Reform of The De Novo Classification Process Needs To Be A Top Priority

By Jeffrey K. Shapiro –

The de novo classification procedure was added to the FD&C Act in 1997 to create a middle pathway between the 510(k) process and full blown premarket application ("PMA") approval.  It is intended for devices that utilize novel technologies that are not risky enough to justify regulation under the burdensome PMA process, but which lack a predicate device that would allow 510(k) clearance.  Under the procedure, a 510(k) submission must first be denied with a “not substantially equivalent” letter.  Then, the submitter may request de novo classification.

With FDA cracking down on the creative use of predicates, particularly split predicates, the threat to device innovation is that novel technologies will inappropriately be subject to the PMA approval process.  This process is so onerous and expensive that some of these novel technologies will simply be abandoned.

One solution would be a timely and efficient de novo classification.  Unfortunately, the device center’s 510(k) Working Group looked at de novo and found that it is not efficiently implemented at present.  Their report showed that average review times for therapeutic devices sky rocketed from 254 days in 2005 to 904 days in 2008, to 752 days in 2009.  The review times for diagnostic devices were somewhat better, increasing from 261 days in 2005 to 308 days in 2008 and 448 days in 2009.

We call your attention to a study published in June 2010 by Boston Medtech Advisors.  The study report can be found here.  It confirms the general tenor of the 510(k) Working Group’s data, but provides greater detail and more sophisticated analysis.  It is also more comprehensive:  Boston Medtech Advisors reviewed FDA’s entire de novo database from 1998 to 2009.   The report is worth reading in its entirety.

We learn from the report that there have been a mere 54 successful de novo classifications in 11 years, 38 diagnostic devices and 16 therapeutic devices.  This figure alone suggests that FDA has used de novo infrequently.  To some extent, that is appropriate if the 510(k) process is sufficiently elastic in the use of predicate devices.  But with FDA tightening up the use of predicate devices, there needs to be a greater use of de novo classification.

FDA committed at the outset to complete review of de novo applications in 60 days.  Until the end of 2006, all but two reviews were completed within 100 days, with an average review time of 62 days (median 51).  Since 2007, however, only 4 of 13 applications were reviewed in less than 100 days, and the average review time has been 241 days (median 217).  When the 510(k) review time is added in, the overall de novo review times are almost twice as long as target review times for panel-track PMAs!  Obviously, this delay lessens the attractiveness of de novo review for a device that has been denied 510(k) clearance because it lacks a predicate device.

The larger point is that FDA has a serious timing problem with its reforms.  If FDA is going to restrict the use of predicate devices, there needs to be a safety valve short of full PMA approval that allows novel technologies to reach patients.  Otherwise, device innovation is simply going to be stifled.  Not every innovation is worth the time and expense of the PMA approval process.  At present, however, the de novo process takes too long and is not used often enough, especially for therapeutic devices. 

Our own experience is that FDA is already tightening up the use of predicate devices, even prior to implementation of the 510(k) Working Group's recommendations.  If FDA wishes to further restrict the use of predicate devices, it should reform the de novo process first, so that there is a viable alternative when 510(k) clearance is denied for lack of a predicate device.  If FDA does not do so, many useful new technologies will simply be abandoned, with patients being the losers.

REMINDER: Register for HPM's free webinar "The Evolution of the Park Doctrine" on October 8th. (link to registration: http://hpmwebinar.eventbrite.com/)