The American Statistical Association (ASA) Statistics and Data Science in Aging (SDSA) Interest Group is pleased to announce our next webinar!
SDSA thanks the ASA and SDSA sponsors and donors:
The University of Maryland Claude D. Pepper Older Americans Independence Center (UM-OAIC)
The University of Maryland School of Medicine Center for Research on Aging
Theresa Kim and Nathan Parrish
Charles B. Hall
Edward C. Hirschland
George Rodriguez
Ginger Holt
Hak‐sing E. Ip
Mary J. Kwasny
Robert A. Oster
Yong‐Fang Kuo
Click here to join SDSA!
Want to become an SDSA sponsor? Visit GiveASA! Under “Additional Details” type “To offset costs of SDSA activities.”
Webinar details:
Speaker: Dr. Roee Gutman, PhD, Professor, Department of Biostatistics, Brown University
Title: Novel Bayesian Record Linkage Methods with Application to Medicare Beneficiaries
When: January 13 2025, 1-2pm ET/12-1pm CT/11am-12pm MT/10am-11am PT
Where: Zoom; Register here -
https://amstat.zoom.us/webinar/register/WN_iI2z0_ksSAeCbZFLnnozAQ
Abstract: Analysis of partially linked datasets is increasingly important as researchers and policy analysts seek to integrate administrative datasets and registries while adapting to privacy regulations that limit access to unique identifiers. Record-linkage tools have been developed to identify records that represent the same entity across multiple datasets in the absence of unique identifiers. Past research mainly focused on the computational efficiency of record-linkage tools. Less attention has been given to features of the data that can improve the linkage and to statistical inferences with linked records.
To address these limitations, we view record linkage as a missing data problem and develop Bayesian procedures that utilize data features that are frequently encountered in public health applications. These procedures improve the linkage, and result in more accurate and precise estimates of scientifically important associations. The first procedure incorporates associations between variables exclusive to one of the datasets in the linkage process. The second procedure ensures that individuals receiving care from the same provider in one dataset are linked to individuals receiving care from a similar provider in the other dataset. This procedure can be implemented even when providers cannot be uniquely linked across datasets. Both procedures generate M datasets in which the links between the two datasets are imputed. The datasets can be analyzed independently and combined using standard multiple imputation rules. This approach minimizes the analytical burden on researchers while offering flexibility for downstream analyses.
We demonstrate the utility of these procedures through two applications. The first links Medicare claims and Vital Statistics Mortality records to examine the relationship between end-of-life medical expenses and causes of death. The second combines the National Trauma Databank and Medicare claims data to investigate how injury characteristics influence successful discharge to the community among patients with traumatic brain injuries.
Speaker Bio: Dr. Roee Gutman is a Professor in the Department of Biostatistics at Brown University. His areas of expertise are record linkage, causal inference, missing data, Bayesian data analysis and their application to big data sources. Dr. Gutman has authored multiple papers in which he developed novel methods to analyze linked data sources, and for estimating causal effects from observational studies. These methods were applied to address clinical, epidemiological, and health services and policy questions, especially among the elderly.
Dr. Gutman has co-authored over 100 publications, including papers in leading statistical and subject-matter journals. His contributions have been recognized with the ISPOR Health Economics and Outcomes Research Methodology Award for his work on estimating the causal effects of Meals on Wheels programs on healthcare utilization using linked datasets. Additionally, he served as an ASA/NSF/BLS Senior Research Fellow, collaborating with researchers at the Bureau of Labor Statistics to develop novel record linkage techniques.
------------------------------
Michelle Shardell
Program Chair, Statistics and Data Science in Aging
Co-Director, Biostatistics and Informatics Core, University of Maryland Claude D. Pepper Older Americans Independence Center
Professor
Institute for Genome Sciences, University of Maryland School of Medicine
------------------------------