Clinical Phenotyping with an Outcomes-driven Mixture of Experts for Patient Matching and Risk Estimation

Observational medical data present unique opportunities for analysis of medical outcomes and treatment decision making. However, because these datasets do not contain the strict pairing of randomized control trials, matching techniques are to draw comparisons among patients. A key limitation to such techniques is verification that the variables used to model treatment decision making are also relevant in identifying the risk of major adverse events. This article explores a deep mixture of experts approach to jointly learn how to match patients and model the risk of major adverse events in patients. Although trained with information regarding treatment and outcomes, after training, the proposed model is decomposable into a network that clusters patients into phenotypes from information available before treatment. This model is validated on a dataset of patients with acute myocardial infarction complicated by cardiogenic shock. The mixture of experts approach can predict the outcome of mortality with an area under the receiver operating characteristic curve of 0.85 ± 0.01 while jointly discovering five potential phenotypes of interest. The technique and interpretation allow for identifying clinically relevant phenotypes that may be used both for outcomes modeling as well as potentially evaluating individualized treatment effects.

Initial Findings From the National Cardiovascular Data Registry of Atrial Fibrillation Ablation Procedures

BACKGROUND

The National Cardiovascular Data Registry (NCDR) AFib Ablation Registry was created to assess real-world prevalence, demographic characteristics, procedural management, and outcomes of patients undergoing atrial fibrillation (AF) ablation procedures.

OBJECTIVES

The goal of this study was to characterize the patient, hospital, and physician characteristics and in-hospital outcomes related to AF ablation in the first 5 years of the registry.

METHODS

This paper describes the AFib Ablation Registry structure and governance, outcome assessment processes, data quality, and data collection processes. The characteristics of the patient population, hospitals, and in-hospital outcomes are also described.

RESULTS

A total of 76,219 patients were included in the registry between January 2016 and December 2020 (mean age 65.5 ± 10.3 years, 65.2% male, 55.8% paroxysmal AF, mean CHA₂DS₂-VASc score 2.7 ± 1.6) treated by 708 physicians in 162 hospitals. Successful isolation of all pulmonary veins was achieved in 92.4% of patients. The prevalence of any complication during procedural admission was 2.50% and major complication was 0.9%, including significant bradycardia in 0.47%, heart failure in 0.47%, and pericardial effusion requiring intervention in 0.44%. Hospitalization >1 day occurred in 11.8% of patients, and in-hospital death was rare (n = 41 [0.05%]).

CONCLUSIONS

The NCDR AFib Ablation Registry is the largest multicenter, prospective cohort study of patients undergoing catheter ablation worldwide. Results in the first 5 years showed that successful pulmonary vein isolation is achieved in the majority of patients, with a low rate of complications. Future studies from the registry will assess practice trends, evaluate treatment patterns associated with different patient outcomes, and support development of evidence-based guidelines.

'Real-world' observational studies in arrhythmia research: data sources, methodology, and interpretation. A position document from European Heart Rhythm Association (EHRA), endorsed by Heart Rhythm Society (HRS), Asia-Pacific HRS (APHRS), and Latin America HRS (LAHRS)

The field of observational studies or "real world studies" is in rapid development with many new techniques introduced and increased understanding of traditional methods. For this reason the current paper provides an overview of current methods with focus on new techniques. Some highlights can be emphasized: We provide an overview of sources of data for observational studies. There is an overview of sources of bias and confounding. Next There is an overview of causal inference techniques that are increasingly used. The most commonly used techniques for statistical modelling are reviewed with focus on the important distinction of risk versus prediction. The final section provides examples of common problems with reporting observational data.

P834Secondary prevention medications of cardiovascular diseases in China: findings from China PEACE million persons project

Background

Cardiovascular diseases (CVD) is the leading cause of death in China. Secondary prevention medications can improve the prognosis of CVD, yet little is known about the current use, variation and associated factors of these therapies in China.

Purpose

The aim of this study was to describe the current use of secondary prevention medications among patients with established CVD in the community setting in China, assess variations across population subgroups, and identify the individual characteristics associated with these therapies.

Methods

We studied 2.6 million participants aged 35–75 years from all 31 provinces in the China Patient-Centered Evaluative Assessment of Cardiac Events (PEACE) Million Persons Project, a government-funded public health program conducted from 2014 to 2018. Participants self-reported their history of ischemic heart disease (IHD) or ischemic stroke (IS) and medication use in an interview. Among participants with IHD and/or IS, we assessed the reported use of secondary prevention medications (anti-platelet drugs and statins) in the overall population and in 1,530,408 population subgroups, defined by all possible combinations of 16 factors (age, sex, urbanity, geographic region, ethnicity, occupation, annual household income, education, marital status, medical insurance, current smoker, current drinker, history of hypertension, history of diabetes, body mass index and years since diagnosis). Multivariable mixed models with a logit link function and community-specific random intercepts were fitted to assess the associations of demographic, socioeconomic and health behavior factors with the reported use of secondary prevention mediations.

Results

Among 2,613,035 screened participants, 2.9% (74,830) had history of IHD and/or IS (1.2% for IHD, 2.4% for IS). Overall, the reported use rate either anti-platelet drugs or statins was 21.9% (18.3% anti-platelet drugs, 11.0% statins, and 7.4% both). Among the 1,530,408 population subgroups, the use of secondary prevention medications varied substantially (3.4% to 52.0%). Multivariable analyses found that that younger people, women, those living in rural areas, current smokers, current drinkers, people without hypertension or diabetes, and those with established CVD for more than 2 years were less likely to take anti-platelet drugs or statins (Figure).

Forest plot of multivariable mixed model

Conclusions

The current use of secondary prevention drugs is suboptimal and varies substantially across population subgroups in China. Our study identifies target populations for interventions to improve secondary prevention of CVD.

Acknowledgement/Funding

This study was supported by the Ministry of Finance of China and National Health Commission of China

Use of Cardiac Resynchronization Therapy Defibrillator in US Hospitals

Importance

Cardiac resynchronization therapy (CRT) provides significant reduction in morbidity and mortality in select patients with left ventricular systolic dysfunction and specific parameters of electrocardiographic evidence of dyssynchrony. Relative to the 2012 American College of Cardiology/American Heart Association/Heart Rhythm Society guideline update for patient selection, little is known about the contemporary use of CRT in the United States.

Objective

To describe the use of CRT defibrillator (CRT-D) in the period around guideline revision.

Design, Setting, and Participants

All patients undergoing new CRT-D implantations in the National Cardiovascular Data Registry for implantable cardioverter-defibrillators from January 1, 2012, to December 31, 2015, at 1710 participating hospitals were identified for this population-based study. Rates of CRT-D implantation that were concordant and discordant with the 2012 American College of Cardiology/American Heart Association/Heart Rhythm Society update of the 2008 guidelines for device-based therapy were determined. Analysis began in January 2012.

Main Outcomes and Measures

Increase in guideline-concordant CRT-D implantation.

Results

Among 135 253 patients undergoing initial CRT-D implantation, 88 923 were included in the study cohort, of which 73 859 implants (83.1%) were guideline concordant. The proportion of guideline-concordant devices increased from 81.2% (16 710 of 20 481) in 2012 to 84.2% (20 515 of 24 356) in 2015 (P for trend < .001). Significant clustering was noted with 33% (565 of 1710) of hospitals accounting for greater than 70% (10 545 of 15 065) of guideline-discordant CRT-D implants. Conduction abnormalities, in particular, underlying right bundle branch block (3597 [23.9%] vs 7425 [10.1%]; P < .001) and nonspecific intraventricular conduction delay (3341 [22.2%] vs 4769 [6.5%]; P < .001) were more common in those who received guideline-discordant devices.

Conclusions and Relevance

Rates of guideline-concordant CRT-D implantation increased during the study. The major fraction of guideline-discordant implants were clustered at a minority of hospitals. Conduction abnormalities, particularly non-left bundle branch block and nonspecific intraventricular conduction delay, correlated with guideline-discordant implants indicating continued opportunity for dissemination and understanding of guideline updates.

Abstract 232: Effect of Oral Anticoagulation Use on Thromboembolic Risk and Bleeding in Women Compared with Men with Atrial Fibrillation: Findings from the PINNACLE Data Registry

Background: Among those with atrial fibrillation (AF), women have a higher risk of thromboembolic events and lower use of oral anticoagulation (OAC) when compared to men. Whether the effect of gender on thromboembolic risk in patients with AF is mediated by lower use of OAC among women is unknown. We examined whether gender differences in OAC use mediated the relationship between gender and thromboembolic risk in patients with AF.

Methods: Adults 65 years and older with AF in the PINNACLE Registry from October 2008 to December 2014 were identified. Adults with valvular AF and contraindications to OAC use were excluded. Thromboembolic events (stroke, systemic embolism) and major bleeding events were determined using linked Medicare claims data. Site-stratified proportional hazard models adjusting for patient and practice-level characteristics compared thromboembolic and bleeding events by gender, as well as the mediating effect of OAC use as a time-dependent covariate on these outcomes.

Results: Among 667,287 patients, the mean ±standard deviation age was 77±8 years and 46% were women. Compared to men, women had higher CHA2DS2-Vasc scores (mean 4.8 vs. 3.8, p<0.001), had similar prevalence of bleeding risk factors, but were prescribed OAC less frequently (51.5% vs. 53.8%, p<0.001). Over a median 2 years of follow up (range 0-6 years), there were 25,947 thromboembolic events and 40,967 major bleeding events. Women had a higher risk of thromboembolic events compared to men (5-year cumulative incidence 8.3% vs. 6.0%, adjusted hazard ratio [HR] 1.24, 95% CI 1.16-1.33) and a slightly lower risk of major bleeding (5-year cumulative incidence 10.4% vs. 11.2%, adjusted HR 0.90, 95% CI 0.83-0.98). Among adult patients with AF, OAC use was associated with lower thromboembolic risk (HR 0.90, 95% CI 0.86-0.94) and greater bleeding risk (HR 1.51, 95% CI 1.42-1.60). Among women, OAC use did not mediate the relationship between gender and either thromboembolic or major bleeding events (HR 1.24 and 0.90 respectively, identical to the unmediated effects).

Conclusions: In a national US sample of older adults with AF, women had lower rates of OAC use, higher risk of thromboembolic events and lower risk of major bleeding. However, the difference in OAC use did not explain the gender differences in these outcomes. These findings underscore that women with AF are independently at higher risk for thromboembolic events and lower OAC use among women does not significantly explain this higher risk.

Abstract 28: Variation in High-Value Cardiovascular Diagnostic Testing: Patient, Payer, and Hospital Effects

Background: In patients hospitalized with 1) newly diagnosed heart failure (HF) or 2) acute myocardial infarction (AMI), assessment of left ventricular systolic function is a high-value test supported by guidelines. We examined the degree to which patient-, payer-, and hospital-level characteristics impact use of testing.

Methods: We analyzed data from the Colorado All-Payer Claims Database, a repository of billing claims from all insurers who provide care in the state. We identified all patients with an index hospitalization for HF and AMI from 2010 to 2014. We excluded patients with a prior diagnosis of HF, and hospitals with fewer than 40 HF or AMI hospitalizations. We determined whether patients had a systolic function assessment performed within 60 days of hospitalization. We calculated adjusted rates of testing at the hospital level, and assessed for correlation of rates between HF and AMI patients. We used multilevel logistic regression to assess patient- and payer- characteristics associated with testing, and used median odds ratios to determine the residual variation in testing attributable to hospitals.

Results: We identified 9,516 patients with HF and 10,315 patients with AMI (mean age 73 years, 48% women) among 36 hospitals. Overall, 74% of HF patients and 73% of AMI patients received testing. Testing rates among hospitals ranged from 56% to 82% for HF and from 42% to 83% for AMI (Figure). Correlation of testing rates for AMI and HF patients among hospitals was moderate (Spearman r=0.58; p<.001). Medicaid insurance was associated with lower likelihood of testing for both AMI and HF (ORs 0.77 [0.67-0.88] and 0.54 [0.47-0.62]; both p<.001). After multivariable adjustment, use of testing across sites varied by a median odds ratio of 1.39 [1.28-1.49] for AMI patients and of 1.25 [1.17-1.34] for HF patients, meaning that on average, patients had 1.39 and 1.25 higher odds of being tested if they received care at a higher performing hospital.

Conclusions: Despite adjustment for patient- and payer-level characteristics, there is 1) significant residual variation in use of high-value cardiac testing and 2) correlation in testing rates for AMI and HF patients among hospitals. These results suggest that hospital-level characteristics and care processes may have a strong influence on use of high-value testing.

Long-term Outcomes Associated With Implantable Cardioverter Defibrillator in Adults With Chronic Kidney Disease

IMPORTANCE

Chronic kidney disease (CKD) is common in adults with heart failure and is associated with an increased risk of sudden cardiac death. Randomized trials of participants without CKD have demonstrated that implantable cardioverter defibrillators (ICDs) decrease the risk of arrhythmic death in selected patients with reduced left ventricular ejection fraction (LVEF) heart failure. However, whether ICDs improve clinical outcomes in patients with CKD is not well elucidated.

OBJECTIVE

To examine the association of primary prevention ICDs with risk of death and hospitalization in a community-based population of potentially ICD-eligible patients who had heart failure with reduced LVEF and CKD.

DESIGN, SETTINGS, AND PARTICIPANTS

This noninterventional cohort study included adults with heart failure and an LVEF of 40% or less and measures of serum creatinine levels available from January 1, 2005, through December 31, 2012, who were enrolled in 4 Kaiser Permanente health care delivery systems. Chronic kidney disease was defined as an estimated glomerular filtration rate of less than 60 mL/min/1.73 m2. Patients who received and did not receive an ICD were matched (1:3) on CKD status, age, and high-dimensional propensity score to receive an ICD. Follow-up was completed on December 31, 2013. Data were analyzed from 2015 to 2017.

EXPOSURES

Placement of an ICD.

MAIN OUTCOMES AND MEASURES

All-cause death, hospitalizations due to heart failure, and any-cause hospitalizations.

RESULTS

A total of 5877 matched eligible adults with CKD (1556 with an ICD and 4321 without an ICD) were identified (4049 men [68.9%] and 1828 women [31.1%]; mean [SD] age, 72.9 [8.2] years). In models adjusted for demographics, comorbidity, and cardiovascular medication use, no difference was found in all-cause mortality between patients with CKD in the ICD vs non-ICD groups (adjusted hazard ratio, 0.96; 95% CI, 0.87-1.06). However, ICD placement was associated with increased risk of subsequent hospitalization due to heart failure (adjusted relative risk, 1.49; 95% CI, 1.33-1.60) and any-cause hospitalization (adjusted relative risk, 1.25; 95% CI, 1.20-1.30) among patients with CKD.

CONCLUSIONS AND RELEVANCE

In a large, contemporary, noninterventional study of community-based patients with heart failure and CKD, ICD placement was not significantly associated with improved survival but was associated with increased risk for subsequent hospitalization due to heart failure and all-cause hospitalization. The potential risks and benefits of ICDs should be carefully considered in patients with heart failure and CKD.

Abstract 137: Clinical Outcomes After Cardiac Stress Testing Among U.S. Patients Younger Than 65, 2006-2012

Background: Recent studies of national practice patterns suggest that there is inconsistency in the appropriateness of referrals for cardiac stress testing. However, little is known about the prevalence of relevant clinical outcomes such as coronary revascularization or acute myocardial infarction (AMI) after stress testing.

Methods: Using administrative claims from a large national private insurer, we conducted an observational cohort study of patients aged 25-64 who underwent de novo stress testing from 2006-2012. We excluded patients with a prior diagnosis of coronary artery disease, heart failure, or stroke. We calculated rates of AMI and coronary angiography with or without revascularization in the year following stress testing. We estimated logistic regression models that included demographic characteristics (age, sex, and race) and relevant comorbidities (diabetes, hypertension, and dyslipidemia). Finally, we stratified the cohort into quintiles based on their risk of having an AMI or revascularization event in order to describe the characteristics of patients at lowest and highest risk.

Results: We identified 564,313 patients (mean age 50, 49% women, 73% white) who underwent stress testing during the study period. Among these patients, 15% had diabetes, 37% had hypertension, and 42% had dyslipidemia. Within one year, 2.5% of the cohort underwent coronary revascularization or were hospitalized for AMI, and 2.7% underwent coronary angiography without revascularization. In the risk-stratified analysis, 0.5% of patients in the lowest risk quintile had an AMI or revascularization event, compared to 6.2% in the highest risk quintile (p<.001). Compared to the highest risk quintile, patients in the lowest risk quintile were younger (mean age 40 versus 58 years; p<.001), more likely to be female (85% versus 0%; p<.001), more likely to be non-white (36% versus 16%; p<.001), and less likely to have comorbidities (Table).

Conclusion: The percentage of commercially insured U.S. patients who had an AMI or revascularization event within 1 year of stress testing was small. In the lowest risk quintile, the a priori prevalence of coronary disease risk factors was low, and the rates of subsequent AMI and revascularization was extremely low, raising questions about the value of stress testing in this subgroup.

Kidney function and appropriateness of device therapies in adults with implantable cardioverter defibrillators

OBJECTIVE

Patients with chronic kidney disease (CKD) have higher risk of sudden cardiac death; however, they may not receive implantable cardioverter defibrillators (ICDs), in part due to higher risk of complications. We evaluated whether CKD is associated with greater risk of device-delivered shocks/antitachycardia pacing (ATP) therapies among patients receiving a primary prevention ICD.

METHODS

We studied participants in the observational Cardiovascular Research Network Longitudinal Study of Implantable Cardioverter Defibrillators. CKD was defined as estimated glomerular filtration rate (eGFR) <60 mL/min/1.73 m². Outcomes included all delivered shocks/ATPs therapies and type of shock/ATP therapies (inappropriate or appropriate, determined by physician adjudication) within the 3 years. We evaluated the associations between CKD and time to first device therapy, burden of device therapy, and inappropriate versus appropriate device therapy, adjusting for demographics, comorbidity, laboratory values and medication use.

RESULTS

Among 2161 participants, 1066 (49.3%) had CKD (eGFR 44±11 mL/min/1.73 m²) at ICD implantation. During mean of 2.26±0.89 years, 9.8% and 18.5% of participants had at least one inappropriate and appropriate shock/ATP therapies, respectively. CKD was not associated with time to first shock/ATP therapies (adjusted HR 0.87, 95% CI 0.73 to 1.05), overall burden of shock/ATP therapies (adjusted relative rate 0.93, 95% CI 0.74 to 1.17) or inappropriate versus appropriate shock/ATP therapies (adjusted relative risk 0.88, 95% CI 0.68 to 1.14) compared with not having CKD.

CONCLUSIONS

In adults receiving a primary prevention ICD, mild-to-moderate CKD was not associated with the timing, burden or appropriateness of subsequent device therapy. Potential concern for inappropriate ICD-delivered therapies should not preclude ICDs among eligible patients with CKD.

Abstract 223: Outcomes of Older Patients Receiving Secondary Prevention Implantable Cardioverter Defibrillators: An Analysis From the NCDR ICD Registry

Background: The clinical trials establishing the efficacy of implantable cardioverter defibrillators (ICDs) for secondary prevention of sudden cardiac death enrolled limited numbers of older patients. We sought to assess the morbidity and mortality of older patients receiving ICDs for secondary prevention in contemporary clinical practice.

Methods: Using the NCDR ICD Registry, we identified 20,751 Medicare beneficiaries > 65 years old who underwent secondary prevention ICD implantation between 2006 and 2010 in 1027 US hospitals. Using a probabilistic method, eligible subjects were matched to Medicare claims to assess longitudinal outcomes, including death, hospitalization, and skilled nursing facility (SNF) admission over 1 year. Outcomes were evaluated in strata according to age (<70, 70-74, 75-79, and >80 years).

Results: The mean age of the cohort was 75 years at the time of implantation; 24.1% were<70 and 26.8% > 80 (Table). Overall 1-year survival was 87%, with 91% and 82% in ages <70 and >=80 respectively. Regardless of age, hospitalizations at one year were frequent, ranging from 47% in <70 to 57% in > 80. Early admission to a SNF was much higher for > 80 (12%) than for <70 (4%) (p <0.01).

Conclusions: In older patients receiving secondary prevention ICDs, 1-year survival was high in all age strata including age > 80. However, rates of hospitalization and SNF admission indicate substantial care needs after device implantation.

Figure:

Achieving Quality in Cardiovascular Imaging

Cardiovascular imaging has enjoyed both rapid technological advances and sustained growth, yet less attention has been focused on quality than in other areas of cardiovascular medicine. To address this deficit, representatives from cardiovascular imaging societies, private payers, government agencies, the medical imaging industry, and experts in quality measurement met, and this report provides an overview of the discussions. A consensus definition of quality in imaging and a convergence of opinion on quality measures across imaging modalities was achieved and are intended to be the start of a process culminating in the development, dissemination, and adoption of quality measures for all cardiovascular imaging modalities.

Consensus statement: Palliative and supportive care in advanced heart failure

BACKGROUND

A consensus conference was convened to define the current state and important gaps in knowledge and needed research on "Palliative and Supportive Care in Advanced Heart Failure."

EVIDENCE

Evidence was drawn from expert opinion and from extensive review of the medical literature, evidence-based guidelines, and reviews.

CONCLUSIONS

The conference identified gaps in current knowledge, practice, and research relating to prognostication, symptom management, and supportive care for advanced heart failure (HF). Specific conclusions include: (1) although supportive care should be integrated throughout treatment of patients with advanced HF, data are needed to understand how to best decrease physical and psychosocial burdens of advanced HF and to meet patient and family needs; (2) prognostication in advanced HF is difficult and data are needed to understand which patients will benefit from which interventions and how best to counsel patients with advanced HF; (3) research is needed to identify which interventions improve quality of life and best achieve the outcomes desired by patients and family members; (4) care should be coordinated between sites of care, and barriers to evidence-based practice must be addressed programmatically; and (5) more research is needed to identify the content and technique of communicating prognosis and treatment options with patients with advanced HF; physicians caring for patients with advanced HF must develop skills to better integrate the patient's preferences into the goals of care.

Effect of alpha tocopheryl succinate on cyclic AMP-dependent protein kinase activity in murine B-16 melanoma cells in culture

d-Alpha tocopheryl succinate (vitamin E succinate), which inhibited growth and survival, and induced differentiation in murine B-16 melanoma cells in culture, increased adenosine 3',5'cyclic monophosphate-(cAMP)-dependent protein kinase (PK) activity without increasing the cellular cAMP level. Prostaglandin (PG)A2, which produced changes in melanoma cells similar to those produced by vitamin E succinate, also increased cAMP-dependent PK activity without changing the intracellular level of cAMP.