Medications that no longer help
Population health approaches to deprescribing
Status: In progress · Funded by: NIH/NIA K01AG076865 · 2022–2027 Most deprescribing happens one patient at a time, when a clinician happens to notice. This project asks whether it can happen at the level of a health system instead: building and validating a model that predicts which older adults are at highest risk from polypharmacy, turning that into a working registry, and testing a pharmacist-led intervention against it.
Role: Michelle Keller, Principal Investigator
Equipping inpatient nurses for deprescribing conversations
Status: In progress · Funded by: U.S. Deprescribing Research Network pilot (NIH/NIA R33AG086944) · 2025–2026
Benzodiazepines and Z-drugs started in the hospital often follow patients home and stay for years. Nurses are present at discharge and are rarely equipped for the conversation. This study designs and pilots a nurse training intervention to change that, built on implementation science rather than education alone.
Role: Michelle Keller, Principal Investigator, Team: Amy Yang
Who is most at risk from polypharmacy
Status: In progress · Funded by: NIH/NIA pilot, Los Angeles Older Americans Independence Center (P30AG094848) · 2025–2026
Age alone is a poor predictor of who gets harmed by taking many medications at once. Using biomarker, medication, and health services data from the Health and Retirement Study, this project tests whether biological age, systemic inflammation, and social conditions identify the people at highest risk more accurately than chronological age does.
Role: Michelle Keller, Principal Investigator, Team: Junxian Liu
How does polypharmacy affect clinical outcomes for patients who get total joint replacements?
Status: In progress · Funded by: Internal funding · 2025–2026
Our team developed and internally validated an evidence-based risk score that uses seven routinely collected preoperative factors to identify patients at elevated risk of hospital revisit within 90 days of total joint arthroplasty, helping care teams focus limited resources on the smaller group most likely to benefit. In a related project, we are using the same preoperative dataset to examine how polypharmacy contributes to hospital readmissions after arthroplasty, with the goal of pinpointing medication-related risks that can be addressed before surgery.
Role: Michelle Keller, Principal Investigator, Team: Junxian Liu
