Observed and predicted event rates across deciles of predicted 10-year ASCVD risk by pooled cohort equations. A, No LTPA (0 metabolic equivalent of task [MET]-min/wk); P = .15. B, Less than guideline-recommended LTPA (1-500 MET-min/wk); P = .11. C, Guideline-recommended LTPA (500-1000 MET-min/wk); P = .26. D, More than guideline-recommended LTPA (>100 MET-min/wk); P = .82. The P values were derived from the Greenwood Nam-D’Agostino χ2 goodness-of-fit test, with P > .05 indicating acceptable calibration.
eTable 1. Atherosclerotic Cardiovascular Disease Event Adjudication Criteria
eTable 2. Leisure Time Physical Activity Estimation
Customize your JAMA Network experience by selecting one or more topics from the list below.
Pandey A, Mehta A, Paluch A, et al. Performance of the American Heart Association/American College of Cardiology Pooled Cohort Equations to Estimate Atherosclerotic Cardiovascular Disease Risk by Self-reported Physical Activity Levels. JAMA Cardiol. 2021;6(6):690–696. doi:10.1001/jamacardio.2021.0948
What is the risk prediction performance of American Heart Association/American College of Cardiology pooled cohort equations across strata of self-reported leisure-time physical activity levels?
In this cohort study of 18 824 participants, pooled cohort equations showed good risk discrimination and calibration performance across the spectrum of estimated leisure-time physical activity levels. Higher leisure-time physical activity level is associated with lower cardiovascular risk; adding reported leisure-time physical activity to pooled cohort equations did not change risk discrimination and reclassification capabilities of the risk prediction model.
In this study, pooled cohort equations appear to be accurate at predicting 10-year atherosclerotic cardiovascular disease risk across all strata of physical activity; the addition of self-reported leisure-time and addition of physical activity to the pooled cohort equation do not meaningfully change the risk prediction model performance.
The American Heart Association/American College of Cardiology pooled cohort equations (PCEs) are used for predicting 10-year atherosclerotic cardiovascular disease (ASCVD) risk. Pooled cohort equation risk prediction capabilities across self-reported leisure-time physical activity (LTPA) levels and the change in model performance with addition of LTPA to the PCE are unclear.
To evaluate PCE risk prediction performance across self-reported LTPA levels and the change in model performance by adding LTPA to the existing PCE model.
Design, Setting, and Participants
Individual-level pooling of data from 3 longitudinal cohort studies—Atherosclerosis Risk in Communities, Multi-Ethnic Study of Atherosclerosis, and Cardiovascular Health Study—was performed. A total of 18 824 participants were stratified into 4 groups based on self-reported LTPA levels: inactive (0 metabolic equivalent of task [MET]-min/wk), less than guideline-recommended (<500 MET-min/wk), guideline-recommended (500-1000 MET-min/week), and greater than guideline-recommended (>1000 MET-min/wk). Pooled cohort equation risk discrimination was studied using the C statistic and reclassification capabilities were studied using the Greenwood Nam-D’Agostino χ2 goodness-of-fit test. Change in risk discrimination and reclassification on adding LTPA to PCEs was evaluated using change in C statistic, integrated discrimination index, and categorical net reclassification index.
Main Outcomes and Measures
Adjudicated ASCVD events during 10-year follow-up.
Among 18 824 participants studied, 10 302 were women (54.7%); mean (SD) age was 57.6 (8.2) years. A total of 5868 participants (31.2%) were inactive, 3849 (20.4%) had less than guideline-recommended LTPA, 3372 (17.9%) had guideline-recommended LTPA, and 5735 (30.5%) had greater than guideline-recommended LTPA level. Higher LTPA levels were associated with a lower risk of ASCVD after adjustment for risk factors (hazard ratio [HR] per 1-SD higher LTPA, 0.91; 95% CI, 0.86-0.96). Across LTPA groups, PCE risk discrimination (C statistic, 0.76-0.78) and risk calibration (all χ2 P > .10) was similar. Addition of LTPA to the PCE model resulted in no significant change in the C statistic (0.0005; 95% CI, −0.0004 to 0.0015; P = .28) and categorical net reclassification index (−0.003; 95% CI, −0.010 to 0.010; P = .95), but a minimal improvement in the integrated discrimination index (0.0008; 95% CI, 0.0002-0.0013; P = .005) was observed. Similar results were noted when cohort-specific coefficients were used for creating the baseline model.
Conclusions and Relevance
Higher self-reported LTPA levels appear to be associated with lower ASCVD risk and increasing LTPA promotes cardiovascular wellness. These findings suggest the American Heart Association/American College of Cardiology PCEs are accurate at estimating the probability of 10-year ASCVD risk regardless of LTPA level. The addition of self-reported LTPA to PCEs does not appear to be associated with improvement in risk prediction model performance.
The American Heart Association (AHA)/American College of Cardiology (ACC) guidelines for atherosclerotic cardiovascular disease (ASCVD) risk assessment recommend using the sex- and race/ethnicity-specific pooled cohort equations (PCEs) for estimating the 10-year risk of developing an ASCVD event in asymptomatic individuals.1 These risk estimates are calculated using a combination of CVD risk factors (age, sex, race/ethnicity, diabetes, smoking, systolic blood pressure, antihypertensive use, total cholesterol level, and high-density lipoprotein cholesterol level) that were examined in selected community-based epidemiologic cohorts.1 Ten-year risk estimated using the PCE serves as the starting point for guiding clinician-patient discussions regarding ASCVD prevention strategies. The AHA/ACC/multisociety cholesterol management guidelines recommend using a priori–defined PCE-estimated risk thresholds to guide statin use in primary prevention settings.2 As such, it is necessary to ensure that PCEs perform adequately in at-risk populations given the association between predicted risk estimates and CVD risk stratification and decision-making for primary ASCVD prevention.
Observational studies have demonstrated that physical inactivity and lower levels of cardiorespiratory fitness are associated with increased CVD risk.3-8 The association between self-reported leisure-time physical activity (LTPA) level and risk for adverse CV events follows an inverse dose-response pattern, wherein higher levels of LTPA were associated with lower ASCVD risk in a curvilinear fashion.7 Although cardiorespiratory fitness was considered as a candidate risk factor in the initial derivation and validation of the PCEs, self-reported LTPA levels were not considered owing to inconsistent availability in the derivation cohorts. It is unclear whether predicted risk estimates based on PCEs can be enhanced by including LTPA levels. Furthermore, to our knowledge, PCE performance among individuals engaging in different levels of LTPA has not been evaluated. To address these knowledge gaps, we studied PCE risk prediction capability across a range of self-reported LTPA levels along with the association between adding LTPA to PCE and model risk discrimination and reclassification capabilities. We hypothesized that addition of LTPA level to the risk prediction equation will improve risk discrimination and reclassification.
Individual-level data were pooled from 3 prospective, observational cohort studies based in the US: the Atherosclerosis Risk in Communities (ARIC), the Multi-Ethnic Study of Atherosclerosis (MESA), and the Cardiovascular Health Study (CHS). Details of study design for these 3 epidemiologic cohorts have been published.9-11 These cohorts are a part of the Cardiovascular Disease Lifetime Risk Pooling Project, a pooled data set of 20 community-based cardiovascular disease cohorts in the US.12 The present analysis included ARIC, MESA, and CHS participants who were free of prevalent cardiovascular disease, had PCE risk factors measured at enrollment, and had self-reported LTPA levels estimated in metabolic equivalent of task (MET) minutes per week at baseline. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cohort studies.
The ARIC, MESA, and CHS studies were approved by institutional review boards at the respective coordinating centers, each field center, and other central agencies. The present study was exempt from review by the Northwestern University Institutional Review Board. All participants provided written informed consent at the time of enrollment. Sex and race/ethnicity were self-reported. The assessment of traditional CV risk factors in each of the study cohort was performed using standard protocols as described previously.9-11 The final study sample consisted of 18 824 participants (ARIC, n = 10 957; MESA, n = 4928; and CHS, n = 2939). The outcome of interest was first ASCVD event during 10 years of follow-up. Atherosclerotic cardiovascular disease was defined as a composite of coronary heart disease–related death, nonfatal myocardial infarction, and fatal and nonfatal stroke. Atherosclerotic cardiovascular disease events were adjudicated using similar criteria in all 3 cohorts as described in eTable 1 in the Supplement.
Self-reported LTPA was estimated using the modified Baecke questionnaire in ARIC,13 the typical week physical activity survey in MESA,14 and the modified Health Interview Survey for LTPA measurement in CHS.15,16 The details of self-reported LTPA estimation in ARIC, MESA, and CHS are described in eTable 2 in the Supplement. Although LTPA ascertainment occurred differently across cohorts, quantitative LTPA estimates were harmonized and calculated for participants (MET-min/wk) using standardized MET values that account for intensities and frequency of reported physical activitiesas a part of PA volume.5,14,16,17
Study participants were divided into 4 groups based on LTPA level at baseline: inactive (0 MET-min/wk), less than guideline recommended (<500 MET-min/week), guideline recommended (500-1000 MET-min/week), and greater than guideline recommended (>1000 MET-min/wk). These cutoffs were based on the 2018 Physical Activity Guidelines that recommend that adults engage in 150 to 300 minutes of moderate-intensity physical activity or 75 to 150 minutes of vigorous-intensity physical activity every week, equivalent to 500 to 1000 MET-min/wk.18 Sex-specific baseline characteristics of study participants are described across LTPA groups. Categorical variables are presented as count (proportion) and continuous variables are presented as mean (SD). Categorical variables were compared across LTPA groups using the χ2 test and continuous variables were compared using the analysis of variance test.
Multivariable-adjusted Cox proportional hazards regression models were constructed to assess the association of self-reported LTPA level with 10-year risk of ASCVD. Cox proportional hazard models were adjusted for following covariates: model 1 included age, sex, and race/ethnicity, and model 2 included the model 1 covariates plus PCE risk factors (diabetes, smoking, treated and untreated increased systolic blood pressure, total cholesterol level, and high-density lipoprotein cholesterol level). This analysis was performed in the overall study population as well as in sex-, race/ethnicity-, and cohort-specific subgroups.
The 10-year ASCVD risk discrimination was studied by calculating the concordance (C) statistic and calibration performance was examined using the Greenwood Nam-D’Agostino χ2 goodness-of-fit test of PCEs across LTPA groups. Change in 10-year ASCVD risk discrimination and reclassification after adding LTPA level to PCEs was studied using change in C statistic, integrated discrimination index, and categorical net reclassification index. Both the original PCEs and the PCEs created using cohort-specific coefficients (cohort-specific PCEs) were used in this analysis. All statistical analyses were performed using SAS, version 9.4 (SAS Inc). A 2-sided P value <.05 was considered statistically significant.
Among 18 824 participants studied, 10 302 were women (54.7%) and 8522 were men (45.3%); mean (SD) age was 57.6 (8.2) years. A total of 5868 participants (31.2%) were inactive, 3849 (20.4%) had less than guideline-recommended LTPA, 3372 (17.9%) had guideline-recommended LTPA, and 5735 (30.5%) had greater than guideline-recommended LTPA level. Other baseline characteristics of study participants stratified by LTPA group are described in Table 1. The burden of CVD risk factors, including prevalence of smoking, diabetes, and low-density lipoprotein level, were higher in the lower LTPA (inactive or less than guideline-recommended) groups compared with the higher LTPA (guideline recommended or greater LTPA) groups (Table 1).
In the pooled cohort of patients with complete data available on all covariates, a total of 1632 adjudicated ASCVD events occurred during the 10-year follow-up period. Higher levels of self-reported LTPA were associated with lower ASCVD risk in age-, sex-, and race/ethnicity-adjusted models (hazard ratio [HR] per 1-SD higher LTPA, 0.87; 95% CI, 0.82-0.92) (Table 2, model 1). The inverse association between LTPA levels and risk of ASCVD attenuated modestly but remained statistically significant after further adjustment for PCE risk factors (HR per 1-SD higher LTPA, 0.91; 95% CI, 0.86-0.96) (Table 2, model 2). A similar pattern of association between higher levels of self-reported LTPA and lower risk of ASCVD was also observed in sex- and race/ethnicity-specific subgroup analysis in the pooled cohort (Table 2). In cohort-stratified analysis, the 10-year ASCVD rate varied considerably across cohorts, with the highest rate in the CHS (14.6%) followed by the MESA (4.4%) and ARIC (3.6%) cohorts. A consistent pattern of association between higher levels of LTPA and lower risk of ASCVD was noted in the CHS and MESA cohorts, but not in the ARIC cohort in adjusted cohort-specific Cox proportional hazards models (Table 2).
Pooled cohort equation risk discrimination performance was comparable across LTPA groups. The risk prediction model C statistics among the groups were 0.768 (95% CI, 0.750-0.788) for the inactive group, 0.771 (95% CI, 0.750-0.792) for the less than guideline-recommended group, 0.763 (95% CI, 0.738-0.788) for the guideline-recommended group, and 0.761 (95% CI, 0.742-0.780) for the more than guideline-recommended group. The PCE risk model demonstrated adequate and comparable calibration performance across all LTPA groups (all Greenwood Nam-D’Agostino goodness-of-fit χ2 P values >.10) (Figure).
Addition of self-reported LTPA levels to the original PCEs did not significantly change the C statistic (0.0005; 95% CI, −0.0004 to 0.0015; P = .28) and categorical net reclassification index NRI (−0.003; 95% CI, −0.010 to 0.010; P = .95) in the overall cohort, 0.0002-0.0013; P = .005) but resulted in minimal integrated discrimination index improvement (0.0008; 95% CI, 0.0002-0.0013; P = .005) similar results were observed when cohort-specific estimates were used (Table 3). In cohort-specific analyses, risk discrimination and reclassification indices did not change markedly after adding LTPA levels to PCEs in the ARIC and MESA cohorts, but modest improvements in integrated discrimination index and categorical net reclassification index were noted in the CHS cohort (Table 3).
In this pooled cohort analysis, we observed several findings. First, higher levels of self-reported LTPA were independently associated with lower ASCVD risk in asymptomatic individuals. This association was consistent across different race/ethnicity- and sex-based groups, highlighting the importance of physical activity promotion in reducing ASCVD risk. Second, the guideline-endorsed PCE demonstrated consistent accuracy and comparable performance in predicting 10-year ASCVD risk across the spectrum of LTPA levels. Third, the addition of LTPA to PCEs did not significantly change risk discrimination and reclassification capabilities of the prediction model.
The sex- and race/ethnicity-specific PCEs for cardiovascular risk assessment were released in 2014.19 Although the equations account for all major cardiovascular risk factors, lifestyle variables, such as LTPA, were not included owing to inconsistent availability across the derivation cohorts. This absence is particularly relevant because previous observational research has shown an association between low self-reported LTPA levels and higher ASCVD risk. In the present study, we observed a higher burden of traditional risk factors in participants with lower self-reported LTPA. Furthermore, we noted a significant association between higher LTPA level and lower risk of ASCVD in the overall pooled cohort. This association was consistent among women and men, as well as Black and non-Black participants. These findings highlight the importance of LTPA as a modifiable ASCVD risk factor. Furthermore, the benefits of higher levels of LTPA and cardiorespiratory fitness extend to lowering risk of nonatherosclerotic CVD (eg, heart failure) and other cardiometabolic diseases(eg, diabetes), highlighting the central the role of physical activity improvement in promoting global cardiovascular wellness.4-6,8,20,21
We observed that the risk discrimination and calibration performance of PCEs in predicting ASCVD risk was consistent across the 4 strata of self-reported LTPA level. These findings underscore the robustness of 10-year ASCVD risk prediction by PCEs. Our findings are consistent with earlier reports in which the PCEs were externally validated and were shown to be well-calibrated near treatment decision thresholds (7.5%-10% 10-year ASCVD risk) among study samples comprised of broad US populations.1,22 In addition, our report lends support to the current AHA/ACC guideline recommendation of using PCEs as the initial tool for stratifying ASCVD risk in diverse primary prevention settings.
Despite the inverse association between self-reported LPTA level and ASCVD risk, the performance of PCEs was not meaningfully changed by inclusion of self-reported LTPA levels in the risk prediction model. Consistent with these findings, a recent study suggested that inclusion of body mass index, another important lifestyle risk factor for CVD, into the PCE model did not improve the risk prediction performance.23 However, like physical inactivity, higher body mass index is associated with a higher burden of other cardiovascular risk factors. Lifestyle modification with intentional weight loss and improvement in LTPA are important modifiable targets for reducing the risk of atherosclerotic and nonatherosclerotic CVD.6,24,25 The lack of improvement in the PCE model performance by incorporation of LTPA suggests that the contribution of LTPA to ASCVD risk may be associated with its favorable effects on the risk factors that are already included in the PCE model. Furthermore, it is plausible that self-reported LTPA or body mass index may not completely capture the biological risk associated with poor exercise behavior and/or obesity.
Previous studies have evaluated whether incorporation of measures of LTPA or cardiorespiratory fitness can improve the performance of CV risk assessment models.3,26-28 Consistent with the observations in our study, in the Copenhagen City Heart Study Graversen et al26 reported a modest change in the performance statistics of the Systemic Coronary Risk Evaluation (SCORE) algorithm, which estimates 10-year risk of CV-associated death based on traditional CV risk factors (blood pressure, smoking, serum cholesterol level, age, and sex) with the addition of LTPA levels. Similarly, in a prospective cohort of 2020 participants from Greece without known CVD at baseline, addition of LTPA levels to the HellenicSCORE model (a calibrated version of the SCORE algorithm), modestly improved its discrimination performance (C statistic improvement, 0.012) in predicting fatal and nonfatal CV events.27
In contrast, in the Cooper Center Longitudinal Study, more robust improvements in the performance of traditional CV risk prediction models have been reported with the addition of objective measures of cardiorespiratory fitness to the models.3,28 Future studies are needed to determine whether incorporation of more integrative and/or accurate measures of exercise capacity, such as cardiorespiratory fitness or objectively measured LTPA, to PCEs may improve ASCVD risk prediction.
This study has several strengths. The pooled cohort sample consisted of more than 18 000 individuals who experienced more than 1600 adjudicated ASCVD events during 10 years of follow-up, providing adequate statistical power to study the association of LTPA levels with ASCVD risk and evaluate the accuracy of PCE across different LTPA categories. The strength of the pooled cohort analysis is highlighted by the observed association between LTPA and ASCVD risk in individual cohorts, whereby the associations did not reach statistical significance in some individual cohorts with lower 10-year event rates. The study also has several limitations. First, the ARIC and CHS cohorts were used for creating the PCEs, making it challenging to improve the PCE risk prediction performance by adding LTPA levels. Second, LTPA levels were self-reported and are prone to recall bias. Third, data on objective measures of LTPA or cardiorespiratory fitness were not available in the pooled cohorts. Fourth, we did not examine sedentary behavior, which is another marker of increased ASCVD risk,29 on PCE risk prediction.
The findings of this study suggest that AHA/ACC PCEs are accurate at predicting 10-year ASCVD risk across the spectrum of estimated LTPA levels, and the addition of self-reported LTPA to the PCEs does not appear to change its risk prediction performance. These observations regarding the performance of PCEs in the context of LTPA do not diminish the importance of physical activity as a core principle for promoting cardiovascular wellness.
Accepted for Publication: March 2, 2021.
Published Online: April 28, 2021. doi:10.1001/jamacardio.2021.0948
Corresponding Author: John T. Wilkins, MD, MS, Department of Medicine (Cardiology), Northwestern University Feinberg School of Medicine, 680 N Lakeshore Dr, Ste 1400, Chicago, IL 60611 (email@example.com).
Author Contributions: Dr Wilkins had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Drs Pandey and Mehta contributed equally.
Concept and design: Pandey, Mehta, Carnethon, Berry, Lloyd-Jones, Wilkins.
Acquisition, analysis, or interpretation of data: Pandey, Mehta, Paluch, Ning, Allen, Michos, Berry, Lloyd-Jones, Wilkins.
Drafting of the manuscript: Pandey, Mehta.
Critical revision of the manuscript for important intellectual content: Pandey, Paluch, Ning, Carnethon, Allen, Michos, Berry, Lloyd-Jones, Wilkins.
Statistical analysis: Pandey, Ning.
Administrative, technical, or material support: Carnethon, Wilkins.
Supervision: Pandey, Berry, Lloyd-Jones, Wilkins.
Conflict of Interest Disclosures: Dr Pandey reported serving in an unpaid position on the advisory board of Roche Diagnostics. Dr Berry reported receiving grants from the National Institutes of Health (NIH), grants from Abbott, and personal fees from Cooper Institute during the conduct of the study; and personal fees from Roche and Astra Zeneca outside the submitted work. Dr Lloyd-Jones reported receiving grants from the NIH and American Hospital Association during the conduct of the study. No other disclosures were reported.
Funding/Support: The Cardiovascular Lifetime Risk Pooling Project was supported in its inception by the NIH National Heart, Lung, and Blood Institute grant R21HL085375 and is currently supported by funds from the Northwestern University Feinberg School of Medicine. Dr Pandey received research funding from Texas Health Resources Clinical Scholarship, Gilead Sciences Research Scholar Program, and National Institute of Aging GEMSSTAR grant (1R03AG067960-01), and an investigator-initiated grant from Applied Therapeutics.
Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.