### Department of Biostatistics Seminar/Workshop Series

# Fixed Center Effects Models for Recurrent Event Data with Large Number of Centers

## Dandan Liu, PhD

### Postdoctoral Research Fellow, Department of Biostatistics & Biomathematics, Fred Hutchinson Cancer Research Center

### Wednesday, March 2, 1:30-2:30pm, MRBIII Conference Room 1220

Fixed center effects models are often used in multi-center studies with center effects treated as fixed factors and estimated simultaneously with treatment effects. In many large registry studies, the evaluation of center performance is often of interest and the number of centers is usually very large. Therefore, the conventional method is difficult to implement due to computation issues. In this talk, I will present an alternative estimation method for fixed center effects models for recurrent event data, with center effects multiplicatively acting on the rate functions. The new estimation method consists of two steps with the regression parameters estimated in the first step and the center effects estimated in the second step. Such estimation method avoids high dimension of the information matrix with the increasing number of centers and thus, is computationally efficient. The proposed center effects estimators are the ratio of the observed cumulative number of events to the corresponding expected quantity in a center. Large sample results are developed for the proposed estimators. The finite sample properties of the proposed estimators are assessed through simulation studies. The method is then applied to national hospitalization data for end stage renal disease patients with more than 5,000 centers being evaluated.