Kenneth M. Lee, PhD
Postdoctoral Researcher
Kenneth Lee is a postdoctoral researcher in the Department of Biostatistics, Epidemiology, and Informatics at the University of Pennsylvania, mentored by Dr. Michael Harhay, Dr. Katherine Courtright, and Dr. Fan Li (Yale). He received his PhD in Biostatistics from Duke-NUS Medical School and specializes in the design and statistical analysis of cluster randomized trials and self-controlled case series.
At Penn, Dr. Lee is developing novel statistical methods for analyzing palliative care outcomes and continues to lead work on methods for time-varying treatment effects, informative sizes, and unmeasured confounding in various cluster randomized trial designs. He also serves as a core instructor for the Pragmatic Clinical Research Institute.
Dr. Lee is a recipient of the FDA-OCE-ASA Oncology Educational Fellowship and was a finalist for the Royal Statistical Society 2023 Statistical Excellence Award for Early-Career Writing.
Research Areas
Biostatistics, Clinical Trials, Cluster Randomized Trials, Stepped-Wedge Designs, Cross-Over Designs, Difference-in-Differences, Multilevel Data, Time-Varying Treatment Effects, Informative Sizes, Self-Controlled Case Series, Causal Inference, Estimands, Palliative Care
Education
- PhD, Biostatistics, Duke-NUS Medical School 2024
- BA, Neuroscience, Vassar College 2018
Contact
Email: kenneth.lee@pennmedicine.upenn.edu
🎓 Google Scholar
đź”— ORCID
Research Highlights
Nudging Effective and Equitable Delivery of Specialty Palliative Care (NEEDS-PC)
Overview: Millions of Americans living with serious illness experience burdensome symptoms and receive goal-discordant care that diminishes their quality of life. Palliative care (PC) is widely endorsed to address these pervasive problems with serious illness care, with its benefits for patients, caregivers, and health systems having been demonstrated in many randomized trials. Despite this, PC delivery remains inefficient and inequitable among patients with serious illness, largely due to clinicians’ challenges in identifying which patients are most likely to benefit from it. To address these challenges, many hospitals have sought to overcome these key barriers by implementing standard referral criteria “triggers” to facilitate patient identification. However, these triggers are commonly based on specific diagnoses or patient locations, such as intensive care units, and therefore may not adequately reflect a patient’s palliative care needs.
This study evaluates an automated, electronic health record (EHR)-based palliative care trigger across nine hospitals in the MedStar health system, assessing its impact on 64,000 seriously ill patients. Over 39 months, the study transitions hospitals from usual care, to implementing automated PC triggers and default consult orders, aiming to enhance care equity and increase hospital-free days. This intervention’s effectiveness and implementation are measured using patient-centered outcomes and hospital collected metrics. Conducted in an established collaboration between Penn’s Palliative and Advanced Illness Research (PAIR) Center, MedStar Health, and Cerner Health (the second-most common EHR in U.S. hospitals), our study seeks to close crucial gaps in our understanding of how to deliver palliative care effectively and equitably. Our central hypotheses are that an EHR-based PC needs trigger will improve both patient-centered outcomes and the equity of PC delivery compared to usual care, and that combining this trigger with default PC consult orders will improve these outcomes further compared to the trigger alone.
Prognosticating Outcomes and Nudging Decisions with Electronic Records in the ICU (PONDER-ICU)
Overview: Most critically ill patients want care that promotes the length and quality of their lives. Too often, their care focuses on only one of these goals. There is a mismatch between the care that patients want and the care they receive. This is because, sometimes, clinicians do not talk to patients about their goals and wants. Or, clinicians do not do it soon enough. This pragmatic, stepped-wedge, cluster-randomized trial tested two electronic health record interventions in 17 intensive care units (ICUs) at 10 hospitals. The interventions were designed to increase ICU clinicians’ engagement of critically ill patients and caregivers in discussions about alternative treatment options, including care focused on comfort. We hypothesized outcomes might improve by requiring ICU clinicians to assess patients’ prognosis at 6 months and provide a justification if they did not offer patients the option of comfort-oriented care.
A Pragmatic Clinical Trial To Improve Inpatient Palliative Care Delivery for Seriously Ill Adults (Palliative Connect)
Overview: Millions of Americans living with serious illness experience burdensome symptoms and receive aggressive care that is not aligned with their goals and preferences. A growing body of evidence suggests that palliative care, which entails a supportive approach to care focused on maximizing quality of life, improves patient-centered, clinical, and economic outcomes. For this reason, national guidelines recommend that clinicians provide palliative care themselves (“primary”) or consult a specialist (“specialty”) as a standard part of caring for seriously ill patients. Yet, many patients with life-limiting illness never receive palliative care, and for those who do, it is typically in the final weeks or days of life, thus limiting its benefits. While efforts focused on increasing specialty palliative care consultation seem sensible, this approach is limited due to a national palliative care workforce shortage. Novel approaches that promote primary along with specialty palliative care are needed. Further, relying on busy clinicians to reliably recognize patients likely to benefit from either primary or specialty palliative care is impractical and is an important source of inequities in palliative care delivery. Thus, this pragmatic clinical trial is being conducted at 6 Penn Medicine acute care hospitals to test a prognosis-triggered, clinician-directed behavioral intervention embedded within the electronic health record to nudge clinicians to either provide palliative care themselves (“primary”) or consult specialists (“specialty”) for seriously ill hospitalized patients. The study will evaluate the intervention’s effect compared to usual care on the primary outcome of hospital-free days through 6 months, and other patient-centered, clinical, and economic outcomes. The study also includes an embedded mixed methods study to understand clinician and hospital contextual factors that influence the intervention’s uptake. This study will provide high-quality evidence regarding the effectiveness of a low-cost, scalable approach to promote primary and specialist palliative care among a large and diverse patient cohort, advance the science of triggers for palliative care, provide new insights into patient groups most likely to benefit from systematic identification for palliative care, and create new knowledge about establishing hospital environments conducive to desired clinician behavior change to improve serious illness care.
Toward Improved Design and Analysis of Stepped Wedge Trials: An Estimand-Aligned and Efficiency-Focused Framework
Overview: This project will contribute new methods and software for planning and analyzing stepped-wedge cluster randomized trials. These new methods will enable investigators to (a) target transparent causal estimands under the counterfactual outcomes framework and (b) to leverage baseline information for achieving higher statistical efficiency.