Evaluation of Risk-Adjusted Home Time After Hospitalization for Heart Failure as a Potential Hospital Performance Metric

Key Points

Question  How does risk-adjusted 30-day home time, defined as time a patient spends alive and out of a hospital or facility after discharge, compare with 30-day readmission and mortality as a hospital-level performance metric for patients hospitalized for heart failure?

Findings  In this cohort study of 2 968 341 patients from 3134 facilities, 30-day readmission, 30-day mortality, and 1-year mortality decreased in a graded fashion across increasing 30-day home time categories. Thirty-day home time was inversely correlated with 30-day risk-standardized readmission rate and 30-day risk-standardized mortality rate and was associated with meaningful reclassification in 30.6% of hospitals compared with 30-day risk-standardized mortality rate and 30-day risk-standardized readmission rate.

Meaning  In this study, 30-day home time was a viable hospital-level quality metric and was associated with 30-day readmission, 30-day mortality, and 1-year mortality outcomes.

Importance  Thirty-day home time, defined as time spent alive and out of a hospital or facility, is a novel, patient-centered performance metric that incorporates readmission and mortality.

Objectives  To characterize risk-adjusted 30-day home time in patients discharged with heart failure (HF) as a hospital-level quality metric and evaluate its association with the 30-day risk-standardized readmission rate (RSRR), 30-day risk-standardized mortality rate (RSMR), and 1-year RSMR.

Design, Setting, and Participants  This hospital-level cohort study retrospectively analyzed 100% of Medicare claims data from 2 968 341 patients from 3134 facilities from January 1, 2012, to November 30, 2017.

Exposures  Home time, defined as time spent alive and out of a short-term hospital, skilled nursing facility, or intermediate/long-term facility 30 days after discharge.

Main Outcomes and Measures  For each hospital, a risk-adjusted 30-day home time for HF was calculated similar to the Centers for Medicare & Medicaid Services risk-adjustment models for 30-day RSRR and RSMR. Hospitals were categorized into quartiles (lowest to highest risk-adjusted home time). The correlations between hospital rates of risk-adjusted 30-day home time and 30-day RSRR, 30-day RSMR, and 1-year RSMR were estimated using the Pearson correlation coefficient. Distribution of days lost from a perfect 30-day home time were calculated. Reclassification of hospital performance using 30-day home time vs 30-day RSRR was also evaluated.

Results  Overall, 2 968 341 patients (mean [SD] age, 81.0 [8.3] years; 53.6% female) from 3134 hospitals were included in this study. The median hospital risk-adjusted 30-day home time for patients with HF was 21.77 days (range, 8.22-28.41 days). Hospitals in the highest quartile of risk-adjusted 30-day home time (best-performing hospitals) were larger (mean [SD] number of beds, 285 [275]), with a higher volume of patients with HF (median, 797 patients; interquartile range, 395-1484) and were more likely academic hospitals (59.9%) with availability of cardiac surgery (51.1%) and cardiac rehabilitation (68.8%). A total of 72% of home time lost was attributable to stays in an intermediate- or long-term care facility (mean [SD], 2.65 [6.44] days) or skilled nursing facility (mean [SD], 3.96 [9.04] days), 13% was attributable to short-term readmissions (mean [SD], 1.25 [3.25] days), and 15% was attributable to death (mean [SD], 1.37 [6.04] days). Among 30-day outcomes, the 30-day RSRR and 30-day RSMR decreased in a graded fashion across increasing 30-day home time categories (correlation coefficients: 30-day RSRR and 30-day home time, −0.23, P < .001; 30-day RSMR and 30-day home time, −0.31, P < .001). Similar patterns of association were also noted for 1-year RSMR and 30-day home time (correlation coefficient, −0.35, P < .001). Thirty-day home time meaningfully reclassified hospital performance in 30% of the hospitals compared with 30-day RSRR and in 25% of hospitals compared with 30-day RSMR.

Conclusions and Relevance  In this study, 30-day home time among patients discharged after a hospitalization for HF was objectively assessed as a hospital-level quality metric using Medicare claims data and was associated with readmission and mortality outcomes and with reclassification of hospital performance compared with 30-day RSRR and 30-day RSMR.

Hospitalizations for heart failure (HF) are associated with a high burden of mortality, functional disability, and health care cost among older adults.^1-7 Accordingly, improvement in care quality, outcomes, and associated health care costs among patients hospitalized with HF has been the focus of recent health policies implemented by the Centers for Medicare & Medicaid Services (CMS). To this end, CMS implemented programs such as the Hospital Readmission Reduction Program (HRRP) and the Hospital Value-Based Purchasing Program to financially incentivize hospitals based on their 30-day clinical outcomes for patients hospitalized with common acute conditions, such as HF, acute myocardial infarction, and pneumonia.⁸ Although the implementation of HRRP has been associated with significant reductions in readmission rates for HF,^9,10 it remains unclear whether current health policies have contributed to improvement in patients’ overall experience or quality of life. Although there has been an increasing emphasis on use of patient-oriented outcomes in evaluation of therapeutic benefits of newer HF therapies in clinical trials, the role of patient-oriented outcomes in defining hospital-level care quality for patients with HF is limited.¹¹ Thus, there is an unmet need for complementary hospital-level performance metrics that focus on patient-centered outcomes that can be easily assessed and better account for the overall health care experience.

Home time is a novel, patient-centered quality metric that is determined by the time a patient spends alive and out of a hospital or skilled nursing facility (SNF).^12,13 This hospital performance metric can be calculated through administrative claims data and is associated with meaningful clinical and patient-reported end points.^12-16 However, the utility of home time as a metric for hospital-level measurement of care quality for patients with HF is not well established. In this study, we assessed home time after hospitalization for HF through Medicare administrative claims data and its association with currently used CMS performance metrics of 30-day risk-standardized readmission rate (RSRR) and risk-standardized mortality rate (RSMR).

This cohort study used 100% CMS Medicare Provider Analysis and Review (MedPAR) data, which include administrative billing claims for 100% of all short-term hospitalizations and intermediate/long-term acute care facility or SNF stays of Medicare fee-for-service beneficiaries. Data include information about a Medicare beneficiary’s age, race/ethnicity, and medical comorbidities and discharge disposition. Primary and secondary diagnosis codes were based on International Classification of Diseases, Ninth Revision (ICD-9) or International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) codes. Data analysis for the study was conducted at The University of Iowa. The University of Iowa Institutional Review Board approved the study. The need for informed consent was waived because of the use of encrypted patient identifiers.

The present analysis included eligible patients who were fee-for-service Medicare beneficiaries older than 65 years and were hospitalized for HF (ICD-9 codes: 402.xx, 404.xx, and 428.xx; ICD-10 codes: I11, I13.xx, and I50.xx) from January 1, 2012, to November 30, 2017. Eligible patients were enrolled in Medicare fee-for-service for 12 months before the index admission. For patients with multiple HF admissions within a 30-day period, only the first hospitalization was included. The exclusion criteria for the present study are detailed in eFigure 1 in the Supplement. Patients who were transferred to another facility were only counted once and assigned to the discharging hospital for estimation of the 30-day home time and readmission metric estimation and to the original admitting hospital for estimation of mortality metric. Hospital characteristics were assessed from the 2016 American Hospital Association Survey, and hospital-level data were used to classify hospitals as academic or nonacademic, for profit or not for profit, private or public, or rural or urban.

The primary exposure of interest was risk-adjusted 30-day home time and was defined as time spent alive and out of an acute care or a subacute care nursing or intermediate/long-term care facility. Stays at facilities were assessed through MedPAR, which included claims for inpatient hospitalization, long-term care facilities, and SNFs. Hospital days during the index HF-related admission were not included in the home time calculation. Patients who died during the index hospitalization were excluded from home time calculation. Patients who were discharged home and remained home were assigned a home time of 30 days. Any part of a day spent after discharge from the hospital that was spent in a facility was considered as a full day for the home time calculation. Home time was modeled using generalized linear mixed models with log link and Poisson distribution using maximum likelihood estimates. These models included hospital site as a random effect and patient variables as fixed effects.

Other hospital performance metrics of interest for the present analysis were 30-day RSRR, 30-day RSMR, and 1-year RSMR. These risk-standardized outcomes were calculated as described previously using hierarchical logistic regression models with each outcome as a dependent variable and included random hospital intercepts. In accordance with the CMS approach, patients who died during index hospitalization were included in the calculation of 30-day and 1-year RSMR but excluded from 30-day RSRR and home time estimation.

Clinical covariates used in risk-adjustment models for estimation of the hospital performance metrics in the present analysis were similar to models suggested by the CMS for 30-day RSRR and RSMR.¹⁷ For variable selection, 25% of patients from each year were randomly selected and pooled to construct a distinct model development and validation cohort. Next, 100 bootstrapped samples were generated from the development cohort, and separate models were run for each sample replicate, including all variables suggested by the CMS for each outcome using a stepwise selection approach. Specifically, logistic regression models were used for 30-day readmission, 30-day mortality, and 1-year mortality outcomes, and general linear models were constructed for 30-day home time to identify the candidate risk-adjustment variables. For each outcome, variables that were selected in more than 80% of the initial 100 bootstrap models were used as final risk-adjustment variables. The hospital-level mean 30-day RSMR, 30-day RSRR, 30-day home time, and 1-year RSMR were calculated separately in the development and validation cohort using the final risk-adjustment variables. Each metric demonstrated a high degree of correlation across the development and validation cohort, highlighting the robustness of our risk-adjusted models (correlation coefficients of 0.88 for 30-day home time, 0.89 for 30-day readmission, 0.88 for 30-day mortality, and 0.89 for 1-year mortality). Consistent with the CMS approach, race/ethnicity and socioeconomic status were not included in risk adjustment. Final candidate variables and parameter estimates for the 30-day home time, 30-day RSMR, 30-day RSRR, and 1-year RSMR are given in eTables 1 to 4 in the Supplement, respectively. Similar to the approach used by the CMS, expected and predicted rates for each outcome were estimated using these models with and without the linear unbiased prediction modeling for estimation of random effects, respectively.^18,19 The risk-standardized rate of each outcome of interest was determined by multiplying the overall unadjusted rate by the ratio of predicted to expected rate of the respective outcome.

The hospital-level distribution of 30-day risk-adjusted home time, 30-day RSRR, 30-day RSMR, and 1-year RSMR in the study cohort was assessed by histogram plots. Hospitals were stratified into 4 quartiles (Q1-Q4) of 30-day risk-adjusted home time, with Q1 identifying worst-performing hospitals and Q4 indicating best-performing hospitals. Baseline hospital-level and patient-level characteristics were reported across quartiles of hospital-level 30-day risk-adjusted home time as medians (interquartile ranges [IQRs]) for continuous variables and percentages for categorical variables. Baseline characteristics were compared across these groups using analysis of variance for continuous variables and χ² tests for categorical variables. Proportion of days lost from perfect 30-day home time because of stays in an intermediate- or long-term care facility or SNF, readmission, or patient death were estimated from the overall study cohort and across quartiles of 30-day home time.

Correlations between the 30-day risk-adjusted home time metric and 30-day RSRR, 30-day RSMR, and 1-year RSMR were calculated using the Pearson coefficient test. Hospital-level measures of 30-day RSRR, 30-day RSMR, and 1-year RSMR were also compared across quartiles of 30-day risk-adjusted home time using an analysis of variance test. Reclassification in hospital performance using 30-day risk-adjusted home time vs the current CMS 30-day RSRR performance measure was also assessed. Specifically, we compared hospital performance based on the 30-day home time (Q1 to Q4: low to high performing) vs 30-day RSRR (Q1 to Q4: high to low performing). Hospitals were assigned positive scores (1-3) if they were in a higher performance group (quartile) for 30-day home time vs 30-day RSRR and a negative score (−1 to −3) if they were in a lower performance group (quartile) for 30-day home time vs 30-day RSRR. Hospitals that ranked in the same performance group on both metrics were assigned a score of 0. A reclassification score of 2 or higher was considered as meaningful up-classification in performance status, and a reclassification score of −2 or lower was considered as meaningful down-classification. A similar analysis was performed to evaluate the reclassification in performance status based on 30-day home time vs 30-day RSMR.

To test the robustness and complementary nature of 30-day risk-adjusted home time as a potential hospital-level performance metric, we performed an additional sensitivity analysis comparing 30-day home time with 2 other hospital-level performance metrics: (1) a composite of 30-day readmission or mortality after discharge and (2) 30-day risk-standardized excess days of acute care (EDAC).^20,21 The composite metric of risk-adjusted 30-day readmission or mortality was estimated for hospitals included in the study cohort using an approach similar to that described for 30-day RSRR and 30-day RSMR. Excess days of acute care is a CMS metric designed to capture the number of days that patients discharged alive from a hospital spend in acute care during the 30-day postdischarge period and includes readmissions, emergency department stays, and short-stay hospitalizations. The details about the methods for EDAC estimation in the study cohort are described in the eMethods in the Supplement. All statistical analyses were performed using SAS software, version 9.4 (SAS Institute Inc) with level of significance set at a 2-sided P < .05.

Overall, 2 968 341 patients (mean [SD] age, 81.0 [8.3] years; 53.6% female) from 3134 hospitals were included in this study. Median hospital risk-adjusted 30-day home time was 21.77 days (range, 8.22-28.41 days). There was substantial variability in hospital-level 30-day risk-adjusted home time (eFigure 2 in the Supplement).

Hospital-level characteristics across quartiles of risk-adjusted 30-day home time are given in the Table. Best-performing hospitals were larger (mean [SD] number of beds, 285 [275]), had higher case-volumes of HF (median, 797 patients; IQR, 395-1484), and were more likely to be teaching hospitals (59.9%) with a greater availability of cardiac surgery (51.1%) and cardiac rehabilitation (68.8%) and greater participation in bundled payment programs (28.9%).

Patient-level characteristics across quartiles of risk-adjusted 30-day home time are given in the Table. No meaningful differences were found in age, sex, or burden of comorbidities, such as acute myocardial infarction and chronic kidney disease, across the 30-day home time–based groups. The proportion of patients discharged home was higher in the best- vs worst-performing groups. Of the days lost from perfect 30-day home time, 72% were attributable to intermediate- or long-term care facility (mean [SD], 2.65 [6.44] days) or SNF stays (mean [SD], 3.96 [9.04] days), with comparable proportions of days lost because of patient mortality (15%; mean [SD], 1.37 [6.04] days) and readmission (13%; mean [SD], 1.25 [3.25] days).

A significant inverse correlation was observed between continuous measures of hospital-level 30-day home time and 30-day RSRR (r = −0.23, P < .001), 30-day RSMR (r = −0.31, P < .001), and 1-year RSRR (r = −0.35, P < .001) (Figure 1). In the categorical analysis, a statistically significant decrease in 30-day RSRR (worst-performing, 0.23; best-performing, 0.21; P < .001) and 30-day RSMR (worst-performing, 0.09; best-performing, 0.07; P < .001) was observed across increasing categories of 30-day home time (Figure 2). Among long-term outcomes, 1-year RSMR also decreased in a graded fashion across increasing 30-day home time categories from 27% in Q1 to 24% in Q4 (Figure 2).

A total of 30.6% of hospitals had a meaningful reclassification in their performance status based on 30-day home time compared with 30-day RSRR with a similar proportion of hospitals up-classified (15.6%) and down-classified (15.0%) (Figure 3). Hospitals with up-classification of performance status based on 30-day home time (vs 30-day RSRR) had significantly lower mortality rates in short-term (30-day mortality: mean [SD], 0.07 [0.25]) and long-term (1-year mortality; mean [SD], 0.24 [0.43]) follow-up compared with those that were down-classified (30-day mortality: mean [SD], 0.09 [0.29]; P < .001; 1-year mortality: mean [SD], 0.27 [0.44]; P < .001) in their performance status (eTable 5 in the Supplement). Similarly, 24.7% of hospitals were reclassified based on the 30-day home time metric compared with 30-day RSMR (12.4% up-classified and 12.3% down-classified).

The median rate of hospital-level composite risk-adjusted 30-day readmission or mortality was 0.25 (range, 0.17-0.37) in the study cohort. The median EDAC in our study cohort was −0.5 days (range, −45.1 to 85.5 days) per 100 admissions (eFigure 3 in the Supplement). Higher risk-adjusted 30-day home time was significantly associated with lower rates of the composite readmission or mortality metric (r = −0.35, P < .001) and lower EDAC (r = −0.19, P < .001) (eFigure 4 in the Supplement). Across increasing quartiles of hospital-level risk-adjusted 30-day home time, a significant graded decrease was noted in the hospital-level composite metric of risk-adjusted 30-day readmission or mortality (Q1: 0.26 [IQR, 0.25-0.28]; Q4: 0.24 [IQR, 0.22-0.25]; P < .001) and EDAC (Q1: 1.07 [IQR, −3.47 to 6.94]; Q4: −4.17 [IQR, −13.08 to 4.66]; P < .001) (eTable 6 in the Supplement). Up to 25% of hospitals had a meaningful reclassification in their performance status based on 30-day risk-adjusted home time vs the composite metric of risk-adjusted 30-day readmission or mortality (12.3% up-classified and 12.8% down-classified). Similarly, up to 44% of hospitals had a meaningful reclassification in their performance status using 30-day risk-adjusted home time vs EDAC (21.9% up-classified and 22.3% down-classified) (Figure 3).

In this study, we found that assessment of hospital-level risk-adjusted 30-day home time for HF using administrative claims data from 100% Medicare hospitalizations was feasible. Second, we observed substantial variability in hospital-level 30-day home time; hospitals with higher 30-day home time were larger, had a higher volume of HF hospitalizations, were more likely to be academic centers, and had greater availability of percutaneous coronary intervention and cardiac rehabilitation facilities. Third, postdischarge stays in SNFs and intermediate- or long-term care facilities were the predominant contributors to the days spent away from home during 30-day follow-up. Fourth, risk-adjusted 30-day home time was significantly associated with postdischarge readmission and mortality rates. Fifth, use of 30-day risk-adjusted home time was associated with meaningful reclassification of hospitals compared with the current CMS performance metrics of 30-day RSRR, 30-day RSMR, and EDAC.

Prior studies^12,15,22 have demonstrated that post-discharge home time may be a meaningful clinical and patient-oriented outcome measure for patients with acute cardiovascular events, such as stroke and HF. A recent study¹⁶ demonstrated the potential utility of risk-adjusted 30-day home time as a hospital-level performance metric for acute myocardial infarction. Findings from the present study add to the existing literature by evaluating 30-day home time as a potential hospital-level performance metric for patients with HF. Consistent with observations for acute myocardial infarction, higher hospital-level 30-day risk-adjusted home time for HF was significantly associated with lower 30-day RSRR and 30-day RSMR rates. Furthermore, up to one-third of hospitals had meaningful reclassification in their performance status associated with 30-day home time compared with 30-day RSRR. Hospitals that were high performing by 30-day RSRR but were reclassified as low performing by 30-day home time had higher inpatient and 30-day mortality and vice versa. This finding is particularly relevant because the 30-day RSRR metric does not account for the competing risk of mortality and therefore can theoretically penalize hospitals that have high HF-related readmission rates but a low HF-associated mortality.^23,24 Compared with a composite metric of 30-day readmission or mortality, there was a meaningful reclassification in performance status of a smaller but still sizable proportion of hospitals (approximately 23%) with use of 30-day home time. Taken together, these findings suggest that the 30-day home time metric complements the existing performance metrics of 30-day RSRR by accounting not only for postdischarge mortality but also for the variability in the post–acute care use of intermediate- and long-term care facilities and SNFs.

In addition to the 30-day readmission metric, the CMS has recently introduced EDAC as a performance metric to capture all acute care facility stays in the postdischarge 30-day period.²¹ However, EDAC largely focuses on days spent in the acute care setting and does not account for postdischarge care provided at other health care facilities. The 30-day risk-adjusted home time complements the EDAC metric by providing a patient-centered measure of health care burden that includes acute care use and postdischarge stays at intermediate- and long-term care facilities and SNFs. We observed meaningful reclassification in more than 40% of hospitals with use of 30-day home time vs EDAC.

Our study findings may have important health policy implications. The 30-day RSRR metric is currently used in the HRRP to financially incentivize hospitals to improve care quality and outcomes among hospitalized patients. Although the implementation of the HRRP has been associated with a significant reduction in the 30-day readmission rate,⁹ its association with patient-oriented outcomes in the postdischarge period is not well established. At the patient level, home time has previously been evaluated as an outcome for patients with acute conditions, such as stroke and HF, and was significantly correlated with health status and patient-reported outcomes.^15,22,25,26 For community-dwelling Medicare beneficiaries, a loss of home time was associated with decreases in patient-reported outcomes, such as difficulty in self-care, mobility impairment, and poor self-rated health.¹³ Our study findings highlight the feasibility of 30-day risk-adjusted home time as a complementary, patient-oriented hospital performance metric that accounts for variation in postdischarge care use and outcomes among patients with HF-related hospitalization. Furthermore, a home time–based hospital performance metric may be more easily understood by patients and practitioners and provides a more comprehensive assessment of the HF-associated burden of morbidity, mortality, and health care use in the postdischarge period.

A 30-day home time–based hospital performance metric may have other important downstream effects on care patterns among patients with HF. We observed that stays at intermediate- or long-term care facilities and SNFs were the main contributors to loss of home time during 30-day follow-up.¹² This finding is particularly relevant considering the increasing use of post–acute care services among patients with HF and substantial variability in care quality and patient outcomes across post–acute care facilities.^27-30 However, the length of stay at SNFs or intermediate- or long-term care facilities may not be directly associated with care provided at the discharging hospital. Thus, the 30-day home time performance metric will hold discharging hospitals accountable, to some extent, for the care received at the post–acute care facilities, which is consistent with the current readmission-based performance model whereby patients readmitted from post–acute care facilities within 30 days of discharge contribute to the discharging hospital’s 30-day RSRR and associated financial penalties. These observations suggest the need for a more shared approach to value-based care and associated financial incentives such that both discharging hospitals and the post–acute care facilities are held accountable for the 30-day outcomes among patients. Furthermore, the home time metric may lead to greater scrutiny into use of post–acute care facilities by discharging hospitals and encourage preferential use of home health care and higher-performing post–acute care facilities.

There may be some unintended consequences of the home time metric for hospital performance that merit consideration. It is plausible that use of home time may disincentivize use of SNFs and intermediate- and long-term care facilities for patients who are frail and require these services in the postdischarge period. However, intentional withholding of indicated post–acute care services may contribute to excess readmission and/or mortality events, which would adversely affect hospital performance based on the home time metric. Furthermore, geographic variations in availability and quality of post–acute care facilities may differentially impact hospital-level 30-day home time performance metric for hospitals based on their location.²⁸

This study has limitations. First, risk-adjusted home time was based on administrative claims data and thus is susceptible to changes in coding practices over time. Second, measures of disease severity, such as ejection fraction, hemodynamic status, and New York Heart Association class, are not available in the claims data and could not be accounted for in the risk-adjustment models, although this is also true for RSRR and RSMR, which are in current use. Third, because our analysis was limited to Medicare fee-for-service beneficiaries, our findings may not be generalizable to younger patients or those with other forms of insurance. Fourth, risk-adjustment variables were derived from inpatient administrative claims data because the data set does not include outpatient clinical data. However, because inpatient claims data can include up to 26 possible patient problems, this approach to risk adjustment should have captured the patient’s clinical comorbidities. Fifth, we did not have detailed information about HF care programs at participating hospitals that may have modified the 30-day home time among patients discharged after an HF-related hospitalization. Future studies are needed to evaluate the effect of HF care quality improvement programs on hospital-level 30-day home time.

In this cohort study of Medicare fee-for-service beneficiaries hospitalized with HF, postdischarge home time was readily calculated using administrative claims and was associated with short-term (30-day readmission and mortality) and longer-term (1-year mortality) outcomes. Furthermore, 30-day home time metric was associated with reclassification of the performance status of up to one-third of hospitals compared with the current CMS standard hospital performance metric of 30-day RSRR.

Accepted for Publication: July 9, 2020.

Published Online: October 28, 2020. doi:10.1001/jamacardio.2020.4928

Corresponding Author: Ambarish Pandey, MD, MSCS, Division of Cardiology, UT Southwestern Medical Center, 5323 Harry Hines Blvd, Dallas, TX 75390-9047 (ambarish.pandey@utsouthwestern.edu).

Author Contributions: Drs Pandey and Keshvani contributed equally to this study and are considered co–first authors. Dr Girotra had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Pandey, Keshvani, Fonarow, Yancy, Girotra.

Acquisition, analysis, or interpretation of data: Pandey, Keshvani, Vaughan-Sarrazin, Gao, Fonarow, Girotra.

Drafting of the manuscript: Pandey, Keshvani, Gao.

Critical revision of the manuscript for important intellectual content: Keshvani, Vaughan Sarrazin, Gao, Fonarow, Yancy, Girotra.

Statistical analysis: Pandey, Vaughan-Sarrazin, Gao.

Obtained funding: Pandey.

Administrative, technical, or material support: Pandey, Girotra.

Supervision: Pandey, Vaughan Sarrazin, Yancy, Girotra.

Conflict of Interest Disclosures: Dr Pandey reported serving on the advisory board of Roche Diagnostics Corporation. Dr Fonarow reported receiving personal fees from Abbott, Amgen, Bayer, CHF Solutions, Janssen, Medtronic, Merck & Co, and Novartis outside the submitted work. No other disclosures were reported.

Disclaimer: Dr Fonarow is an associate editor of JAMA Cardiology and Dr Yancy is a deputy editor of JAMA Cardiology but neither was involved in the editorial review or the decision to accept the manuscript for publication.