Racial and Sex Inequities in the Use of and Outcomes After Left Ventricular Assist Device Implantation Among Medicare Beneficiaries.

Importance: While left ventricular assist devices (LVADs) increase survival for patients with advanced heart failure (HF), racial and sex access and outcome inequities remain and are poorly understood. Objectives: To assess risk-adjusted inequities in access and outcomes for both Black and female patients and to examine heterogeneity in treatment decisions among patients for whom clinician discretion has a more prominent role. Design, Setting, and Participants: This retrospective cohort study of 12 310 Medicare beneficiaries used 100% Medicare Fee-for-Service administrative claims. Included patients had been admitted for heart failure from 2008 to 2014. Data were collected from July 2007 to December 2015 and analyzed from August 23, 2020, to May 15, 2022. Exposures: Beneficiary race and sex. Main Outcomes and Measures: The propensity for LVAD implantation was based on clinical risk factors from the 6 months preceding HF admission using XGBoost and the synthetic minority oversampling technique. Beneficiaries with a 5% or greater probability of receiving an LVAD were included. Logistic regression models were estimated to measure associations of race and sex with LVAD receipt adjusting for clinical characteristics and social determinants of health (eg, distance from LVAD center, Medicare low-income subsidy, neighborhood deprivation). Next, 1-year mortality after LVAD was examined. Results: The analytic sample included 12 310 beneficiaries, of whom 22.9% (n = 2819) were Black and 23.7% (n = 2920) were women. In multivariable models, Black beneficiaries were 3.0% (0.2% to 5.8%) less likely to receive LVAD than White beneficiaries, and women were 7.9% (5.6% to 10.2%) less likely to receive LVAD than men. Individual poverty and worse neighborhood deprivation were associated with reduced use, 2.9% (0.4% to 5.3%) and 6.7% (2.9% to 10.5%), respectively, but these measures did little to explain observed disparities. The racial disparity was concentrated among patients with a low propensity score (propensity score <0.52). One-year survival by race and sex were similar on average, but Black patients with a low propensity score experienced improved survival (7.2% [95% CI, 0.9% to 13.5%]). Conclusions and Relevance: In this cohort study of Medicare beneficiaries hospitalized for HF, disparities in LVAD use by race and sex existed and were not explained by clinical characteristics or social determinants of health. The treatment and post-LVAD survival by race were equivalent among the most obvious LVAD candidates. However, there was differential use and outcomes among less clear-cut LVAD candidates, with lower use but improved survival among Black patients. Inequity in LVAD access may have resulted from differences in clinician decision-making because of systemic racism and discrimination, implicit bias, or patient preference.

Discovering anomalous patterns in large digital pathology images.

Advances in medical imaging technology have created opportunities for computer-aided diagnostic tools to assist human practitioners in identifying relevant patterns in massive, multiscale digital pathology slides. This work presents Hierarchical Linear Time Subset Scanning, a novel statistical method for pattern detection. Hierarchical Linear Time Subset Scanning exploits the hierarchical structure inherent in data produced through virtual microscopy in order to accurately and quickly identify regions of interest for pathologists to review. We take a digital image at various resolution levels, identify the most anomalous regions at a coarse level, and continue to analyze the data at increasingly granular resolutions until we accurately identify its most anomalous subregions. We demonstrate the performance of our novel method in identifying cancerous locations on digital slides of prostate biopsy samples and show that our methods detect regions of cancer in minutes with high accuracy, both as measured by the ROC curve (measuring ability to distinguish between benign and cancerous slides) and by the spatial precision-recall curve (measuring ability to pick out the malignant areas on a slide which contains cancer). Existing methods need small scale images (small areas of a slide preselected by the pathologist for analysis, eg, 32 × 32 pixels) and may not work effectively on large, raw digitized images of size 100K × 100K pixels. In this work, we provide a methodology to fill this significant gap by analyzing large digitized images and identifying regions of interest that may be indicative of cancer.