Rachel Richesson, associate professor, recently published a paper entitled "Statistical Code for Clinical Research Papers" in Annals of Internal Medicine. Co-authors include Gregory Simon of Kaiser Permanente Washington Health Research Institute, Kevin Weinfurt of Duke Clinical Research Institute, Adrian Hernandez of Duke Clinical Research Institute and Lesley Curtis of Duke Clinical Research Institute. 

As members of the National Institutes of Health Healthcare Systems Research Collaboratory Coordinating Center, we are ambivalent about Assel and Vickers' report (1) calling for sharing statistical code. We enthusiastically agree that all researchers should share detailed code used for statistical analyses and would expand this concept to include sharing other methods and technical tools, such as those used for data collection and code used to define research data elements from electronic health records. Transparent sharing of all methods and tools should improve the rigor and reproducibility of research and accelerate the development of improved methods. However, we are concerned that the authors criticized contributors of shared code for repetition and deficiencies in formatting and documentation. Criticizing those who share statistical code or other tools will probably have a chilling effect on the sharing behavior that we hope to encourage. Although we hope that all analytic or data management code is elegant and well documented, most (including much of the code from our own work) is simply not.