Abstract
Background
high-sensitivity C-reactive protein (CRP) has been associated with cardiovascular events and mortality, but the association of CRP with functional status is not well defined. We hypothesised that serum levels of high-sensitivity CRP are associated with long-term trajectories of functional status independently of vascular risk factors and stroke and myocardial infarction (MI) occurring during follow-up.
Design
prospective, population-based.
Setting
northern Manhattan Study.
Participants
stroke-free participants aged ≥40 years.
Measurements
annual assessments of disability with the Barthel index (BI) for a median of 13 years. BI was analysed as a continuous variable (range 0–100). Baseline demographics, risk factors and laboratory studies were collected, including CRP (n = 2,240). Separate generalised estimating equation models estimated standardised associations between CRP and (i) baseline functional status and (ii) change in function over time, adjusting for demographics, vascular risk factors, social variables, cognition, and depression measured at baseline, and stroke and MI occurring during follow-up.
Results
mean age was 69 (SD 10) years, 36% were male, 55% Hispanic, 75% hypertensive and 21% diabetic; 337 MIs and 369 first strokes occurred during follow-up. Mean CRP level was 5.24 mg/l (SD 8.86). logCRP was associated with baseline BI (−0.34 BI points per unit logCRP, 95% confidence interval −0.62, −0.06) but not with change over time.
Conclusions
in this large population-based study, higher serum CRP levels were associated with higher baseline disability, even when adjusting for baseline covariates and stroke and MI occurring during follow-up. Systemic inflammation may contribute to disability independently of clinical vascular events.
Keywords: older people, C-reactive protein, disability, epidemiology
Introduction
C-reactive protein (CRP) is an acute phase reactant that reflects acute inflammatory states and tissue injury [1]. CRP may also be directly implicated in pro-atherogenic processes, and several studies have shown associations between higher CRP levels and vascular events and mortality [2–4]. In the Cardiovascular Health Study (CHS), CRP was associated with white-matter lesions [5], and other studies have found associations between CRP and stroke severity and mortality. [6]. Beyond the association with vascular outcomes, inflammation has been associated with quality of life (QOL) in a limited number of studies [7, 8], but the association of inflammatory markers with disability has not been well studied, particularly among minority populations.
We hypothesised that elevated serum CRP levels independently predict worse functional status in those free of stroke at baseline. We estimated two components of functional trajectories: baseline functional status (which can be considered as a change in overall or mean function) and change over time. To our knowledge, this is the first analysis in which the association of CRP was assessed with both of these components of functional trajectories.
Methods
In the Northern Manhattan Study (NOMAS) population-based prospective cohort, 3,298 participants were recruited in a race/ethnically diverse community by random digit dialing of published and unpublished telephone numbers between 1993 and 2001. Participants were enrolled if they: (i) were ≥40 years of age; (ii) lived in northern Manhattan for ≥3 months in a household with a telephone and (iii) had no history of stroke. The study was approved by the institutional review boards of Columbia University and the University of Miami, and informed consent was obtained from all participants. Further characteristics of the cohort have previously described [9–11].
Baseline evaluation
Bilingual research assistants interviewed participants and collected data using standardised questions [12]. Baseline examination included comprehensive medical history, physical examination, medical record reviews, functional status assessed by the Barthel Index (BI), QOL assessed by the Spitzer QOL index and fasting blood samples.
Follow-up
All participants were followed annually via phone screening to detect change in vital status, new neurological or cardiac symptoms and events, interval hospitalisations, cognitive function and functional status via the BI. Only two participants were lost to follow-up after their baseline examination, and the average annual contact rate was 99%.
A positive screen for any potential cardiac or neurological event was followed by an in-person assessment. Also, all admissions and discharges were screened for outcomes that may not have been captured by telephone interview. Nearly 70% of vascular events led to hospitalisations at Columbia University Medical Center. Hospital records were reviewed to classify outcomes as previously reported [11]. Stroke included ischaemic stroke, intracerebral haemorrhage (ICH) and subarachnoid haemorrhage (SAH). At least two stroke neurologists verified and classified all stroke cases. Cardiologist-adjudicated myocardial infarction (MI) required at least two of the following criteria [13, 14]: (i) ischaemic cardiac pain determined to be typical angina; (ii) cardiac marker abnormalities (abnormal CK-MB fraction or troponin I values) and (iii) ischaemic EKG abnormalities.
There were 333 first MIs that occurred during follow-up, 184 (55.3%) definite, 81 (24.3%) probable and 68 (20.4%) possible. Only definite and probable MI were included. There were 369 first strokes that occurred during follow-up, 322 (87.3%) infarcts, 35 (9.5%) ICHs, 8 (2.2%) SAHs and 4 (1.1%) unknown, and all were included.
Study outcome
The BI [15] (Granger version) measures an individual's performance in 10 activities of daily living (ADLs) and ranges from 0 to 100 in 5-point increments, with 100 indicating normal physical functioning. Previous research has demonstrated the reliability of phone BI assessments [16]. Although it is an ordinal scale, several publications have advocated analysing the scale as a continuous variable due to increased power to detect associations, ability to describe the course of change over time in linear form, and avoidance of potential misclassification due to crude categorisation [17–19].
Explanatory variable: CRP
Blood samples were collected from 2,240 participants at baseline, and CRP levels were determined using a BNII nephelometer (Dade Behring, Deerfield, IL) by technicians blinded to clinical status. Serum CRP samples were drawn into EDTA tubes, spun immediately at 3,000 g at 4 °C for 20 min, and frozen at –70 °C for later analysis.
Since CRP levels were skewed, log transformation was performed. Additionally, CRP values were analysed per standard deviation, and categorised according to the three Centers of Disease Control (CDC)/American Heart Association (AHA) risk stratification levels: <1 mg/l; 1–3 mg/l and >3 mg/l [20].
Covariates
We adjusted for the following variables, measured at baseline: (i) demographics: age, sex, race/ethnicity, (ii) medical risk factors: body mass index (weight in kilograms divided by the square of height in metres), self-reported hypercholesterolaemia, diabetes mellitus (defined by self-report, fasting blood glucose level ≥126 mg/dl, or insulin/oral hypoglycaemia use), hypertension (defined as a systolic blood pressure recording ≥140 mmHg or a diastolic blood pressure recording ≥90 mmHg based on the average of two blood pressure measurements or the patient's self-report of a history of hypertension or antihypertensive use), smoking (defined as either non-smoker or smoker within the last year), alcohol use (with moderate alcohol use classified as 1 drink/month to 2 drinks/day), physical activity (defined as any versus none), (iii) social variables: marital status, insurance status (classified as uninsured, Medicaid, Medicare or private insurance), number of friends (individuals whom the participant knows well enough to visit in their homes), years living in the community and (iv) cognitive and mood factors: depression (defined as a Hamilton Depression scale score of ≥12), performance on mini-mental state examination (analysed as a continuous variable) and Spitzer quality of life scale score [21].
Statistical analysis
The goal of this analysis was to determine whether serum CRP levels were associated with baseline BI and a steeper slope of decline over time. We first calculated distributions of explanatory variables, baseline covariates and BI, and we compared distributions of variables based on the availability of CRP data to detect differences in the inflammatory lab cohort compared with the full NOMAS prospective cohort. In order to assess for selection bias in functional status related to availability of CRP measurements, we compared the time trend of BI in the cohort with CRP measurements to the time trend in the cohort without CRP measurements, in unadjusted and fully adjusted models, fitting a model with a cohort by time interaction.
Due to correlations among repeated measures of outcomes in the same individual, regression models based upon generalised estimating equations (GEE) with an identity link function were used to assess the association between CRP levels and repeated measurements of BI, adjusting successively for baseline demographic variables, medical risk factors, social variables and cognitive/mood factors.
In order to assess whether CRP levels were associated with change in outcomes over time, we included interaction terms between time of follow-up assessment and CRP. We also pursued an alternative modelling strategy using mixed models. We used QIC for GEE models and AIC for mixed models as model selection criteria. Various model diagnostics including tests of linearity, residual plots and goodness of fit measures were used to evaluate the final model. As a working correlation structure for the GEE models, we chose the exchangeable (intraclass) structure and compared the QIC obtained with this model with one using the unstructured working correlation structure. In order to assess whether interval vascular events such as clinical stroke and MI were implicated in the trajectory of functional status, we ran a second set of models in which stroke and MI were included as time-varying covariates.
In a sensitivity analysis, we examined whether the relationship between CRP levels and functional status differed for mobility (transfers, mobility and stair use) and non-mobility (feeding, bathing, grooming, dressing, bowels, bladder and toilet use) BI domains. We also performed three other sensitivity analyses, but models did not significantly differ from the primary analyses and are not presented here: (i) adding possible MI to the MI definition; (ii) excluding those with baseline coronary artery disease and (iii) excluding those with baseline BI <100 (n = 761).
Results
Table 1 shows distributions of variables stratified by availability of CRP. There were significant differences in the distributions of several variables based on the availability of CRP, but there were no differences in major vascular risk factors. There were a total of 25,924 BI assessments. There were no differences in the time trend of BI when the cohort with CRP measurements was compared with the cohort without CRP measurements, in unadjusted and fully adjusted models, suggesting no significant selection bias. The mean CRP level was 5.24 mg/l (median = 2.55, interquartile range = 4.82 and standard deviation = 8.86). Log-transformed CRP levels followed a normal distribution. In terms of AHA/CDC CRP categories, 524 (23.4%) had levels <1 mg/l, 710 (31.7%) 1–3 mg/l and 1,006 (44.9%) >3 mg/l. There was no evidence to suggest lack of linearity in the final models. We chose the exchangeable model, with the lower QIC, as the final model.
Table 1.
Baseline characteristics of the cohort, by availability of CRP
| Variable | Cohort with CRP | Cohort without CRP | P-value |
|---|---|---|---|
| Number of participants, No. (%) | 2,240 (67.9) | 1,058 (32.1) | -- |
| Biological characteristics | |||
| Age, mean (SD), years | 69.4 (10.1) | 70.5 (10.6) | 0.005 |
| Body mass index, mean (SD), kg/m2 | 27.9 (5.5) | 27.7 (5.5) | 0.2 |
| Demographics | |||
| Male, No. (%) | 804 (35.9) | 423 (40.0) | 0.02 |
| Race-ethnicity | <0.0001 | ||
| Non-Hispanic white, No. (%) | 420 (18.8) | 270 (18.8) | |
| Non-Hispanic black, No. (%) | 526 (23.5) | 277 (26.2) | |
| Hispanic, No. (%) | 1,237 (55.2) | 491 (46.4) | |
| Other, No. (%) | 57 (2.5) | 20 (1.9) | |
| Received at least high school education, No. (%) | 996 (44.5) | 515 (48.7) | 0.02 |
| Marital status, No. (%) married | 721 (32.2) | 321 (30.3) | 0.3 |
| Health insurance, No. (%) | <0.0001 | ||
| Medicaid or no insurance | 1,028 (46.2) | 407 (38.7) | |
| Medicare or private insurance | 1,197 (53.8) | 644 (61.3) | |
| Vascular risk factors, No. (%) | |||
| Hypertension | 1,671 (74.6) | 758 (71.6) | 0.07 |
| Alcohol consumption | 0.5 | ||
| Never drank | 538 (24.0) | 283 (26.8) | |
| Past drinker | 553 (24.7) | 246 (23.3) | |
| Light drinker | 293 (13.1) | 128 (12.1) | |
| Moderate drinker | 745 (33.3) | 341 (32.2) | |
| Intermediate drinker | 77 (3.4) | 43 (4.1) | |
| Heavy drinker | 34 (1.5) | 17 (1.6) | |
| Physical activity, any | 1,226 (54.7) | 683 (64.6) | <0.0001 |
| Diabetes mellitus | 479 (21.4) | 237 (22.4) | 0.5 |
| Smoking | 0.3 | ||
| Never | 1,070 (47.8) | 475 (44.9) | |
| Former | 829 (37.0) | 420 (39.7) | |
| Current | 339 (15.2) | 163 (15.4) | |
| Hypercholesterolaemia | 1,414 (63.1) | 636 (60.1) | 0.1 |
| History of atrial fibrillation | 100 (4.5) | 43 (4.1) | 0.6 |
| History of coronary heart disease | 475 (21.2) | 229 (21.6) | 0.8 |
| Spitzer quality of scale score, mean (SD) | 9.1 (1.3) | 9.1 (1.3) | 0.8 |
| Other medical conditions, No. (%) | |||
| Hamilton depression scale score, mean (SD) | 3.3 (3.8) | 3.0 (3.9) | 0.04 |
| Hamilton depression score ≥12 | 90 (4.1) | 49 (4.7) | 0.4 |
| Mini-mental state score, mean (SD) | 25.9 (3.8) | 26.2 (3.6) | 0.02 |
| Social variables, No. (%) | |||
| Number of people known well enough to visit with in their homes | 0.5 | ||
| None | 85 (3.8) | 45 (4.3) | |
| 1 or 2 | 238 (10.6) | 129 (12.2) | |
| 3 or 4 | 443 (19.8) | 210 (19.9) | |
| 5 or more | 1,472 (65.8) | 673 (63.7) | |
| Years lived in community | 28.5 (16.3) | 31.4 (17.2) | <0.0001 |
Table 2 shows associations between logCRP levels and trajectories of functional status, in unadjusted and adjusted models. In most models, there was a significant annual decline in functional status of ~1 BI point per year, and of 0.39 points per year in a model adjusted for stroke and MI occurring during follow-up. In addition, in most models, including the fully adjusted model, higher logCRP levels were associated with a lower baseline mean functional score (−0.34 BI points per unit logCRP, 95% confidence interval (CI) −0.6, −0.06). The largest drop in the magnitude of this estimate occurred when vascular risk factors were added to the model. Finally, logCRP levels were not significantly associated with change in functional status over time. Similarly to the GEE models, using adjusted mixed models, increasing logCRP levels were associated with lower overall functional status (−0.41, 95% CI −0.82, 0.002; P = 0.051) but no change in function over time (0.05 per year, 95% CI −0.06, 0.16; P = 0.4).
Table 2.
Associations between log of CRP levels and trajectories of functional status
| Variable | Change in BI score | 95% CI | P-value |
|---|---|---|---|
| Unadjusted model | |||
| Annual change in BI score | −1.01 | −1.09, −0.93 | <0.0001 |
| Change in BI score per unit increase in log of CRP levels | −0.72 | −1.13, −0.31 | 0.0006 |
| Additional annual change in BI score per unit increase in log of CRP levels | −0.03 | −0.10, 0.05 | 0.5 |
| Adjusted for demographicsa | |||
| Annual change in BI score | −1.02 | −1.10, −0.94 | <0.0001 |
| Change in BI score per unit increase in log of CRP levels | −0.79 | −1.22, −0.36 | 0.0003 |
| Additional annual change in BI score per unit increase in log of CRP levels | −0.03 | −0.10, 0.04 | 0.4 |
| Adjusted for vascular risk factorsb | |||
| Annual change in BI score | −1.03 | −1.11, −0.94 | <0.0001 |
| Change in BI score per unit increase in log of CRP levels | −0.40 | −0.85, 0.05 | 0.08 |
| Additional annual change in BI score per unit increase in log of CRP levels | −0.03 | −0.11, 0.04 | 0.4 |
| Adjusted for social variablesc | |||
| Annual change in BI score | −1.01 | −1.09, −0.93 | <0.0001 |
| Change in BI score per unit increase in log of CRP levels | −0.33 | −0.78, 0.11 | 0.14 |
| Additional annual change in BI score per unit increase in log of CRP levels | −0.03 | −0.11, 0.04 | 0.4 |
| Adjusted for mood and cognitive variablesd | |||
| Annual change in BI score | −0.57 | −0.72, −0.43 | <0.0001 |
| Change in BI score per unit increase in log of CRP levels | −0.30 | −0.58, −0.02 | 0.04 |
| Additional annual change in BI score per unit increase in log of CRP levels | 0.00 | −0.12, 0.13 | 0.96 |
| Adjusted for stroke and MIe | |||
| Annual change in BI score | −0.39 | −0.53, −0.26 | <0.0001 |
| Change in BI score per unit increase in log of CRP levels | −0.34 | −0.62, −0.06 | 0.02 |
| Additional annual change in BI score per unit increase in log of CRP levels | 0.02 | −0.11, 0.14 | 0.8 |
aAdjusted for: baseline age, sex and race-ethnicity.
bAdditionally adjusted for: diabetes, hypertension, coronary artery disease, hypercholesterolaemia, physical activity, alcohol use, smoking and body mass index.
cAdditionally adjusted for: marital status, insurance, number of friends and years lived in the neighbourhood.
dAdditionally adjusted for: depression, mini-mental state score and Spitzer quality of life index.
eAdditionally adjusted for: stroke or myocardial infarction occurring during follow-up.
Patterns of association were similar when standardised CRP levels were tested as the main predictor—namely, a significant overall annual decline in function in the overall cohort (−0.39 BI points per year, 95% CI −0.52, −0.25), a lower baseline functional level with higher CRP levels (−0.41 BI points per SD increase in CRP levels, 95% CI −0.82, −0.002) and no significant association between CRP levels and change in functional status over time (0.04 BI points per SD increase in CRP levels, 95% CI −0.06, 0.15). When the AHA/CDC categorisation of CRP levels was used, similar trends were seen, with reduced power from the use of categorical instead of continuous variables.
In domain analysis, logCRP levels were significantly associated with baseline functional status in mobility (−0.15, 95% CI −0.30, −0.01, P = 0.037) and non-mobility domains (−0.18, 95% CI −0.35, −0.01, P = 0.04), in fully adjusted models. There were no associations with BI change over time.
Discussion
NOMAS is a large, population-based study with frequent, regular measurements of functional status, and it is well suited to estimate functional trajectories and identify the key influencing factors. Among NOMAS participants with inflammatory biomarker data at baseline, the prevalence of elevated CRP levels was high. There was an overall decline in functional status over time in the entire cohort. Higher CRP levels, examined using different cutoffs and variable definitions, were associated with lower baseline functional score but not with change in slope of function over time. For each unit increase in logCRP levels, there was a decrease in baseline functional score equivalent to the decline in function associated with a year of ageing. In addition to GEE models, we also used mixed models to estimate associations, which confirmed the significance and magnitude of effects seen with GEE models. Also, we adjusted for the occurrence of stroke and MI during follow-up, and the relationship of CRP to functional status was independent of these events. When we examined mobility and non-mobility domains of the BI separately, we found associations for both domains. In addition, there was evidence for vascular functional impairment, with a significant reduction in the magnitude of associations when vascular risk factors were adjusted for. However, the associations between CRP levels and functional status remained even after this adjustment.
Prior studies have examined the association between CRP and vascular outcomes, mortality, disability and frailty. In a meta-analysis of 160,309 individuals without vascular disease from 54 prospective studies [2], higher logCRP levels were independently associated with coronary heart disease, ischaemic stroke and vascular and non-vascular death. Among 807 consecutively admitted ICH patients in Finland [22], elevated CRP levels at admission were associated with worse modified Rankin scores at 3 months. In contrast, we examined functional status over a longer time period. In a population-based sample of 1255 individuals [23], inflammation (including CRP) partially mediated the relationship between risk factors and disability. In a cross-sectional analysis using 10 years of data among 1,729 adults with diabetes [24], elevated CRP was independently associated with disability and lower extremity mobility. In CHS, higher CRP levels were associated with incident frailty up to 9 years of follow-up [25]. Other studies have shown variable associations between CRP and frailty but have not examined trajectories of functional status over time.[26, 27]
In the only study prior to ours that examined change of function over time, among 624 individuals ≥70 years of age [28], elevated CRP was associated with decreased mobility in the entire cohort, as well as incident disability and gait speed decline among those without vascular disease, over a median of 2 years of follow-up. In contrast, we adjusted for vascular events occurring during follow-up, had longer follow-up and examined a larger cohort both with and without vascular disease.
There are several possible mechanisms that could explain the association between CRP and disability seen in this study. CRP may be a measure of subclinical cerebrovascular disease that is responsible for the decreased functional status [29]. Or, CRP may be a marker of a diffuse vascular disease burden that contributes to decreased functional status, i.e. diffuse atherosclerosis with impaired perfusion of all vessels in the body. Alternatively, CRP may be measuring frailty and not necessarily a specific vascular process; indeed, CRP has been associated with non-vascular mortality and non-vascular conditions [2]. Finally, the findings may have been due to residual confounding; elevated CRP levels may be an epiphenomenon of other risk factors that are incompletely measured but are responsible for disability, such as diabetes, which has varying severities and has been previously associated with trajectories of disability [30].
Strengths of this analysis include a large, population-based cohort with repeated, regular measurements of functional status with a validated scale over long-term follow-up. There was also regular surveillance for vascular events and hospitalisation and expert adjudication of events. One limitation of the proposed study involves deficiencies in the primary outcome measure. The BI, as a measure of ADLs and not additionally instrumental ADLs, is subject to ceiling effects and is insensitive to small changes in disability [31]. Analysing the BI as a continuous variable can capture and quantify the variance and course of change over time [18, 19]. Multiple prior studies have demonstrated associations between CRP and vascular outcomes, mortality, and disability measured at single time points. However, the current analysis is the only one, among studies of inflammatory biomarkers, in which both baseline functional status and the trajectory of change over time were analysed. Most prior studies examining disability have focused on single measurements at discrete time points, which ignores ongoing decline in functional status that can be accelerated by various factors [30, 32].
Key points.
C-reactive protein is associated with disability.
This effect is independent of vascular events.
This effect is seen over a long follow-up.
Supplementary Material
Acknowledgements
Dr Elkind receives compensation for providing consultative services for Biotelemetry/Cardionet, BMS-Pfizer Partnership, Boehringer-Ingelheim and Sanofi-Regeneron; serves as an expert witness for Merck/Organon (Nuvaring and stroke), BMS-Sanofi Partnership (Plavix and stroke) and Hi-Tech (DMAA and stroke); serves on the National, Founders Affiliate and New York City chapter boards of the American Heart Association/American Stroke Association; and receives royalties from UpToDate for chapters related to cryptogenic stroke.
Conflicts of interest
None declared.
Funding
This work was supported by grants from the National Institute of Neurological Disorders and Stroke (R01 NS48134, M.S.V.E.; R37 29993, R.L.S./M.S.V.E.; K23NS079422, M.S.D.).
References
- 1. Koenig W. High-sensitivity c-reactive protein and atherosclerotic disease: from improved risk prediction to risk-guided therapy. Int J Cardiol 2013; 168: 5126–34. [DOI] [PubMed] [Google Scholar]
- 2. Kaptoge S, Di Angelantonio E, Lowe G et al. C-reactive protein concentration and risk of coronary heart disease, stroke, and mortality: an individual participant meta-analysis. Lancet 2010; 375: 132–40. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Elkind MSV, Moon YP, Cespedes S et al. High-sensitivity c-reactive protein and risk of stroke, vascular events, and mortality in a multi-ethnic cohort: the Northern Manhattan Study. Neurology 2008; 70(Suppl 1):A399. [Google Scholar]
- 4. Elkind MS, Luna JM, Moon YP et al. High-sensitivity c-reactive protein predicts mortality but not stroke: the Northern Manhattan Study. Neurology 2009; 73: 1300–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Fornage M, Chiang YA, O'Meara ES, Psaty BM, Reiner AP, Siscovick DS et al. Biomarkers of inflammation and MRI-defined small vessel disease of the brain: the Cardiovascular Health Study. Stroke 2008; 39: 1952–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Idicula TT, Brogger J, Naess H, Waje-Andreassen U, Thomassen L. Admission c-reactive protein after acute ischemic stroke is associated with stroke severity and mortality: the ‘Bergen Stroke Study’. BMC Neurol 2009; 9: 18. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Cummings DM, King DE, Mainous AG III. C-reactive protein, antiinflammatory drugs, and quality of life in diabetes. Ann Pharmacother 2003; 37: 1593–7. [DOI] [PubMed] [Google Scholar]
- 8. Unden AL, Andreasson A, Elofsson S et al. Inflammatory cytokines, behaviour and age as determinants of self-rated health in women. Clin Sci (Lond) 2007; 112: 363–73. [DOI] [PubMed] [Google Scholar]
- 9. Elkind MS, Sciacca R, Boden-Albala B, Rundek T, Paik MC, Sacco RL. Moderate alcohol consumption reduces risk of ischemic stroke: the Northern Manhattan Study. Stroke 2006; 37: 13–9. [DOI] [PubMed] [Google Scholar]
- 10. Willey JZ, Paik MC, Sacco R, Elkind MS, Boden-Albala B. Social determinants of physical inactivity in the Northern Manhattan Study (NOMAS). J Community Health 2010; 35: 602–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Sacco RL, Anand K, Lee HS et al. Homocysteine and the risk of ischemic stroke in a triethnic cohort: the Northern Manhattan Study. Stroke 2004; 35: 2263–9. [DOI] [PubMed] [Google Scholar]
- 12. Gentry EM, Kalsbeek WD, Hogelin GC et al. The behavioral risk factor surveys: II. Design, methods, and estimates from combined state data. Am J Prev Med 1985; 1: 9–14. [PubMed] [Google Scholar]
- 13. Echt DS, Liebson PR, Mitchell LB et al. Mortality and morbidity in patients receiving encainide, flecainide, or placebo. The cardiac arrhythmia suppression trial. N Engl J Med 1991; 324: 781–8. [DOI] [PubMed] [Google Scholar]
- 14. Schaefer EJ, Lamon-Fava S, Jenner JL et al. Lipoprotein(a) levels and risk of coronary heart disease in men. The lipid research clinics coronary primary prevention trial. JAMA 1994; 271: 999–1003. [DOI] [PubMed] [Google Scholar]
- 15. Granger CV, Dewis LS, Peters NC, Sherwood CC, Barrett JE. Stroke rehabilitation: analysis of repeated barthel index measures. Arch Phys Med Rehabil 1979; 60: 14–7. [PubMed] [Google Scholar]
- 16. Shinar D, Gross CR, Bronstein KS et al. Reliability of the activities of daily living scale and its use in telephone interview. Arch Phys Med Rehabil 1987; 68: 723–8. [PubMed] [Google Scholar]
- 17. Bath PM, Gray LJ, Collier T, Pocock S, Carpenter J. Can we improve the statistical analysis of stroke trials? Statistical reanalysis of functional outcomes in stroke trials. Stroke 2007; 38: 1911–5. [DOI] [PubMed] [Google Scholar]
- 18. Saver JL. Optimal end points for acute stroke therapy trials: Best ways to measure treatment effects of drugs and devices. Stroke 2011; 42: 2356–62. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Song F, Jerosch-Herold C, Holland R, Drachler Mde L, Mares K, Harvey I. Statistical methods for analysing barthel scores in trials of poststroke interventions: a review and computer simulations. Clin Rehabil 2006; 20: 347–56. [DOI] [PubMed] [Google Scholar]
- 20. Pearson TA, Mensah GA, Alexander RW et al. Markers of inflammation and cardiovascular disease: application to clinical and public health practice: a statement for healthcare professionals from the Centers for Disease Control and Prevention and the American Heart Association. Circulation 2003; 107: 499–511. [DOI] [PubMed] [Google Scholar]
- 21. Spitzer WO, Dobson AJ, Hall J et al. Measuring the quality of life of cancer patients: a concise ql-index for use by physicians. J Chronic Dis 1981; 34: 585–97. [DOI] [PubMed] [Google Scholar]
- 22. Lopponen P, Qian C, Tetri S et al. Predictive value of c-reactive protein for the outcome after primary intracerebral hemorrhage. J Neurosurg 2014; 121: 1374–9. [DOI] [PubMed] [Google Scholar]
- 23. Friedman EM, Christ SL, Mroczek DK. Inflammation partially mediates the association of multimorbidity and functional limitations in a national sample of middle-aged and older adults: the Midus Study. J Aging Health 2015; 27: 843–63. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Kuo HK, Al Snih S, Kuo YF, Raji MA. Cross-sectional associations of albuminuria and c-reactive protein with functional disability in older adults with diabetes. Diabetes Care 2011; 34: 710–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Barzilay JI, Blaum C, Moore T et al. Insulin resistance and inflammation as precursors of frailty: The cardiovascular health study. Arch Intern Med 2007; 167: 635–41. [DOI] [PubMed] [Google Scholar]
- 26. Fernandez-Garrido J, Ruiz-Ros V, Buigues C, Navarro-Martinez R, Cauli O. Clinical features of prefrail older individuals and emerging peripheral biomarkers: a systematic review. Arch Gerontol Geriatr 2014; 59: 7–17. [DOI] [PubMed] [Google Scholar]
- 27. Arts MH, Collard RM, Comijs HC et al. Relationship between physical frailty and low-grade inflammation in late-life depression. J Am Geriatr Soc 2015; 63: 1652–7. [DOI] [PubMed] [Google Scholar]
- 28. Verghese J, Holtzer R, Lipton RB, Wang C. High-sensitivity c-reactive protein and mobility disability in older adults. Age Ageing 2012; 41: 541–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Wright CB, Moon Y, Paik MC et al. Inflammatory biomarkers of vascular risk as correlates of leukoariosis. Stroke 2009; 40: 3466–71. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Dhamoon MS, Moon YP, Paik MC, Sacco RL, Elkind MS. Diabetes predicts long-term disability in an elderly urban cohort: the Northern Manhattan Study. Ann Epidemiol 2014; 24: 362–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Dromerick AW, Edwards DF, Diringer MN. Sensitivity to changes in disability after stroke: a comparison of four scales useful in clinical trials. J Rehabil Res Dev 2003; 40: 1–8. [DOI] [PubMed] [Google Scholar]
- 32. Dhamoon MS, Moon YP, Paik MC, Sacco RL, Elkind MS. Trajectory of functional decline before and after ischemic stroke: the Northern Manhattan Study. Stroke 2012; 43: 2180–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.