Tuesday, April 5th, 2016, 4:10pm, 1010 Valley Hall
Speaker: Robert Obenchain (Risk Benefit Statistics LLC)
http://localcontrolstatistics.org
Title: “Fair” Treatment Comparisons for Continuous Exposures
Abstract: My published material on “Fair” Treatment Comparisons (FTCs) in OCER (Observational Comparative Effectiveness Research) covers only the case where two alternative treatments for a single disease or medical condition are to be compared either on effectiveness or on cost. In this talk, I first review published materials on theoretical FTC estimands and corresponding unsupervised, nonparametric estimators, then I discuss natural generalizations of basic concepts from binary treatment choices to analyses of continuous exposure levels. Both types of analysis are illustrated using a single dataset amassed from published government statistics for 2,881 US counties that encompass measures of indoor Radon level, lung cancer mortality and three X-confounder measures (age, smoking habits and obesity.) Surprisingly, perhaps, these data show that lung cancer mortality tends to be lower in counties where prevalent indoor Radon levels are higher ...with mortality heterogeneity directly predictable from (i.e. moderated by) the three X-confounding factors.
Obenchain “Radon Study” Collaborators:
Stanley Young, PhD, FASA; CEO, CGStat, Raleigh, NC, USA
Goran Krstic, PhD, RPBio; Human Health Risk Assessment Specialist, Fraser Health Authority, Vancouver, BC, Canada
Presentation materials for this talk consists of 37 PowerPoint slides, many of which contain explanatory "Notes." Two slides listing general references or related internet downloads (plus 6 backup slides) complete a PDF file totaling 45 pages.