What should FDA regulations mean for me?

June 2014

Focus on statistical analysis plans (SAPs)

Submitting a new drug or medical device application to the FDA for approval can be a major challenge. The FDA review process is long and rigorous, as meeting the FDA’s high standards is not simple. To make the process all the more complex, every type of submission has its own regulations and for many indications the FDA has very specific criteria for how to carry out safety and efficacy analyses in clinical studies. In order to be properly prepared for an FDA submission, extremely careful planning of statistical analyses is essential. Every year dozens of clinical trials fail to meet the requirements of the FDA on the basis of insufficient or sloppy statistical analysis planning. Here is a list of the top 10 mistakes made in planning.

Top 10 mistakes in planning statistical analyses:

  1.  Planning of clinical trials is not coordinated with FDA reviewers
  2. Statistical planning begins after the study is already underway and changes to the protocol are difficult if not impossible; changes made can end up significantly complicating the FDA approval process
  3. Protocol statistical section is written by a scientist and not by an experienced statistician, thus leaving out portions that are critical to the FDA
  4. Study is designed without including specific parameters or analyses required by the FDA
  5. Development of a case report form (CRF) is not coordinated with the analyses planned resulting in collecting the wrong data, potentially leaving out parameters required by the FDA
  6. Analyses detailed in the statistical analysis plan (SAP) do not answer the questions the FDA reviewers are asking
  7. SAP is written in an ambiguous manner, causing mistakes in analyses executed
  8. SAP is not consistent with the study protocol’s statistical section, raising questions at the FDA
  9. SAP is not followed, again raising questions by FDA reviewers
  10. Proper blinding in accordance with FDA standards is not maintained throughout the design and execution of analyses, calling into question the accuracy of the analyses performed

Effective statistical planning from the very first steps of designing the study can ensure passing the regulatory expectations of the FDA and even save money in the process.