ABOUT


Purpose of this blog

Dr Ben Goldacre has argued in favour of sharing the complete programming code alongside research results:
I propose that The BMJ should change its policy and require all analytic codes to be published, as an appendix, alongside all quantitative papers. This would do more than simply enhance transparency and reproducibility. It would also create a growing archive of codes, with immediate clinical relevance, from which researchers could learn new techniques and shortcuts, and from which whole routines could be re-used, avoiding unnecessary duplication of effort. It would allow each publication to contribute more to advancing science and accelerate the discovery of important signals in patient data.
I'd be reluctant to send reams of code to medical journals for various reasons, and data sharing is the real concern for GPP. However, I sympathise with Dr Goldacre's position and, to the extent that it makes sense, I include my programming code for published studies here along with some explanatory notes. There are a few options for researchers hoping to publish code, such as JMASM, CPMB, JSS, F1000 and the R journal. However, these journals obviously limit what can be made available. Therefore, a blog seems to be a decent way to go about it, supplemented by GitLab. In this way, code can be shared and is under continual development, rather than static and merely archived. The blog is focused primarily on SAS code, however it may also include e.g. MATLAB, WinBUGS etc.

About me

I'm internationally trained in biostatistics, pharmacology, toxicology, drug development and regulatory affairs, clinical trial design and medicine. I've worked in academic research units (Australia, England and Canada) and the pharmaceutical industry (Denmark and Germany). I was an independent consultant to the pharma industry in Europe and have experience across various diseases (primarily diabetes, oncology, heart failure etc.) and all phases of research. My primary research interests include the design and interpretation of clinical trials, adaptive designs and Bayesian methods, and survival analysis and meta-analysis. I've been using SAS for over 20 years. Early in my career I was the sole statistician/programmer at a small company. I can see the value in supporting an online community, sharing code and SAS tips, and connecting solitary programmers who may be isolated within a company or reside in remote areas.

Paul M. Brown, Alberta Children's Hospital Research Institute
MSc Biometry, MSc Clinical Research, PhD Medicine
ORCID iD iconorcid.org/0000-0002-2411-3312

Acknowledgements 

Much of the recent work on composite endpoints was carried out at the Canadian VIGOUR Centre (thecvc.ca), University of Alberta.