Is the Malfunction MDR Reporting Requirement Unconstitutional?

By Jeffrey K. Shapiro –

In 2012, the U.S. Supreme Court held that agency regulations and enforcement policy violate due process if they do not provide fair notice of what is prohibited or required, or are so standardless that they permit seriously discriminatory enforcement.  FCC v. Fox Television, Inc., 132 S.Ct. 2307 (2012) (“Fox”).  (We previously posted about the case here.)  

There are two ways an agency may fail to provide fair notice.  One is to change its policy and apply it without adequate notice to the regulated community.  Fox, 132 S.Ct. at 2319.  Another is to impose requirements based upon subjective judgment without definitions, narrowing context, or settled legal meanings, thereby forcing the regulated community to guess at what is expected of them.  See id. at 2817; U.S. v. Williams, 128 S.Ct. 1830, 1846 (2008). 

If fair notice is not provided, even a warning letter can violate due process if the public rebuke could cause reputational harm and/or could be considered in determining punishment of a subsequent violation.  Fox, 132 S.Ct. at 2318.

As an example of an apparent FDA violation of due process under Fox, consider the 2011 warning letter to the Animas Corporation.  (We previously blogged about the warning letter here.)

As background, under FDA’s Medical Device Reporting (“MDR”) regulation (21 C.F.R. Part 803), a malfunction must be reported to FDA if is “likely” to cause serious injury or death upon recurrence.  21 C.F.R. § 803.50.  In the preamble to this regulation, FDA explains its interpretation of “likely” as meaning a “not remote” probability of serious injury or death.  60 Fed. Reg. 63,577, 63,585 (Dec. 11, 1995). In 1997 guidance, FDA states that once a malfunction has caused a serious injury or death, the agency will presume it is likely to do so again unless two years pass with no additional serious injuries or deaths.

The Animas Corporation incorporated this two year presumption in its MDR procedure.  FDA’s warning letter acknowledged the 1997 guidance, but indicated that the two year expiration of the presumption was no longer acceptable.  This abrupt departure from written guidance, announced in a warning letter, constitutes a failure to provide fair notice very similar to the agency’s conduct in Fox.  As in that case, the lack of fair notice should be considered a due process violation. 

As another example, consider FDA’s general malfunction reporting requirement.  As noted, a malfunction MDR is reportable if its recurrence is “likely” to cause serious injury or death, which FDA interprets as a “not remote” probability.  In this context, the terms “remote” or “not remote” are probability estimates that, unless applied correctly, can mean the difference between not being required to report and committing a civil and criminal violation by failing to report.

Unfortunately, nowhere in the statute, regulation or FDA guidance is “remote” or “not remote” defined or explained.  Nor does either term have a settled legal meaning.  Certainly, FDA has never pointed the regulatory community toward any such settled meaning. 

A dictionary definition of “remote” is “small in degree” or “slight,” as in slight possibility.  This definition does not indicate what the probability cutoff is, nor does it provide adequate qualitative guidance as to how this term should be applied in the context of MDR reporting.

The fact is, the regulated community has been left to guess what the probability cutoff is that makes a malfunction reportable versus not reportable.  What is “not remote”?  Is it 1 in 100?  1 in 1,000?  1 in 10 thousand?  1 in 10 million?  No one knows for sure.  This uncertainty as to what is required, as the Court held in Fox, is a violation of due process.

One last point – although not necessarily a due process concern: FDA’s two year presumption stretches the statutory term “likely” to the breaking point, resting on the premise that a malfunction is “likely” to cause serious injury or death if even a single serious injury or death has occurred, without regard for the denominator.  A device might be used 100 times annually or 100 million times annually and a single serious injury trips the presumption in both cases.  This result effectively equates “likely” with any probability of occurrence above zero, no matter how vanishingly small, and “remote” with a nullity.  As subjective as these terms are, there are limits, and a court could well find that this interpretation of “likely” exceeds FDA’s statutory authority.