FDA Finalizes Guidance on Use of Statistics for Medical Device Approvals

The Food and Drug Administration is the critical regulator for most medical products companies–drugs, devices and equipment, even software.  Understanding the FDA’s position on issues can be difficult and interacting with the agency is often frustrating to medical product firms.  As the agency reviews its 510(k) process, and responds to other criticisms, many have been concerned that the outcome will lengthen approval and other processes and make them more expensive.  A longer time through a regulatory process means slower speed to market, a likely lower valuation for a company and more difficulty in obtaining funding during research, development and market launch.

A recently finalized guidance on use of Bayesian statistics should actually allow firms to get necessary studies completed more quickly and accelerate their regulatory processes.   (FDA Guidance) The practical aspect of Bayesian statistics is that they allow prior information, especially prior clinical trial data, to be combined with that from a current study.  This should allow for smaller and/or shorter trials, speeding up the time, and lessening the cost, to develop data for filing with the FDA.    Even when there is not prior information, application of Bayesian approaches can be useful in adaptive trial designs, which may change certain parameters as they proceed.

Bayesian statistics have begun more popular as the computational power necessary for their application has become more widely available.  While the guidance seems positive, how it actually is applied by the FDA staff will be critical in seeing if it actually helps firms get to market faster.  In any event, the guidance provides useful information for trial designers and statisticians.