A Bayesian analysis of bivariate survival data from a multicentre cancer clinical trial

I present an analysis of data from a clinical trial of new chemotherapies for colorectal cancer. I model participating clinical sites via random effects, to allow for heterogeneous patient responses across centres. Patient response is measured by both progression and survival times. The data suggest that patient responses are homogeneous before, but heterogeneous after, disease progression. For one new therapy, an increase in efficacy relative to standard therapy is detected before, but not after, progression.