A team of researchers led by statisticians at the U.S. Food and Drug Administration (FDA) have published guidelines that show how to design presentations of clinical trial safety data so they tell clear, compelling stories with minimal explanatory text. The guidelines appear in the June 25, 2015 issue of Statistics in Medicine.
The authors write that researchers can present even large amounts of complex data in a way that makes it easy for readers to understand, without extensive captions and heavily annotated layouts. This is especially important when communicating clinical trial safety data to readers who make decisions based on that data, such as editorial staffs at medical journals and regulatory officials reviewing applications in support of the product.
The authors are members of the Safety Graphics Working Group, a team of collaborators from FDA, industry, and academe who created a publicly available database of graphics designs for reporting safety data. The new database includes a decision tree that instructs scientists how to choose the most effective graphic designs for particular types of data and the computer code for creating them. For example, the designs enable clear presentation of a wide variety of potential adverse effects (AEs) data from medical treatments that enables readers to clearly see important information, such as which AEs are elevated in treatment vs control, which AEs could be a safety signal, and which AEs are elevated in specific patient subgroups.
Describing presentations with sometimes fanciful names, such as spaghetti and lasagna graphs and violin and forest plots, the authors also demonstrate how to produce effective statistical graphs that use effective self-explanatory symbols or annotations. Such data can include any of a wide variety of measurements, from liver enzymes to prolonged Q-T intervals (abnormal electrocardiogram results associated with heart beat abnormalities, fainting, and sudden death).
The authors, who are in the Office of Biostatistics and Epidemiology, also offer guidance in choosing colors to differentiate key data, such as treatment results among different groups of individuals in the same study.
Seeing is believing: good graphic design principles for medical research
Statistics in Medicine
Susan P. Duke (GlaxoSmithKline), Fabrice Bancken (Novartis), Brenda Crowe (Eli Lilly and Company), Mat Soukup (FDA-CDER), Taxiarchis Botsis (FDA-CBER) and Richard Forshee (FDA-CBER)