[Skip to Content]
Access to paid content on this site is currently suspended due to excessive activity being detected from your IP address Please contact the publisher to request reinstatement.
[Skip to Content Landing]
Views 776
Citations 0
January 2, 2013

Statistical Analysis Plans in Observational Research—Reply

Author Affiliations

Letters Section Editor: Jody W. Zylke, MD, Senior Editor.

Author Affiliations: Department of Biostatistics and Bioinformatics, Duke University School of Medicine, Durham, North Carolina (Dr Thomas); and Duke Clinical Research Institute, Durham, North Carolina (Dr Peterson; eric.peterson@dm.duke.edu).

JAMA. 2013;309(1):32. doi:10.1001/jama.2012.56917

In Reply: Dr Trinh and Ms Sun contend that a prospective SAP process, outlined in our Viewpoint, is not feasible and does not go far enough to regulate the current observational research environment. Although there are elements of Trinh and Sun's comments we agree with, we reject their main premise. Our Viewpoint recognized the importance of flexibility in observational research, even in the context of a rigorous analytic process. While we agree that the wide availability of data sources and statistical analytic packages can enable fishing expeditions, we propose that using our SAP process (if done honestly) would limit such activity. Researchers could potentially be deceitful regarding what was prospectively defined, but thankfully willful deceit remains rare in science.