Faculty Bibliography

INTRODUCTION/UNASSIGNED:Compared to stably housed peers, people experiencing homelessness (PEH) have lower rates of ideal glycemic control, and experience premature morbidity and mortality. High rates of behavioral health comorbidities and trauma add to access barriers driving poor outcomes. Limited evidence guides behavioral approaches to support the needs of PEH with diabetes. Lay coaching models can improve care for low-resource populations with diabetes, yet we found no evidence of programs specifically tailored to the needs of PEH. METHODS/UNASSIGNED:We used a multistep, iterative process following the ORBIT model to develop the Diabetes Homeless Medication Support (D-HOMES) program, a new lifestyle intervention for PEH with type 2 diabetes. We built a community-engaged research team who participated in all of the following steps of treatment development: (1) initial treatment conceptualization drawing from evidence-based programs, (2) qualitative interviews with affected people and multi-disciplinary housing and healthcare providers, and (3) an open trial of D-HOMES to evaluate acceptability (Client Satisfaction Questionnaire, exit interview) and treatment engagement (completion rate of up to 10 offered coaching sessions). RESULTS/UNASSIGNED:= 10) overall found the program acceptable, however, we saw better program satisfaction and treatment engagement among more stably housed people. We developed adapted treatment materials for the target population and refined recruitment/retention strategies and trial procedures sensitive to prevalent discrimination and racism to better retain people of color and those with less stable housing. DISCUSSION/UNASSIGNED:The research team has used these findings to inform an NIH-funded randomized control pilot trial. We found synergy between community-engaged research and the ORBIT model of behavioral treatment development to develop a new intervention designed for PEH with type 2 diabetes and address health equity gaps in people who have experienced trauma. We conclude that more work and different approaches are needed to address the needs of participants with the least stable housing.

BACKGROUND:The environment in which clinicians provide care and think about their patients is a crucial and undervalued component of the diagnostic process. CONTENT:In this paper, we propose a new conceptual model that links work conditions to clinician responses such as stress and burnout, which in turn impacts the quality of the diagnostic process and finally patient diagnostic outcomes. The mechanism for these interactions critically depends on the relationship between working memory (WM) and long-term memory (LTM), and ways WM and LTM interactions are affected by working conditions. SUMMARY:We propose a conceptual model to guide interventions to improve work conditions, clinician reactions and ultimately diagnostic process, accuracy and outcomes. OUTLOOK:Improving diagnosis can be accomplished if we are able to understand, measure and increase our knowledge of the context of care.

UNLABELLED:BACKGROUND  : Hospitalist physician stress was exacerbated by the pandemic, yet there have been no large scale studies of contributing factors. OBJECTIVE:Assess remediable components of burnout in hospitalists. PARTICIPANTS, STUDY DESIGN AND MEASURES:In this Coping with COVID study, we focused on assessment of stress factors among 1022 hospital-based clinicians surveyed between April to December 2020. We assessed variables previously associated with burnout (anxiety/depression due to COVID-19, work overload, fear of exposure or transmission, mission/purpose, childcare stress and feeling valued) on 4 point Likert scales, with results dichotomized with the top two categories meaning "present"; burnout was assessed with the Mini Z single item measure (top 3 choices = burnout). Quantitative analyses utilized multilevel logistic regression; qualitative analysis used inductive and deductive methods. These data informed a conceptual model. KEY RESULTS:Of 58,408 HCWs (median response rate 32%), 1022 were hospital-based clinicians (906 (89%) physicians; 449 (44%) female; 469 (46%) White); 46% of these hospital-based clinicians reported burnout. Work overload was associated with almost 5 times the odds of burnout (OR 4.9, 95% CIs 3.67, 6.85, p < 0.001), and those with anxiety or depression had 4 times the odds of burnout (OR 4.2, CIs 3.21, 7.12, p < 0.001), while those feeling valued had half the burnout odds (OR 0.43, CIs 0.31, 0.61, p < 0.001). Regression models estimated 42% of burnout variance was explained by these variables. In open-ended comments, leadership support was helpful, with "great leadership" represented by transparency, regular updates, and opportunities to ask questions. CONCLUSIONS:In this national study of hospital medicine, 2 variables were significantly related to burnout (workload and mental health) while two variables (feeling valued and leadership) were likely mitigators. These variables merit further investigation as means of reducing burnout in hospital medicine.

Palliative care clinicians (PCCs) in the United States face the combination of increasing burnout and a growing need for their services based on demographic changes and an increasing burden of serious illness. In addition to efforts to increase the number of PCCs and to train other clinicians in "primary palliative skills," we must address the burnout in the field to address the growing gap between need for this care and capacity to provide it. To address burnout in PCCs, we must develop solutions with the unique contributors to burnout in this field in mind. PCCs are particularly susceptible to moral distress and moral injury faced by all clinicians, and these states are inextricably linked to burnout. We propose three solutions to address moral distress and moral injury in PCCs to reduce burnout. These solutions are grounded in the dilemmas particular to palliative care and in best evidence: first, to create space for PCCs to confront moral challenges head-on; second, to integrate ethics consultations into care of some patients cared for by PCCs; and third, to reassess care models for PCCs. These approaches can mitigate burnout and thus address the growing gap in our ability to provide high-quality palliative care for those patients in need.

BACKGROUND AND AIMS:Covert HE (CHE) is a common early stage of HE associated with poor outcomes. Available neuropsychiatric diagnostic testing is underutilized and has significant clinical limitations. Sleep deterioration is consistently associated with CHE and HE; however, objective data is sparse and it has not been studied longitudinally. We longitudinally study and describe an association of sleep metrics with CHE as detected by a commercial wearable technology. METHODS:We monitored sleep for 6 months using a commercial fitness tracker in 25 participants with cirrhosis, hypothesizing that CHE as diagnosed by psychometric testing would be associated with significant reductions in sleep quality, especially restorative sleep (deep sleep + rapid eye movement). Mixed-effects modeling was performed to evaluate sleep factors associated with CHE and developed and internally validated a score based on these sleep metrics for associated CHE. RESULTS:Across 2862 nights with 66.3% study adherence, we found that those with CHE had consistently worse sleep, including an average of 1 hour less of nightly restorative sleep, driven primarily by reductions in rapid eye movement. A model including albumin, bilirubin, rapid eye movement, sleep disturbances, and sleep consistency showed good discrimination (area under the receiver operating curve=0.79) for CHE status with a sensitivity of 76% and specificity of 69%. CONCLUSIONS:Our large longitudinal study of sleep in cirrhosis suggests that sleep derangements in CHE can be detected using wearable technology. Given the known importance of sleep to overall health and CHE/HE to prognosis in cirrhosis, the ability to associate dynamic sleep metrics with CHE may in the future help with the detection and passive monitoring as factors that precipitate decompensation of cirrhosis become better understood and mobile health data validation and integration improves.

Breast ultrasound is a valuable adjunctive tool to mammography in detecting breast cancer, especially in women with dense breasts. Ultrasound also plays an important role in staging breast cancer by assessing axillary lymph nodes. However, its utility is limited by operator dependence, high recall rate, low positive predictive value and low specificity. These limitations present an opportunity for artificial intelligence (AI) to improve diagnostic performance and pioneer novel uses of ultrasound. Research in developing AI for radiology has flourished over the past few years. A subset of AI, deep learning, uses interconnected computational nodes to form a neural network, which extracts complex visual features from image data to train itself into a predictive model. This review summarizes several key studies evaluating AI programs' performance in predicting breast cancer and demonstrates that AI can assist radiologists and address limitations of ultrasound by acting as a decision support tool. This review also touches on how AI programs allow for novel predictive uses of ultrasound, particularly predicting molecular subtypes of breast cancer and response to neoadjuvant chemotherapy, which have the potential to change how breast cancer is managed by providing non-invasive prognostic and treatment data from ultrasound images. Lastly, this review explores how AI programs demonstrate improved diagnostic accuracy in predicting axillary lymph node metastasis. The limitations and future challenges in developing and implementing AI for breast and axillary ultrasound will also be discussed.

Background Obesity and illicit drugs are independent risk factors for developing heart failure (HF). However, recent studies have suggested that patients who already have HF and are obese have better clinical outcomes. We aim to study the effect of cocaine use on this obesity paradox phenomenon as it pertains to HF readmissions. Methodology In a retrospective chart analysis, we reviewed patients with a diagnosis of HF with reduced ejection fraction (HFrEF) admitted to Metropolitan Hospital in New York. We studied the association between body mass index (BMI) categories, namely, non-obese (<30 kg/m²) and obese (≥30 kg/m²), cocaine use, and the primary outcome (time to readmission for HF within 30 days after discharge). The interaction between cocaine and obesity status and its association with the primary outcome was also assessed. Results A total of 261 patients were identified. Non-obese status and cocaine use were associated with an increased hazard of readmission in 30 days (hazard ratio (HR) = 2.28, p = 0.049 and HR = 3.12, p = 0.004, respectively). Furthermore, cocaine users who were non-obese were over six times more likely to be re-admitted in 30 days compared to non-cocaine users who were obese (HR = 6.45, p = 0.0002). Conclusions Non-obese status and continued use of cocaine have a negative additive effect in impacting HF readmissions.