Rationale and design of the Henry Ford Exercise Testing Project (the FIT project)

The Role of Electrocardiographic Exercise Testing for the Possibility of Permanent Pacemaker Implantation in Patients with Sinus Bradycardia

Objective

Study on the role of electrocardiographic (ECG) exercise testing for the possibility of permanent pacemaker implantation (PPI) in patients with sinus bradycardia (SB).

Methods

Cross-sectional descriptive, prospective study. The study on 60 patients with SB below 50 beats/minute on 12-lead electrocardiogram at rest, with or without symptoms of SB, conducted at the Vietnam National Heart Institute and Hanoi Heart Hospital from January 2020 to September 2021.

Results

Sixty patients with SB were studied, 36 male (60%) and 24 female patients (40%), p > 0.05. The average age was 55.12 ± 13.89 years old. Maximum exercise capacity (MEC) is low and only reaches 7.78 ± 3.59 metabolic equivalents (METs); Not reaching 85% of predicted MEC accounts for 53.5%; Maximum exercise time is 10.53 ± 0.46 minutes; Impaired heart rate (HR) variability in patients with SB is high: Chronotropic Index <0.8 accounts for 53.5%, not reaching 85% of predicted HR max accounts for 45%. The average HR max was 129.90 ± 29.22 beats per minute (BPM). The average maximum workload systolic blood pressure was 155.23 ± 20.59 mmHg. The average value of maximum exercise diastolic blood pressure was 88.10 ± 9.11 mmHg. The HR decreased by 27.87 ± 16.82 BPM in the first minute. Not achieving 85% of predicted MEC (p = 0.062), so it is not an independent factor predicting the ability for PPI. Only the HR variability index <0.8 is an independent predictor for PPI in bradycardic patients, which has OR = 21.521 (95% CI: 2.27-04.34, p < 0.05).

Conclusion

Results can be seen that Chronotropic Index <0.8 is an important marker for physicians to decide on PPI in ECG during exercise testing in SB patients and is a potential prognostic factor for the need for PPI.

FIT calculator: a multi-risk prediction framework for medical outcomes using cardiorespiratory fitness data

Accurately predicting patients' risk for specific medical outcomes is paramount for effective healthcare management and personalized medicine. While a substantial body of literature addresses the prediction of diverse medical conditions, existing models predominantly focus on singular outcomes, limiting their scope to one disease at a time. However, clinical reality often entails patients concurrently facing multiple health risks across various medical domains. In response to this gap, our study proposes a novel multi-risk framework adept at simultaneous risk prediction for multiple clinical outcomes, including diabetes, mortality, and hypertension. Leveraging a concise set of features extracted from patients' cardiorespiratory fitness data, our framework minimizes computational complexity while maximizing predictive accuracy. Moreover, we integrate a state-of-the-art instance-based interpretability technique into our framework, providing users with comprehensive explanations for each prediction. These explanations afford medical practitioners invaluable insights into the primary health factors influencing individual predictions, fostering greater trust and utility in the underlying prediction models. Our approach thus stands to significantly enhance healthcare decision-making processes, facilitating more targeted interventions and improving patient outcomes in clinical practice. Our prediction framework utilizes an automated machine learning framework, Auto-Weka, to optimize machine learning models and hyper-parameter configurations for the simultaneous prediction of three medical outcomes: diabetes, mortality, and hypertension. Additionally, we employ a local interpretability technique to elucidate predictions generated by our framework. These explanations manifest visually, highlighting key attributes contributing to each instance's prediction for enhanced interpretability. Using automated machine learning techniques, the models simultaneously predict hypertension, mortality, and diabetes risks, utilizing only nine patient features. They achieved an average AUC of 0.90 ± 0.001 on the hypertension dataset, 0.90 ± 0.002 on the mortality dataset, and 0.89 ± 0.001 on the diabetes dataset through tenfold cross-validation. Additionally, the models demonstrated strong performance with an average AUC of 0.89 ± 0.001 on the hypertension dataset, 0.90 ± 0.001 on the mortality dataset, and 0.89 ± 0.001 on the diabetes dataset using bootstrap evaluation with 1000 resamples.

Determination of Peak Oxygen Uptake in Patients with Acute Myocardial Infarction: The Role of Arterial Stiffness in Cardio-Vascular-Skeletal Muscle Coupling

The relationship between arterial stiffness and oxygen uptake (VO2) in patients with acute myocardial infarction (AMI) remains unclear. We aimed to investigate this relationship and factors contributing to VO2 in patients with AMI. The role of arterial stiffness in cardio-skeletal muscle coupling during exercise was then elucidated. Upon discharge, we measured exercise capacity using cardiopulmonary exercise testing (CPX), assessed arterial stiffness with the cardio-ankle vascular index (CAVI), and determined body composition to assess the skeletal muscle mass of 101 patients with AMI. Patients were categorized based on their CAVI scores into three groups: (i) normal (CAVI: ≤7.9), (ii) borderline (CAVI: 8.0-8.9), and (iii) abnormal (CAVI: ≥9.0). Subsequently, VO2 was compared among these groups. The relationship between the CAVI and VO2 Peak during CPX and factors contributing to VO2 Peak were investigated. The abnormal CAVI group had a significantly lower VO2 Peak than the normal and borderline groups. The CAVI was associated with VO2 Peak. Furthermore, the CAVI was found to be a factor contributing to VO2 Peak. These findings suggest that arterial stiffness in tissue blood distribution and blood supply causes systemic exercise limits in patients with AMI. This suggests that arterial stiffness plays a significant role in cardio-vascular-skeletal muscle coupling.

Improved cardiorespiratory fitness is associated with lower incident ischemic stroke risk: Henry Ford FIT project

BACKGROUND

Change in cardiorespiratory fitness (CRF) modulates vascular disease risk; however, it's unclear if this adds further prognostic information, particularly for ischemic stroke. The objective of this analysis is to describe the association between the change in CRF over time and subsequent incident ischemic stroke.

METHODS

This is a retrospective, longitudinal, observational cohort study of 9,646 patients (age=55±11 years; 41% women; 25% black) who completed 2 clinically indicated exercise tests (> 12 months apart) and were free of any stroke at the time of test 2. CRF was expressed as metabolic-equivalents-of-task (METs). Incident ischemic stroke was identified using ICD codes. The adjusted hazard ratio (aHR) was determined for risk of ischemic stroke associated with change in CRF.

RESULTS

Mean time between tests was 3.7 years (IQR, 2.2, 6.0). During a median of 5.0 years (IQR, 2.7, 7.6 y) of follow-up, there were 873 (9.1%) ischemic stroke events. Each 1 MET increase between tests was associated with a 9% lower ischemic stroke risk (aHR 0.91 [0.88-0.94]; n = 9.646). There was an interaction effect by baseline CRF category, but not for sex or race. A sensitivity analysis which removed those who experienced an incident diagnosis known to be associated with an increased risk of ischemic vascular disease, validated our primary findings (aHR 0.91 [0.88, 0.95]; n= 6,943).

CONCLUSIONS

Improvement in CRF over time is independently and inversely associated with a lower risk of ischemic stroke. Encouragement of regular exercise focused on improving CRF may reduce ischemic stroke risk.

Exercise and Experiments of Nature

In this article, we highlight the contributions of passive experiments that address important exercise-related questions in integrative physiology and medicine. Passive experiments differ from active experiments in that passive experiments involve limited or no active intervention to generate observations and test hypotheses. Experiments of nature and natural experiments are two types of passive experiments. Experiments of nature include research participants with rare genetic or acquired conditions that facilitate exploration of specific physiological mechanisms. In this way, experiments of nature are parallel to classical "knockout" animal models among human research participants. Natural experiments are gleaned from data sets that allow population-based questions to be addressed. An advantage of both types of passive experiments is that more extreme and/or prolonged exposures to physiological and behavioral stimuli are possible in humans. In this article, we discuss a number of key passive experiments that have generated foundational medical knowledge or mechanistic physiological insights related to exercise. Both natural experiments and experiments of nature will be essential to generate and test hypotheses about the limits of human adaptability to stressors like exercise. © 2023 American Physiological Society. Compr Physiol 13:4879-4907, 2023.

Coronary Artery Calcium Dispersion and Cause-Specific Mortality

Coronary artery calcium (CAC) measures subclinical atherosclerosis and improves risk stratification. CAC characteristics-including vessel(s) involved, number of vessels, volume, and density-have been shown to differentially impact risk. We assessed how dispersion-either the number of calcified vessels or CAC phenotype (diffuse, normal, and concentrated)-impacted cause-specific mortality. The CAC Consortium is a retrospective cohort of 66,636 participants without coronary heart disease (CHD) who underwent CAC scoring. This study included patients with CAC >0 (n = 28,147). CAC area, CAC density, and CAC phenotypes (derived from the index of diffusion = 1 - [CAC in most concentrated vessel/total Agatston score]) were calculated. The associations between CAC characteristics and cause-specific mortality were assessed. The participant details included (n = 28,147): mean age 58.3 years, 25% female, 89.6% White, and 66% had 2+ calcified vessels. Diabetes, hypertension, and hyperlipidemia were predictors of multivessel involvement (p <0.001). After controlling for the overall CAC score, those with 4-vessel CAC involvement had more CAC area and less dense calcifications than those with 1-vessel. There was a graded increase in all-cause and cardiovascular disease (CVD)- and CHD-specific mortality as the number of calcified vessels increased. Among those with ≥2 vessels involved (n = 18,516), a diffuse phenotype was associated with a higher CVD-specific mortality and had a trend toward higher all-cause and CHD-specific mortality than a concentrated CAC phenotype. Diffuse CAC involvement was characterized by less dense calcification, more CAC area, multiple coronary vessel involvement, and presence of certain traditional risk factors. There is a graded increase in all-cause and CVD- and CHD-specific mortality with increasing CAC dispersion.

Discordance Between Coronary Artery Calcium Area and Density Predicts Long-Term Atherosclerotic Cardiovascular Disease Risk

BACKGROUND

Coronary artery calcium (CAC) is commonly quantified as the product of 2 generally correlated measures: plaque area and calcium density.

OBJECTIVES

The authors sought to determine whether discordance between calcium area and density has long-term prognostic importance in atherosclerotic cardiovascular disease (ASCVD) risk.

METHODS

The authors studied 10,373 primary prevention participants from the CAC Consortium with CAC >0. Based on their median values, calcium area and mean calcium density were divided into 4 mutually exclusive concordant/discordant groups. Cox proportional hazards regression assessed the association of calcium area/density groups with ASCVD mortality over a median of 11.7 years, adjusting for traditional risk factors and the Agatston CAC score.

RESULTS

The mean age was 56.7 years, and 24% were female. The prevalence of plaque discordance was 19% (9% low calcium area/high calcium density, 10% high calcium area/low calcium density). Female sex (odds ratio [OR]: 1.48 [95% CI: 1.27-1.74]) and body mass index (OR: 0.81 [95% CI: 0.76-0.87], per 5 kg/m2 higher) were significantly associated with high calcium density discordance, whereas diabetes (OR: 2.23 [95% CI: 1.85-3.19]) was most strongly associated with discordantly low calcium density. Compared to those with low calcium area/low calcium density, individuals with low calcium area/high calcium density had a 71% lower risk of ASCVD death (HR: 0.29 [95% CI: 0.09-0.95]).

CONCLUSIONS

For a given CAC score, high calcium density relative to plaque area confers lower long-term ASCVD risk, likely serving as an imaging marker of biological resilience for lesion vulnerability. Additional research is needed to define a robust definition of calcium area/density discordance for routine clinical risk prediction.

Relation of Exercise Capacity to Incident Heart Failure Among Men and Women With Coronary Heart Disease (from the Henry Ford Exercise Testing [FIT] Project)

Exercise capacity (EC) is inversely related to the risk of cardiovascular disease and incident heart failure (HF) in healthy subjects. However, there are no present studies that exclusively evaluate EC and the risk of incident HF in patients with known coronary heart disease (CHD). We aimed to determine the relation between EC and incident HF in patients with an established clinical diagnosis of CHD. We retrospectively identified 8,387 patients (age 61 ± 12 years; 30% women; 33% non-White) with a history of myocardial infarction (MI) or coronary revascularization procedure and no history of HF at the time of a clinically indicated exercise stress test completed between 1991 and 2009. EC was quantified in metabolic equivalents of task (METs) estimated from treadmill testing. Incident HF was identified through June 2010 from administrative databases based on ≥3 encounters with International Classification of Diseases, Ninth Revision 428.x. Cox regression analysis was used to evaluate the risk of incident HF associated with METs. Covariates included age; gender; race; hypertension, diabetes, hyperlipidemia, smoking, and MI; medications for CHD and lung diseases; and clinical indication for treadmill testing. During a median follow-up of 8.2 years (interquartile range 4.7 to 12.4 years) after the exercise test, 23% of the cohort experienced a new HF diagnosis. Lower EC categories were associated with higher HF incidence compared with METs ≥12, with nearly fourfold greater adjusted risk among patients with METs <6. Per unit increase in METs of EC was associated with a 12% lower adjusted risk for HF. There was no significant interaction based on race (p = 0.06), gender (p = 0.88), age ≤61 years (p = 0.60), history of MI (p = 0.31), or diabetes (p = 0.38). This study reveals that among men and women with CHD and no history of HF, EC is independently and inversely related to the risk of future HF.

Estimating Cardiorespiratory Fitness Without Exercise Testing or Physical Activity Status in Healthy Adults: Regression Model Development and Validation

Background

Low cardiorespiratory fitness (CRF) is an independent predictor of morbidity and mortality. Most health care settings use some type of electronic health record (EHR) system. However, many EHRs do not have CRF or physical activity data collected, thereby limiting the types of investigations and analyses that can be done.

Objective

This study aims to develop a nonexercise equation to estimate and classify CRF (in metabolic equivalent tasks) using variables commonly available in EHRs.

Methods

Participants were 42,676 healthy adults (female participants: n=9146, 21.4%) from the Aerobics Center Longitudinal Study examined from 1974 to 2005. The nonexercise estimated CRF was based on sex, age, measured BMI, measured resting heart rate, measured resting blood pressure, and smoking status. A maximal treadmill test measured CRF.

Results

After conducting nonlinear feature augmentation, separate linear regression models were used for male and female participants to calculate correlation and regression coefficients. Cross-classification of actual and estimated CRF was performed using low CRF categories (lowest quintile, lowest quartile, and lowest tertile). The multiple correlation coefficient (R) was 0.70 (mean deviation 1.33) for male participants and 0.65 (mean deviation 1.23) for female participants. The models explained 48.4% (SE estimate 1.70) and 41.9% (SE estimate 1.56) of the variance in CRF for male and female participants, respectively. Correct category classification for low CRF (lowest tertile) was found in 77.2% (n=25,885) of male participants and 74.9% (n=6,850) of female participants.

Conclusions

The regression models developed in this study provided useful estimation and classification of CRF in a large population of male and female participants. The models may provide a practical method for estimating CRF derived from EHRs for population health research.

Prognostic Interplay Between Coronary Artery Calcium Scoring and Cardiorespiratory FItness: The CAC-FIT Study

OBJECTIVE

To assess the incremental prognostic role of coronary artery calcium score (CACS) and exercise capacity (EC), two independent prognostic tests in the assessment of patients with coronary artery disease.

METHODS

The cohort consisted of patients who had clinically indicated exercise stress testing and CACS assessment from January 1, 2015, to September 30, 2021, with a median of 27 days between each other. Exercise capacity was defined by peak metabolic equivalents of task (METs) achieved during exercise stress test. The CACS was determined by the Agatston method. Patients were observed from the latest test date to incident major adverse cardiac events (inclusive of all-cause death, nonfatal myocardial infarction, late revascularization, and admission for heart failure).

RESULTS

There were a` total of 1932 patients in the study population (mean age, 56±12 years; 42% female, 48% hypertension, 21% diabetes, 48% dyslipidemia). Peak METs below 6 was achieved in 8% of patients, and the median (interquartile range) CACS was 9 (0-203). In multivariable Cox regression models, both CACS (1 unit increase in log CACS: hazard ratio, 1.19; 95% CI, 1.06 to 1.34; P=.003;) and EC (1 unit increase in peak METs: hazard ratio, 0.89; 95% CI, 0.81 to 0.97; P=.01) were independently associated with outcomes. Using CACS+EC added incremental prognostic value over clinical and fitness models (C index increase from 0.68 to 0.75; P=.015). Incident event rates increased across categories of CACS and EC.

CONCLUSION

Our analysis found that CACS and EC have complementary risk-stratifying roles in coronary artery disease.

Equipment-free, unsupervised high intensity interval training elicits significant improvements in the physiological resilience of older adults

Background

Reduced cardiorespiratory fitness (CRF) is an independent risk factor for dependency, cognitive impairment and premature mortality. High-intensity interval training (HIIT) is a proven time-efficient stimulus for improving both CRF and other facets of cardiometabolic health also known to decline with advancing age. However, the efficacy of equipment-free, unsupervised HIIT to improve the physiological resilience of older adults is not known.

Methods

Thirty independent, community-dwelling older adults (71(SD: 5) years) were randomised to 4 weeks (12 sessions) equipment-free, supervised (in the laboratory (L-HIIT)) or unsupervised (at home (H-HIIT)) HIIT, or a no-intervention control (CON). HIIT involved 5, 1-minute intervals of a bodyweight exercise each interspersed with 90-seconds recovery. CRF, exercise tolerance, blood pressure (BP), body composition, muscle architecture, circulating lipids and glucose tolerance were assessed at baseline and after the intervention period.

Results

When compared to the control group, both HIIT protocols improved the primary outcome of CRF ((via anaerobic threshold) mean difference, L-HIIT: +2.27, H-HIIT: +2.29, both p < 0.01) in addition to exercise tolerance, systolic BP, total cholesterol, non-HDL cholesterol and m. vastus lateralis pennation angle, to the same extent. There was no improvement in these parameters in CON. There was no change in diastolic BP, glucose tolerance, whole-body composition or HDL cholesterol in any of the groups.

Conclusions

This is the first study to show that short-term, time-efficient, equipment-free, HIIT is able to elicit improvements in the CRF of older adults irrespective of supervision status. Unsupervised HIIT may offer a novel approach to improve the physiological resilience of older adults, combating age-associated physiological decline, the rise of inactivity and the additional challenges currently posed by the COVID-19 pandemic.

Trial registration

This study was registered at clinicaltrials.gov and coded: NCT03473990.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12877-022-03208-y.

Mean Versus Peak Coronary Calcium Density on Non-Contrast CT: Calcium Scoring and ASCVD Risk Prediction

OBJECTIVES

This study sought to assess the relationship between mean vs peak calcified plaque density and their impact on calculating coronary artery calcium (CAC) scores and to compare the corresponding differential prediction of atherosclerotic cardiovascular disease (ASCVD) and coronary heart disease (CHD) mortality.

BACKGROUND

The Agatston CAC score is quantified per lesion as the product of plaque area and a 4-level categorical peak calcium density factor. However, mean calcium density may more accurately measure the heterogenous mixture of lipid-rich, fibrous, and calcified plaque reflective of ASCVD risk.

METHODS

We included 10,373 individuals from the CAC Consortium who had CAC >0 and per-vessel measurements of peak calcium density factor and mean calcium density. Area under the curve and continuous net reclassification improvement analyses were performed for CHD and ASCVD mortality to compare the predictive abilities of mean calcium density vs peak calcium density factor when calculating the Agatston CAC score.

RESULTS

Participants were on average 53.4 years of age, 24.4% were women, and the median CAC score was 68 Agatston units. The average values for mean calcium density and peak calcium density factor were 210 ± 50 HU and 3.1 ± 0.5, respectively. Individuals younger than 50 years of age and/or those with a total plaque area <100 mm2 had the largest differences between the peak and mean density measures. Among persons with CAC 1-99, the use of mean calcium density resulted in a larger improvement in ASCVD mortality net reclassification improvement (NRI) (NRI = 0.49; P < 0.001 vs. NRI = 0.18; P = 0.08) and CHD mortality discrimination (Δ area under the curve (AUC) = +0.169 vs +0.036; P < 0.001) compared with peak calcium density factor. Neither peak nor mean calcium density improved mortality prediction at CAC scores >100.

CONCLUSION

Mean and peak calcium density may differentially describe plaque composition early in the atherosclerotic process. Mean calcium density performs better than peak calcium density factor when combined with plaque area for ASCVD mortality prediction among persons with Agatston CAC 1-99.

Prediction of High-Altitude Cardiorespiratory Fitness Impairment Using a Combination of Physiological Parameters During Exercise at Sea Level and Genetic Information in an Integrated Risk Model

Insufficient cardiorespiratory compensation is closely associated with acute hypoxic symptoms and high-altitude (HA) cardiovascular events. To avoid such adverse events, predicting HA cardiorespiratory fitness impairment (HA-CRFi) is clinically important. However, to date, there is insufficient information regarding the prediction of HA-CRFi. In this study, we aimed to formulate a protocol to predict individuals at risk of HA-CRFi. We recruited 246 volunteers who were transported to Lhasa (HA, 3,700 m) from Chengdu (the sea level [SL], <500 m) through an airplane. Physiological parameters at rest and during post-submaximal exercise, as well as cardiorespiratory fitness at HA and SL, were measured. Logistic regression and receiver operating characteristic (ROC) curve analyses were employed to predict HA-CRFi. We analyzed 66 pulmonary vascular function and hypoxia-inducible factor- (HIF-) related polymorphisms associated with HA-CRFi. To increase the prediction accuracy, we used a combination model including physiological parameters and genetic information to predict HA-CRFi. The oxygen saturation (SpO₂) of post-submaximal exercise at SL and EPAS1 rs13419896-A and EGLN1 rs508618-G variants were associated with HA-CRFi (SpO₂, area under the curve (AUC) = 0.736, cutoff = 95.5%, p < 0.001; EPAS1 A and EGLN1 G, odds ratio [OR] = 12.02, 95% CI = 4.84–29.85, p < 0.001). A combination model including the two risk factors—post-submaximal exercise SpO₂ at SL of <95.5% and the presence of EPAS1 rs13419896-A and EGLN1 rs508618-G variants—was significantly more effective and accurate in predicting HA-CRFi (OR = 19.62, 95% CI = 6.42–59.94, p < 0.001). Our study employed a combination of genetic information and the physiological parameters of post-submaximal exercise at SL to predict HA-CRFi. Based on the optimized prediction model, our findings could identify individuals at a high risk of HA-CRFi in an early stage and reduce cardiovascular events.

Prognostic Value of Cardiorespiratory Fitness in Patients with Chronic Kidney Disease: The FIT (Henry Ford Exercise Testing) Project

PURPOSE

We conducted this study to investigate the association of cardiorespiratory fitness and all-cause mortality among patients with chronic kidney disease.

METHODS

We studied a retrospective cohort of patients from the Henry Ford Health System who underwent clinically indicated exercise stress testing with baseline cardiorespiratory fitness and estimated glomerular filtration rate measurement. Cardiorespiratory fitness was expressed as metabolic equivalents of task, and kidney function was categorized into stages according to estimated glomerular filtration rate. Multivariable-adjusted Cox proportional hazard models were used to examine the association between metabolic equivalents of task and all-cause mortality among patients with chronic kidney disease stages 3-5. Discrimination of mortality was assessed using receiver operating characteristic curves, while reclassification was evaluated using net reclassification index (NRI).

RESULTS

Among 50,121 participants, the mean age was 55 ± 12.6 years; 47.5% were women, 64.5% were white, and 6877 (13.7%) participants had chronic kidney disease stage 3-5. Over a median follow-up of 6.7 years, 6308 participants died (12.6%). Each 1-unit higher metabolic equivalents of task was associated with a significant 15% reduction in all-cause mortality (hazard ratio 0.85; 95% confidence interval [CI], 0.84-0.87). Metabolic equivalents of task improved discriminatory ability of mortality prediction when added to traditional risk factors and estimated glomerular filtration rate (area under the curve 0.7996; 95% CI, 0.789-0.810 vs 0.759; 95% CI, 0.748-0.770, respectively; P < .001). The addition of metabolic equivalents of task to traditional risk factors resulted in significant reclassification (6% for events, 5% for non-events: NRI = 0.13, P < .001).

CONCLUSIONS

Cardiorespiratory fitness improves mortality risk prediction among patients with chronic kidney disease. Cardiorespiratory fitness provides incremental prognostic information when added to traditional risk factors and may help guide treatment options among patients with renal dysfunction.

Fitness and Mortality Among Persons 70 Years and Older Across the Spectrum of Cardiovascular Disease Risk Factor Burden: The FIT Project

OBJECTIVE

To determine whether fitness could improve mortality risk stratification among older adults compared with cardiovascular disease (CVD) risk factors.

METHODS

We examined 6509 patients 70 years of age and older without CVD from the Henry Ford ExercIse Testing Project (FIT Project) cohort. Patients performed a physician-referred treadmill stress test between 1991 and 2009. Traditional categorical CVD risk factors (hypertension, hyperlipidemia, diabetes, and smoking) were summed from 0 to 3 or more. Fitness was grouped as low, moderate, and high (<6, 6 to 9.9, and ≥10 metabolic equivalents of task). All-cause mortality was ascertained through US Social Security Death Master files. We calculated age-adjusted mortality rates, multivariable adjusted Cox proportional hazards, and Kaplan-Meier survival models.

RESULTS

Patients had a mean age of 75±4 years, and 3385 (52%) were women; during a mean follow-up of 9.4 years, there were 2526 deaths. A higher fitness level (P<.001), not lower CVD risk factor burden (P=.31), was associated with longer survival. The age-adjusted mortality rate per 1000 person-years was 56.7 for patients with low fitness and 0 risk factors compared with 24.9 for high fitness and 3 or more risk factors. Among patients with 3 or more risk factors, the adjusted mortality hazard was 0.68 (95% CI, 0.61 to 0.76) for moderate and 0.51 (95% CI, 0.44 to 0.60) for high fitness compared with the least fit.

CONCLUSION

Among persons aged 70 years and older, there was no significant difference in survival of patients with 0 vs 3 or more risk factors, but a higher fitness level identified older persons with good long-term survival regardless of CVD risk factor burden.

Fitness and prostate cancer screening, incidence, and mortality: Results from the Henry Ford Exercise Testing (FIT) Project

BACKGROUND

The relation between cardiorespiratory fitness (CRF) and prostate cancer is not well established. The objective of this study was to determine whether CRF is associated with prostate cancer screening, incidence, or mortality.

METHODS

The Henry Ford Exercise Testing Project is a retrospective cohort study of men aged 40 to 70 years without cancer who underwent physician-referred exercise stress testing from 1995 to 2009. CRF was quantified in metabolic equivalents of task (METs) (<6 [reference], 6-9, 10-11, and ≥12 METs), estimated from the peak workload achieved during a symptom-limited, maximal exercise stress test. Prostate-specific antigen (PSA) testing, incident prostate cancer, and all-cause mortality were analyzed with multivariable adjusted Poisson regression and Cox proportional hazard models.

RESULTS

In total, 22,827 men were included, of whom 739 developed prostate cancer, with a median follow-up of 7.5 years. Men who had high fitness (≥12 METs) had an 28% higher risk of PSA screening (95% CI, 1.2-1.3) compared with those who had low fitness (<6 METs. After adjusting for PSA screening, fitness was associated with higher prostate cancer incidence (men aged <55 years, P = .02; men aged >55 years, P ≤ .01), but not with advanced prostate cancer. Among the men who were diagnosed with prostate cancer, high fitness was associated with a 60% lower risk of all-cause mortality (95% CI, 0.2-0.9).

CONCLUSIONS

Although men with high fitness are more likely to undergo PSA screening, this does not fully account for the increased incidence of prostate cancer seen among these individuals. However, men with high fitness have a lower risk of death after a prostate cancer diagnosis, suggesting that the cancers identified may be low-risk with little impact on long-term outcomes.

Relation of Absence of Coronary Artery Calcium to Cardiovascular Disease Mortality Risk Among Individuals Meeting Criteria for Statin Therapy According to the 2018/2019 ACC/AHA Guidelines

The 2013 American College of Cardiology and the American Heart Association (ACC/AHA) guidelines resulted in broad recommendations for preventive statin therapy allocation in patients without known cardiovascular disease (CVD). Subsequent studies demonstrated significant heterogeneity of atherosclerotic cardiovascular disease risk across the primary prevention population. In 2018/2019, the guidelines were revised to optimize risk assessment and cholesterol management. We sought to evaluate the heterogeneity of risk in statin-recommended patients, using coronary artery calcium (CAC) according to 2018/2019 ACC/AHA guidelines in a primary prevention cohort. We evaluated 5,800 statin-naive patients aged 40 to 75 years without known coronary heart disease from the Cedars-Sinai Medical Center study cohort. All participants underwent clinical CAC scoring for risk stratification and were followed for all-cause and CVD-specific mortality. A total of 181 deaths occurred including 54 CVD deaths over a follow-up of 9.5 years. Overall, 1,939 participants would have been recommended statin therapy, 32% of whom had no detectable CAC. CAC = 0 participants had the lowest all-cause and CVD mortality rates in both statin-recommended and nonrecommended groups (0.2 and 0.4 CVD deaths per 1,000 person-years, respectively). Absence of CAC in statin-naive patients portends an approximately 12-fold lower CVD mortality (0.2% vs 2.4%) in those recommended for statin therapy compared with any CAC present. In conclusion, in a cohort of patients meeting the 2018/2019 ACC/AHA guidelines for statin therapy for primary prevention, there was a marked heterogeneity of CAC scores, with about one-third of the statin recommended population having no detectable CAC (CAC = 0) with a significantly lower CVD mortality compared with CAC>0.

Coronary Artery Calcium and the Age-Specific Competing Risk of Cardiovascular Versus Cancer Mortality: The Coronary Artery Calcium Consortium

BACKGROUND

Coronary artery calcium (CAC) is a guideline recommended cardiovascular disease (CVD) risk stratification tool that increases with age and is associated with non-cardiovascular disease outcomes including cancer. We sought to define the age-specific change in the association between CAC and cause-specific mortality.

METHODS

The Coronary Artery Calcium Consortium includes 59,502 asymptomatic patients age 40-75 without known CVD. Age-stratified mortality rates and parametric survival regression modeling was performed to estimate the age-specific CAC score at which CVD and cancer mortality risk were equal.

RESULTS

The mean age was 54±8 years (67% men) and there were 2,423 deaths over a mean 12±3 years follow-up. Among individuals with CAC = 0, cancer was the leading cause of death, with low CVD mortality rates for both younger (40-54 years) 0.2/1,000 person-years and older participants (65-75 years) 1.3/1,000 person-years. When CAC ≥400, CVD was consistently the leading cause of death among younger (71% of deaths) and older participants (56% of deaths). The CAC score at which CVD overtook cancer as the leading cause of death increased exponentially with age and was approximately 115 at age 50 and 380 at age 65.

CONCLUSIONS

Regardless of age, when CAC = 0 cancer was the leading cause of death and the cardiovascular disease mortality rate was low. Our age-specific estimate for the CAC score at which CVD overtakes cancer mortality allows for a more precise approach to synergistic prediction and prevention strategies for CVD and cancer.

Exercise capacity is associated with hospital readmission among patients with diabetes

INTRODUCTION

Patients with diabetes are at greater risk of hospital readmission than patients without diabetes. There is a need to identify more modifiable risk factors for readmission as potential targets for intervention. Cardiorespiratory fitness is a predictor of morbidity and mortality. The purpose of this study was to examine whether there is an association between exercise capacity based on the maximal workload achieved during treadmill stress testing and readmission among patients with diabetes.

RESEARCH DESIGN AND METHODS

This retrospective cohort study included adult patients with diabetes discharged from an academic medical center between July 1, 2012 and December 31, 2018 who had a stress test documented before the index discharge. Univariate analysis and multinomial multivariable logistic regressions were used to evaluate associations with readmission within 30 days, 6 months, and 1 year of discharge. Exercise capacity was measured as metabolic equivalents (METs).

RESULTS

A total of 580 patients with 1598 hospitalizations were analyzed. Mean METs of readmitted patients were significantly lower than for non-readmitted patients (5.7 (2.6) vs 6.7 (2.6), p<0.001). After adjustment for confounders, a low METs level (<5) was associated with higher odds of readmission within 30 days (OR 5.46 (2.22-13.45), p<0.001), 6 months (OR 2.78 (1.36-5.65), p=0.005), and 1 year (OR 2.16 (1.12-4.16), p=0.022) compared with medium (5-7) and high (>7) METs level. During the 6.5-year study period, patients with low METs had a mean of 3.2±3.6 hospitalizations, while those with high METs had 2.5±2.4 hospitalizations (p=0.007).

CONCLUSIONS

Lower exercise capacity is associated with a higher risk of readmission within 30 days, 6 months, and 1 year, as well as a greater incidence of hospitalization, in patients with diabetes. Future studies are needed to explore whether exercise reduces readmission risk in this population.

Association of Body Mass Index With Coronary Artery Calcium and Subsequent Cardiovascular Mortality: The Coronary Artery Calcium Consortium

BACKGROUND

Obesity is associated with higher risk for coronary artery calcium (CAC), but the relationship between body mass index (BMI) and mortality is complex and frequently paradoxical.

METHODS

We analyzed BMI, CAC, and subsequent mortality using data from the CAC Consortium, a multi-centered cohort of individuals free of established cardiovascular disease (CVD) who underwent CAC testing. Mortality was assessed through linkage to the Social Security Death Index and cause of death from the National Death Index. Multivariable logistic regression was used to determine odds ratios for the association of clinically relevant BMI categories and prevalent CAC. Cox proportional hazards regression modeling was used to determine hazard ratios for coronary heart disease, CVD, and all-cause mortality according to categories of BMI and CAC.

RESULTS

Our sample included 36 509 individuals, mean age 54.1 (10.3) years, 34.4% female, median BMI 26.6 (interquartile range, 24.1-30.1), 46.6% had zero CAC, and 10.5% had CAC ≥400. Compared with individuals with normal BMI, the multivariable adjusted odds of CAC >0 were increased in those overweight (odds ratio, 1.13 [95% CI, 1.1-1.2]) and obese (odds ratio, 1.5 [95% CI, 1.4-1.6]). Over a median follow-up of 11.4 years, there were 1550 deaths (4.3%). Compared with normal BMI, obese individuals had a higher risk of coronary heart disease, CVD, and all-cause mortality while overweight individuals, despite a higher odds of CAC, showed no significant increase in mortality. In a sex-stratified analysis, the increase in coronary heart disease, CVD, and all-cause mortality in obese individuals appeared largely limited to men, and there was a lower risk of all-cause mortality in overweight women (hazard ratio, 0.79 [95% CI, 0.63-0.98]).

CONCLUSIONS

In a large sample undergoing CAC scoring, obesity was associated with a higher risk of CAC and subsequent coronary heart disease, CVD, and all-cause mortality. However, overweight individuals did not have a higher risk of mortality despite a higher risk for CAC.

Cardiorespiratory Fitness and Incident Stroke Types: The FIT (Henry Ford ExercIse Testing) Project

OBJECTIVE

To study the association between cardiorespiratory fitness (CRF) and incident stroke types.

PATIENTS AND METHODS

We studied a retrospective cohort of patients referred for treadmill stress testing in the Henry Ford Health System (Henry Ford ExercIse Testing Project) without history of stroke. CRF was expressed by metabolic equivalents of task (METs). Using appropriate International Classification of Diseases, Ninth Revision codes, incident stroke was ascertained through linkage with administrative claims files and classified as ischemic, hemorrhagic, and subarachnoid hemorrhage (SAH). Multivariable-adjusted Cox proportional hazards models examined the association between CRF and incident stroke.

RESULTS

Among 67,550 patients, mean ± SD age was 54±13 years, 46% (n=31,089) were women, and 64% (n=43,274) were white. After a median follow-up of 5.4 (interquartile range 2.7-8.5) years, a total of 7512 incident strokes occurred (6320 ischemic, 2481 hemorrhagic, and 275 SAH). Overall, there was a graded lower incidence of stroke with higher MET categories. Patients with METs of 12 or more had lower risk of overall stroke [0.42 (95% CI, 0.36-0.49)], ischemic stroke [0.69 (95% CI, 0.58-0.82)], and hemorrhagic stroke [0.71 (95% CI, 0.52-0.95)].

CONCLUSION

In a large ethnically diverse cohort of patients referred for treadmill stress testing, CRF is inversely associated with risk for ischemic and hemorrhagic stroke.

Coronary Artery Calcium as a Synergistic Tool for the Age- and Sex-Specific Risk of Cardiovascular and Cancer Mortality: The Coronary Artery Calcium Consortium

Background

Coronary artery calcium (CAC) is a predictor for the development of cardiovascular disease (CVD) and to a lesser extent cancer. The age‐ and sex‐specific relationship of CAC with CVD and cancer mortality is unknown.

Methods and Results

Asymptomatic patients aged 40 to 75 years old without known CVD were included from the CAC Consortium. We calculated sex‐specific mortality rates per 1000 person‐years’ follow‐up. Using parametric survival regression modeling, we determined the age‐ and sex‐specific CAC score at which the risk of death from CVD and cancer were equal. Among the 59 502 patients included in this analysis, the mean age was 54.9 (±8.5) years, 34% were women, and 89% were white. There were 671 deaths attributable to CVD and 954 deaths attributable to cancer over a mean follow‐up of 12±3 years. Among patients with CAC=0, cancer was the leading cause of death, the total mortality rate was low (women, 1.8; men, 1.5), and the CVD mortality rate was exceedingly low for women (0.3) and men (0.3). The age‐specific CAC score at which the risk of CVD and cancer mortality were equal had a U‐shaped relationship for women, while the relationship was exponential for men.

Conclusions

The age‐ and sex‐specific relationship of CAC with CVD and cancer mortality differed significantly for women and men. Our age‐ and sex‐specific CAC score provides a more precise estimate and further facilitates the use of CAC as a synergistic tool in strategies for the prediction and prevention of CVD and cancer mortality.

Myocardial Perfusion Imaging Using Positron Emission Tomography

Coronary artery disease (CAD), also known as ischemic heart disease, is a major cause of morbidity and mortality worldwide, and timely noninvasive diagnosis of clinical and subclinical CAD is imperative to mitigate its burden on individual patients and populations. Positron emission tomography (PET) is a versatile tool that can perform relative myocardial perfusion imaging (MPI) with high accuracy; furthermore, it provides valuable information about the coronary microvasculature using rest and stress myocardial blood flow (MBF) and coronary flow reserve (CFR) measurements. Several radiotracers are approved by the US Food and Drug Administration to help with MPI, MBF, and CFR evaluation. A large body of evidence indicates that evaluation of the coronary microcirculation using MBF and CFR provides strong diagnostic and prognostic data in a multitude of patient populations. This review describes the technical aspects of PET compared to other modalities and discusses its clinical uses for diagnosis and prognosis of coronary arterial epicardial and microcirculatory disease.

Association of BMI, Fitness, and Mortality in Patients With Diabetes: Evaluating the Obesity Paradox in the Henry Ford Exercise Testing Project (FIT Project) Cohort

OBJECTIVE

To determine the effect of fitness on the association between BMI and mortality among patients with diabetes.

RESEARCH DESIGN AND METHODS

We identified 8,528 patients with diabetes (self-report, medication use, or electronic medical record diagnosis) from the Henry Ford Exercise Testing Project (FIT Project). Patients with a BMI <18.5 kg/m² or cancer were excluded. Fitness was measured as the METs achieved during a physician-referred treadmill stress test and categorized as low (<6), moderate (6-9.9), or high (≥10). Adjusted hazard ratios for mortality were calculated using standard BMI (kilograms per meter squared) cutoffs of normal (18.5-24.9), overweight (25-29.9), and obese (≥30). Adjusted splines centered at 22.5 kg/m² were used to examine BMI as a continuous variable.

RESULTS

Patients had a mean age of 58 ± 11 years (49% women) with 1,319 deaths over a mean follow-up of 10.0 ± 4.1 years. Overall, obese patients had a 30% lower mortality hazard (P < 0.001) compared with normal-weight patients. In adjusted spline modeling, higher BMI as a continuous variable was predominantly associated with a lower mortality risk in the lowest fitness group and among patients with moderate fitness and BMI ≥30 kg/m². Compared with the lowest fitness group, patients with higher fitness had an ∼50% (6-9.9 METs) and 70% (≥10 METs) lower mortality hazard regardless of BMI (P < 0.001).

CONCLUSIONS

Among patients with diabetes, the obesity paradox was less pronounced for patients with the highest fitness level, and these patients also had the lowest risk of mortality.

Comparing Risk Scores in the Prediction of Coronary and Cardiovascular Deaths: Coronary Artery Calcium Consortium

OBJECTIVES

This study compared risk discrimination for the prediction of coronary heart disease (CHD) and cardiovascular disease (CVD) deaths for the Pooled Cohort Equations (PCE), the MESA (Multi-Ethnic Study of Atherosclerosis) Risk Score (with and without coronary artery calcium [CAC]), and of simple addition of CAC to the PCE.

BACKGROUND

The PCE predict 10-year risk of atherosclerotic CVD events, and the MESA Risk Score predicts risk of CHD. Their comparative performance for the prediction of fatal events is poorly understood.

METHODS

We evaluated 53,487 patients ages 45 to 79 years from the CAC Consortium, a retrospective cohort study of asymptomatic individuals referred for clinical CAC scoring. Risk discrimination was measured using C-statistics.

RESULTS

Mean age was 57 years, 35% were women, and 39% had CAC of 0. There were 421 CHD and 775 CVD deaths over a mean 12-year follow-up. In the overall study population, discrimination with the MESA Risk Score with CAC and the PCE was almost identical for both outcomes (C-statistics: 0.80 and 0.79 for CHD death, 0.77 and 0.78 for CVD death, respectively). Addition of CAC to the PCE improved risk discrimination, yielding the largest C-statistics. The MESA Risk Score with CAC and the PCE plus CAC showed the best discrimination among the 45% of patients with 5% to 20% estimated risk. Secondary analyses by estimated CVD risk strata showed modestly improved risk discrimination with CAC also among low- and high-estimated risk groups.

CONCLUSIONS

Our findings support the current guideline recommendation to use, among available risk scores, the PCE for initial risk assessment and to use CAC for further risk assessment in a broad borderline and intermediate risk group. Also, in select individuals at low or high estimated risk, CAC modestly improved discrimination. Studies in unselected populations will lead to further understanding of the potential value of tools combining risk scores and CAC for optimal risk assessment.

Validation of the Coronary Artery Calcium Data and Reporting System (CAC-DRS): Dual importance of CAC score and CAC distribution from the Coronary Artery Calcium (CAC) consortium

BACKGROUND

The Coronary Artery Calcium Data and Reporting System (CAC-DRS), which takes into account the Agatston score category (A) and the number of calcified vessels (N) has not yet been validated in terms of its prognostic significance.

METHODS

We included 54,678 patients from the CAC Consortium, a large retrospective clinical cohort of asymptomatic individuals free of baseline cardiovascular disease (CVD). CAC-DRS groups were derived from routine, cardiac-gated CAC scans. Cox proportional hazards regression models, adjusted for traditional CVD risk factors, were used to assess the association between CAC-DRS groups and CHD, CVD, and all-cause mortality. CAC-DRS was then compared to CAC score groups and regional CAC distribution using area under the curve (AUC) analysis.

RESULTS

The study population had a mean age of 54.2 ± 10.7, 34.4% female, and mean ASCVD score 7.3% ± 9.0. Over a mean follow-up of 12 ± 4 years, a total of 2,469 deaths (including 398 CHD deaths and 762 CVD deaths) were recorded. There was a graded risk for CHD, CVD and all-cause mortality with increasing CAC-DRS groups ranging from an all-cause mortality rate of 1.2 per 1,000 person-years for A0 to 15.4 per 1,000 person-years for A3/N4. In multivariable-adjusted models, those with CAC-DRS A3/N4 had significantly higher risk for CHD mortality (HR 5.9 (95% CI 3.6-9.9), CVD mortality (HR4.0 (95% CI 2.8-5.7), and all-cause mortality a (HR 2.5 (95% CI 2.1-3.0) compared to CAC-DRS A0. CAC-DRS had higher AUC than CAC score groups (0.762 vs 0.754, P < 0.001) and CAC distribution (0.762 vs 0.748, P < 0.001).

CONCLUSION

The CAC-DRS system, combining the Agatston score and the number of vessels with CAC provides better stratification of risk for CHD, CVD, and all-cause death than the Agatston score alone. These prognostic data strongly support new SCCT guidelines recommending the use CAC-DRS scoring.

Interplay of Coronary Artery Calcium and Risk Factors for Predicting CVD/CHD Mortality: The CAC Consortium

OBJECTIVES

This study sought to evaluate the association and burden of coronary artery calcium (CAC) with long-term, cause-specific mortality across the spectrum of baseline risk.

BACKGROUND

Although CAC is a known predictor of short-term, all-cause mortality, data on long-term and cause-specific mortality are inadequate.

METHODS

The CAC Consortium cohort is a multicenter cohort of 66,636 participants without coronary heart disease (CHD) who underwent CAC testing. The following risk factors (RFs) were considered: 1) current cigarette smoking; 2) dyslipidemia; 3) diabetes mellitus; 4) hypertension; and 5) family history of CHD.

RESULTS

During the 12.5-years median follow-up, 3,158 (4.7%) deaths occurred; 32% were cardiovascular disease (CVD) deaths. Participants with CAC scores ≥400 had a significantly increased risk for CHD and CVD mortality (hazard ratio [HR]: 5.44; 95% confidence interval [CI]: 3.88 to 7.62; and HR: 4.15; 95% CI: 3.29 to 5.22, respectively) compared with CAC of 0. Participants with ≥3 RFs had a smaller increased risk for CHD and CVD mortality (HR: 2.09; 95% CI: 1.52 to 2.85; and HR: 1.84; 95% CI: 1.46 to 2.31, respectively) compared with those without RFs. Across RF strata, CAC added prognostic information. For example, participants without RFs but with CAC ≥400 had significantly higher all-cause, non-CVD, CVD, and CHD mortality rates compared with participants with ≥3 RFs and CAC of 0.

CONCLUSIONS

Across the spectrum of RF burden, a higher CAC score was strongly associated with long-term, all-cause mortality and a greater proportion of deaths due to CVD and CHD. Absence of CAC identified people with a low risk over 12 years of follow-up, with most deaths being non-CVD in nature, regardless of RF burden.

Race/Ethnicity and the Prognostic Implications of Coronary Artery Calcium for All-Cause and Cardiovascular Disease Mortality: The Coronary Artery Calcium Consortium

Background

Coronary artery calcium (CAC) predicts cardiovascular disease (CVD) events; however, less is known about how its prognostic implications vary by race/ethnicity.

Methods and Results

A total of 38 277 whites, 1621 Asians, 977 blacks, and 1349 Hispanics from the CAC Consortium (mean age 55 years, 35% women) were followed over a median of 11.7 years. Modeling CAC in continuous and categorical (CAC=0; CAC 1–99; CAC 100–399; CAC ≥400) forms, we assessed its predictive value for all‐cause and CVD mortality by race/ethnicity using Cox proportional hazards and Fine and Gray competing‐risk regression, respectively. We also assessed the impact of race/ethnicity on risk within individual CAC strata, using whites as the reference. Models were adjusted for traditional cardiovascular risk factors. Increased CAC was associated with higher total and CVD mortality risk in all race/ethnicity groups, including Asians. However, the risk gradient with increasing CAC was more pronounced in blacks and Hispanics. In Fine and Gray subdistribution hazards models adjusted for traditional cardiovascular risk factors and CAC (continuous), blacks (subdistribution hazard ratio 3.4, 95% confidence interval, 2.5–4.8) and Hispanics (subdistribution hazard ratio 2.3, 95% confidence interval, 1.6–3.2) showed greater risk of CVD mortality when compared with whites, while Asians had risk similar to whites. These race/ethnic differences persisted when CAC=0.

Conclusions

CAC predicts all‐cause and CVD mortality in all studied race/ethnicity groups, including Asians and Hispanics, who may be poorly represented by the Pooled Cohort Equations. Blacks and Hispanics may have greater mortality risk compared with whites and Asians after adjusting for atherosclerosis burden, with potential implications for US race/ethnic healthcare disparities research.

The Interplay of the Global Atherosclerotic Cardiovascular Disease Risk Scoring and Cardiorespiratory Fitness for the Prediction of All-Cause Mortality and Myocardial Infarction: The Henry Ford ExercIse Testing Project (The FIT Project)

Cardiorespiratory fitness (CRF) is inversely associated with atherosclerotic cardiovascular disease (ASCVD) risk. It is unclear whether the prognostic value of CRF differs by baseline estimated ASCVD risk. We studied a retrospective cohort of patients without known heart failure or myocardial infarction (MI) who underwent treadmill stress testing. CRF was measured by metabolic equivalents of task (METs) and ASCVD risk was calculated using the Pooled Cohorts Equations. Multivariable-adjusted Cox regressions analyses examined the association between METs and incident all-cause mortality and MI outcomes stratified by baseline ASCVD risk. The C-index evaluated risk discrimination while net reclassification improvement evaluated reclassification with CRF added to the ASCVD risk score. Our study population consisted of 57,999 patients of mean age 53 (13) years, 49% women, 64% white, 29% black. Over a median follow-up 11 years (interquartile range 8 to 14 years) there were 6,670 (11%) deaths, while there were 1,757 (3.0%) MIs over a median follow-up of 6 years (interquartile range 3 to 8 years). Among patients with ASCVD risk ≥20%, those with METs ≥12 had a 77% lower risk of all-cause mortality (Hazard ratio 0.23 95% confidence interval = 0.20, 0.27) and 67% lower risk of MI (Hazard ratio 0.33 95% confidence interval = 0.24, 0.46) compared to METs <6. Similar results were obtained for those with ASCVD risk <5%. Addition of METs to ASCVD risk score improved the C-statistic from 0.778 to 0.798 for all-cause mortality and 0.726 to 0.733 for MI (both p <0.001). Addition of METs to ASCVD risk score significantly reclassified risk of all-cause mortality (p <0.001) but not MI (p = 0.052). In conclusion, CRF is inversely associated with risk of all-cause mortality and MI at all levels of ASCVD risk, and provides incremental risk discrimination and reclassification beyond the ASCVD risk score.

Cardiorespiratory fitness and incident lung and colorectal cancer in men and women: Results from the Henry Ford Exercise Testing (FIT) cohort

BACKGROUND

To the authors' knowledge, the relationship between cardiorespiratory fitness (CRF) and lung and colorectal cancer outcomes is not well established.

METHODS

A retrospective cohort study was performed of 49,143 consecutive patients who underwent clinician-referred exercise stress testing from 1991 through 2009. The patients ranged in age from 40 to 70 years, were without cancer, and were treated within the Henry Ford Health System in Detroit, Michigan. CRF, measured in metabolic equivalents of task (METs), was categorized as <6 (reference), 6 to 9, 10 to 11, and ≥12. Incident cancer was obtained through linkage to the cancer registry and all-cause mortality from the National Death Index.

RESULTS

Participants had a mean age of 54 ± 8 years. Approximately 46% were female, 64% were white, 29% were black, and 1% were Hispanic. The median follow-up was 7.7 years. Cox proportional hazard models, adjusted for age, race, sex, body mass index, smoking history, and diabetes, found that those in the highest fitness category (METs ≥12) had a 77% decreased risk of lung cancer (hazard ratio [HR], 0.23; 95% CI, 0.14-0.36) and a 61% decreased risk of incident colorectal cancer (HR, 0.39; 95% CI, 0.23-0.66; with additional adjustment for aspirin and statin use). Among those diagnosed with lung and colorectal cancer, those with high fitness had a decreased risk of subsequent death of 44% and 89%, respectively (HR, 0.56 [95% CI, 0.32-1.00] and HR, 0.11 [95% CI, 0.03-0.37], respectively).

CONCLUSIONS

In what to the authors' knowledge is the largest study performed to date, higher CRF was associated with a lower risk of incident lung and colorectal cancer in men and women and a lower risk of all-cause mortality among those diagnosed with lung or colorectal cancer.

On the interpretability of machine learning-based model for predicting hypertension

BACKGROUND

Although complex machine learning models are commonly outperforming the traditional simple interpretable models, clinicians find it hard to understand and trust these complex models due to the lack of intuition and explanation of their predictions. The aim of this study to demonstrate the utility of various model-agnostic explanation techniques of machine learning models with a case study for analyzing the outcomes of the machine learning random forest model for predicting the individuals at risk of developing hypertension based on cardiorespiratory fitness data.

METHODS

The dataset used in this study contains information of 23,095 patients who underwent clinician-referred exercise treadmill stress testing at Henry Ford Health Systems between 1991 and 2009 and had a complete 10-year follow-up. Five global interpretability techniques (Feature Importance, Partial Dependence Plot, Individual Conditional Expectation, Feature Interaction, Global Surrogate Models) and two local interpretability techniques (Local Surrogate Models, Shapley Value) have been applied to present the role of the interpretability techniques on assisting the clinical staff to get better understanding and more trust of the outcomes of the machine learning-based predictions.

RESULTS

Several experiments have been conducted and reported. The results show that different interpretability techniques can shed light on different insights on the model behavior where global interpretations can enable clinicians to understand the entire conditional distribution modeled by the trained response function. In contrast, local interpretations promote the understanding of small parts of the conditional distribution for specific instances.

CONCLUSIONS

Various interpretability techniques can vary in their explanations for the behavior of the machine learning model. The global interpretability techniques have the advantage that it can generalize over the entire population while local interpretability techniques focus on giving explanations at the level of instances. Both methods can be equally valid depending on the application need. Both methods are effective methods for assisting clinicians on the medical decision process, however, the clinicians will always remain to hold the final say on accepting or rejecting the outcome of the machine learning models and their explanations based on their domain expertise.

Association of Coronary Artery Calcium With Long-term, Cause-Specific Mortality Among Young Adults

IMPORTANCE

The level of coronary artery calcium (CAC) can effectively stratify cardiovascular risk in middle-aged and older adults, but its utility for young adults is unclear.

OBJECTIVES

To determine the prevalence of CAC in adults aged 30 to 49 years and the subsequent association of CAC with coronary heart disease (CHD), cardiovascular disease (CVD), and all-cause mortality.

DESIGN, SETTING, AND PARTICIPANTS

A multicenter retrospective cohort study was conducted among 22 346 individuals from the CAC Consortium who underwent CAC testing (baseline examination, 1991-2010, with follow-up through June 30, 2014; CAC quantified using nonconrast, cardiac-gated computed tomography scans) for clinical indications and were followed up for cause-specific mortality. Participants were free of clinical CVD at baseline. Statistical analysis was performed from June 1, 2017, to May 31, 2018.

MAIN OUTCOMES AND MEASURES

The prevalence of CAC and the subsequent rates of CHD, CVD, and all-cause mortality. Competing risks regression modeling was used to calculate multivariable-adjusted subdistribution hazard ratios for CHD and CVD mortality.

RESULTS

The sample of 22 346 participants (25.0% women and 75.0% men; mean [SD] age, 43.5 [4.5] years) had a high prevalence of hyperlipidemia (49.6%) and family history of CHD (49.3%) but a low prevalence of current smoking (11.0%) and diabetes (3.9%). The prevalence of any CAC was 34.4%, with 7.2% having a CAC score of more than 100. During follow-up (mean [SD], 12.7 [4.0] years), there were 40 deaths related to CHD, 84 deaths related to CVD, and 298 total deaths. A total of 27 deaths related to CHD (67.5%) occurred among individuals with CAC at baseline. The CHD mortality rate per 1000 person-years was 10-fold higher among those with a CAC score of more than 100 (0.69; 95% CI, 0.41-1.16) compared with those with a CAC score of 0 (0.07; 95% CI, 0.04-0.12). After multivariable adjustment, those with a CAC score of more than 100 had a significantly increased risk of CHD (subdistribution hazard ratio, 5.6; 95% CI, 2.5-12.7), CVD (subdistribution hazard ratio, 3.3; 95% CI, 1.8-6.2), and all-cause mortality (hazard ratio, 2.6; 95% CI, 1.9-3.6) compared with those with a CAC score of 0.

CONCLUSIONS AND RELEVANCE

In a large sample of young adults undergoing CAC testing for clinical indications, 34.4% had CAC, and those with elevated CAC scores had significantly higher rates of CHD and CVD mortality. Coronary artery calcium may have potential utility for clinical decision-making among select young adults at elevated risk of cardiovascular disease.

Coronary artery calcium scoring in low risk patients with family history of coronary heart disease: Validation of the SCCT guideline approach in the coronary artery calcium consortium

BACKGROUND

The Society of Cardiovascular Computed Tomography (SCCT) recommends consideration of coronary artery calcium (CAC) scoring among individuals with a family history (FH) of coronary heart disease (CHD) and atherosclerotic cardiovascular disease (ASCVD) risk <5%. No dedicated study has examined the prognostic significance of CAC scoring among this population.

METHODS

The CAC Consortium is a multi-center observational cohort study from four clinical centers linked to long-term follow-up for cause-specific mortality. All CAC scans were physician referred and performed in patients without a history of CHD. Our analysis includes 14,169 patients with ASCVD scores <5% and self-reported FH of CHD.

RESULTS

This cohort had a mean age of 48.1 (SD 7.4), was 91.3% white, 47.4% female, had an average ASCVD score of 2.3% (SD 1.3), and 59.4% had a CAC = 0. The event rate for all-cause mortality was 1.2 per 1000 person-years, 0.3 per 1000 person-years for CVD-specific mortality, and 0.2 per 1000 person-years for CHD-specific mortality. In multivariable Cox proportional hazard models, those with CAC>100 had a 2.2 (95% CI 1.5-3.3) higher risk of all-cause mortality, 4.3 (95% CI 1.9-9.5) times higher risk of CVD-specific mortality, and a 10.4 (95% CI 3.2-33.7) times higher risk of CHD-specific mortality compared to individuals with CAC = 0. The NNS to detect CAC >100 in this sample was 9.

CONCLUSION

In otherwise low risk patients with FH of CHD, CAC>100 were associated with increased risk of all-cause and CHD mortality with event rates in a range that may benefit with preventive pharmacotherapy. These data strongly support new SCCT recommendations regarding testing of patients with a family history of CHD.

Relation of Isolated Low High-Density Lipoprotein Cholesterol to Mortality and Cardiorespiratory Fitness (from the Henry Ford Exercise Testing Project [FIT Project])

Isolated low high-density lipoprotein cholesterol (HDL-C) is associated with lower fitness and increased mortality. Whether the association between isolated low HDL-C and mortality differs by fitness is uncertain. Patients in the Henry Ford ExercIse Testing Project (FIT Project) completed a physician-referred treadmill stress test and those prescribed lipid-lowering medications or with known cardiovascular disease were excluded. Isolated low HDL-C was defined as HDL-C <40 mg/dl for men and <50 mg/dl for women with low-density lipoprotein cholesterol (LDL-C) and triglycerides <100 mg/dl (n = 688). An optimal lipid panel was defined as HDL-C ≥40 mg/dl for men and ≥50 mg/dl for women with LDL-C and triglycerides <100 mg/dl (n = 2,923). Mortality was ascertained through Social Security Death Index linkage. Patients with isolated low HDL-C had a mean age of 48.9 ± 12.9 years and 62.9% were women. Over a mean follow-up of 10.3 ± 5 years, 12.8% of patients with isolated low HDL-C and 8.7% with optimal lipids died. Compared to individuals with optimal lipids, those with isolated low HDL-C who achieved <6 METs had a lower survival (p = 0.02), whereas there was no mortality difference for those who achieved 6 to 10 METs (p = 0.13) or ≥10 METs (p = 0.66). In adjusted Cox models, the mortality hazard for those with isolated low HDL-C compared with optimal lipids was 1.73 (95% confidence interval [CI] 1.18 to 2.54), 1.90 (95% CI 1.19 to 3.04), and 0.97 (95% CI 0.53 to 1.78) for the METS categories of <6, 6 to 10, and ≥10. In conclusion, individuals with isolated low HDL-C fitness significantly improved risk stratification and only those with lower fitness had an increased totality mortality risk.

Trends in cardiorespiratory fitness: The evolution of exercise treadmill testing at a single Academic Medical Center from 1970 to 2012

OBJECTIVE

To identify temporal trends in the use of exercise treadmill testing (ETT) and cardiorespiratory fitness (CRF) estimated by ETT in metabolic equivalents (METs).

PATIENTS AND METHODS

We compiled an ETT database of all available treadmill tests-including those with concomitant stress echocardiography and nuclear perfusion imaging studies-performed at Duke University Hospital from January 1, 1970- December 31, 2012. Six different ramp protocols were used in these combined modalities. CRF at maximal exertion was estimated using established metrics. Eligible patients were required to have no missing data on maximal treadmill speed, grade, and protocol.

RESULTS

The most commonly used ETT protocol was the Bruce (n = 28,877), followed by manual test (n = 7390). Since the 1980's, the use of ETT for clinical purposes declined substantially; there was a decreased trend in utilization of 9.4% over the decades 1990-1999 and 2000-2009. When standard protocol (Bruce) was assessed in isolation, trends in CRF decreased progressively from 1970 to 2012 (mean METs (standard deviation): 11.7 (4.3) to 10.5 (3.5)). After adjusting for baseline comorbidities, the trend was reduced to a lesser degree.

CONCLUSIONS

The use of ETT at our institution has declined over time, perhaps due to changes in clinical practice. In patients undergoing ETT using the standard Bruce protocol, CRF decreased progressively over the last five decades. Future studies are needed to clarify the etiology of the decrease in use of such a powerful predictor of clinical outcomes in our medical care environment.

Long-Term All-Cause and Cause-Specific Mortality in Asymptomatic Patients With CAC ≥1,000: Results From the CAC Consortium

OBJECTIVES

This study thoroughly explored the demographic and imaging characteristics, as well as the all-cause and cause-specific mortality risks of patients with a coronary artery calcium (CAC) score ≥1,000 in the largest dataset of this population to date.

BACKGROUND

CAC is commonly used to quantify cardiovascular risk. Current guidelines classify a CAC score of >300 or 400 as the highest risk group, yet little is known about the potentially unique imaging characteristics and mortality risk in individuals with a CAC score ≥1,000.

METHODS

A total of 66,636 asymptomatic adults were included from the CAC consortium, a large retrospective multicenter clinical cohort. Mean patient follow-up was 12.3 ± 3.9 years for patients with cardiovascular disease (CVD), coronary heart disease (CHD), cancer, and all-cause mortality. Multivariate Cox proportional hazards regression models adjusted for age, sex, and conventional risk factors were used to assess the relative mortality hazard of individuals with CAC ≥1,000 compared with, first, a CAC reference of 0, and second, with patients with a CAC score of 400 to 999.

RESULTS

There were 2,869 patients with CAC ≥1,000 (86.3% male, mean 66.3 ± 9.7 years of age). Most patients with CAC ≥1,000 had 4-vessel CAC (mean: 3.5 ± 0.6 vessels) and had greater total CAC area, higher mean CAC density, and more extracoronary calcium (79% with thoracic artery calcium, 46% with aortic valve calcium, and 21% with mitral valve calcium) than those with CAC scores of 400 to 999. After full adjustment, those with CAC ≥1,000 had a 5.04- (95% confidence interval [CI]: 3.92 to 6.48), 6.79- (95% CI: 4.74 to 9.73), 1.55- (95% CI:1.23 to 1.95), and 2.89-fold (95% CI: 2.53 to 3.31) risk of CVD, CHD, cancer, and all-cause mortality, respectively, compared to those with CAC score of 0. The CAC ≥1,000 group had a 1.71- (95% CI: 1.41 to 2.08), 1.84- (95% CI: 1.43 to 2.36), 1.36- (95% CI:1.07 to 1.73), and 1.51-fold (95% CI: 1.33 to 1.70) increased risk of CVD, CHD, cancer, and all-cause mortality compared to those with CAC scores 400 to 999. Graphic analysis of CAC ≥1,000 patients revealed continued logarithmic increase in risk, with no clear evidence of a risk plateau.

CONCLUSIONS

Patients with extensive CAC (CAC ≥1,000) represent a unique very high-risk phenotype with mortality outcomes commensurate with high-risk secondary prevention patients. Future guidelines should consider CAC ≥1,000 patients to be a distinct risk group who may benefit from the most aggressive preventive therapy.

Higher cardiorespiratory fitness predicts long-term survival in patients with heart failure and preserved ejection fraction: the Henry Ford Exercise Testing (FIT) Project

INTRODUCTION

Higher cardiorespiratory fitness (CRF) is associated with improved exercise capacity and quality of life in heart failure with preserved ejection fraction (HFpEF), but there are no large studies evaluating the association of HFpEF, CRF, and long-term survival. We therefore aimed to determine the association between CRF and all-cause mortality, in patients with HFpEF.

MATERIAL AND METHODS

In the Henry Ford Exercise Testing (FIT) Project, 167 patients had baseline HFpEF, defined as a clinical diagnosis of heart failure with ejection fraction ≥ 50% on echocardiogram. The CRF was estimated from the peak workload (in METs) from a clinician-referred treadmill stress test and categorized as poor (1-4 METs), intermediate (5-6 METs), and moderate-high (≥ 7 METs). Additional analyses assessing the effect of HFpEF and CRF on mortality were also conducted, matching HFpEF patients to non-HFpEF patients using propensity scores.

RESULTS

Mean age was 64 ±13 years, with 55% women, and 46% Black. Over a median follow-up of 9.7 (5.2-18.9) years, there were 103 deaths. In fully adjusted models, moderate-high CRF was associated with 63% lower mortality risk (HR = 0.37, 95% CI: 0.18-0.73) compared to the poor-CRF group. In the propensity-matched cohort, HFpEF was associated with a HR of 2.3 (95% CI: 1.7-3.2) for mortality compared to non-HFpEF patients, which was attenuated to 1.8 (95% CI: 1.3-2.5) after adjusting for CRF.

CONCLUSIONS

Moderate-high CRF in patients with HFpEF is associated with improved survival, and differences in CRF partly explain the intrinsic risk of HFpEF. Randomized trials of interventions aimed at improving CRF in HFpEF are needed.

Exercise Capacity and the Obesity Paradox in Heart Failure: The FIT (Henry Ford Exercise Testing) Project

OBJECTIVES

To assess the influence of exercise capacity and body mass index (BMI) on 10-year mortality in patients with heart failure (HF) and to synthesize these results with those of previous studies.

PATIENTS AND METHODS

This large biracial sample included 774 men and women (mean age, 60±13 years; 372 [48%] black) with a baseline diagnosis of HF from the Henry Ford Exercise Testing (FIT) Project. All patients completed a symptom-limited maximal treadmill stress test from January 1, 1991, through May 31, 2009. Patients were grouped by World Health Organization BMI categories for Kaplan-Meier survival analyses and stratified by exercise capacity (<4 and ≥4 metabolic equivalents [METs] of task). Associations of BMI and exercise capacity with all-cause mortality were assessed using multivariable-adjusted Cox proportional hazards models.

RESULTS

During a mean follow-up of 10.1±4.6 years, 380 patients (49%) died. Kaplan-Meier survival plots revealed a significant positive association between BMI category and survival for exercise capacity less than 4 METs (log-rank, P=.05), but not greater than or equal to 4 METs (P=.76). In the multivariable-adjusted models, exercise capacity (per 1 MET) was inversely associated, but BMI was not associated, with all-cause mortality (hazard ratio, 0.89; 95% CI, 0.85-0.94; P<.001 and hazard ratio, 0.99; 95% CI, 0.97-1.01; P=.16, respectively).

CONCLUSION

Maximal exercise capacity modified the relationship between BMI and long-term survival in patients with HF, upholding the presence of an exercise capacity-obesity paradox dichotomy as observed over the short-term in previous studies.

Prognostic value of exercise capacity among patients with treated depression: The Henry Ford Exercise Testing (FIT) Project

BACKGROUND

Exercise capacity is associated with survival in the general population. Whether this applies to patients with treated depression is not clear.

HYPOTHESIS

High exercise capacity remains associated with lower risk of all-cause mortality (ACM) and nonfatal myocardial infraction (MI) among patients with treated depression.

METHODS

We included 5128 patients on antidepressant medications who completed a clinically indicated exercise stress test between 1991 and 2009. Patients were followed for a median duration of 9.4 years for ACM and 4.5 years for MI. Exercise capacity was estimated in metabolic equivalents of tasks (METs). Cox proportional hazards regression models were used.

RESULTS

Patients with treated depression who achieved ≥12 METs (vs those achieving <6 METs) were younger (age 46 ± 9 vs 61 ± 12 years), more often male (60% vs 23%), less often black (10% vs 27%), and less likely to be hypertensive (51% vs 86%), have DM (9% vs 38%), or be obese (11% vs 36%) or dyslipidemic (45% vs 54%). In the fully adjusted Cox proportional hazard regression model, exercise capacity was associated with a lower ACM (HR per 1-MET increase in exercise capacity: 0.82, 95% CI: 0.79-0.85, P < 0.001) and nonfatal MI (HR: 0.92, 95% CI: 0.87-0.97, P = 0.004).

CONCLUSIONS

Exercise capacity had an inverse association with both ACM and nonfatal MI in patients with treated depression, independent of cardiovascular risk factors. These results highlight the potential impact of assessing exercise capacity to identify risk, as well as promoting an active lifestyle among treated depression patients.

Using machine learning on cardiorespiratory fitness data for predicting hypertension: The Henry Ford ExercIse Testing (FIT) Project

This study evaluates and compares the performance of different machine learning techniques on predicting the individuals at risk of developing hypertension, and who are likely to benefit most from interventions, using the cardiorespiratory fitness data. The dataset of this study contains information of 23,095 patients who underwent clinician- referred exercise treadmill stress testing at Henry Ford Health Systems between 1991 and 2009 and had a complete 10-year follow-up. The variables of the dataset include information on vital signs, diagnosis and clinical laboratory measurements. Six machine learning techniques were investigated: LogitBoost (LB), Bayesian Network classifier (BN), Locally Weighted Naive Bayes (LWB), Artificial Neural Network (ANN), Support Vector Machine (SVM) and Random Tree Forest (RTF). Using different validation methods, the RTF model has shown the best performance (AUC = 0.93) and outperformed all other machine learning techniques examined in this study. The results have also shown that it is critical to carefully explore and evaluate the performance of the machine learning models using various model evaluation methods as the prediction accuracy can significantly differ.

Exercise Training Workloads Upon Exit From Cardiac Rehabilitation in Men and Women: THE HENRY FORD HOSPITAL EXPERIENCE

PURPOSE

To describe exercise training workloads, estimated as metabolic equivalents of task (METs) both upon exit from cardiac rehabilitation (CR) and as the change in MET level following CR, stratified by age, sex, initial MET level, number of sessions completed, and qualifying event at entry into CR.

METHODS

A retrospective study involving 8319 (31% female) patients who completed ≥9 exercise training sessions in the early outpatient CR program at Henry Ford Hospital. Exercise training MET levels achieved during CR were estimated on the basis of the speed and grade recorded from a treadmill. Exercise training METs at the start of CR were defined as the average of the second and third sessions, whereas MET level upon exit from CR was determined from the average of the last 2 patient encounters.

RESULTS

The overall mean MET level while training just prior to exit from CR was 3.9 ± 1.4 (4.1 ± 1.4 and 3.3 ± 1.0 in men and women, respectively). The mean change in METs after CR was 1.3 ± 1.1 (+45% ± 37%) and 0.9 ± 0.7 (+40% ± 32%) in men and women, respectively.

CONCLUSIONS

In a large and demographically diverse cohort of patients who participated in CR, increases in mean workload (ie, METs) during exercise training were observed that approximated 45% in men and 40% in women. These data could be considered when establishing benchmarks for program-related performance outcome measures.

Cardiorespiratory Fitness and Cardiovascular Disease Prevention: an Update

PURPOSE OF REVIEW

Cardiovascular diseases account for nearly one third of all deaths globally. Improving exercise capacity and cardiorespiratory fitness (CRF) has been an important target to reduce cardiovascular events. In addition, the American Heart Association defined decreased physical activity as the fourth risk factor for coronary artery disease. Multiple large cohort studies have evaluated the impact of CRF on outcomes. In this review, we will discuss the role of CRF in reducing cardiovascular morbidity and mortality.

RECENT FINDINGS

Recent data suggest that CRF has an important role in reducing not only cardiovascular and all-cause mortality, but also incident myocardial infarction, hypertension, diabetes, atrial fibrillation, heart failure, and stroke. Most recently, its role in cancer prevention started to emerge. CRF protective effects have also been seen in patients with prior comorbidities like prior coronary artery disease, heart failure, depression, end-stage renal disease, and stroke. The prognostic value of CRF has been demonstrated in various patient populations and cardiovascular conditions. Higher CRF is associated with improved survival and decreased incidence of cardiovascular diseases (CVD) and other comorbidities including hypertension, diabetes, heart failure, and atrial fibrillation.

Comparison of machine learning techniques to predict all-cause mortality using fitness data: the Henry ford exercIse testing (FIT) project

BACKGROUND

Prior studies have demonstrated that cardiorespiratory fitness (CRF) is a strong marker of cardiovascular health. Machine learning (ML) can enhance the prediction of outcomes through classification techniques that classify the data into predetermined categories. The aim of this study is to present an evaluation and comparison of how machine learning techniques can be applied on medical records of cardiorespiratory fitness and how the various techniques differ in terms of capabilities of predicting medical outcomes (e.g. mortality).

METHODS

We use data of 34,212 patients free of known coronary artery disease or heart failure who underwent clinician-referred exercise treadmill stress testing at Henry Ford Health Systems Between 1991 and 2009 and had a complete 10-year follow-up. Seven machine learning classification techniques were evaluated: Decision Tree (DT), Support Vector Machine (SVM), Artificial Neural Networks (ANN), Naïve Bayesian Classifier (BC), Bayesian Network (BN), K-Nearest Neighbor (KNN) and Random Forest (RF). In order to handle the imbalanced dataset used, the Synthetic Minority Over-Sampling Technique (SMOTE) is used.

RESULTS

Two set of experiments have been conducted with and without the SMOTE sampling technique. On average over different evaluation metrics, SVM Classifier has shown the lowest performance while other models like BN, BC and DT performed better. The RF classifier has shown the best performance (AUC = 0.97) among all models trained using the SMOTE sampling.

CONCLUSIONS

The results show that various ML techniques can significantly vary in terms of its performance for the different evaluation metrics. It is also not necessarily that the more complex the ML model, the more prediction accuracy can be achieved. The prediction performance of all models trained with SMOTE is much better than the performance of models trained without SMOTE. The study shows the potential of machine learning methods for predicting all-cause mortality using cardiorespiratory fitness data.

Using Machine Learning to Define the Association between Cardiorespiratory Fitness and All-Cause Mortality (from the Henry Ford Exercise Testing Project)

Previous studies have demonstrated that cardiorespiratory fitness is a strong marker of cardiovascular health. Machine learning (ML) can enhance the prediction of outcomes through classification techniques that classify the data into predetermined categories. The aim of the analysis is to compare the prediction of 10 years of all-cause mortality (ACM) using statistical logistic regression (LR) and ML approaches in a cohort of patients who underwent exercise stress testing. We included 34,212 patients (55% males, mean age 54 ± 13 years) free of coronary artery disease or heart failure who underwent exercise treadmill stress testing between 1991 and 2009 and had complete 10-year follow-up. The primary outcome of this analysis was ACM at 10 years. The probability of 10-years ACM was calculated using statistical LR and ML, and the accuracy of these methods was calculated and compared. A total of 3,921 patients died at 10 years. Using statistical LR, the sensitivity to predict ACM was 44.9% (95% confidence interval [CI] 43.3% to 46.5%), whereas the specificity was 93.4% (95% CI 93.1% to 93.7%). The sensitivity of ML to predict ACM was 87.4% (95% CI 86.3% to 88.4%), whereas the specificity was 97.2% (95% CI 97.0% to 97.4%). The ML approach was associated with improved model discrimination (area under the curve for ML [0.923 (95% CI 0.917 to 0.928)]) compared with statistical LR (0.836 [95% CI 0.829 to 0.846], p<0.0001). In conclusion, our analysis demonstrates that ML provides better accuracy and discrimination of the prediction of ACM among patients undergoing stress testing.