Letters Barriers to access of clinical data

GlaxoSmithKline on its remote desktop interface for independent reanalysis of drug trial data

BMJ 2014; 349 doi: https://doi.org/10.1136/bmj.g4898 (Published 30 July 2014) Cite this as: BMJ 2014;349:g4898
  1. James Shannon, chief medical officer1
  1. 1GlaxoSmithKline, Brentford TW8 9GS, UK
  1. james.s.shannon{at}gsk.com

We launched our system in May 2013 to provide researchers with access to patient level data from our studies in the belief that greater access to these data would facilitate further research to advance medical science and lead to improvements in patient care. I am sorry that Jureidini and his team have had difficulty working within the system.1 We worked hard to develop an approach that would provide access while ensuring that the patient confidentiality of those who take part in our trials isn’t compromised.

The system has been live for more than 12 months and, although originally developed to provide access to GlaxoSmithKline’s patient level data, now has support from seven other industry study sponsors. Proposals are independently reviewed for scientific merit by an independent panel. In addition, access to anonymised data and a variety of common analytical tools is provided in a secure environment in which researchers conduct their analyses—the data cannot be downloaded. This is different to our approach to clinical study reports—these documents, without individual patient data, can be freely downloaded.

We are pleased by the amount of interest from the research community in conducting research with our patient level data. More than 50 research proposals have been received, most of which have been approved, and many research teams now have access to data in the system. When researchers are granted access, because we recognise the size and complexity of the data, we proactively offer support to help them effectively navigate the datasets. We always knew that as researchers started to use the system, questions about its capabilities and suitability to support their work would arise. Feedback, good or bad, from real world users will help with future improvements to the system and guide how we can best support research teams with their analyses.


Cite this as: BMJ 2014;349:g4898



View Abstract

Log in

Log in through your institution


* For online subscription