid chromatography (23). Levels of total cholesterol were obtained by using a standard biochemistry method.
Analysis
The distributions of CRP levels in the population were right skewed. Therefore, we used natural-log-transformed values, which provided the best-fitting model for analysis in which the plasma levels of CRP were treated as continuous variables. Standard-deviation scores of CRP were obtained from the formula (Xi − Xm) ÷ SD, where Xi is the natural-log-transformed level of CRP in the individual participant, Xm the mean natural-log-transformed level of CRP in the study cohort, and SD the standard deviation of the natural-log-transformed levels of CRP in the study cohort. This calculation allowed us to determine the changes in functional implications for each increment of 1SD in the natural-log-transformed CRP levels. In addition, we divided CRP levels into quartiles, and participants with CRP level in the lowest quartile were the reference group.
Multiple logistic regression was used to examine the relation of CRP levels to odds of functional disability in ADL, IADL, LSA, LEM, or GPA. We used an extended-model approach for covariates adjustment: Model 1 = age, sex, race, and BMI categories; Model 2 = Model 1 + chronic diseases (hypertension, diabetes, heart disease, COPD, arthritis, anemia, and cognitive impairment); Model 3 = Model 2 + health behaviors (smoking status and alcohol intake) + nutritional markers (natural-log-transformed levels of total cholesterol, folate, vitamin B12, and homocysteine); Model 4 = Model 3 + knee extensor power; Model 5 = Model 3 + habitual walking speed; and Model 6 = Model 3 + knee extensor power + habitual walking speed.
We also assessed the relation of CRP levels, treated as a continuous variable or quartiles, to performance-based physical measures, specifically leg power and habitual walking speed, by using multiple linear regression while adjusting for covariates in Model 3. Given the fact that power is a proximal pathway component in the impairment-disability pathway, we additionally adjusted for leg power in the model for the association between CRP and habitual walking speed to observe possible change of association.
Because the NHANES population weights are only applicable to analyses that use the entire population and because we limited our analyses to a special subset of participants, we did not use the NHANES 1999–2002 population weights for the purposes of this study. Data management and analysis were performed using STATA 8.0 software (STATA Corporation, College Station, TX).
Results
Characteristics of Study Sample
Selected baseline characteristics of the study sample as a whole (N = 1680, mean age 70.2 years) and by quartiles of CRP levels are summarized in Table 2. More than half of the study sample was non-Hispanic white (59.1%), and the mean BMI was 27.9 kg/m2. In the context of chronic conditions, 66.6% of the participants had evidence of hypertension, 15.4% diabetes, 16.5% heart diseases, 8.3% COPD, 6.4% anemia, and 42.4% arthritis.
Table 2.
Characteristics of Study Population According to Levels of C-Reactive Protein (N = 1680)
| Quartiles of C-Reactive Protein (mg/dL) | ||||||
|---|---|---|---|---|---|---|
| Characteristics | Q1 (<0.13) | Q2 (0.13–.27) | Q3 (0.28–.52) | Q4 (>0.52) | Total | p Value |
| Continuous variables* | ||||||
| Age, y | 70.1 (7.7) | 70.5 (7.5) | 71.1 (7.6) | 69.2 (7.1) | 70.2 (7.5) | .209 |
| Blood pressure, mmHg | ||||||
| Systolic | 137.0 (21.0) | 139.0 (21.4) | 139.2 (20.4) | 139.1 (21.1) | 138.6 (21.0) | .165 |
| Diastolic | 70.0 (14.6) | 69.6 (15.8) | 70.5 (15.1) | 68.9 (17.0) | 69.7 (15.6) | .472 |
| Body mass index, kg/m2 | 25.9 (4.2) | 27.8 (4.4) | 28.6 (4.7) | 29.5 (5.7) | 27.9 (4.9) | <.001 |
| Total cholesterol, mg/dL† | 204 (52) | 210 (47) | 218.5 (53) | 214 (53) | 212 (50) | <.001 |
| Homocysteine, μmol/L† | 9.02 (3.45) | 9.27 (3.93) | 9.06 (3.78) | 9.03 (4.24) | 9.07 (3.84) | .614 |
| Folate, ng/mL† | 16.7 (11.4) | 15.9 (9.9) | 15.8 (11.5) | 15.7 (12.1) | 16 (11.3) | .019 |
| Vitamin B12, pg/mL† | 473 (323) | 464 (277) | 477 (281) | 497 (300) | 478 (289) | .224 |
| Digit Symbol Substitution, numbers of correct | 46.2 (18.3) | 45.3 (18.2) | 43.8 (18.5) | 42.7 (18.7) | 44.5 (18.4) | .005 |
| Knee extensor power, Watts | 116.7 (43.6) | 114.1 (40.7) | 106.7 (38.7) | 103.5 (38.1) | 110.3 (40.7) | <.001 |
| Habitual walking speed, meter-second−1 | 1.018 (0.231) | 0.986 (0.214) | 0.962 (0.233) | 0.958 (0.265) | 0.982 (0.237) | <.001 |
| Categorical variables‡ | ||||||
| Female | 155 (36.6) | 183 (41.7) | 204 (51.1) | 235 (56.1) | 777 (46.3) | <.001 |
| Non-Hispanic white | 256 (60.5) | 279 (63.6) | 241 (60.4) | 217 (51.8) | 993 (59.1) | <.001 |
| Hypertension | 252 (59.6) | 292 (66.5) | 277 (69.4) | 297 (70.9) | 1118 (66.6) | .004 |
| Diabetes mellitus | 64 (15.1) | 52 (11.9) | 69 (17.3) | 74 (17.7) | 259 (15.4) | .071 |
| Heart diseases | 72 (17.0) | 68 (15.5) | 67 (16.8) | 70 (16.7) | 277 (16.5) | .931 |
| Chronic obstructive pulmonary disease | 22 (5.2) | 35 (8.0) | 36 (9.0) | 46 (11.0) | 139 (8.3) | .022 |
| Arthritis | 154 (36.4) | 183 (41.7) | 169 (42.4) | 206 (49.2) | 712 (42.4) | .008 |
| Anemia | 23 (5.4) | 24 (5.5) | 24 (6.0) | 36 (8.6) | 107 (6.4) | .328 |
| Current smoker | 49 (11.6) | 54 (12.3) | 52 (13.0) | 69 (16.5) | 224 (13.3) | .163 |
| Alcohol intake ≥12 drinks/y | 274 (64.8) | 286 (65.2) | 248 (62.2) | 250 (59.7) | 1058 (63.0) | .119 |
| Self-report disability§ | ||||||
| Activities of daily living | 57 (13.5) | 57 (13.0) | 63 (15.8) | 82 (19.6) | 259 (15.4) | .032 |
| Instrumental activities of daily living | 64 (15.1) | 73 (16.6) | 78 (19.6) | 102 (24.3) | 317 (18.9) | .014 |
| Leisure and social activities | 42 (9.9) | 43 (9.8) | 53 (13.3) | 74 (17.7) | 212 (12.6) | .001 |
| Lower extremity mobility | 71 (16.8) | 105 (23.9) | 119 (29.8) | 118 (28.2) | 413 (24.6) | <.001 |
| General physical activities | 188 (44.4) | 216 (49.2) | 211 (52.9) | 253 (60.4) | 868 (51.7) | <.001 |
Participants in the upper quartile of CRP level seemed more likely to be female and hypertensive. They tended to have lower quadriceps power, lower habitual gait speed, and higher prevalence of disability.
CRP and Disability
The results of multivariable adjusted logistic regression with CRP as a continuous variable are provided in Table 3. After adjustment for age, sex, race, and BMI categories (Model 1), elevated levels of CRP were associated with increased odds of disability in ADL, IADL, LSA, LEM, and GPA. After additionally controlling for covariates including chronic diseases, smoking status, alcohol intake, as well as nutritional markers (Model 2 and Model 3), the association between CRP levels and ADL disability was no longer significant. Yet, we still observed statistically significant relationships of elevated CRP to disability in IADL, LSA, LEM, and GPA after controlling for covariates in Model 2 and Model 3. Knee extensor power and habitual walking speed were introduced as covariates from Model 4 to Model 6. For disability outcomes in IADL, LSA, LEM, and GPA, the sizes of the odds ratios were substantially reduced compared with the previous models, and their confidence intervals always included 1, suggesting that leg power, usual gait speed, or both, to a large extent, might explain the association between elevated CRP and disability.
Table 3.
Logistic Regression Models Testing the Association Between C-Reactive Protein and Disability: Models With C-Reactive Protein As a Continuous Variable
| ADL Disability | IADL Disability | LSA Disability | LEM Disability | GPA Disability | ||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Model | OR† (95% CI) | p Value | OR† (95% CI) | p Value | OR† (95% CI) | p Value | OR† (95% CI) | p Value | OR† (95% CI) | p Value |
| 1 | 1.15 (1.00–1.33) | .043 | 1.22 (1.07–1.39) | .003 | 1.25 (1.07–1.46) | .004 | 1.22 (1.08–1.38) | .002 | 1.21 (1.09–1.34) | <.001 |
| 2 | 1.12 (0.97–1.29) | .117 | 1.18 (1.03–1.35) | .019 | 1.21 (1.03–1.42) | .018 | 1.20 (1.06–1.37) | .004 | 1.18 (1.05–1.32) | .004 |
| 3 | 1.10 (0.95–1.28) | .197 | 1.18 (1.02–1.35) | .024 | 1.18 (1.00–1.39) | .05 | 1.17 (1.03–1.33) | .019 | 1.17 (1.05–1.31) | .006 |
| 4 | 1.00 (0.85–1.16) | .952 | 1.08 (0.93–1.25) | .306 | 1.06 (0.89–1.25) | .534 | 1.08 (0.94–1.24) | .261 | 1.10 (0.98–1.24) | .099 |
| 5 | 1.01 (0.86–1.18) | .899 | 1.09 (0.94–1.26) | .267 | 1.06 (0.89–1.26) | .506 | 1.10 (0.95–1.27) | .191 | 1.12 (0.99–1.26) | .056 |
| 6 | 0.98 (0.84–1.15) | .824 | 1.06 (0.91–1.23) | .467 | 1.03 (0.86–1.23) | .760 | 1.07 (0.92–1.23) | .375 | 1.10 (0.98–1.24) | .108 |
Notes: Adjusted covariates: Model 1 = age, sex, race, and body mass index; Model 2 = Model 1 + chronic diseases (hypertension, diabetes, heart disease, chronic obstructive pulmonary disease (COPD), arthritis, anemia, and cognitive impairment); Model 3 = Model 2 + health behaviors (smoking status and alcohol intake) + nutritional markers (levels of total cholesterol, folate, vitamin B12, and homocysteine); Model 4 = Model 3 + knee extensor power; Model 5 = Model 3 + habitual walking speed; and Model 6 = Model 3 + knee extensor power + habitual walking speed.
Disability was categorized into five major domains based on abilities to perform the following activities: 1) ADL (e.g., eating, walking, dressing, and getting out of bed); 2) IADL (e.g., managing money, keeping house, and preparing food); 3) LSA (e.g., reading, watching television, going out to movies, visiting friends, attending clubs or meetings); 4) LEM (e.g., difficulty in walking a quarter mile or up 10 steps); and 5) GPA (e.g., stooping, bending, standing, sitting, lifting, reaching, grasping). Disability in a specific functional domain was defined if a person reported any difficulty in one or more activities in a given disability domain.
OR values are for each standard deviation increase in the natural-log-transformed C-reactive protein levels.
ADL = activities of daily living; CI = confidence interval; GPA = general physical activities; IADL = instrumental activities of daily living; LEM = lower extremity mobility; LSA leisure and social activities; OR = odds ratio.
We subsequently analyzed the CRP levels divided into quartiles; the results of quartile-based multiple logistic regression are shown in Table 4. From Model 1 to Model 3, we observed positive associations between CRP levels and disability in IADL, LSA, LEM, and GPA. Participants in the higher quartiles of CRP concentrations tended to have higher odds of disability in IADL, LSA, LEM, and GPA. The trends of disability in IADL, LSA, LEM, and GPA across CRP quartiles were all statistically significant. Likewise, when leg power, usual walking speed, or both were additionally introduced to the quartile-based logistic regression models, the sizes of odds ratios dropped substantially and statistical significance disappeared (Model 4 to Model 6). In terms of ADL disability, we did not observe any significant association in the quartile-based approach.
Table 4.
Logistic Regression Models Testing the Association Between Quartiles of C-Reactive Protein and Disability*
| ADL Disability | IADL Disability | LSA Disability | LEM Disability | GPA Disability | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Model | Quartiles Comparison | OR† (95% CI) | p for Trend | OR† (95% CI) | p for Trend | OR† (95% CI) | p for Trend | OR† (95% CI) | p for Trend | OR† (95% CI) | p for Trend |
| 1 | Q2 vs Q1 | 0.90 (0.60–1.35) | .065 | 1.10 (0.75–1.60) | .004 | 0.92 (0.58–1.46) | .003 | 1.43 (1.00–2.04) | .005 | 1.04 (0.78–1.38) | .003 |
| Q3 vs Q1 | 1.08 (0.72–1.61) | 1.25 (0.85–1.84) | 1.24 (0.79–1.94) | 1.75 (1.23–2.50) | 1.14 (0.85–1.53) | ||||||
| Q4 vs Q1 | 1.36 (0.92–2.00) | 1.66 (1.14–2.40) | 1.74 (1.13–2.68) | 1.64 (1.15–2.36) | 1.54 (1.15–2.07) | ||||||
| 2 | Q2 vs Q1 | 0.92 (0.61–1.40) | .187 | 1.15 (0.77–1.72) | .035 | 0.99 (0.61–1.59) | .017 | 1.48 (1.03–2.14) | .013 | 1.03 (0.76–1.38) | .023 |
| Q3 vs Q1 | 1.08 (0.71–1.63) | 1.26 (0.84–1.88) | 1.27 (0.79–2.03) | 1.76 (1.22–2.54) | 1.13 (0.83–1.53) | ||||||
| Q4 vs Q1 | 1.24 (0.83–1.86) | 1.51 (1.02–2.23) | 1.60 (1.02–2.51) | 1.59 (1.10–2.31) | 1.41 (1.04–1.93) | ||||||
| 3 | Q2 vs Q1 | 0.90 (0.59–1.37) | .242 | 1.17 (0.78–1.76) | .043 | 0.99 (0.61–1.62) | .036 | 1.42 (0.97–2.06) | .039 | 0.99 (0.73–1.34) | .042 |
| Q3 vs Q1 | 1.06 (0.69–1.62) | 1.27 (0.84–1.92) | 1.26 (0.77–2.04) | 1.68 (1.15–2.45) | 1.09 (0.80–1.50) | ||||||
| Q4 vs Q1 | 1.21 (0.80–1.83) | 1.51 (1.01–2.26) | 1.53 (0.96–2.44) | 1.48 (1.02–2.17) | 1.36 (0.99–1.87) | ||||||
| 4 | Q2 vs Q1 | 0.80 (0.52–1.25) | .899 | 1.06 (0.70–1.62) | .476 | 0.89 (0.54–1.47) | .494 | 1.29 (0.88–1.90) | .372 | 0.92 (0.67–1.25) | .296 |
| Q3 vs Q1 | 0.92 (0.59–1.43) | 1.11 (0.72–1.71) | 1.06 (0.64–1.75) | 1.49 (1.01–2.20) | 1.00 (0.72–1.39) | ||||||
| Q4 vs Q1 | 0.91 (0.59–1.41) | 1.16 (0.76–1.78) | 1.11 (0.68–1.82) | 1.19 (0.80–1.78) | 1.16 (0.84–1.62) | ||||||
| 5 | Q2 vs Q1 | 0.83 (0.53–1.30) | .966 | 1.03 (0.67–1.57) | .368 | 0.87 (0.52–1.45) | .402 | 1.23 (0.83–1.82) | .274 | 0.91 (0.66–1.24) | .170 |
| Q3 vs Q1 | 0.94 (0.60–1.47) | 1.10 (0.71–1.70) | 1.03 (0.62–1.72) | 1.47 (0.99–2.18) | 1.01 (0.73–1.41) | ||||||
| Q4 vs Q1 | 0.95 (0.61–1.48) | 1.20 (0.78–1.84) | 1.15 (0.70–1.89) | 1.22 (0.82–1.83) | 1.22 (0.88–1.70) | ||||||
| 6 | Q2 vs Q1 | 0.76 (0.50–1.20) | .790 | 1.01 (0.66–1.54) | .650 | 0.84 (0.51–1.41) | .688 | 1.20 (0.81–1.78) | .521 | 0.89 (0.65–1.22) | .299 |
| Q3 vs Q1 | 0.88 (0.56–1.37) | 1.04 (0.67–1.62) | 0.97 (0.58–1.62) | 1.40 (0.94–2.08) | 0.98 (0.71–1.37) | ||||||
| Q4 vs Q1 | 0.88 (0.56–1.37) | 1.09 (0.71–1.69) | 1.04 (0.63–1.72) | 1.12 (0.75–1.69) | 1.16 (0.83–1.62) | ||||||
Notes: Adjusted covariates: Model 1 = age, sex, race, and body mass index; Model 2 = Model 1 + chronic diseases (hypertension, diabetes, heart disease, chronic obstructive pulmonary disease [COPD], arthritis, anemia, and cognitive impairment); Model 3 = Model 2 + health behaviors (smoking status and alcohol intake) + nutritional markers (levels of total cholesterol, folate, vitamin B12, and homocysteine); Model 4 = Model 3 + knee extensor power; Model 5 = Model 3 + habitual walking speed; and Model 6 = Model 3 + knee extensor power + habitual walking speed.
Disability was categorized into five major domains based on abilities to perform the following activities: 1) ADL (e.g., eating, walking, dressing, and getting out of bed); 2) IADL (e.g., managing money, housekeeping, and preparing food); 3) LSA (e.g., reading, watching television, going out to movies, visiting friends, attending clubs or meetings); 4) LEM (e.g., difficulty in walking a quarter mile or up 10 steps); and 5) GPA (e.g., stooping, bending, standing, sitting, lifting, reaching, grasping). Disability in a specific functional domain was defined if a person reported any difficulty in one or more activities in a given disability domain.
ORs were for disability in a specific domain comparing participants in the 2nd, 3rd, or 4th quartiles of C-reactive protein to those in the lowest quartile.
ADL = activities of daily living; CI = confidence interval; GPA = general physical activities; IADL = instrumental activities of daily living; LEM = lower extremity mobility; LSA = leisure and social activities; OR = odds ratio.
CRP and Performance-Based Physical Measures
In the models with CRP as a continuous variable, there were inverse relationships between CRP concentrations and knee extensor power, as well as habitual walking speed. After adjustment for age, sex, race, BMI categories, chronic diseases, health-related behaviors (smoking and alcohol intake), as well as levels of nutritional markers including total cholesterol, folate, vitamin B12, and homocysteine, there was a 3.4 Watt decrease (p < .001) and a 0.013 meter-second−1 decrease (p = .017) in peak leg power and habitual gait speed, respectively, for each SD increase in the natural-log-transformed CRP concentrations (Table 5). We additionally adjusted for knee extensor power in the association between CRP and habitual walking speed and found that the association diminished, suggesting that power might mediate the association between CRP and habitual walking speed.
Table 5.
Associations Between Levels of C-Reactive Protein and Performance-Based Physical Measures
| CRP as a Continuous Variable | |||||
|---|---|---|---|---|---|
| Models* | β† (SE) | p Value | |||
| Power | −3.4 (0.7) | <.001 | |||
| Walking speed | −0.013 (0.005) | .017 | |||
| Walking speed while controlling for power | −0.008 (0.005) | .148 | |||
| CRP by Increasing Quartiles | |||||
|
|
|||||
| Models* | Quartile | β‡ (SE) | p Value | Adjusted Means | p for Trend |
|
| |||||
| Power | Q1 | Reference | 115.4 | <.001 | |
| Q2 | −3.3 (1.9) | .08 | 112.2 | ||
| Q3 | −5.9 (1.9) | .002 | 109.5 | ||
| Q4 | −10.4 (1.9) | <.001 | 105.0 | ||
| Walking speed | Q1 | Reference | 1.005 | .033 | |
| Q2 | −0.028 (0.014) | .054 | 0.977 | ||
| Q3 | −0.031 (0.015) | .042 | 0.974 | ||
| Q4 | −0.034 (0.015) | .025 | 0.971 | ||
| Walking speed while controlling for power | Q1 | Reference | 0.998 | .273 | |
| Q2 | −0.023 (0.014) | .105 | 0.975 | ||
| Q3 | −0.022 (0.015) | .139 | 0.976 | ||
| Q4 | −0.018 (0.015) | .216 | 0.980 | ||
Notes: Units of measure: muscle power (Watts), and walking speed (m/s).
All models were adjusted for age, sex, race, body mass index, hypertension, diabetes, heart disease, COPD, arthritis, anemia, cognitive impairment, self-reported health conditions, smoking status, alcohol intake, as well as levels of nutritional markers including total cholesterol, folate, vitamin B12, and homocysteine.
The plasma CRP levels were analyzed as continuous variables. The β coefficients are changes of knee extensor power or walking speed per increment of 1SD in the natural-log-transformed CRP values.
Coefficients ( (β) can be interpreted as differences in mean knee extensor power or walking speed comparing subjects in the 2nd, 3rd, or 4th quartiles of CRP levels to those in the lowest quartile.
β = regression coefficient; COPD = chronic obstructive pulmonary disease; CRP = C-reactive protein; Q = C-reactive protein quartile; SE = standard error.
In the quartile-based approach, we found that participants in the higher quartiles of CRP had poorer knee extensor power and habitual walking speed. Participants in the highest quartile of CRP had lower knee extensor power (β = −10.4, p < .001) and slower walking speed (β = −0.034, p = .025) compared to those in the lowest quartile of CRP. Likewise, additional adjustment for leg power diminished the association between CRP and habitual walking speed to a nonsignificant level. For every corresponding model, the adjusted means of leg power (Watts) and habitual walking speed (meter-second−1) in each quartile of CRP levels are provided in Table 5. We did not find any effect modification of sex on the relationship of CRP to leg power or to habitual gait speed as assessed by sex and CRP interaction.
Discussion
Among U.S. noninstitutionalized older adults, elevated levels of CRP were associated with disability in IADL, LSA, LEM, as well as GPA. In addition, there were inverse associations between performance-based physical measures, including knee extensor power and habitual gait speed, and levels of CRP. Moreover, our results suggested that the association between CRP and disability, to a large extent, can be explained by magnitudes of leg power, walking speed, or both.
According to the theoretical disablement model first described by Nagi (24), disability is the result of a pathway from pathology to physical impairments (for example, in strength and balance) to functional limitations (for example, in walking). In this context, our study provided an excellent opportunity to revisit this model in which chronic inflammation with CRP as a marker indicated “pathology,” peak leg power “impairment,” usual walking speed “functional limitation” and self-reported functional status in terms of ADL, IADL, LSA, LEM, and GPA “disability.” We found that elevated CRP was associated with disability in IADL, LSA, LEM, and GPA independent of basic demographics, chronic medical diseases, health behaviors, as well as nutritional markers. The statistical significance of the association vanished after additional adjustment for leg power, habitual walking speed, or both, indicating that both leg power and walking speed may mediate the association and that both are intrinsic, intermediate components of the disablement process from chronic inflammation to disability. We also demonstrated that elevated CRP was associated with lower quadriceps power (impairment) and lower gait speed (function). Likewise, leg power seemed to mediate the association between CRP and gait speed. Our study sheds new light in illustrating the disablement process related to inflammation.
Several population-based studies have demonstrated that markers of chronic inflammation, including interleukin-1 (25), interleukin-6 (26,27), tumor necrosis factor-α (27), or CRP (26), were cross-sectionally associated with muscle strength (26,27), muscle mass (27), physical performance (26), or disability (25). The MacArthur Study of Successful Aging, however, produced somewhat conflicting findings. Although an inverse relationship of inflammatory markers to physical performance (28) or self-reported recreational activity (29) was found in the cross-sectional analyses, the prospective analysis did not demonstrate a predictive role of inflammatory markers in the change of physical performance. However, the prospective MacArthur Study had a weakness in generalizability because the analysis was limited to a subgroup of participants with high physical and cognitive functioning who survived 7-year follow-up. Although results from the Women's Health and Aging Study suggested that high interleukin-6 predicted disability and that decline in muscle strength might mediate the association, all participants had some disability at baseline and there were no men in the study (10). Ravaglia and colleagues (11) cross-sectionally examined 739 elderly Italian community dwellers from the Conselice Study of Brain Aging and demonstrated that increased CRP was inversely associated with ADL, IADL, as well as the Tinetti test score, independently of demographics, lifestyle, and comorbidity. However, they did not address the possible mediating effect of objective physical performance in the association between CRP and disability measures. To the best of our knowledge, this is the first report to describe the association between CRP and late-life disability while considering the roles of both functional impairment and physical limitations by using a geographically dispersed and ethnically diverse representative sample of community-dwelling elderly persons living in the United States.
In addition to causing protein catabolism, chronic inflammation also plays a crucial role in the pathogenesis of atherosclerosis. Inflammatory markers, such as CRP and proinflammatory cytokines, are sensitive measures of the burden of systemic atherosclerosis and extent of atherosclerotic activity (30,31). In participants with increased loads of cardiovascular risks, physical performance such as walking speed, muscle strength, or power can be impaired due to atherosclerotic changes centrally affecting the heart and peripherally afflicting the vasculature. Moreover, cerebral vascular changes, including large observable stroke or leukoaraiosis, may develop and thus interrupt the descending motor fibers arising from medial cortical areas important for lower extremity motor control, as well as debilitate the frontal-subcortical circuits responsible for normal gait and balance (32). In fact, impairment or limitation in balance, gait speed, or muscle strength have been shown to predict disability (such as dependence on ADL) among older adults (20,33–36).
Our results have several clinical implications. First, in addition to being an independent predictor of cardiovascular disease, an elevated level of CRP may be an important indicator of risk for impairment, functional limitation, as well as disability. Measurement of CRP levels may be useful in identifying and targeting elderly individuals who may require intervention to prevent loss of function and disability. Second, our findings may serve as theoretical basis for novel therapeutic approaches. Pharmacologic therapies such as statin therapy have been shown to be effective strategies to lower levels of CRP (37). Moreover, statin improves coronary endothelial function (38,39) and effectively reduces the risk of developing cardiovascular events. Perhaps these pharmacological therapies in combination with established approaches such as exercise may augment our therapeutic approach for those persons at risk for disability.
Our study has potential limitations that deserve comment. First, due to the cross-sectional design of the study, causal relationship from CRP to functional disability can not be established. The relationship should be study prospectively. Second, although we demonstrated important implications of CRP in elderly people, it is not clear whether elevated CRP level is a mechanistic factor influencing disablement factors or is simply a marker of an underlying physiologic process which is the true cause of decline.
Conclusion
High levels of CRP are associated with physical disability mediated by knee extensor power, habitual gait speed, or both. Our study provides new information about the potential role of inflammation in the disablement process and underscores the importance of inflammatory markers in their association with disablement outcomes, having important implications for clinical care and future research.
Acknowledgments
We thank Dr. Richard N. Jones of the Research and Training Institute, Hebrew Rehabilitation Center for Aged for statistical expertise and helpful comments on the manuscript.
References
- 1.Ridker PM. Clinical application of C-reactive protein for cardiovascular disease detection and prevention. Circulation. 2003;107:363–369. doi: 10.1161/01.cir.0000053730.47739.3c. [DOI] [PubMed] [Google Scholar]
- 2.Gabay C, Kushner I. Acute-phase proteins and other systemic responses to inflammation. N Engl J Med. 1999;340:448–454. doi: 10.1056/NEJM199902113400607. [DOI] [PubMed] [Google Scholar]
- 3.Whitehead AS, Bruns GA, Markham AF, Colten HR, Woods DE. Isolation of human C-reactive protein complementary DNA and localization of the gene to chromosome 1. Science. 1983;221:69–71. doi: 10.1126/science.6857266. [DOI] [PubMed] [Google Scholar]
- 4.Kuta AE, Baum LL. C-reactive protein is produced by a small number of normal human peripheral blood lymphocytes. J Exp Med. 1986;164:321–326. doi: 10.1084/jem.164.1.321. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Cannon JG. Cytokines in aging and muscle homeostasis. J Gerontol A Biol Sci Med Sci. 1995;50A(Spec Iss):120–123. doi: 10.1093/gerona/50a.special_issue.120. [DOI] [PubMed] [Google Scholar]
- 6.Tisdale MJ. New cachexic factors. Curr Opin Clin Nutr Metab Care. 1998;1:253–256. doi: 10.1097/00075197-199805000-00003. [DOI] [PubMed] [Google Scholar]
- 7.Janssen I, Heymsfield SB, Ross R. Low relative skeletal muscle mass (sarcopenia) in older persons is associated with functional impairment and physical disability. J Am Geriatr Soc. 2002;50:889–896. doi: 10.1046/j.1532-5415.2002.50216.x. [DOI] [PubMed] [Google Scholar]
- 8.Fisher AL. Of worms and women: sarcopenia and its role in disability and mortality. J Am Geriatr Soc. 2004;52:1185–1190. doi: 10.1111/j.1532-5415.2004.52320.x. [DOI] [PubMed] [Google Scholar]
- 9.Ferrucci L, Harris TB, Guralnik JM, et al. Serum IL-6 level and the development of disability in older persons. J Am Geriatr Soc. 1999;47:639–646. doi: 10.1111/j.1532-5415.1999.tb01583.x. [DOI] [PubMed] [Google Scholar]
- 10.Ferrucci L, Penninx BW, Volpato S, et al. Change in muscle strength explains accelerated decline of physical function in older women with high interleukin-6 serum levels. J Am Geriatr Soc. 2002;50:1947–1954. doi: 10.1046/j.1532-5415.2002.50605.x. [DOI] [PubMed] [Google Scholar]
- 11.Ravaglia G, Forti P, Maioli F, et al. Peripheral blood markers of inflammation and functional impairment in elderly community-dwellers. Exp Gerontol. 2004;39:1415–1422. doi: 10.1016/j.exger.2004.06.010. [DOI] [PubMed] [Google Scholar]
- 12.Bassey EJ, Fiatarone MA, O'Neill EF, Kelly M, Evans WJ, Lipsitz LA. Leg extensor power and functional performance in very old men and women. Clin Sci (Lond) 1992;82:321–327. doi: 10.1042/cs0820321. [DOI] [PubMed] [Google Scholar]
- 13.Skelton DA, Greig CA, Davies JM, Young A. Strength, power and related functional ability of healthy people aged 65-89 years. Age Ageing. 1994;23:371–377. doi: 10.1093/ageing/23.5.371. [DOI] [PubMed] [Google Scholar]
- 14.Bean JF, Kiely DK, Herman S, et al. The relationship between leg power and physical performance in mobility-limited older people. J Am Geriatr Soc. 2002;50:461–467. doi: 10.1046/j.1532-5415.2002.50111.x. [DOI] [PubMed] [Google Scholar]
- 15.Metter EJ, Conwit R, Tobin J, Fozard JL. Age-associated loss of power and strength in the upper extremities in women and men. J Gerontol Biol Sci. 1997;52A:B267–B276. doi: 10.1093/gerona/52a.5.b267. [DOI] [PubMed] [Google Scholar]
- 16.National Health and Nutrition Examination Survey (NHANES) Bethesda, MD: National Center for Health Statistics; 1999–2000. Available at http://www.cdc.gov/nchs/about/major/nhanes/currentnhanes.htm. [Google Scholar]
- 17.National Health and Nutrition Examination Survey (NHANES) Bethesda, MD: National Center for Health Statistics; 2001–2002. Available at http://www.cdc.gov/nchs/about/major/nhanes/current_nhanes_01_02.htm. [Google Scholar]
- 18.Suzuki T, Bean JF, Fielding RA. Muscle power of the ankle flexors predicts functional performance in community-dwelling older women. J Am Geriatr Soc. 2001;49:1161–1167. doi: 10.1046/j.1532-5415.2001.49232.x. [DOI] [PubMed] [Google Scholar]
- 19.Wilson G. Strength and power assessment. In: Bloomfield J, Ackland TR, Elliott BC, editors. Applied Anatomy and Biomechanics in Sport. Blackwell Publishing Inc.; 1994. [Google Scholar]
- 20.Guralnik JM, Ferrucci L, Pieper CF, et al. Lower extremity function and subsequent disability: consistency across studies, predictive models, and value of gait speed alone compared with the short physical performance battery. J Gerontol A Biol Sci Med Sci. 2000;55A:M221–M231. doi: 10.1093/gerona/55.4.m221. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.National Heart, Lung, and Blood Institute. Obesity Education Initiative: The Evidence Report. Bethesda, MD: National Institutes of Health; 1998. [Google Scholar]
- 22.World Health Organization (WHO) Nutritional Anemia: Report of a WHO Scientific Group. Geneva, Switzerland: World Health Organization; 1968. [Google Scholar]
- 23.Pfeiffer CM, Huff DL, Smith SJ, Miller DT, Gunter EW. Comparison of plasma total homocysteine measurements in 14 laboratories: an international study. Clin Chem. 1999;45(8 Pt 1):1261–1268. [PubMed] [Google Scholar]
- 24.Nagi SZ. An epidemiology of disability among adults in the United States. Milbank Mem Fund Q Health Soc. 1976;54:439–467. [PubMed] [Google Scholar]
- 25.Cohen HJ, Pieper CF, Harris T, Rao KM, Currie MS. The association of plasma IL-6 levels with functional disability in community-dwelling elderly. J Gerontol Med Sci. 1997;52A:M201–M208. doi: 10.1093/gerona/52a.4.m201. [DOI] [PubMed] [Google Scholar]
- 26.Cesari M, Penninx BW, Pahor M, et al. Inflammatory markers and physical performance in older persons: the InCHIANTI study. J Gerontol A Biol Sci Med Sci. 2004;59A:242–248. doi: 10.1093/gerona/59.3.m242. [DOI] [PubMed] [Google Scholar]
- 27.Visser M, Pahor M, Taaffe DR, et al. Relationship of interleukin-6 and tumor necrosis factor-alpha with muscle mass and muscle strength in elderly men and women: the Health ABC Study. J Gerontol Med Sci. 2002;57A:M326–M332. doi: 10.1093/gerona/57.5.m326. [DOI] [PubMed] [Google Scholar]
- 28.Taaffe DR, Harris TB, Ferrucci L, Rowe J, Seeman TE. Cross-sectional and prospective relationships of interleukin-6 and C-reactive protein with physical performance in elderly persons: MacArthur studies of successful aging. J Gerontol Med Sci. 2000;55A:M709–M715. doi: 10.1093/gerona/55.12.m709. [DOI] [PubMed] [Google Scholar]
- 29.Reuben DB, Judd-Hamilton L, Harris TB, Seeman TE. The associations between physical activity and inflammatory markers in high-functioning older persons: MacArthur Studies of Successful Aging. J Am Geriatr Soc. 2003;51:1125–1130. doi: 10.1046/j.1532-5415.2003.51380.x. [DOI] [PubMed] [Google Scholar]
- 30.Alexander RW. Inflammation and coronary artery disease. N Engl J Med. 1994;331:468–469. doi: 10.1056/NEJM199408183310709. [DOI] [PubMed] [Google Scholar]
- 31.Rajavashisth TB, Xu XP, Jovinge S, et al. Membrane type 1 matrix metalloproteinase expression in human atherosclerotic plaques: evidence for activation by proinflammatory mediators. Circulation. 1999;99:3103–3109. doi: 10.1161/01.cir.99.24.3103. [DOI] [PubMed] [Google Scholar]
- 32.Masdeu JC, Wolfson L, Lantos G, et al. Brain white-matter changes in the elderly prone to falling. Arch Neurol. 1989;46:1292–1296. doi: 10.1001/archneur.1989.00520480034016. [DOI] [PubMed] [Google Scholar]
- 33.Brown M, Sinacore DR, Host HH. The relationship of strength to function in the older adult. J Gerontol A Biol Sci Med Sci. 1995;50A(Spec Iss):55–59. doi: 10.1093/gerona/50a.special_issue.55. [DOI] [PubMed] [Google Scholar]
- 34.Guralnik JM, Ferrucci L, Simonsick EM, Salive ME, Wallace RB. Lower-extremity function in persons over the age of 70 years as a predictor of subsequent disability. N Engl J Med. 1995;332:556–561. doi: 10.1056/NEJM199503023320902. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Giampaoli S, Ferrucci L, Cecchi F, et al. Hand-grip strength predicts incident disability in non-disabled older men. Age Ageing. 1999;28:283–288. doi: 10.1093/ageing/28.3.283. [DOI] [PubMed] [Google Scholar]
- 36.Ostir GV, Markides KS, Black SA, Goodwin JS. Lower body functioning as a predictor of subsequent disability among older Mexican Americans. J Gerontol Med Sci. 1998;53:M491–M495. doi: 10.1093/gerona/53a.6.m491. [DOI] [PubMed] [Google Scholar]
- 37.Ridker PM, Rifai N, Pfeffer MA, Sacks F, Braunwald E. Long-term effects of pravastatin on plasma concentration of C-reactive protein. The Cholesterol and Recurrent Events (CARE) Investigators. Circulation. 1999;100:230–235. doi: 10.1161/01.cir.100.3.230. [DOI] [PubMed] [Google Scholar]
- 38.Effect of nifedipine and cerivastatin on coronary endothelial function in patients with coronary artery disease: the ENCORE I Study (Evaluation of Nifedipine and Cerivastatin On Recovery of coronary Endothelial function) Circulation. 2003;107:422–428. doi: 10.1161/01.cir.0000046488.52939.bf. [DOI] [PubMed] [Google Scholar]
- 39.Vita JA, Yeung AC, Winniford M, et al. Effect of cholesterol-lowering therapy on coronary endothelial vasomotor function in patients with coronary artery disease. Circulation. 2000;102:846–851. doi: 10.1161/01.cir.102.8.846. [DOI] [PubMed] [Google Scholar]