Machine-Learning Modeling to Predict Hospital Readmission Following Discharge to Post-Acute Care Settings

Abstract Increased attention to post-acute care (PAC) settings and available services to meet patients’ needs following acute hospital discharge is needed as these settings are being utilized increasingly in models of care delivery. The primary purpose was to generate a model to identify the most predictive factors relevant to hospital readmission within 90 days following discharge to one of three types of PAC sites: home with home care services (HC), skilled nursing facility (SNF), in-patient rehabilitation facility (IRF). Specific aims were to (1) examine number and characteristics of older adults discharged to the 3 PAC sites; (2) compare 90 day hospital readmission rate across sites and acuity level; and (3) examine assessment items across population and subgroups to identify variables most predictive of hospital readmission. 2015 assessment data from 3,592,995 Medicare beneficiaries were analyzed representing 1,536,908 from SNFs, 306,878 from IRFs, and 1,749,209 receiving HC services. Total sample 90-day readmission was 25.8 % . Patients discharged to IRF had lowest readmission rate (23.34%), and those receiving HC services had highest readmission rate (29.34%). Creation of risk subgroups however, revealed alternative outcomes. Among all patients in the low, intermediate and high risk groups, the lowest readmission rates occurred among SNF patients. Factor analysis of assessment variables indicated bladder and bowel incontinence and functional limitations were the most distinguishing factors between the very low and very high risk subgroups.

Elder mistreatment (EM) complexity, while described anecdotally, lacks an empirical foundation for measurement. Improved knowledge on the range and nature of concurrent issues that complicate EM intervention would inform the development of more effective solutions and enable greater precision of evaluation research. The purpose of this qualitative study was to explore factors contributing to complexity in a sample of EM cases that was selected based on difficulty reaching resolution. The sample was drawn from those reviewed by an experienced EM Multidisciplinary team (MDT) and determined to require long-term case management (n=39) beyond the capacity of the MDT's usual response. Case manager narrative documentation of ongoing assessment and social service records were qualitatively coded by two researchers. Inductive content analysis, with iterative code reconciliation, was used to identify issues and problems both related and concurrent to EM. Eighteen themes and 74 sub-themes emerged, with 93% initial coding agreement between researchers. The most frequent themes were problems with Caregiving (80%), Cognition (80%), Physical Health (80%), Behavioral Health (69%), Socialization (64%), and Finances (62%). Refusal of formal services was common (90%), yet all accepted visitation by the case manager, suggesting informal support may be effective. Diversity, interconnectedness, and emergence of issues along the duration of case management indicates a system approach to intervention design and evaluation is warranted. This research underscores the need for holistic intervention for highly complex EM, and lays the foundation for objective measure of complexity to standardize selection for specialized intervention. Pia Kontos, Alisa Grigorovich, and Romeo Colobong, The Kite Research Institute -University Health Network, Toronto, Ontario, Canada There have been important advances in research on creativity that have provided a more inclusive view of everyday and ordinary creativity, including that of persons living with dementia. However, these developments are limited by a lack of engagement with scholarship on embodiment, relationality, and citizenship. We address these limitations by drawing on a relational model of citizenship that offers a critical rethinking of the nature of creativity and the imperative that these be supported in long-term dementia care. We draw on transcribed video-recorded interactions between elder-clowns and residents living with dementia in one long-term care home in central Canada. These are analyzed with reference to key theoretical tenets of the relational model of citizenship. Embodied selfhood (i.e., the primordial and socio-cultural dispositions of the body that are fundamental sources of self-expression and relationality), are identified as key to the creativity of persons living with dementia. We argue that creativity is not an individual cognitive trait but rather emerges from the complex intersection of enabling environments and the embodied intentionality of all involved. We conclude that creativity must be supported in everyday life through organizational practices and socio-political institutions that more fully support the relational, interpersonal, and affective dimensions of care.

EVALUATION FINDINGS OF A COMMUNITY-BASED INTERVENTION FOR OLDER ADULTS WITH A HISTORY OF TRAUMA
Carmel Rabin, Karen Edell Yoskowitz, and Barbara Bedney, The Jewish Federations of North America, Washington, District of Columbia, United States Between 70% and 90% of Americans aged 65 and older have experienced at least one traumatic event such as a sexual or physical assault, disaster, illness, or terrorism. Trauma exposure in older adult populations is linked to physical, mental, and cognitive decline. A new approach to improve outcomes of trauma-affected older adults is Person-Centered, Trauma-Informed (PCTI) Care, which promotes the dignity, strength, and empowerment of traumaaffected individuals by incorporating knowledge about trauma into agency programs, policies, and procedures. The Administration for Community Living/Administration on Aging has awarded The Jewish Federations of North America (JFNA) a grant to develop innovative PCTI interventions for Holocaust survivors. This includes a community-based intervention whereby local leadership councils are developed to identify Holocaust survivor needs, distribute grant funding, train caregivers in PCTI care, and forge partnerships to advance community-led Holocaust survivor care. This program has been implemented in eight major US cities where 168 community leaders dispersed 25 grants serving approximately 500 Holocaust survivors. JFNA conducted an evaluation of the first six of the eight cities to determine the impact of this community-based model on participants and Holocaust survivors and investigate the process by which a community-based model can be replicated. This evaluation used surveys and semi-structured interviews to collect data on variables including understanding of PCTI care, awareness of Holocaust survivor needs, strength of community partnerships, and leadership council sustainability. This session will review evaluation findings including best practices for community-based models of PCTI care, and applicability of findings to other older populations.

MACHINE-LEARNING MODELING TO PREDICT HOSPITAL READMISSION FOLLOWING DISCHARGE TO POST-ACUTE CARE SETTINGS
Elizabeth Howard, 1 John N Morris, 2 and Erez Schachter, 3 1. Boston College,Dover,Massachusetts,United States,2. Hebrew SeniorLife,Boston,Massachusetts,United States,3. Profility,Boston,Massachusetts,United States Increased attention to post-acute care (PAC) settings and available services to meet patients' needs following acute hospital discharge is needed as these settings are being utilized increasingly in models of care delivery. The primary purpose was to generate a model to identify the most predictive factors relevant to hospital readmission within 90 days following discharge to one of three types of PAC sites: home with home care services (HC), skilled nursing facility (SNF), in-patient rehabilitation facility (IRF). Specific aims were to (1) examine number and characteristics of older adults discharged to the 3 PAC sites; (2) compare 90 day hospital readmission rate across sites and acuity level; and (3) examine assessment items across population and subgroups to identify variables most predictive of hospital readmission. 2015 assessment data from 3,592,995 Medicare beneficiaries were analyzed representing 1,536,908 from SNFs, 306,878 from IRFs, and 1,749,209 receiving HC services. Total sample 90-day readmission was 25.8 % . Patients discharged to IRF had lowest readmission rate (23.34%), and those receiving HC services had highest readmission rate (29.34%). Creation of risk subgroups however, revealed alternative outcomes. Among all patients in the low, intermediate and high risk groups, the lowest readmission rates occurred among SNF patients. Factor analysis of assessment variables indicated bladder and bowel incontinence and functional limitations were the most distinguishing factors between the very low and very high risk subgroups.

Oregon Health & Science University, Portland, Oregon, United States, 2. Portland State University, Portland, Oregon, United States
Oregon's Behavioral Health Initiative for Older Adults and People with Disabilities is entering its fifth year. This novel state-level Initiative seeks to better coordinate services and resources for older adults and people with disabilities who have behavioral health needs by assigning a Behavioral Health Specialist (BHS) for every 60,000 adults 65 + and embedding them within local service agencies around Oregon. BHS primary job functions include improving coordination and collaboration between local service agencies, providing complex case consultations (CCC), and delivering workforce development training and community education. Five years of data from Portland State University's Institute on Aging's ongoing evaluation of the Initiative suggests significant impact in terms of workforce development trainings, community education, and new community partnerships. Data are collected from BHS and Initiative stakeholders (e.g., aging services agencies). Data collection tools include quarterly reports from the BHS, including a CCC reporting instrument; semi-structured interviews with stakeholders assessing Initiative involvement; and an electronic post-training survey (and two-month follow-up survey) for stakeholders attending BHS trainings. After five years, the evaluation appears to show the Initiative has delivered an abundance of innovative collaborations, workforce trainings, and educational opportunities aimed at better supporting the behavioral health of older Oregonians. It also highlights several persistent systemic barriers including a need for additional public funding of behavioral health, the challenges of accessing Medicare for behavioral health, and siloed agencies and organizations. Future evaluative efforts could explore adding outcomes-based assessments of the Initiative, including local-level quality improvement projects initiated by BHS within their communities.

UNDERSTANDING NURSING TURNOVER: THE CASE OF HOME HEALTH CARE
Joanne Spetz, 1 Alon Bergman, 2 Hummy Song, 2 Amber Rose, 1 and Guy David, 2 1. University of California,San Francisco,San Francisco,California,United States,2. University of Pennsylvania,Philadelphia,Pennsylvania,United States Only a few studies of nursing turnover have examined post-acute home health care. This study examines factors that are associated with home health licensed nurse turnover using linked employee-level and patient-level data from one of the five largest home health companies in the US. The data include variables from human resources and payroll systems, visit logs, discharge records, physical and mental health assessments, care plans, and patient encounters and is organized at the employee-day level. We measured turnover using human resources data, including measures of voluntary and involuntary job separation, and from exit interviews that allow classification of whether turnover was associated with agency-related factors (e.g., pay, schedule, supervisor, coworkers) versus personal factors (e.g., family needs, relocation). In bivariate and multivariate analyses, explanatory variables included nurse demographics, patient population characteristics, and the degree to which nurses can delegate tasks to home care aides. We found a downward trend in turnover for licensed nurses between 2016 and 2019. Attrition in the first year was 34% for full-time nurses and 45% for part-time nurses, most of it occurring in the first 180 days of employment. The rate of voluntary turnover was nearly four times as great as involuntary turnover. We found that agency factors accounted for 26% of monthly turnover on average, while personal factors accounted for 74%. In states in which licensed nurses could delegate more tasks to home care aides, turnover rates were slightly higher than in states with little delegation.

CONSIDERING LTSS THROUGH THE LENSES OF SOCIAL CONSTRUCTION OF TARGET POPULATIONS: THEORY AND NEOLIBERALISM
Tommy Buckley, 1 Carole Cox, 2 and Israel (Issi) Doron, 3 1. Virginia Commonwealth University, Richmond,Virginia,United States,2. Fordham University,New York,New York,United States,3. University of Haifa,Haifa,Israel The global increase of the older population has led to a greater demand for long term support services (LTSS) that address their rights and needs for care. However, policies among countries remain diverse with varying options, services, and recognition of human rights. This study applies the Social Construction of Target Population (SCTP) theory which relates to the perception of older adults and Neoliberalism, a political theory associated with policies of economic privatization, deregulation, and free market activity to the analysis of LTSS systems. As an example, in the United States, the concept of Successful Aging, conflicts with the need for LTSS while present Neoliberal policies that stress Innovation in Aging, 2020, Vol. 4, No. S1