Enhancing prognosis communication for older adults in skilled nursing facilities

University Of California, San Francisco, San Francisco CA

Investigators

Abstract

PROJECT SUMMARY/ABSTRACT This is an application for a Beeson K76 award for Dr. James Deardorff, a geriatrician and Assistant Professor at the University of California, San Francisco. Dr. Deardorff's career goal is to become an independent investigator focused on trialing methods to optimize prognosis communication for older adults receiving post-acute care at a skilled nursing facility (SNF). Nearly 20% of hospitalized older adults are discharged to a SNF, representing 1.8 million SNF stays annually. While SNF admissions are often focused on short-term rehabilitation with the goal of returning home, they represent a particularly vulnerable time for older adults: ~20% are re-admitted to the hospital and ~15% transition to long-term nursing home care. Many patients receive overly optimistic information about their health trajectory and feel blindsided when their goals are not met. SNF patients could benefit from a communication tool that provides accurate and personalized prognostic information about the likelihood of various outcomes, yet that tool does not yet exist. Dr. Deardorff has a strong background in developing prognostic calculators and has created a prognostic model for older adults at SNFs using logistic regression through an R03 GEMSSTAR, which predicts the likelihood of hospital re-admission, nursing home transition for long-term care, and 6-month mortality. However, this model could be improved upon through novel machine-learning methods, and simply developing prediction models is not enough; they must be used to impact change. This K76 proposal addresses a critical gap in SNF care by developing and implementing a prognosis communication tool containing personalized prognostic information. Mentored by an extraordinary team led by Dr. Sei Lee, Dr. Deardorff will: 1) develop and compare novel multi-outcome machine learning-based models to traditional logistic regression to ensure highly accurate risk estimates (Aim 1); 2) assess prognosis communication needs through semi-structured interviews with key SNF informants and develop and refine a SNF prognosis communication tool containing prognostic estimates from the best performing model in Aim 1, anticipatory guidance for SNF care, and questions to elicit patients' values and preferences (Aim 2); and 3) pilot the prognosis communication tool to evaluate its feasibility, acceptability, and preliminary impact on patient-centered outcomes (Aim 3). These Aims correspond with career development activities focused on 1) advanced machine learning methods to develop high-performing prognostic models, 2) qualitative research skills to understand facilitators and barriers to communicating prognostic estimates, 3) development and implementation of communication tools in SNFs, and 4) leadership skills. Results will inform an R01 proposal to evaluate the effectiveness of the prognosis communication tool on patient-centered outcomes in a randomized clinical trial within SNFs. Dr. Deardorff will receive the necessary training and experience in machine learning, qualitative research, implementation science, and pragmatic trial design to launch his career as a leading investigator in post-acute care for older adults at SNFs.