American Statistical Association
Physicians have large discretion in the way they use medical services, leading to well-documented variation between providers in both utilization and outcomes. While part of this variation is natural, some of it is unwarranted, a potential indication of both undertreatment and overtreatment. Health services data are characterized by large numbers of clusters (providers) and small cluster sizes (patients per provider). Random effect models, which are the standard way to study variation, are known to be unstable in this setting. We performed simulation studies to investigate the ability of random effects models to detect between-provider variation under these conditions. An alternative approach to test between-provider variation will also be discussed.
|Date:||Wednesday, December 10, 2014|
|Time:||4:00 - 5:00 P.M.|
Memorial Sloan-Kettering Cancer Center
Department of Epidemiology and Biostatistics
307 East 63rd Street
(between First and Second Avenues)
3rd Floor Conference Room
New York, New York
Note: To gain access to the building, please follow the directions by the telephone in the foyer.