On 10th December 2013 the RSS Merseyside local group hosted Erik Von Elm from the Institute for Social and Preventative Medicine, University Hospital Lausanne, Switzerland and Cochrane Switzerland. There were 31 people in attendance, including RSS members (39%) and guest representation from employers such as the University of Liverpool, University of Central Lancashire, Liverpool School of Tropical Medicine and the NHS.
Discontinuation of trials often represents a waste of scarce research resources and raises ethical concerns. The single most important reason for trial discontinuation is problems with recruitment of participants. In the first part of Erik’s talk, he presented recent empirical evidence on trial discontinuation from the DISCO study, a cohort of about 1,000 trials in Switzerland, Germany, and Canada. This study identified trial characteristics associated with discontinuation due to poor recruitment and non-publication of trials. Additionally, Erik considered reasons for trial discontinuation and he evaluated why it is a problem when trials are discontinued.
In the second part of the talk, Erik discussed approaches to statistical modelling of trial recruitment and established criteria to identify trials at risk of recruitment problems at an early stage. The modelling approaches considered included conditional and unconditional strategies as well as Bayesian, Poisson and Markov chain models. The outlook Erik proposed included assembling a set of recruitment data for multicentre trials from several countries incorporating a broad range of trials and testing existing modelling approaches.
Finally, instead of aiming for precision in predicting accrual rate or overall accrual period, Erik proposed focusing on criteria defining 'failure', potentially using the framework of diagnostic studies.