A Bayesian Analysis of Institutional Effects in a Multicenter Cancer Clinical Trial

This paper examines a Bayesian method for investigating the amount of institutional variation in a multicenter clinical trial with a censored failure time endpoint. A hierarchical structure is used to model the institutional effects in a proportional hazards model, with the posterior distributions calculated using Gibbs sampling. The methods are applied to data from a lung cancer trial conducted by the Eastern Cooperative Oncology Group. In this trial there appears to be substantial variation in the treatment effect across institutions. Although the reasons for this have been identified, it would be possible to investigate this further through a detailed examination of the data from institutions with extreme effects.