The BioImage Study: Novel approaches to risk assessment in the primary prevention of atherosclerotic cardiovascular disease—study design and objectives

Article Outline

• Abstract
• Methods
  • BioImage Study
  • Study objectives
  • Targeted at-risk population
  • Recruitment population stratified by claims
  • Study population
  • Generalization of study results
  • Baseline examination
  • Notification of baseline findings
  • Surveillance and event detection
  • Statistical analysis and power considerations
• Discussion
  • Limitations
• Conclusions
• Disclosures
• Appendix. 
• References
• Copyright

The identification of asymptomatic individuals at risk for near-term atherothrombotic events to ensure optimal preventive treatment remains a challenging goal. In the BioImage Study, novel approaches are tested in a typical health-plan population. Based on certain demographic and risk characteristics on file with Humana Inc, a total of 7,687 men 55 to 80 years of age and women 60 to 80 years of age without evidence of atherothrombotic disease but presumed to be at risk for near-term atherothrombotic events were enrolled between January 2008 and June 2009. Those who met the prespecified eligibility criteria were randomized to a telephonic health survey only (survey only: n = 865), standard risk assessment (Framingham only: n = 718), or comprehensive risk assessment in a dedicated mobile facility equipped with advanced imaging tools (n = 6,104). Baseline examination included assessment of cardiovascular risk factors and screening for subclinical (asymptomatic) atherosclerosis with quantification of coronary artery calcification by computed tomography (CT), measurement of intima-media thickness, presence of carotid atherosclerotic plaques and abdominal aortic aneurysm by ultrasound, and ankle brachial index. Participants with one or more abnormal screening test results underwent advanced imaging with contrast-enhanced magnetic resonance imaging for carotid and aortic plaques, contrast-enhanced coronary CT angiography for luminal stenosis and noncalcified plaques, and 18F-fluorodeoxyglucose–positron emission tomography/CT for carotid and aortic plaque inflammation. Plasma, PAXgene RNA, and DNA samples were obtained, frozen, and stored for future biomarker discovery studies. All individuals will be followed until 600 major atherothrombotic events have occurred in those undergoing imaging. The BioImage Study will help identify those patients with subclinical atherosclerosis who are at risk for near-term atherothrombotic events and enable a more personalized management of care.

Atherosclerosis is a major cause of death and disability in industrialized and, increasingly, in developing countries and, as such, has a large economic and public health impact.¹ Current primary prevention strategies rely almost exclusively on risk factor recognition and management.², ³ The importance of modifiable risk factors cannot be overstated; they account for most of the risk of atherosclerotic cardiovascular disease worldwide.⁴ A major problem is, however, that most people destined for a myocardial infarction or stroke are unaware of their risk because their traditional risk factor levels are not unusually high.⁵ Conversely, many individuals with an apparently adverse risk factor profile remain asymptomatic.

Most myocardial infarctions and strokes occur in individuals who would be classified as low or intermediate risk by the traditional risk factor–based approach, such as the Framingham Risk Score.², ³ For example, in the Framingham Heart Study,⁶ the Physicians' Health Study,⁷ the Women's Health Study,⁸ and the Northwick Park Heart Study,⁹ >75% of all hard coronary events occurred in people classified at low or intermediate risk and, consequently, not offered optimal preventive therapy.

An alternative strategy of potentially greater impact would be the detection of subclinical (asymptomatic) atherosclerosis and, in particular, the type of atherosclerotic lesion known as “vulnerable” or “high-risk” plaque that is associated with a high risk of thrombosis.¹⁰ In recent years, studies using advanced technologies have provided insights into atherosclerotic plaque development and its progression to vulnerable plaque and rupture as a cause for acute atherothrombosis leading to myocardial infarction or stroke.¹¹, ¹², ¹³, ¹⁴, ¹⁵, ¹⁶, ¹⁷ These insights are providing impetus for the discovery and development of novel screening and diagnostic tools and potential therapies.

Identifying subclinical atherosclerosis should be considered complementary to conventional risk factor assessment. When a high-risk person is identified, the risk factor burden needs to be reduced to the extent possible. Yet, identifying subclinical atherosclerosis and its treatment have the potential to surpass conventional risk factor assessment and management in terms of overall impact on cardiovascular morbidity and mortality.

Back to Article Outline

Methods 

BioImage Study 

The BioImage Study (NCT00738725) is a prospective, observational study designed to evaluate associations among imaging and circulating biomarkers (cross-sectional) and their ability to predict atherothrombotic events (longitudinal) in asymptomatic at-risk subjects in the primary prevention of myocardial infarction and stroke. This study is part of the High-Risk Plaque (HRP) Initiative (www.hrpinitiative.com), an industry-supported program of research studies aimed at advancing the understanding, recognition, and management of atherosclerotic disease. In particular, the Initiative aims to identify and validate novel imaging and blood biomarkers of asymptomatic but high-risk atherosclerosis that could be used to spur the development of in vitro diagnostic tests, medical imaging devices, and therapeutic products for the more effective identification, assessment, and treatment of patients at high near-term risk of major atherothrombotic events.

In collaboration with Humana, a large US health benefits company, the BioImage Study used a unique and innovative approach to identify and recruit a typical at-risk population. In contrast to other large-scale epidemiologic studies that used either a network of academic research centers¹⁸ or a well-defined population in a limited geographic area,⁶ the BioImage Study has recruited subjects from 2 geographic locations intended to represent the US population-at-large. In each location, we established a dedicated temporary research facility with mobile imaging equipment. The mobile equipment and dedicated staff moved from one location to the next, facilitating consistency of image acquisition.

Age and gender remain the most meaningful discriminators of those at risk from those not at risk.¹⁹ Hence, in designing the BioImage Study, we considered all subjects over a certain age without known cardiovascular disease as potential candidates. We included currently available noninvasive modalities for vascular imaging, such as ultrasound (US), computed tomography (CT), magnetic resonance imaging (MRI), and positron emission tomography (PET). Certain modalities and examinations are limited to certain subsets, whereas others are used on the overall study population.

The BioImage Study and method of selection and recruitment were approved by the Western Institutional Review Board, Olympia, WA. All study participants provided written informed consent and Health Insurance Portability and Accountability Act authorization before enrollment.

Study objectives 

The primary objective of the BioImage Study is to identify imaging biomarkers that predict near-term (3-year) atherothrombotic events, with incremental improvement over traditional risk assessment (Framingham Risk Score). The secondary objectives of the study are to (1) correlate imaging data of subclinical atherosclerosis and blood-based markers, (2) provide standardized high-quality image sets for the development and refinement of quantitative digital image analysis, (3) provide standardized high-quality image sets for the optimization of image protocols and correction algorithms, (4) compare event rates in patients randomized to undergo imaging studies with those in patients who do not undergo such studies, (5) collect and store biological specimens for future evaluation, (6) postulate a novel screening and diagnostic pathway based on the discovered new risk markers, and (7) acquire pertinent data for the development of an economic and health impact model in the primary prevention of cardiovascular disease.

Targeted at-risk population 

We wanted to enrich the study population with individuals with a meaningful probability of developing events in the near term. To do so, we identified members in the Humana database who were 55 to 80 years of age (men >55 years and women >60 years), as this is the group where most cardiovascular events occur and where many were expected to have at least one additional risk factor.¹⁹ Study participants were recruited among members of the Humana Health Plan (>11 million members nationwide) with a male-female ratio of 1:1 and a racial/ethnic distribution corresponding to US Census data (approximately 69% white, 12% African American, 13% Hispanic, 4% Asian, and 2% other).²⁰ To ensure a diverse study population, enrollment occurred in 2 different cities: Chicago (IL) and Ft Lauderdale (FL).

Recruitment population stratified by claims 

Preliminary observations indicate it is possible to risk-stratify Humana members using information on file, including medical and pharmacy health insurance claims (Figure 1). Candidates had to be free of claims-based evidence of prior major cardiovascular disease, active cancer treatment, or certain other claims-based indicators of major intercurrent disease. We used medical and pharmacy claims to create claims-based risk indicators of established cardiovascular risk, such as medical claims or prescription medication use for hypertension, diabetes, or hyperlipidemia. One of the limitations of claims-based risk assessment is that certain risk factors, such as smoking and physical inactivity, cannot be determined or inferred from claims. The most common claims-based risk indicators were hypertension (65%) and hyperlipidemia (57%). The majority (57%) had ≥2 claims-based risk indicators (Supplementary Appendix-Table I).

• 
  • View Large Image
  • Download to PowerPoint

• Figure 1. 

  BioImage Study flowchart.

We used medical claims for acute myocardial infarction and stroke to identify putative major cardiovascular events. The relationships between claims-based risk indicators and claims-based event rates are significant (P < .001) (Figure 2, A). With the exception of hyperlipidemia, the presence or absence of specific claims-based risk indicators is also strongly related with claims-based event rates (P < .001) (Figure 2, B). Thus, despite the limitations of claims data, the concept of using claims-based risk indicators to estimate the average risk of a study population appears to be valid in individuals in the age range of 55 to 80 years without prior history of cardiovascular disease.

• 
  • View Large Image
  • Download to PowerPoint

• Figure 2. 

  Putative relative event rates (A) by age, gender, and number of claims-based risk indicators and (B) by age in relation to presence or absence of specific claims-based risk indicators. Data are based on 2-year follow-up in subjects from population described in Table I. The event rate for the total population =1.00. Pearson χ² tests for each age group, gender, and indicator (except hyperlipidemia) were significant at P < .001.

Two historical cohorts of Humana members unrelated to the BioImage Study population are being identified and will be used to evaluate claims-based monitoring of event triggers and cardiovascular event rates. One cohort (control 1: n = 12,208) will include a random sample of Humana members that meet the selection criteria for the BioImage Study except for the absence of claims-based evidence of major prior cardiovascular disease. The other cohort (control 2: n = 12,208) will match those included in the BioImage Study, including the claims-based evidence criteria. These cohorts will serve as controls for analyses of any bias introduced by the selection criteria (control 1) or the recruitment process and participation in the baseline visit (control 2).

Study population 

Potential study participants were contacted by phone and interviewed to evaluate eligibility, obtain verbal informed consent and authorization, and obtain selected personal and health information. Of those meeting the entrance criteria (Table I), approximately 750 were randomly selected to set aside as “survey-only group.” All other individuals meeting the entrance criteria were invited to participate in the main study. Those who agreed to participate were randomized into (1) Framingham-only group (target n = 750) and (2) the full study group (target n = 6,000), which includes imaging at the baseline visit and eligibility of certain of these individuals for advanced imaging studies. The latter 2 groups were invited to undergo baseline examination in a dedicated mobile facility.

Table I. Eligibility for the BioImage Study: inclusion and exclusion criteria

COPD, Chronic obstructive pulmonary disease.

As part of the informed consent procedure, participants granted authorization and consent to access their medical-pharmacy claims and related financial information. Pharmacy claims will be used to construct the medication history. In addition, the participants granted consent and authorization to verify outcomes on the basis of medical chart review at the facility where such care was provided.

At completion of enrollment in June of 2009, a total of 24,149 Humana members had been surveyed by telephone or mail, of which 9,866 met the eligibility criteria for the study and agreed to participate. Of this population, 7,687 subjects completed enrollment, of which 865 were enrolled as “survey-only” participants, 718 were enrolled in the Framingham-only group, and 6,104 were enrolled in the full study group (Figure 1). Demographic characteristics of the participants are summarized in Table II of the Supplementary Appendix.

Generalization of study results 

Preliminary data from the Chicago location indicate that the BioImage Study population is representative of the eligible Chicago population with respect to age, gender, business line (commercial vs Medicare), median household income, prevalence of risk factors, clinical conditions, comorbidities, and prescription drug use, as determined based on medical or pharmacy claims (Figure 3, A-C).

• 
  • View Large Image
  • Download to PowerPoint

• Figure 3. 

  Enrolled study population in comparison to the eligible population (Chicago location, preliminary data) for age, gender, income, and risk indicators. A, Distribution of female and male gender among BioImage Study participants (P) by age group in comparison with the pool of eligible members (E) from which they were recruited. B, Household income as determined based on US Census information using census track mapping for BioImage Study participants, those eligible to participate, and those living in the Chicago metropolitan statistical area (Chicago-at-large) mapped to the same zip codes from which the eligible population is drawn. C, Presence of claims-based risk indicators among BioImage Study participants in comparison with the pool of eligible members from which they were recruited.

Baseline examination 

The examination approach is also novel and proactive, bringing the necessary research facilities, experienced staff, and advanced imaging equipments to the study participants rather than the opposite. The mobile facility consists of 2 dedicated trailers, one containing a 64-slice CT scanner and the other, a 3.0T MRI scanner. In addition, a fixed building is used for the other baseline study procedures, including blood sampling and US imaging.

All participants underwent measurement of customary physical parameters, including height, weight, waist-hip ratio, brachial blood pressure, ankle blood pressure, and a 12-lead electrocardiogram. A venous blood sample was processed for the collection of plasma, serum, RNA, and DNA (stored for later biomarker discovery studies) and routine chemistry tests.

The full study (imaging) group was also screened for subclinical atherosclerosis, including assessment of coronary artery calcification score (CACS) by CT, measurement of carotid intima-media thickness (IMT), presence of carotid atherosclerotic plaques and abdominal aortic aneurysm (AAA) by US, and ankle brachial index (ABI) (Figure 1). Subjects who meet one or more predefined criteria for CACS, carotid IMT, presence of carotid plaque and AAA, or ABI were offered advanced imaging for subclinical atherosclerosis (Table II). The choice of advanced imaging method was based on logistical considerations, patient preference, and presence/absence of certain clinical criteria that would render a particular method unsuitable or contraindicated. The advanced imaging methods included contrast-enhanced or noncontrast MRI for carotid and aortic plaques, contrast-enhanced coronary CT angiography for noncalcified plaques and stenoses, and 18F-fluorodeoxyglucose PET/CT for carotid and aortic inflammation.

Table II. Criteria for advanced imaging substudy eligibility

Notification of baseline findings 

Study participants were informed that these study procedures were conducted for research purposes only and that only those obvious findings requiring immediate medical attention would be reported back to the participant and their physicians.

Surveillance and event detection 

The study population will be followed until 600 major atherothrombotic events have occurred in the imaging group (Figure 1). Primary end points for the study include fatal and nonfatal myocardial infarction (excluding procedure-related), coronary death (excluding procedure-related), hospitalization for unstable angina (angina at rest with documented electrocardiographic changes), ischemic stroke (fatal and nonfatal), and arterial revascularization procedures (either percutaneous or surgical). Potential cardiovascular events will be identified by querying the Humana claims and member databases for predefined event triggers at regular intervals throughout the follow-up period. The claims monitoring for event triggers will be conducted using Diagnosis-Related Groups (DRG), International Classification of Diseases (ICD)-9, and Current Procedure Terminology (CPT) codes for myocardial infarction, cerebrovascular events, unstable angina, peripheral vascular disease, revascularization procedures, and all deaths. Use of claims information for these conditions has previously been validated in numerous peer-reviewed studies comparing administrative claims information with clinical information obtained through chart reviews.²¹, ²² Any study participants leaving the Humana health plan before the end of the follow-up period will be followed by telephone contact only. The study participants randomized to the “telephone survey-only” group and the Humana members included in the 2 control cohorts will also be followed through the evaluation of claims and member data, as described above.

Statistical analysis and power considerations 

To address the primary objective of the study, a number of statistical analyses will be conducted, including proportional-hazard regression, discrimination evaluated by the c statistic, and net reclassification improvement. A multivariable Cox proportional-hazards model will be used to evaluate the association of biomarker levels with the risk of cardiovascular events, upon confirmation of the assumption of proportional hazard. Regression models will be adjusted for the Framingham risk score covariates, and likelihood ratio tests will be used to determine a multivariable P value. To address multiple hypotheses testing, statistical significance will account for false discovery rates.²³ Backward elimination will be used to select a parsimonious set of biomarkers. In addition, the discriminatory ability of biomarkers and conventional risk factors will be evaluated using the c statistic, as previously described.²⁴ Receiver operating characteristic curves will be plotted for models with blood and imaging biomarkers and for those without biomarkers based on follow-up data. The effect of biomarkers on risk reclassification of participants will be evaluated using net reclassification improvement assessment approaches, as described previously.⁶, ²⁵

The number of participants required in the full study cohort is based upon the primary objective of identifying biomarkers for prediction of atherothrombotic events during the observation period with improvement over traditional risk factors. In the current study, with a 30% inability to verify some of the end points and a 10% proportion of patients characterized by an unfavorable biomarker value, it is calculated that 6,000 participants will yield 90% power to detect a hazard ratio of 1.50 associated with a novel biomarker at a significance level of .05 within a proportional-hazards regression framework. Calculations were made using R software, version 2.6.

To address the secondary objectives of the study, a number of statistical analyses are appropriate, including descriptive statistics, multivariable regression, and parametric and nonparametric correlation analyses for associating imaging data with blood-based marker concentration, and survival analyses for comparison of event rates in patients based on imaging randomization group.

Back to Article Outline

Discussion 

More than 200 million people in the United States belong to some form of health insurance plan. The individual's personal information and most health care and pharmacy transactions are known to the plan to adjudicate and pay the claims. Although the potential for doing studies within the health care system has long been recognized, few examples exist where the health care and research chasms have been successfully bridged, allowing health plans to play a critical role in addressing an important medical research question.

The BioImage Study actively recruits volunteer participants from Humana's health plan members, linking their research data with their broader health care data and then prospectively using their health care data to monitor for putative events. This represents a unique and novel approach that affords many advantages:

One of the benefits of this approach is the ability to judge how the participants compare with nonparticipants, including those who were not invited or those who could not be reached or declined to participate. The BioImage Study will be able to answer questions about selection bias as well as whether the study induced health care changes. In particular, the risks of selection bias or study-induced provider or patient behavioral changes are important factors to consider. As an example, the preliminary results from the Chicago location regarding risk factor distribution (Figure 3, C) demonstrate a significant difference in the proportions of individuals with zero risk factors between study participants and eligible subjects (P < .05). This may indicate some selection bias, presumably due to those volunteering being more concerned about their health status than other similar individuals from the general population.

The use of dedicated, temporary, and mobile research facilities allows us to use multiple advanced modalities within a single convenient visit for the participants. This setup also allows for the use of few highly trained and dedicated technicians avoiding operator or equipment variability. The BioImage Study involves a single baseline set of imaging, clinical, and biological measurements in a group of US participants with mostly moderate or high risk. Therefore, conclusions reached in this study will apply to relatively healthy elderly individuals, generally with above-average cardiovascular risk. Recently, the HRP Initiative has been selected to partner with Centro Nacional de Investigaciones Cardiovasculares Carlos III of Madrid, Spain, on a long-term observational study involving 5,000 younger employees (aged 40-55 years) who will have 3 similar imaging studies over a follow-up period of 8 years.

It is difficult to overstate the importance of finding asymptomatic individuals at near-term risk for atherothrombosis. A simple screening blood test with similar performance as brain natriuretic peptide in heart failure or troponin in acute myocardial infarction or, alternatively, an imaging biomarker could enable effective primary prevention in the months or years before the first event and has a potential impact on life expectancy that seems unattainable for any other major medical condition. Only recognition and treatment of at-risk atherosclerosis, not only risk factor assessment, offer the potential to reduce cardiovascular morbidity and mortality from its current level.

Limitations 

One limitation of our study is that only individuals with health benefit coverage with Humana were eligible for enrollment and, consequently, no uninsured individuals were enrolled. When we compared the household incomes of the insured participants from the Chicago area, it mirrored those for the state as a whole (insured and noninsured individuals). However, the uninsured individuals, independent of socioeconomic characteristics, have multiple disadvantages that in themselves are associated with poor health; hence, the findings of this study may not be applicable to uninsured individuals. Another limitation is the absence of traditional physician follow-up and reliance on administrative data to identify primary outcomes. However, and as mentioned in the text, claims-based outcomes have been extensively validated. Finally, we could not randomize people for the advanced imaging studies; and personal preference (dislike for tight spaces, concerns about administration of contrast agents) or personal health concerns (i.e., desire to know as much as possible) could have biased participation.

Back to Article Outline

Conclusions 

The BioImage Study represents a new model that brings important medical research to a health plan population and leverages the health plan data and systems to the benefit of research. This research model is broadly applicable to important medical questions that require study of a large population representing the population-at-large. The BioImage Study has the potential to materially contribute to our understanding of who is at risk for near-term atherothrombosis and enable a new paradigm of early disease recognition and treatment to augment what is potentially achievable through risk factor reduction.

Back to Article Outline

Disclosures 

Pieter Muntendam is an employee of BG Medicine and owns stock options and shares in BG Medicine; and Carol McCall is an employee of Humana Inc, who owns stock options and shares in BG Medicine. Javier Sanz, MD, has no conflicts of interest to disclose. Erling Falk, MD, PhD, and Valentin Fuster, MD, PhD, are co-chairman of the HRP Initiative's Scientific Program Board and received honoraria from the HRP Initiative (administered by BG Medicine).

Funding sources: The study was funded by BG Medicine Inc on behalf of the HRP Initiative. The HRP Initiative is a precompetitive industry collaboration funded by Abbott, AstraZeneca, Merck, Philips, and Takeda.

Back to Article Outline

Appendix. 

HRP Scientific Program Board: Valentin Fuster, MD, PhD (PI); Erling Falk, MD, PhD (Co-PI); Juan José Badimon, PhD; René M. Botnar, PhD; Mat JAP Daemen, MD, PhD; Zahi A. Fayad, PhD (MRI Core Laboratory); Mario Garcia, MD (CT Core Laboratory); Geoffrey S. Ginsburg, MD, PhD; Stanley L. Hazen, MD, PhD; Spencer B. King, MD; Pedro R. Moreno, MD; Børge G. Nordestgaard, MD, DMSc; James H. F. Rudd, MD, PhD (PET Core Laboratory); Predimon K. Shah, MD; Henrik Sillesen, MD, DMSc (Ultrasound Core Laboratory); Antonius F. W. van der Steen, PhD; Magdi H. Yacoub, DSc, FRS; and Chun Yuan, PhD (MRI Core Laboratory).

Clinical Investigators: James T. Bui, MD (Chicago, IL); Christopher J. Chen, MD (South Florida); and Albert Seow, MD (Louisville, KY).

HRP Joint Steering Committee: Pieter Muntendam, MD (BG Medicine); Andrew Plump, MD, PhD (Merck); Joel Raichlen, MD (AstraZeneca); Paul Smit (Philips); James C. Stolzenbach (Abbott); and Richard Urquhart (Takeda).

HRP Program Office: BG Medicine, Inc, 610N Lincoln St, Waltham, MA, 02451 USA.

Table I. Baseline characteristics of the US Humana-covered population meeting age and gender criteria for participation

Population characteristic	Entire population
No. of subjects	407€497
Mean age in years (SD)	68.1 ± 6.5
Female (%)	51%
Age in years (%)
55-69	58%
70-80	42%
Type of insurance (%)
Commercial (HMO or PPO)	23%
Medicare	77%
Prevalence of claims-based risk indicators
Hypertension	65%
Hyperlipidemia	57%
Diabetes	23%
Obesity (claims-coded/inferred)	6%/31%
No. of risk indicators per person⁎
0	18%
1	25%
2	28%
3	19%
4	10%

Data are based on the period of July 1, 2005, through June 30, 2008, and include commercial- and Medicare-insured members with medical and pharmacy benefits and continuous coverage for ≥4 months in the first 12-month period who met the preselection eligibility criteria. Risk-based markers are based on medical claims indicative of the condition or the use of prescription medications predominantly used to treat them. Estimated prevalence of obesity uses Humana's Obesity Inference Model as supplement to the claims-based indicator of obesity. Certain known risk factors, such as smoking and physical inactivity, cannot be determined based on available health plan data.

HMO, Health maintenance organization; PPO, preferred provider organization.

^⁎The risk indicator count uses Humana's Obesity Inference Model.

Table II. Demographic characteristics of the study participants

	Imaging group	Framingham only	Survey only	Total
Total	6104	718	865	7687
Female	3441	384	516	4341
Male	2663	334	349	3346
% Male	43.63%	46.52%	40.35%	43.53%
Age (y), mean ± SD	68.8 ± 6.0	68.7 ± 6.3	70.05 ± 6.0	68.9 ± 6.0
Ethnicity (%)
African American	15.3%	14.5%	22.1%	16.0%
Asian	2.0%	2.1%	2.3%	2.0%
Hispanic	6.1%	4.7%	6.4%	6.0%
Other	2.6%	3.2%	2.7%	2.7%
White	74.0%	75.5%	66.6%	73.3%

Back to Article Outline

References 

1. Woolf SH. A closer look at the economic argument for disease prevention. JAMA. 2009;301:536–538
   • View In Article
   • CrossRef
2. Third report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III) final report. National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III). Circulation. 2002;106:3143–3421Available at: www.nhlbi.nih.gov/guidelines/cholesterol/atp3full.pdf
   • View In Article
3. Grundy SM, Cleeman JI, Merz CN, et al. Implications of recent clinical trials for the National Cholesterol Education Program Adult Treatment Panel III guidelines. Circulation. 2004;110:227–239Available at: www.nhlbi.nih.gov/guidelines/cholesterol/atp3upd04.pdf
   • View In Article
4. Yusuf S, Hawken S, Ounpuu S, et al. Effect of potentially modifiable risk factors associated with myocardial infarction in 52 countries (the INTERHEART study): case-control study. Lancet. 2004;364:937–952
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (257 KB)
   • CrossRef
5. Lauer MS. Primary prevention of atherosclerotic cardiovascular disease: the high public burden of low individual risk. JAMA. 2007;297:1376–1378
   • View In Article
   • CrossRef
6. Wilson PWF, Pencina M, Jacques P, et al. C-reactive protein and reclassification of cardiovascular risk in the Framingham Heart Study. Circ Cardiovasc Qual Outcomes. 2008;1:92–97
   • View In Article
   • CrossRef
7. Ridker PM, Paynter NP, Rifai N, et al. C-reactive protein and parental history improve global cardiovascular risk prediction. The Reynolds Risk Score for men. Circulation. 2008;118:2243–2251
   • View In Article
   • CrossRef
8. Ridker PM, Buring JE, Rifai N, et al. Development and validation of improved algorithms for the assessment of global cardiovascular risk in women: the Reynolds Risk Score. JAMA. 2007;297:611–619
   • View In Article
   • CrossRef
9. Shah T, Casas JP, Cooper JA, et al. Critical appraisal of CRP measurement for the prediction of coronary heart disease events: new data and systematic review of 31 prospective cohorts. Int J Epidemiol. 2009;38:217–231
   • View In Article
   • CrossRef
10. Schaar JA, Muller JE, Falk E, et al. Terminology for high-risk and vulnerable coronary artery plaques. Report of a meeting on the vulnerable plaque, June 17 and 18, 2003, Santorini, Greece. Eur Heart J. 2004;25:1077–1082
    • View In Article
    • MEDLINE
    • CrossRef
11. Detrano R, Guerci AD, Carr JJ, et al. Coronary calcium as a predictor of coronary events in four racial or ethnic groups. N Engl J Med. 2008;358:1336–1345
    • View In Article
    • CrossRef
12. Motoyama S, Sarai M, Harigaya H, et al. Computed tomographic angiography characteristics of atherosclerotic plaques subsequently resulting in acute coronary syndrome. J Am Coll Cardiol. 2009;54:49–57
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (744 KB)
    • CrossRef
13. de Groot E, van Leuven SI, Duivenvoorden R, et al. Measurement of carotid intima-media thickness to assess progression and regression of atherosclerosis. Nat Clin Pract Cardiovasc Med. 2008;5:280–288
    • View In Article
    • CrossRef
14. Sanz J, Fayad ZA. Imaging of atherosclerotic cardiovascular disease. Nature. 2008;451:953–957
    • View In Article
    • CrossRef
15. Fuster V, Fayad ZA, Moreno PR, et al. Atherothrombosis and high-risk plaque: part II: approaches by noninvasive computed tomographic/magnetic resonance imaging. J Am Coll Cardiol. 2005;46:1209–1218
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (407 KB)
    • CrossRef
16. Saam T, Hatsukami TS, Takaya N, et al. The vulnerable, or high-risk, atherosclerotic plaque: noninvasive MR imaging for characterization and assessment. Radiology. 2007;244:64–77
    • View In Article
    • MEDLINE
    • CrossRef
17. Rudd JH, Warburton EA, Fryer TD, et al. Imaging atherosclerotic plaque inflammation with [18F]-fluorodeoxyglucose positron emission tomography. Circulation. 2002;105:2708–2711
    • View In Article
    • CrossRef
18. Bild DE, Bluemke DA, Burke GL, et al. Multi-ethnic study of atherosclerosis: objectives and design. Am J Epidemiol. 2002;156:871–881
    • View In Article
    • MEDLINE
    • CrossRef
19. Wald NJ, Law MR. A strategy to reduce cardiovascular disease by more than 80%. BMJ. 2003;326:1419–1424
    • View In Article
    • CrossRef
20. U.S. Census Bureau. 2000 Census of population and housing, demographic profile. Profiles of general demographic characteristics.
    • View In Article
21. Heckbert SR, Kooperberg C, Safford MM, et al. Comparison of self-report, hospital discharge codes, and adjudication of cardiovascular events in the Women's Health Initiative. Am J Epidemiol. 2004;160:1152–1158
    • View In Article
    • MEDLINE
    • CrossRef
22. Varas-Lorenzo C, Castellsague J, Stang MR, et al. Positive predictive value of ICD-9 codes 410 and 411 in the identification of cases of acute coronary syndromes in the Saskatchewan Hospital automated database. Pharmacoepidemiol Drug Saf. 2008;17:842–852
    • View In Article
    • CrossRef
23. Storey JD. A direct approach to false discovery rates. J R Stat Soc B. 2002;64:479–498
    • View In Article
24. Pencina MJ, D'Agostino RB. Overall C as a measure of discrimination in survival analysis: model specific population value and confidence interval estimation. Stat Med. 2004;23:2109–2123
    • View In Article
    • MEDLINE
    • CrossRef
25. Pencina MJ, D'Agostino RB, D'Agostino RB, et al. 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
    • View In Article
    • CrossRef