Incorporating external information in analyses of clinical trials with binary outcomes