For a decision that affirmed not just a verdict, but an award of punitive damages, against a device manufacturer, Nicholson v. Biomet, Inc., 46 F.4th 757 (8th Cir. 2022), is not as bad as it could have been. That’s because most of the decision was about evidentiary rulings that the court found to be “harmless” error, and one might even have some benefit for our clients in the long run.
Still, Nicholson was pretty darn depressing.
The most blatant problem with Nicholson has to do with the treatment of certain data from the FDA’s medical device adverse reporting system (called “MAUDE”). Generally, we’re skeptical about the evidentiary value of data from what is, in large part, a voluntary reporting system. We discussed those issues at length here, and the FDA also acknowledges them:
The submission of an MDR itself is not evidence that the device caused or contributed to the adverse outcome or event. . . . [T]his passive surveillance system has limitations. The incidence, prevalence, or cause of an event cannot be determined from this reporting system alone due to under-reporting of events, inaccuracies in reports, lack of verification that the device caused the reported event, and lack of information about frequency of device use.
FDA, “Overview of Medical Device Reporting.” Usually, MAUDE data is being introduced by plaintiffs seeking to do precisely the no-no that the FDA identifies – use them as evidence of “incidence, prevalence, or cause” – but in Nicholson the defendant offered MAUDE data for the opposite purpose: to contradict the plaintiff’s expert.
Because, at the time of the plaintiff’s surgery, there weren’t many reports at all. It was apparently uncontradicted that “on the week of [the] surgery in 2007, the MAUDE database . . . showed only one complaint of the [device’s] loosening out of approximately 25,000 devices sold.” Nicholson, 2022 WL 3642917, at *3. Thus, the defendant sought to admit MAUDE data to discredit the other side’s claim, based entirely on post-surgical data, that this type of device “had higher rates of revision surgery” than another type of device. Id. The trial court excluded it, exactly why we’re not sure from the opinion, but the cursory description of the defendant’s arguments (concerning “critiques concerning the data’s meaning and value,” id.), we’re pretty sure it was for the same reasons that defendants often argue.
In a brief discussion, the Eighth Circuit did not even try to defend the district court’s ruling – instead it affirmed on harmless error:
The MAUDE data suggesting the [device’s] success in 2007 does not refute any of [plaintiff’s] evidence. Success of the [device] does not mean the design did not have foreseeable risks and that those risks could not have been prevented with an alternative design. We thus hold that any alleged error in excluding the MAUDE data was harmless.
Nicholson, 46 F.4th at 764. We question that rationale, because Iowa is a Restatement Third state, and under that restatement “[w]hen a defendant demonstrates that its product design was the safest in use at the time of sale, it may be difficult for the plaintiff to prove that an alternative design could have been practically adopted. Restatement (Third) of Torts, Products Liability §2, comment d (1998). But our critique of this aspect of Nicholson is quite different.
This evidence – that at the time of the surgery there was only one known adverse event claiming the a similar type of device failure out of 25,000 implantations – has to do with punitive damages.
Iowa’s punitive damages statute, Iowa C. §668A, does not require bifurcated punitive damages trials, nor does anything else indicate that punitive damages were tried separately in Nicholson. Thus the exclusion of this MAUDE data goes directly to the intent element of punitive damages. Like most states, Iowa requires “willful and wanton” conduct as a prerequisite to punitive damages. 46 F.4th at 769. “Willful and wanton” is defined as:
an act of an unreasonable character in disregard of a known or obvious risk that was so great as to make it highly probable that harm will follow, and which thus is usually accompanied by a conscious indifference to the consequences.
Id. (quoting Mercer v. Pittway Corp., 616 N.W.2d 602, 617 (Iowa 2000)) (emphasis added). Further, by statute Iowa requires plaintiffs to establish punitive damages by “clear, convincing, and satisfactory” evidence. Id. (quoting Iowa Code § 668A.1(1)(a)).
So, in Nicholson there seems to be a complete disconnect between the MAUDE discussion and the punitive damages analysis. Regardless of how reliable the MAUDE data is as a matter of assessing the relative safety of alternative designs, we don’t see how the then-available MAUDE data of one incident of the relevant risk in 25,000 implantations can possibly not be relevant to whether that same risk was “known” or “obvious” to the defendant or that it was “highly probable.”
Even if incidents were underreported by a factor of ten, hell, by a factor of one hundred, the minimal MAUDE data establishes that there was nothing for the defendant to “disregard” at the time its conduct was to be judged. Nicholson even recites that in Iowa “punitive damages [a]re inappropriate where the risk of injury from defect was not so great as to make it highly probable that an injury would occur.” 46 F.4th at 770 (citation omitted). Because the MAUDE evidence goes directly to the peculiar requirements of punitive damages, and the jury imposed such damages, we can’t see how its outright exclusion could possibly be “harmless” – particularly under a clear and convincing evidence standard.
But Nicholson ruled the other way. In the context of that case, we think the result was an injustice, but we generally oppose the admission of MAUDE data (and similar reporting data concerning drugs) in any evidentiary context. So in the converse case, where a plaintiff seeks to admit evidence of a large number of reports, the result in Nicholson is something we would cite in support of excluding that evidence. If MAUDE data is inadmissible when a defendant offers it for lack of similar incidents then it should be equally inadmissible when a plaintiff offers it as purportedly showing many similar incidents. So if Nicholson is a dark cloud, it comes with a distinct silver lining.
That’s not the only thing we question about Nicholson. Also on punitive damages, the court rejected the defendant’s position that punitive damages were unavailable because the defendant had warned about the risk at issue. Id. Its response – that this argument would “allow[ defendant to] evade liability for known defects of the [device] before marketing the product simply by issuing safety warnings would defy the purpose of design defect claims,” id. – was simply a non sequitur. The defendant raised this argument in connection with punitive damages, not liability for design defects. Once again, intent is the issue. The question is whether providing a warning, even if inadequate, precludes a finding that the defendant acted with the requisite bad intent.
We took a look at that issue here. Many cases recognize that, even though a defendant “failed to adequately warn of these dangers,” where, the “evidence establishes” that “warnings were given,” “there was no evidence to support punitive damages” since that means there is “no indication of malice, wantonness, or reckless indifference to the consequences from which malice could be inferred.” DeLuryea v. Winthrop Laboratories, 697 F.2d 222, 231 (8th Cir. 1983) (applying Arkansas law). None of the cases (more than 20 of them) distinguished between warning or design defects being the underlying basis for liability in finding that even negligent attempts at giving warnings precluded a finding of the state of mind necessary to permit punitive damages.
Perhaps the worst aspect of Nicholson is that it is published. Iowa didn’t have any law on either of these points, so now we have a federal court filling that void with slapdash reasoning and illogical arguments that will haunt litigants in that state for who knows how many years to come.