Cardiac Rehabilitation Use After Heart Failure Hospitalization Associated With Advanced Heart Failure Center Status.

PURPOSE: Cardiac rehabilitation (CR) is an evidence-based, guideline-endorsed therapy for patients with heart failure with reduced ejection fraction (HFrEF) but is broadly underutilized. Identifying structural factors contributing to increased CR use may inform quality improvement efforts. The objective here was to associate hospitalization at a center providing advanced heart failure (HF) therapies and subsequent CR participation among patients with HFrEF. METHODS: A retrospective analysis was performed on a 20% sample of Medicare beneficiaries primarily hospitalized with an HFrEF diagnosis between January 2008 and December 2018. Outpatient claims were used to identify CR use (no/yes), days to first session, number of attended sessions, and completion of 36 sessions. The association between advanced HF status (hospitals performing heart transplantation or ventricular assist device implantations) and CR participation was evaluated with logistic regression, accounting for patient, hospital, and regional factors. RESULTS: Among 143 392 Medicare beneficiaries, 29 487 (20.6%) were admitted to advanced HF centers (HFCs) and 5317 (3.7%) attended a single CR session within 1 yr of discharge. In multivariable analysis, advanced HFC status was associated with significantly greater relative odds of participating in CR (OR = 2.20: 95% CI, 2.08-2.33; P < .001) and earlier initiation of CR participation (-8.5 d; 95% CI, -12.6 to 4.4; P < .001). Advanced HFC status had little to no association with the intensity of CR participation (number of visits or 36 visit completion). CONCLUSIONS: Medicare beneficiaries hospitalized for HF were more likely to attend CR after discharge if admitted to an advanced HFC than a nonadvanced HFC.

Determinants of survival following heart transplantation in adults with congenital heart disease.

BACKGROUND: Adult patients surviving with congenital heart disease (ACHD) is growing. We examine the factors associated with heart transplant outcomes in this challenging population with complex anatomy requiring redo-surgeries. METHODS: We reviewed the United Network for Organ Sharing-Standard Transplant Analysis and Research database and analyzed 35,952 heart transplants from January 1st, 2000, to September 30th, 2018. We compared transplant characteristics for ischemic cardiomyopathy (ICM) (n = 14,236), nonischemic cardiomyopathy (NICM) (n = 20,676), and ACHD (n = 1040). Mean follow-up was 6.20 ± 4.84 years. Kaplan-Meier survival curves and Cox-proportional hazards analysis were used to analyze survival data. RESULTS: Multivariable analysis confirmed that ACHD was associated greater in-hospital death compared to ICM (HR = 0.54, P < 0.001) and NICM (HR = 0.46, P < 0.001). Notable factors associated with increased mortality were history of cerebrovascular disease (HR = 1.11, P = 0.026), prior history of malignancy (HR = 1.12, P = 0.006), pre-transplant biventricular support (HR = 1.12, P = 0.069), postoperative stroke (HR = 1.47, P < 0.001) and postoperative dialysis (HR = 1.71, P < 0.001). ACHD transplants had a longer donor heart ischemic time (P < 0.001) and trend towards more deaths from primary graft dysfunction (P = 0.07). In-hospital deaths were more likely with ACHD and use of mechanical support such as use of right ventricular assist device (HR = 2.20, P = 0.049), biventricular support (HR = 1.62, P < 0.001) and extracorporeal membrane oxygenation (HR = 2.36, P < 0.001). Conditional survival after censoring hospital deaths was significantly higher in ACHD (P < 0.001). CONCLUSION: Heart transplant in ACHD is associated with a higher post-operative mortality given anatomical complexity but a better long-term conditional survival. Normothermic donor heart perfusion may improve outcomes in the ACHD population by reducing the impact of longer ischemic times.

Enhancing heart failure treatment decisions: interpretable machine learning models for advanced therapy eligibility prediction using EHR data.

Timely and accurate referral of end-stage heart failure patients for advanced therapies, including heart transplants and mechanical circulatory support, plays an important role in improving patient outcomes and saving costs. However, the decision-making process is complex, nuanced, and time-consuming, requiring cardiologists with specialized expertise and training in heart failure and transplantation. In this study, we propose two logistic tensor regression-based models to predict patients with heart failure warranting evaluation for advanced heart failure therapies using irregularly spaced sequential electronic health records at the population and individual levels. The clinical features were collected at the previous visit and the predictions were made at the very beginning of the subsequent visit. Patient-wise ten-fold cross-validation experiments were performed. Standard LTR achieved an average F1 score of 0.708, AUC of 0.903, and AUPRC of 0.836. Personalized LTR obtained an F1 score of 0.670, an AUC of 0.869 and an AUPRC of 0.839. The two models not only outperformed all other machine learning models to which they were compared but also improved the performance and robustness of the other models via weight transfer. The AUPRC scores of support vector machine, random forest, and Naive Bayes are improved by 8.87%, 7.24%, and 11.38%, respectively. The two models can evaluate the importance of clinical features associated with advanced therapy referral. The five most important medical codes, including chronic kidney disease, hypotension, pulmonary heart disease, mitral regurgitation, and atherosclerotic heart disease, were reviewed and validated with literature and by heart failure cardiologists. Our proposed models effectively utilize EHRs for potential advanced therapies necessity in heart failure patients while explaining the importance of comorbidities and other clinical events. The information learned from trained model training could offer further insight into risk factors contributing to the progression of heart failure at both the population and individual levels.

Gastrointestinal bleeding following Heartmate 3 left ventricular assist device implantation: The Michigan Bleeding Risk Model.

BACKGROUND: Gastrointestinal bleeding (GIB) results in frequent hospitalizations and impairs quality of life in durable left ventricular assist device (LVAD) recipients. Anticipation of these events before implantation could have important implications for patient selection and management. METHODS: The study population included all adult HeartMate 3 (HM3) primary LVAD recipients enrolled in the STS Intermacs registry from January 2017 to December 2020. Using multivariable modeling methodologies, we investigated the relationships between preimplantation characteristics and postimplant bleeding, bleeding and death, and additional bleeding episodes on subsequent bleeding episodes and created a risk score to predict the likelihood of post-LVAD GIB based solely on preimplantation factors. RESULTS: Of 6,425 patients who received an HM3 LVAD, 1,010 (15.7%) patients experienced GIB. Thirteen preimplantation factors were independent predictors of post-LVAD GIB. A risk score was created from these factors and calculated for each patient. By 3 years postimplant, GIB occurred in 11%, 26%, and 43% of low-, medium- and high-risk patients, respectively. Experiencing 1 post-LVAD GIB event was associated with an increased risk for further GIB events, with 33.9% of patients experiencing at least 1 recurrence. While post-LVAD GIB was associated with mortality, there was no relationship between number of GIB events and death. CONCLUSIONS: The Michigan Bleeding Risk Model is a simple tool, which facilitates the prediction of post-LVAD GIB in HM3 recipients using 13 preimplant variables. The implementation of this tool may help in the risk stratification process and may have therapeutic and clinical implications in HM3 LVAD recipients.

Cytomegalovirus infection is associated with impaired myocardial flow reserve after heart transplantation.

BACKGROUND: Cardiac allograft vasculopathy (CAV) limits long-term survival after heart transplantation (HT). This study evaluates the relationship between clinically significant cytomegalovirus infection (CS-CMVi) and CAV using cardiac positron emission tomography (PET). METHODS: We retrospectively evaluated HT patients from 2005 to 2019 who underwent cardiac PET for CAV evaluation. Multivariable linear and logistic regression models were used to evaluate the association between CS-CMVi and myocardial flow reserve (MFR). Kaplan-Meier and Cox regression analyses were used to assess the relationship between CS-CMV, MFR, and clinical outcomes. RESULTS: Thirty-two (31.1%) of 103 HT patients developed CS-CMVi at a median 9 months after HT. Patients with CS-CMVi had a significantly lower MFR at year 1 and 3, driven by reduction in stress myocardial blood flow. Patients with CS-CMVi had a faster rate of decline in MFR compared to those without infection (-0.10 vs -0.06 per year, p < 0.001). CS-CMVi was an independent predictor of abnormal MFR (<2.0) (odds ratio: 3.8, 95% confidence intervals (CI): 1.4-10.7, p = 0.001) and a lower MFR (ß = -0.39, 95% CI: -0.63 to -0.16, p = 0.001) at year 3. In adjusted survival analyses, both abnormal MFR (log-rank p < 0.001; hazard ratio [HR]: 5.7, 95% CI: 4.2-7.2) and CS-CMVi (log-rank p = 0.028; HR: 3.3, 95% CI: 1.8-4.8) were significant predictors of the primary outcome of all-cause mortality, retransplantation, heart failure hospitalization, and acute coronary syndrome. CONCLUSIONS: CS-CMVi is an independent predictor of reduced MFR following HT. These findings suggest that CMV infection is an important risk factor in the development and progression of CAV.

Predicting need for heart failure advanced therapies using an interpretable tropical geometry-based fuzzy neural network.

BACKGROUND: Timely referral for advanced therapies (i.e., heart transplantation, left ventricular assist device) is critical for ensuring optimal outcomes for heart failure patients. Using electronic health records, our goal was to use data from a single hospitalization to develop an interpretable clinical decision-making system for predicting the need for advanced therapies at the subsequent hospitalization. METHODS: Michigan Medicine heart failure patients from 2013-2021 with a left ventricular ejection fraction ≤ 35% and at least two heart failure hospitalizations within one year were used to train an interpretable machine learning model constructed using fuzzy logic and tropical geometry. Clinical knowledge was used to initialize the model. The performance and robustness of the model were evaluated with the mean and standard deviation of the area under the receiver operating curve (AUC), the area under the precision-recall curve (AUPRC), and the F1 score of the ensemble. We inferred membership functions from the model for continuous clinical variables, extracted decision rules, and then evaluated their relative importance. RESULTS: The model was trained and validated using data from 557 heart failure hospitalizations from 300 patients, of whom 193 received advanced therapies. The mean (standard deviation) of AUC, AUPRC, and F1 scores of the proposed model initialized with clinical knowledge was 0.747 (0.080), 0.642 (0.080), and 0.569 (0.067), respectively, showing superior predictive performance or increased interpretability over other machine learning methods. The model learned critical risk factors predicting the need for advanced therapies in the subsequent hospitalization. Furthermore, our model displayed transparent rule sets composed of these critical concepts to justify the prediction. CONCLUSION: These results demonstrate the ability to successfully predict the need for advanced heart failure therapies by generating transparent and accessible clinical rules although further research is needed to prospectively validate the risk factors identified by the model.

Longitudinal Change and Predictors of Myocardial Flow Reserve by Positron Emission Tomography for the Evaluation of Cardiac Allograft Vasculopathy Following Heart Transplantation.

BACKGROUND: Positron emission tomography (PET) myocardial flow reserve (MFR) is a noninvasive method of detecting cardiac allograft vasculopathy in recipients of heart transplants (HTs). There are limited data on longitudinal change and predictors of MFR following HT. METHODS: We conducted a retrospective analysis of HT recipients undergoing PET myocardial perfusion imaging at an academic center. Multivariable linear and Cox regression models were constructed to identify longitudinal trends, predictors and the prognostic value of MFR after HT. RESULTS: Of HT recipients, 183 underwent 658 PET studies. The average MFR was 2.34 ± 0.70. MFR initially increased during the first 3 years following HT (+ 0.12 per year; P = 0.01) before beginning to decline at an annual rate of -0.06 per year (P < 0.001). MFR declines preceding acute rejection and improves after treatment. Treatment with mammalian target of rapamycin (mTOR) inhibitors (37.2%) slowed the rate of annual MFR decline (P = 0.03). Higher-intensity statin therapy was associated with improved MFR. Longer time post-transplant (P < 0.001), hypertension (P < 0.001), chronic kidney disease (P < 0.001), diabetes mellitus (P = 0.038), antibody-mediated rejection (P = 0.040), and cytomegalovirus infection (P = 0.034) were associated with reduced MFR. Reduced MFR (HR: 7.6, 95% CI: 4.4-13.4; P < 0.001) and PET-defined ischemia (HR: 2.3, 95% CI: 1.4-3.9; P < 0.001) were associated with a higher risk of the composite outcome of mortality, retransplantation, heart failure hospitalization, acute coronary syndrome, or revascularization. CONCLUSION: MFR declines after the third post-transplant year and is prognostic for cardiovascular events. Cardiometabolic risk-factor modification and treatment with higher-intensity statin therapy and mechanistic target of rapamycin inhibitors are associated with a higher MFR.

Heart Failure With Recovered Ejection Fraction in Patients With Nonischemic Cardiomyopathy: Characteristics, Outcomes, and Long-term Follow-up.

BACKGROUND: Duration of recovery and long-term outcomes have not been well-described in a large cohort of patients with heart failure with recovered ejection fraction (HFrecEF) owing to nonischemic cardiomyopathy. The aim of the study was to characterize the duration of recovery and long-term outcomes of patients with HFrecEF. METHODS AND RESULTS: We performed a retrospective analysis of our institution's databases. Only patients with nonischemic cardiomyopathy, a chronic HF diagnosis, and a previous left ventricular ejection fraction (LVEF) of ≤35% who had a subsequent LVEF of ≥50% were considered to have recovery. Patients with an LVEF of ≤35% who did not recover served as the comparison group. Included were 2319 patients with an LVEF of ≤35%, of whom 465 (20% [18.4%-21.7%]) met the above criteria for recovery (HFrecEF group). Recovery in the HFrecEF group was temporary in most cases, with 50% of patients experiencing a decline in LVEF to <50% within 3.5 [interquartile range 2.4-4.9] years after the day of recovery. Age and sex adjusted death and hospitalization were lower in the HFrecEF group than the HFrEF group (HR 0.29 [interquartile range 0.20-0.41] for death and 0.44 [interquartile range 0.32-0.60] for HF hospitalization, P < .0001 for both). Longer recovery was associated with better survival, with patients spending >5 years in recovery (LVEF of ≥50%) displaying the highest survival rates (83% alive at 10 years after recovery). Survival after recurrence of LV dysfunction was longer for those whose recovery duration was >1 year. CONCLUSIONS: Patients with nonischemic HFrecEF display a unique clinical course. Although recovery is temporary in most cases, patients with HFrecEF display lower mortality and hospitalization rates, with the more durable the recovery of LV systolic function, the longer survival can be anticipated.

Gastrointestinal bleeding on continuous-flow left ventricular assist device therapy.

While technological advances in the field of continuous-flow left ventricular assist device (CF-LVAD) are constantly being made, CF-LVAD recipients are still subjected to a relatively high rate of LVAD-related adverse events, with post-LVAD gastrointestinal bleeding (GIB) being the most common one. GIB is associated with a significant impairment in quality of life, multiple hospital admissions, blood transfusion requirements and possibly death. Furthermore, of those patients who bled once, many will experience recurrent GIB events, which further aggravates their discomfort. While some medical and endoscopic treatment options are available, evidence regarding their benefit remains largely equivocal, with all related studies based on data from registries rather than clinical trials. Although having a major impact on LVAD recipients, an effective and validated pre-implant screening options to predict GIB events post-implant are scarce. This review focuses on the aetiology, incidence, risk factors, treatment options and the effect of new generation devices on post-LVAD GIB.

Infections following left ventricular assist device implantation and 1-year health-related quality of life.

BACKGROUND: Left ventricular assist device (LVAD) implantation leads to substantial and sustained improvement in health-related quality of life (HRQOL) among patients. Infection following device implantation remains an important and frequent complication and adversely affects patient-reported HRQOL. METHODS: Patients in The Society of Thoracic Surgeons' Interagency Registry for Mechanically Assisted Circulatory Support receiving a primary LVAD between April 2012 to October 2016 were included. The primary exposure was one-year post-implant infection, characterized by: (1) any infection; (2) total number of infections and (3) type (LVAD-specific, LVAD-related, non-LVAD). The association between infection and the primary composite adverse outcome (defined as EuroQoL Visual Analog Scale< 65, too sick to complete the survey, or death at 1-year) was estimated using inverse probability weighting and Cox regression. RESULTS: The study cohort included 11,618 patients from 161 medical centers with 4,768 (41.0%) patients developing an infection, and 2,282 (19.6%) patients having> 1 infection during the follow up period. The adjusted odds ratio for the primary composite adverse outcome was 1.22 (95% CI, 1.19-1.24, p < 0.001) for each additional infection. Each additional infection was associated with a 3.49% greater probability of the primary composite outcome and was associated with worse performance across multiple dimensions of HRQOL as assessed by the EQ-5D for patients who survived to 1 year. CONCLUSIONS: For patients undergoing LVAD implantation, each additional infection within the first post-implantation year was associated with an incremental negative effect on survival free of impaired HRQOL.

Translation of immunomodulatory therapy to treat chronic heart failure: Preclinical studies to first in human.

BACKGROUND: Inflammation has been associated with progression and complications of chronic heart failure (HF) but no effective therapy has yet been identified to treat this dysregulated immunologic state. The selective cytopheretic device (SCD) provides extracorporeal autologous cell processing to lessen the burden of inflammatory activity of circulating leukocytes of the innate immunologic system. AIM: The objective of this study was to evaluate the effects of the SCD as an extracorporeal immunomodulatory device on the immune dysregulated state of HF. HF. METHODS AND RESULTS: SCD treatment in a canine model of systolic HF or HF with reduced ejection fraction (HFrEF) diminished leukocyte inflammatory activity and enhanced cardiac performance as measured by left ventricular (LV) ejection fraction and stroke volume (SV) up to 4 weeks after treatment initiation. Translation of these observations in first in human, proof of concept clinical study was evaluated in a patient with severe HFrEFHFrEF ineligible for cardiac transplantation or LV LV assist device (LVAD) due to renal insufficiency and right ventricular dysfunction. Six hour SCD treatments over 6 consecutive days resulted in selective removal of inflammatory neutrophils and monocytes and reduction in key plasma cytokines, including tumor necrosis factor-alpha (TNF-α),), interleukin (IL)-6, IL-8, and monocyte chemoattractant protein (MCP)-1. These immunologic changes were associated with significant improvements in cardiac power output, right ventricular stroke work index, cardiac index and LVSV index…. Stabilization of renal function with progressive volume removal permitted successful LVAD implantation. CONCLUSION: This translational research study demonstrates a promising immunomodulatory approach to improve cardiac performance in HFrEFHFrEF and supports the important role of inflammation in the progression of HFHF.

Effectiveness of the Family Heart Talk Communication Tool in Improving Family Member Screening for Dilated Cardiomyopathy: Results of a Randomized Trial.

BACKGROUND: Managing disease risk among first-degree relatives of probands diagnosed with a heritable disease is central to precision medicine. A critical component is often clinical screening, which is particularly important for conditions like dilated cardiomyopathy (DCM) that remain asymptomatic until severe disease develops. Nonetheless, probands are frequently ill-equipped to disseminate genetic risk information that motivates at-risk relatives to complete recommended clinical screening. An easily implemented remedy for this key issue has been elusive. METHODS: The DCM Precision Medicine Study developed Family Heart Talk, a booklet designed to help probands with DCM communicate genetic risk and the need for cardiovascular screening to their relatives. The effectiveness of the Family Heart Talk booklet in increasing cardiovascular clinical screening uptake among first-degree relatives was assessed in a multicenter, open-label, cluster-randomized, controlled trial. The primary outcome measured in eligible first-degree relatives was completion of screening initiated within 12 months after proband enrollment. Because probands randomized to the intervention received the booklet at the enrollment visit, eligible first-degree relatives were limited to those who were alive the day after proband enrollment and not enrolled on the same day as the proband. RESULTS: Between June 2016 and March 2020, 1241 probands were randomized (1:1) to receive Family Heart Talk (n=621) or not (n=620) within strata defined by site and self-identified race/ethnicity (non-Hispanic Black, non-Hispanic White, or Hispanic). Final analyses included 550 families (n=2230 eligible first-degree relatives) in the Family Heart Talk arm and 561 (n=2416) in the control arm. A higher percentage of eligible first-degree relatives completed screening in the Family Heart Talk arm (19.5% versus 16.0%), and the odds of screening completion among these first-degree relatives were higher in the Family Heart Talk arm after adjustment for proband randomization stratum, sex, and age quartile (odds ratio, 1.30 [1-sided 95% CI, 1.08-∞]). A prespecified subgroup analysis did not find evidence of heterogeneity in the adjusted intervention odds ratio across race/ethnicity strata (P=0.90). CONCLUSIONS: Family Heart Talk, a booklet that can be provided to patients with DCM by clinicians with minimal additional time investment, was effective in increasing cardiovascular clinical screening among first-degree relatives of these patients. REGISTRATION: URL: https://www. CLINICALTRIALS: gov; Unique identifier: NCT03037632.

Association of Days Alive and Out of the Hospital After Ventricular Assist Device Implantation With Adverse Events and Quality of Life.

Importance: There is a need to better assess the cumulative effect on morbidity and mortality in patients undergoing durable left ventricular assist device (LVAD) implantation. This study evaluates a patient-centered performance metric (days alive and out of hospital [DAOH]) for durable LVAD therapy. Objective: To determine the incidence of percent of DAOH before and after LVAD implantation and (2) explore its association with established quality metrics (death, adverse events [AEs], quality of life). Design, Settings, and Participants: This was a retrospective national cohort study of Medicare beneficiaries implanted with a durable continuous-flow LVAD between April 2012 and December 2016. The data were analyzed from December 2021 to May 2022. Follow-up was 100% complete at 1 year. Data from The Society of Thoracic Surgeons Intermacs registry were linked to Medicare claims. Main Outcomes and Measures: The number of DAOH 180 days before and 365 days after LVAD implantation and daily patient location (home, index hospital, nonindex hospital, skilled nursing facility, rehabilitation center, hospice) were calculated. Percent of DAOH was indexed to each beneficiary's pre- (percent DAOH-BF) and postimplantation (percentage of DAOH-AF) follow-up time. The cohort was stratified by terciles of percentage of DAOH-AF. Results: Among the 3387 patients included (median [IQR] age: 66.3 [57.9-70.9] years), 80.9% were male, 33.6% and 37.1% were Interfaces Patient Profile 2 and 3, respectively, and 61.1% received implants as destination therapy. Median (IQR) percent of DAOH-BF was 88.8% (82.7%-93.8%) and 84.6% (62.1-91.5%) for percent of DAOH-AF. While DAOH-BF was not associated with post-LVAD outcomes, patients in the low tercile of percentage of DAOH-AF had a longer index hospitalization stay (mean, 44 days; 95% CI, 16-77), were less likely to be discharged home (mean. -46.4 days; 95% CI, 44.2-49.1), and spent more time in a skilled nursing facility (mean, 27 days; 95% CI, 24-29), rehabilitation center (mean, 10 days; 95% CI, 8-12), or hospice (mean, 6 days; 95% CI, 4-8). Increasing percentage of DAOH-AF was associated with patient risk, AEs, and indices of HRQoL. Patients experiencing non-LVAD-related AEs had the lowest percentage of DAOH-AF. Conclusions and Relevance: Significant variability existed in the percentage of DAOH within a 1-year time horizon and was associated with the cumulative AEs burden. This patient-centered measure may assist clinicians in informing patients about expectations after durable LVAD implantation. Validation of percentage DAOH as a quality metric for LVAD therapy across centers should be explored.

Racial Inequities in Access to Ventricular Assist Device and Transplant Persist After Consideration for Preferences for Care: A Report From the REVIVAL Study.

BACKGROUND: Racial disparities in access to advanced therapies for heart failure (HF) patients are well documented, although the reasons remain uncertain. We sought to determine the association of race on utilization of ventricular assist device (VAD) and transplant among patients with access to care at VAD centers and if patient preferences impact the effect. METHODS: We performed an observational cohort study of ambulatory chronic systolic HF patients with high-risk features and no contraindication to VAD enrolled at 21 VAD centers and followed for 2 years in the REVIVAL study (Registry Evaluation of Vital Information for VADs in Ambulatory Life). We used competing events cause-specific proportional hazard methodology with multiple imputation for missing data. The primary outcomes were (1) VAD/transplant and (2) death. The exposures of interest included race (Black or White), additional demographics, captured social determinants of health, clinician-assessed HF severity, patient-reported quality of life, preference for VAD, and desire for therapies. RESULTS: The study included 377 participants, of whom 100 (26.5%) identified as Black. VAD or transplant was performed in 11 (11%) Black and 62 (22%) White participants, although death occurred in 18 (18%) Black and 36 (13%) White participants. Black race was associated with reduced utilization of VAD and transplant (adjusted hazard ratio, 0.45 [95% CI, 0.23-0.85]) without an increase in death. Preferences for VAD or life-sustaining therapies were similar by race and did not explain racial disparities. CONCLUSIONS: Among patients receiving care by advanced HF cardiologists at VAD centers, there is less utilization of VAD and transplant for Black patients even after adjusting for HF severity, quality of life, and social determinants of health, despite similar care preferences. This residual inequity may be a consequence of structural racism and discrimination or provider bias impacting decision-making. REGISTRATION: URL: https://www. CLINICALTRIALS: gov; Unique identifier: NCT01369407.

Failure to rescue: A candidate quality metric for durable left ventricular assist device implantation.

OBJECTIVE: Failure to rescue (FTR), defined as death after a complication, is recognized as a principal driver of variation in mortality among hospitals. We evaluated FTR as a quality metric in patients who received durable left ventricular assist devices (LVADs) using the Society of Thoracic Surgeons Interagency Registry for Mechanically Assisted Circulatory Support. METHODS: Data on 13,617 patients who received primary durable LVADs from April 2012 to October 2017 at 131 hospitals that performed at least 20 implants were analyzed from the Society of Thoracic Surgeons Interagency Registry for Mechanically Assisted Circulatory Support. Rates of major complications and FTR were compared across risk-adjusted in-hospital mortality terciles (low, medium, high) and hospital volume. Logistic regression was used to estimate expected FTR rates on the basis of patient factors for each major complication. RESULTS: The overall unadjusted in-hospital mortality rate was 6.96%. Risk-adjusted in-hospital mortality rates varied 3.1-fold across terciles (low, 3.3%; high, 10.3%; P trend <.001). Rates of major complications varied 1.1-fold (low, 34.0%; high, 38.8%; P < .0001). Among patients with a major complication, 854 died in-hospital for an FTR rate of 17.7%, with 2.8-fold variation across mortality terciles (low, 8.5%; high, 23.9%; P < .0001). FTR rates were highest for renal dysfunction requiring dialysis (45.3%) and stroke (36.5%). Higher average annual LVAD volume was associated with higher rates of major complications (<10 per year, 26.7%; 10-20 per year, 34.0%; 20-30 per year, 34.0%; >30 per year, 40.1%; P trend <.0001) whereas hospitals implanting <10 per year had the highest FTR rate (<10 per year, 23.5%; 10-20 per year, 16.5%; 20-30 per year, 17.0%; >30 per year, 17.9%; P = .03). CONCLUSIONS: FTR might serve as an important quality metric for durable LVAD implant procedures, and identifying strategies for successful rescue after complications might reduce hospital variations in mortality.

Mortality following durable left ventricular assist device implantation by timing and type of first infection.

OBJECTIVE: Although infections are common after left ventricular assist device implantation, the relationship between timing and type of first infection with regard to mortality is less well understood. METHODS: The Society of Thoracic Surgeons Interagency Registry for Mechanically Assisted Circulatory Support patients receiving a primary left ventricular assist device from April 2012 to May 2017 were included. The primary exposure was defined 3 ways: any infection, timing of first infection (early: ≤90 days; intermediate: 91-180 days; late: >180 days), and type (ventricular assist device specific, ventricular assist device related, non-ventricular assist device). The association between first infection and all-cause mortality was estimated using Cox regression. RESULTS: The cohort included 12,957 patients at 166 centers (destination therapy: 47.4%, bridge-to-transplant: 41.2%). First infections were most often non-ventricular assist device (54.2%). Rates of first infection were highest in the early interval (10.7/100 person-months). Patients with any infection had a significantly higher adjusted hazard of death (hazard ratio, 2.63; 2.46-2.86). First infection in the intermediate interval was associated with the largest increase in adjusted hazard of death (hazard ratio, 3.26; 2.82-3.78), followed by late (hazard ratio, 3.13; 2.77-3.53) and early intervals (hazard ratio, 2.37; 2.16-2.60). Ventricular assist device-related infections were associated with the largest increase in hazard of death (hazard ratio, 3.02; 2.69-3.40), followed by ventricular assist device specific (hazard ratio, 2.92; 2.57-3.32) and non-ventricular assist device (hazard ratio, 2.42; 2.20-2.65). CONCLUSIONS: Relative to those without infection, patients with any postimplantation infection had an increased risk of death. Ventricular assist device-related infections and infections occurring in the intermediate interval were associated with the largest increase in risk of death. After left ventricular assist device implantation, infection prevention strategies should target non-ventricular assist device infections in the first 90 days, then shift to surveillance/prevention of driveline infections after 90 days.

Mitral regurgitation severity at left ventricular assist device implantation is associated with distinct myocardial transcriptomic signatures.

OBJECTIVES: We examined for differences in pre-left ventricular assist device (LVAD) implantation myocardial transcriptome signatures among patients with different degrees of mitral regurgitation (MR). METHODS: Between January 2018 and October 2019, we collected left ventricular (LV) cores during durable LVAD implantation (n = 72). A retrospective chart review was performed. Total RNA was isolated from LV cores and used to construct cDNA sequence libraries. The libraries were sequenced with the NovaSeq system, and data were quantified using Kallisto. Gene Set Enrichment Analysis (GSEA) and Gene Ontology analyses were performed, with a false discovery rate <0.05 considered significant. RESULTS: Comparing patients with preoperative mild or less MR (n = 30) and those with moderate-severe MR (n = 42), the moderate-severe MR group weighted less (P = .004) and had more tricuspid valve repairs (P = .043), without differences in demographics or comorbidities. We then compared both groups with a group of human donor hearts without heart failure (n = 8). Compared with the donor hearts, there were 3985 differentially expressed genes (DEGs) for mild or less MR and 4587 DEGs for moderate-severe MR. Specifically altered genes included 448 DEGs for specific for mild or less MR and 1050 DEGs for moderate-severe MR. On GSEA, common regulated genes showed increased immune gene expression and reduced expression of contraction and energetic genes. Of the 1050 genes specific for moderate-severe MR, there were additional up-regulated genes related to inflammation and reduced expression of genes related to cellular proliferation. CONCLUSIONS: Patients undergoing durable LVAD implantation with moderate-severe MR had increased activation of genes related to inflammation and reduction of cellular proliferation genes. This may have important implications for myocardial recovery.

A Novel Tropical Geometry-Based Interpretable Machine Learning Method: Pilot Application to Delivery of Advanced Heart Failure Therapies.

A model's interpretability is essential to many practical applications such as clinical decision support systems. In this article, a novel interpretable machine learning method is presented, which can model the relationship between input variables and responses in humanly understandable rules. The method is built by applying tropical geometry to fuzzy inference systems, wherein variable encoding functions and salient rules can be discovered by supervised learning. Experiments using synthetic datasets were conducted to demonstrate the performance and capacity of the proposed algorithm in classification and rule discovery. Furthermore, we present a pilot application in identifying heart failure patients that are eligible for advanced therapies as proof of principle. From our results on this particular application, the proposed network achieves the highest F1 score. The network is capable of learning rules that can be interpreted and used by clinical providers. In addition, existing fuzzy domain knowledge can be easily transferred into the network and facilitate model training. In our application, with the existing knowledge, the F1 score was improved by over 5%. The characteristics of the proposed network make it promising in applications requiring model reliability and justification.