Clustering of Rehospitalization Risk Factors and Identification of Associated Home Health Interventions in Aging Sepsis Survivors Receiving Home Healthcare

About This Grant

Project Summary This F99/K00 Transition to Aging Research for Predoctoral Student Award application will support the applicant’s research training to become an independent researcher focused on improving outcomes for older sepsis survivors with complex needs transitioning from hospital to home healthcare (HH) using advanced data science methods. The study aims to understand complex associations of coexisting rehospitalization risk factors (including social support, personal, and contextual factors) with rehospitalization among older sepsis survivors and to identify specific HH interventions that improve care quality and outcomes. In the F99 phase, the applicant will (1) describe distinct and shared profiles of clusters among older sepsis survivors with coexisting rehospitalization risk factors and (2) examine the associations between these clusters and rehospitalization. This retrospective cohort study will use existing national Medicare claims data to identify HH patients aged 65+ with a sepsis diagnosis during an inpatient stay. In the K00 phase, the applicant will build on F99 findings by validating the clustering algorithm with a new, unseen sample and expanding training for the applicant in natural language processing (NLP) to analyze clinical notes by (1) developing an NLP algorithm to identify HH interventions, (2) describing differences in HH interventions across distinct clusters, and (3) assessing the association between identified HH interventions and rehospitalization. Research training will take place at two research-intensive institutions, the University of Pennsylvania (F99) and Columbia University (K00), with ample resources and support from a multidisciplinary mentorship team with expertise in aging, sepsis survivorship, transitions in care, multimorbidity (i.e., multiple chronic conditions or clinical conditions), machine learning, NLP, and quantitative statistics. The training plan includes coursework, practical training, research meetings, seminars, professional and career development, and research dissemination skills. Training will focus on conceptual knowledge (e.g., sepsis survivorship, aging, multimorbidity), methodological skills (e.g., clustering, NLP), rigorous and ethical research conduct, and professional development. To ensure a smooth transition from F99 to K00 to independence, the applicant will acquire the foundational knowledge for each phase. This proposal aligns with the priorities of the National Institute on Aging to understand multilevel aging factors and develop efficacious, cost-effective strategies for older adults. The findings of these proposed studies will provide insights into designing, testing, and implementing cluster-specific care pathways for the growing population of older sepsis survivors.