LAUSR

Dear Editor-in-Chief, We would like to respond to questions raised in the editorials by Goyal et al., and von Kummer in the December 2017 issue of this journal on the design of new randomized clinical trials (RCT) of endovascular stroke therapy, and on the risk of selective inclusion in such trials [1, 2]. Goyal et al. state that after a series of overwhelmingly positive RCTs, clinicians are expected to go beyond the limits of the patient population in which mechanical thrombectomy has proven to be effective. They encourage the performance of trials seeking to extend the indications for endovascular therapy, while expressing their concern about selective randomization in such trials, a concern repeated by von Kummer who considers a slow recruitment rate in our trial a potential sign of this occurring. Goyal et al. wonder if selective treatment of eligible patients outside trials (cherry picking) will reduce the effect size to such an extent that the efficacy of treatment in these subpopulations can never been shown. Fortunately, the status of the BASICS trial offers hope for future trials [3]. Over the last 6 years, we have randomized 189 patients, 30 of whom in the last 3 months. Randomization rates have markedly increased after the publication of the MR CLEAN trial results, and more recently after the presentation of the DAWN trial results, supporting the notion that we can use the success and the momentum of successful trials to show definitive proof for relevant subpopulations. In the BASICS trial, participating centers enter data on all non-randomized patients with a basilar artery occlusion in a screening log stating the reason why a patient was not eligible, or if eligible, why not randomized. We used the screening log to monitor for cherry picking, but also to check the validity of our inclusion criteria. With a randomization rate of eligible patients of around 80% throughout the trial, we feel confident that cherry picking, although present, will not be a major factor in the outcome of our trial. Especially taking into account that the reason why eligible patients were not randomized often was logistic in nature, such as unavailability of study team or inability to obtain informed consent. Thanks to the screening log, we were able to broaden inclusion criteria allowing for the randomization of patients with a minor stroke (NIHSS < 10), age above 80, and patients with a contra-indication for IV thrombolysis, as the screening log showed the existence of equipoise in these subgroups of patients in most participating centers. We did not find evidence of an overall lack of equipoise in subgroups of trial eligible patients. Although several centers voiced reluctance to randomize patients with a contra-indication for IVT, we preferred to include them in the trial, while monitoring the screening log to make sure a sufficient number is randomized. We agree with Goyal et al. and von Kummer that the treatment of eligible patients, deemedmore likely to respond to the treatment tested, outside the trial (cherry picking) is a potential threat for any intervention trial by lowering the effect size. However, as the effect sizes in future trials are likely to be more modest than those in previous trials, as stated by Goyal et al., wewould expect that clinicians would be less tempted to succumb to their Bnon-scientific desire^ for cherry picking. A screening log can be used to monitor selection bias caused by treatment of trial eligible patients outside the trial. Together with the publication of baseline characteristics of the randomized patients, data from the screening log will allow clinicians to decide to which degree final trial results are applicable to their patients in every day clinical practice. * Wouter J. Schonewille [email protected]