麻豆约拍

Rebecca Hubbard, PhD

Carl Kawaja and Wendy Holcombe Professor of Public Health, Professor of Biostatistics and Data Science
School of Public Health |
Brown University

WHEN: Wednesday, October 2, 2024, from 3:30 to 4:30 p.m.
WHERE:聽Hybrid | 2001 麻豆约拍 College Avenue, Room 1201;
NOTE: Rebecca Hubbard will be presenting in-person

Abstract

Data are captured in electronic health records (EHRs) as a direct result of patient interactions with the healthcare system. Consequently, EHR data for patients with more intensive healthcare utilization are captured more frequently and provide more detail about the patient鈥檚 health. This connection between patterns of healthcare utilization and data quantity and quality, termed informed presence bias, violates the common statistical assumption of independence between observation and outcome processes. The complex processes giving rise to EHR data, which hinge on patient interactions with the healthcare system, medical provider practice patterns, and health system coding conventions, must be accounted for in analysis to avoid biased inference. In this setting, grounding statistical analyses in the context of the relevant medical practice patterns and scientific questions of interest is key to remediating bias. In this presentation, I will discuss the roots of informed presence bias in EHR data and illustrate the importance of contextual thinking in addressing this challenge using several real-world examples. I will quantify the magnitude of bias resulting from alternative patterns of dependence between outcome and exposure data capture and healthcare utilization intensity and demonstrate methodologic solutions to this problem. My overarching objective is to highlight the role of contextual thinking in EHR-based research and the importance of bridging science and statistics in analyses of modern data sources.

Speaker bio

Dr. Hubbard is Professor of Biostatistics and Data Science at the Brown University School of Public Health. Her research focuses on development and application of statistical methodology for studies using data from electronic health records (EHR) and medical claims, including issues of data availability and quality, and has been applied to studies in oncology and pharmacoepidemiology. She is a Fellow of the American Statistical Association, Co-Editor of the journal Biostatistics and a Statistical Editor for the New England Journal of Medicine.