Evaluation of the Pooled Cohort Equations for Prediction of Cardiovascular Risk in a Contemporary Prospective Cohort

doi:10.1016/j.amjcard.2016.11.042

The American Journal of Cardiology

Volume 119, Issue 6, 15 March 2017, Pages 881-885

https://doi.org/10.1016/j.amjcard.2016.11.042 Get rights and content

Most guidelines suggest a baseline risk assessment to guide atherosclerotic cardiovascular disease (ASCVD) prevention strategies. The American Heart Association/American College of Cardiology Pooled Cohort Equations (PCEs) is one tool to assess baseline risk; however, the accuracy of this tool has been called into question. We aimed to examine the calibration and discrimination of the PCEs in the BioImage study, a contemporary multiethnic cohort of asymptomatic adults enrolled from 2008 to 2009 in the Humana Health System in Chicago, Illinois, and Fort Lauderdale, Florida. Our primary end point was hard ASCVD, defined as cardiovascular death, myocardial infarction, and stroke. A total of 3,635 adults who were not on lipid-lowering therapy at baseline were followed for a maximum of 4.6 years. The mean age was 68.6 years; 2000 (55%) participants were women and 935 patients reported being of non-white race (26%). Although 74 ASCVD events were observed over a median follow-up of 2.7 years, 198 events were predicted by the PCEs. The observed event rate was 7.9 per 1,000 participant-years (95% confidence interval [CI] 6.1 to 9.8), whereas the predicted rate by the PCEs was 21 per 1,000 participant-years (95% CI 20.7 to 21.8). This represents an overestimation of 167% (Hosmer–Lemeshow chi-square = 173; p <0.001). With regard to discrimination, the C-statistic of the PCEs was 0.65 (CI 0.58 to 0.71). In an analysis restricted to 3,080 participants without diabetes mellitus and with low-density lipoprotein cholesterol between 70 and 189 mg/dl, the PCEs similarly overestimated risk by 181% (152 predicted events vs 54 observed events; p <0.001). The PCEs substantially overestimate ASCVD risk in this middle-aged adult insured population. Refinement of existing risk prediction functions may be warranted.

Section snippets

Methods

The methods used in the BioImage study (NCT00738725) have been previously described.⁵ In brief, men (aged 55 to 80 years) and women (aged 60 to 80 years) who were members of the Humana Health System and did not have clinical ASCVD were eligible for inclusion in the study and enrolled from January 2008 to June 2009. Participants were required to have at least 1 cardiovascular risk factor of diabetes mellitus, hypertension, obesity, or dyslipidemia. The BioImage study was approved by

Results

A total of 3,635 participants were included in this analysis. The mean age of the participants was 68.6 ± 6 years (Table 1). Fifty-five percent of participants were women, while 10% of participants had type 2 diabetes. In self-report of race, 74% of participants identified as white, 16% identified as African American, 6% identified as Hispanic, and 2% identified as Asian. Thirty-two percent reported current use of aspirin, whereas 22% reported use of blood pressure–lowering medication.

Discussion

In a contemporary prospective cohort from the United States, the PCEs overestimated ASCVD risk by 167%. Overestimation of risk was observed across predicted risk categories, among women and men, among participants for whom the PCEs were developed (i.e., consideration of statin therapy) and among nontreated participants.

Previous research on the PCEs has broadly, although not universally, suggested that the equations are poorly calibrated and that the equations overestimate ASCVD risk. In the 5

Acknowledgment

Author contributions: Dr. Emdin had full access to all the data in the study and took responsibility for the integrity of the data and the accuracy of the data analysis. Drs. Emdin, Khera, and Kathiresan contributed to study concept and design. Acquisition of data was done by Drs. Fuster, Baber, and Mehran. Analysis and interpretation of data were done by all authors. Drs. Emdin, Khera, and Kathiresan drafted the manuscript. Critical revision of the manuscript for important intellectual content

References (14)

D.C. Goff et al.
2013 ACC/AHA guideline on the assessment of cardiovascular risk: a report of the American College of Cardiology/American Heart association Task Force on Practice Guidelines
J Am Coll Cardiol
(2014)
U. Baber et al.
Prevalence, impact, and predictive value of detecting subclinical coronary and carotid atherosclerosis in asymptomatic adults: the BioImage study
J Am Coll Cardiol
(2015)
A.S. Agatston et al.
Quantification of coronary artery calcium using ultrafast computed tomography
J Am Coll Cardiol
(1990)
P.M. Ridker et al.
Statins: new American guidelines for prevention of cardiovascular disease
Lancet
(2013)
J.S. Rana et al.
Accuracy of the atherosclerotic cardiovascular risk equation in a large contemporary, multiethnic population
J Am Coll Cardiol
(2016)
M.J. Blaha
The critical importance of risk score calibration: time for transformative approach to risk score validation?
J Am Coll Cardiol
(2016)
A.P. DeFilippis et al.
An analysis of calibration and discrimination among multiple cardiovascular risk scores in a modern multiethnic cohort
Ann Intern Med
(2015)

There are more references available in the full text version of this article.

Cited by (28)

Impact of Social Vulnerability and Demographics on Ischemic Heart Disease Mortality in the United States
2023, JACC: Advances
Cardiovascular disease is a leading cause of morbidity and mortality, largely dominated by ischemic heart diseases (IHDs). Social determinants of health, including geographic, psychosocial, and socioeconomic factors, influence the development of IHD.
This study aimed to evaluate yearly trends and disparities in IHD mortality and to assess the impact of social vulnerability.
We performed cross-sectional analyses using United States county-level mortality data and social vulnerability index (SVI) obtained from the Centers for Disease Control and Prevention databases. Age-adjusted mortality rates (AAMRs) per 100,000 population were compared between aggregated U.S. county groups, stratified by demographic information and SVI quartiles. Log-linear regression models were used to identify mortality trends from 1999 to 2020, with inflection points determined through the Monte-Carlo permutation test.
We identified a total of 9,108,644 deaths related to IHD between 1999 and 2020. Overall AAMR decreased from 194.6 in 1999 to 91.8 in 2020. Males (AAMR: 161.51) and Black (AAMR: 141.49) populations exhibited higher AAMR compared to females (AAMR: 93.16) and White (AAMR: 123.34) populations, respectively. Disproportionate AAMRs were observed among nonmetropolitan (AAMR: 136.17) and Northeastern (AAMR: 132.96) regions. Counties with a higher SVI experienced a greater AAMR, with a cumulative excess of 20.91 deaths per 100,000 person-years associated with increased social vulnerability.
Despite a decline in IHD mortality from 1999 to 2020, disparities persisted among racial, gender, and geographic subgroups. A higher SVI was linked to increased IHD mortality. Policy interventions should prioritize integrating the SVI into health care delivery systems to effectively address these disparities.
Integrating Indices of Genetic Risk for Cardiovascular Disease
2023, JACC: Advances
Detection of small coronary calcifications in patients with Agatston coronary artery calcium score of zero
2022, Journal of Cardiovascular Computed Tomography
Citation Excerpt :
The Agatston CAC score (CACS) is most commonly used as the reference standard to define the presence of CAC.14 For more than 30 years, this score has been successfully used in clinical practice to assess cardiovascular risk in patients with intermediate risk of atherosclerotic cardiovasulary disease beyond traditional methods9,14–17 With the Agatston method, a CT attenuation threshold of 130 Hounsfield units (HU) is used for the detection of CAC, with only contiguous voxels (3 pixels) totalling 1 mm2 in area regarded as a calcified lesion in order to reduce the influence of image noise.14 These size and attenuation thresholds for calcium were developed at the era of electron beam CT (EBCT).
A coronary artery calcium score (CACS) of 0 is associated with a very low risk of cardiac event. However, the Agatston CACS may fail to detect very small or less dense calcifications. We investigated if an alteration of the Agatston criteria would affect the ability to detect such plaques.
We evaluated 322 patients, 161 who had a baseline scan with CACS = 0 and a follow-up scan with CACS>0 and 161 with two serial CACS = 0 scans (control group), to identify subtle calcification not detected in the baseline scan because it was not meeting the Agatston size and HU thresholds (≥1 mm² and ≥130HU). Size threshold was set to <1 mm² and the HU threshold modified in a stepwise manner to 120, 110, 100 and 90. New lesions were classified as true positive or false positive(noise) using the follow-up scan.
We identified 69 visually suspected subtle calcified lesions in 65/322 (20.2%) patients with CAC = 0 by the Agatston criteria. When size threshold was set as <1 mm² and HU ≥ 130, 36 lesions scored CACS>0, 34 (94.4%) true positive and 2 (5.6%) false positive. When decrease in HU (120HU, 110HU, 100HU, and 90HU) threshold was added to the reduced size threshold, the number of lesions scoring>0 increased (46, 55, 59, and 69, respectively) at a cost of increased false positive rate (8.7%, 20%, 22%, and 30.4% respectively). Eliminating size or both size and HU threshold to ≥120HU correctly reclassified 9.6% and 12.1% of patients respectively.
Eliminating size and reducing HU thresholds to ≥120HU improved the detection of subtle calcification when compared to the Agatston CACS method.
Pharmacogenomics and circadian rhythms as mediators of cardiovascular drug-drug interactions
2021, Current Research in Pharmacology and Drug Discovery
Citation Excerpt :
The interplays between genetic and environmental factors and drugs may predispose to resilience, i.e., the ability to adapt to insults from DDI adverse impacts. A recent study of 55,685 individuals reported that a polygenic risk score consisting of 50 single nucleotide polymorphisms (SNPs) is associated with a high genetic risk of coronary artery disease (CAD) and that the relative risk of coronary events decreases in those with a healthy lifestyle (Emdin et al., 2016, 2017; Khera et al., 2016a, 2016b). Genome-wide association studies (GWAS) have been widely employed to genetically map disease association to genomic regions contributing to disease pathogenesis (Russo et al., 2018; Ross et al., 2004).
This article summarizes the current literature and documents new evidence concerning drug-drug interactions (DDI) stemming from pharmacogenomic and circadian rhythm determinants of therapies used to treat common cardiovascular diseases (CVD), such as atherosclerosis and hypertension. Patients with CVD often have more than one pathophysiologic condition, namely metabolic syndromes, hypertension, hyperlipidemia, and hyperglycemia, among others, which necessitate polytherapeutic or polypharmaceutic management. Interactions between drugs, drugs and food/food supplements, or drugs and genetic/epigenetic factors may have adverse impacts on the cardiovascular and other systems of the body. The mechanisms underlying cardiovascular DDI may involve the formation of a complex pharmacointeractome, including the absorption, distribution, metabolism, and elimination of drugs, which affect their respective bioavailability, efficacy, and/or harmful metabolites. The pharmacointeractome of cardiovascular drugs is likely operated with endogenous rhythms controlled by circadian clock genes. Basic and clinical investigations have improved the knowledge and understanding of cardiovascular pharmacogenomics and pharmacointeractomes, and additionally they have presented new evidence that the staging of deterministic circadian rhythms, according to the dosing time of drugs, e.g., upon awakening vs. at bedtime, cannot only differentially impact their pharmacokinetics and pharmacodynamics but also mediate agonistic/synergetic or antagonistic DDI. To properly manage CVD patients and avoid DDI, it is important that clinicians have sufficient knowledge of their multiple risk factors, i.e., age, gender, and life style elements (like diet, smoking, psychological stress, and alcohol consumption), and comorbidities, such as diabetes, hypertension, dyslipidemia, and depression, and the potential interactions between genetic or epigenetic background of their prescribed therapeutics.
2019 ACC/AHA Guideline on the Primary Prevention of Cardiovascular Disease: A Report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines
2019, Journal of the American College of Cardiology
Negative Risk Markers for Cardiovascular Events in the Elderly
2019, Journal of the American College of Cardiology
Citation Excerpt :
In a secondary analysis, we also assessed NRI across a 7.5% 10-year CVD risk threshold using a BioImage-derived, well-calibrated logistic risk model (developed in the BioImage cohort). Because all participants in the BioImage study had <10 years of follow-up, we used an exponential survival function to scale participants’ risk to their length of follow-up, as previously performed (18–20). All analyses were performed using Stata version 14 (StataCorp, College Station, Texas).
Cardiovascular risk increases dramatically with age, leading to nearly universal risk-based statin eligibility in the elderly population. To limit overtreatment, elderly individuals at truly low risk need to be identified.
Discovering “negative” risk markers able to identify elderly individuals at low short-term risk for coronary heart disease and cardiovascular disease.
In 5,805 BioImage participants (mean age 69 years; median follow-up 2.7 years), the authors evaluated 13 candidate markers: coronary artery calcium (CAC) = 0, CAC ≤10, no carotid plaque, no family history, normal ankle-brachial index, test result <25th percentile (carotid intima-media thickness, apolipoprotein B, galectin-3, high-sensitivity C-reactive protein, lipoprotein(a), N-terminal pro–B-type natriuretic peptide, and transferrin), and apolipoprotein A1 >75th percentile. Negative risk marker performance was compared using patient-specific diagnostic likelihood ratio (DLR) and binary net reclassification index (NRI).
CAC = 0 and CAC ≤10 were the strongest negative risk markers with mean DLRs of 0.20 and 0.20 for coronary heart disease (i.e., ≈80% lower risk than expected from traditional risk factor assessment) and 0.41 and 0.48 for cardiovascular disease, respectively, followed by galectin-3 <25th percentile (DLR 0.44 and 0.43, respectively) and absence of carotid plaque (DLR 0.39 and 0.65, respectively). Results obtained by other candidate markers were less impressive. Accurate downward risk reclassification across the Class I statin-eligibility threshold defined by the American College of Cardiology/American Heart Association was largest for CAC = 0 (NRI 0.23) and CAC ≤10 (NRI 0.28), followed by galectin-3 <25th percentile (NRI 0.14) and absence of carotid plaque (NRI 0.08).
Elderly individuals with CAC = 0, CAC ≤10, low galectin-3, or no carotid plaque had remarkable low cardiovascular risk, calling into question the appropriateness of a treat-all approach in the elderly population.

View all citing articles on Scopus

: Drs. Emdin and Khera contributed equally to this work.

: Funding: This 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.

: See page 885 for disclosure information.

View full text

Preventive CardiologyEvaluation of the Pooled Cohort Equations for Prediction of Cardiovascular Risk in a Contemporary Prospective Cohort

Section snippets

Methods

Results

Discussion

Acknowledgment

J Am Coll Cardiol

J Am Coll Cardiol

J Am Coll Cardiol

Lancet

J Am Coll Cardiol

J Am Coll Cardiol

An analysis of calibration and discrimination among multiple cardiovascular risk scores in a modern multiethnic cohort

Ann Intern Med

Preventive Cardiology
Evaluation of the Pooled Cohort Equations for Prediction of Cardiovascular Risk in a Contemporary Prospective Cohort