Biostatistics Seminar Series: Charles E. McCulloch, PhD

"Diagnostic methods for uncovering outcome-dependent visit processes" Charles E. McCulloch, PhD Professor and Head, Division of Biostatistics Vice Chair, Department of Epidemiology and Biostatistics University of California, San Francisco   Abstract:  With the advent of electronic health records, information collected in the course of regular healthcare is increasingly being used for clinical research.  The hope is that the wealth of clinical data and the realistic setting (compared to information derived from highly controlled experiments like randomized trials) will aid in the investigation of determinants of disease and understanding of which treatments are effective for which patients.  The availability of information in such databases is often driven by how a patient feels and may therefore be associated with the health outcomes being considered. We call this an outcome-dependent visit process and it has the potential to lead to bias if data are analyzed using standard statistical methods.  We give conditions under which bias may arise, derive a local score test to motivate diagnostic methods, and evaluate their performance.