The Danger Within Us

by Jeanne Lenzer

  It was during this slowed-down period, in May of 2008, when he pulled a letter from his mailbox and was surprised by the return address. The letter was from an unexpected source—the law firm of Winckler & Harvey, LLP, in Austin, Texas. He had contacted them more than a year earlier, along with “nearly a hundred” other law firms. None wanted to take his case. But now here was a response. Finally someone was following up. He found himself almost unwittingly giving in to hope. He knew better. He’d been disappointed so many times before. But still…he opened the letter. In amazement, he read that representatives from Cyberonics, who had previously disavowed any responsibility for Fegan’s near-death experience, advised the law firm that they were “reconsidering their position on settlement.”180

  Fegan’s hopes soared. A settlement! That was good. And while it wasn’t as good as a trial, it was better than nothing.

  A week or two later, Fegan received a call from one of the lawyers. He had no idea what to expect.

  Fegan doesn’t remember the exact amount of money the lawyer told him Cyberonics was offering. He does recall that it was less than $20,000. An amount that didn’t even cover the hospital costs he and his insurance company paid.

  Then, three weeks later, he pulled another letter from his mailbox. He hadn’t given the lawyers a definite answer about the offer. Maybe that was a good thing. Maybe the pathetic settlement offer by Cyberonics had persuaded the attorneys to take his case to court. He recalls, “I tore the letter open. I wanted my day in court. I was so anxious to move forward. Now maybe the FDA would have to listen.”

  But his excitement turned to disbelief as he read the letter, signed by attorney Jay Winckler: “I must inform you that you have no basis to file a lawsuit, and I therefore cannot take legal action on your behalf due to the recent U.S. Supreme Court ruling, In re: Riegel v. Medtronic, Inc., 552 U.S. 312 (2008), which bars claims against a manufacturer regarding the safety or effectiveness of a medical device.”181, 182

  Fegan had to read the sentence several times. The new ruling meant that no one could sue a manufacturer over a device if the device had been approved by the FDA. Riegel v. Medtronic meant it was all over. That was it. There would be no settlement. No suit. No nothing. Fegan’s last avenue of redress had been shut down.

  * * *

  The Supreme Court ruling, based on a suit brought by Charles R. Riegel and his wife, Donna, against the medical device company Medtronic,182 had been issued just months before the lawyers finally sent their letters to Fegan. Riegel had been undergoing angioplasty (a procedure to open a blocked coronary artery) with the Evergreen catheter, a device manufactured by Medtronic,181 consisting of a catheter with a balloon on its leading edge. The catheter was threaded through an artery into the left side of Riegel’s heart and from there into a blocked coronary artery. When the doctor inflated the balloon to open the artery, the balloon ruptured. Riegel went into complete heart block and lost consciousness. Placed on life support and rushed into emergency surgery to repair the damage, he was left with severe and permanent disabilities and died several years later, in 2004.

  The Riegels contended that the catheter was defective and had caused grave harm. Their lawyers argued that this was a classic example of the kind of case for which a personal-injury lawsuit is the ideal—and often the only—remedy. They asserted in oral argument that “FDA regulation alone may not adequately protect consumer safety. Without the threat of lawsuits, manufacturers may hide information about device harms from the FDA during and after the [premarket approval (PMA)] process. Because the FDA does not conduct its own studies into device safety, the PMA process typically relies on those provided in the manufacturer’s application, which may exclude those with unfavorable results.”

  For most of its history, the FDA itself had supported cases like the one brought by the Riegels. FDA administrators had long held that the right to sue was necessary to help compensate patients harmed by drugs and devices and potentially prevent future deaths and injuries and protect the public.183 But by the time the Riegel case reached the Supreme Court, something had changed. Dan Troy, the general counsel for the FDA, had submitted an unprecedented friend-of-the-court brief in the Riegel case supporting the medical device industry’s position, which advocated the so-called doctrine of preemption.183 This doctrine would prohibit suits against makers of devices that have previously been approved by the FDA on the grounds that the agency’s approval process makes any additional health or safety requirements unnecessary and, indeed, onerous and unfair. Since lawsuits in state courts (where personal-injury suits are filed) effectively threaten to establish requirements that are different from or additional to those of the FDA, Troy argued that such suits should be preempted.

  That’s exactly the logic that the Supreme Court relied upon in its Riegel ruling. Justice Antonin Scalia, writing for the eight-to-one majority, said that states can’t impose requirements that are different from or additional to those established by the FDA. He concluded that the twelve hundred hours spent by the FDA reviewing premarket applications for high-risk devices offer “reasonable assurance” that they are safe and effective.

  Called the ten-thousand-pound gorilla of device rulings, Riegel v. Medtronic deprived many individuals such as Riegel (and Dennis Fegan) of redress.184, 185 What’s more, by preempting lawsuits that could bring safety problems to light, it deprived the general public of important information that could protect other healthcare consumers.

  The ruling on preemption relies entirely on the FDA’s device-approval process to ensure that devices are safe and effective. So let’s examine the agency approval process—in particular what the FDA calls its “stringent” processes used for high-risk and implanted devices.

  * * *

  Medical devices didn’t come under regulatory control by the FDA until 1976, when Congress passed the Medical Device Amendments (MDA) to the Food, Drug, and Cosmetic Act. The act followed on the heels of a scandal involving the Dalkon Shield, an intrauterine device designed to prevent pregnancy that was first marketed in 1971 as a safer alternative to birth control pills.186 Just three years after the device went on the market, two and a half million women were implanted with the device. A design flaw allowed bacteria to enter the uterus, and thousands of women had to be hospitalized with serious infections. There were also a number of fatalities.187, 188

  Because devices like tongue depressors and crutches don’t need the same oversight as cardiac pacemakers, the FDA stratifies devices according to level of risk and need for clinical studies: low-risk (class I) devices include items such as scalpels and bandages; medium-risk (class II) devices include endoscopes and most total joint implants (such as most artificial hips and knees); high-risk (class III) devices include cardiac pacemakers, implanted defibrillators, deep-brain stimulators—and the VNS device.

  Most class I and some class II devices are exempt from clearance or approval and can simply be registered with the FDA. For the remaining devices in classes I and II, the manufacturer simply has to notify the FDA of the product’s class and its intent to market the device ninety days before distribution. The process, known as the 510(k) process, does not require clinical trials, and such devices are said to be cleared for market rather than approved.

  Class III, or high-risk, devices are generally expected to undergo a premarket approval (PMA) process in which the manufacturer must provide “reasonable assurance” of safety and efficacy. However, one of the concessions won by the medical device industry was an allowance that all devices, including high-risk devices, on the market before 1976 could continue to be sold under the 510(k) provision. In addition, any new device that a manufacturer deems to be “substantially equivalent” to an existing device (known as a predicate device) could be eligible for clearance under the 510(k) provision. Even devices that are substantially equivalent to other devices cleared under the 510(k) process can be similarly cleared by the FDA, allowing potentially infinite iterations, a problem known as “predicate creep.�
��123 In this way, devices can change over time, the same way that a game of telephone can distort an original statement.

  Addressing the 2009 annual meeting of the Medical Device Manufacturers Association, the director of the FDA’s Center for Devices and Radiological Health, Daniel Schultz, acknowledged the problem with the 510(k) process, saying, “[There are situations] where we started with one device a long time ago and ended up some place very, very different. And it is really hard to explain that entire complicated path that got us from where we were in 1976 to 2009.”189

  The 1976 Medical Device Amendments were intended to ensure that high-risk devices would undergo scientifically valid clinical testing. Yet three decades later, only 16 percent (170 of 1,062) of the highest-risk devices approved by the FDA had gone through the PMA process. The rest had been either cleared without clinical testing through the 510(k) process (21 percent) or approved by supplemental filing (63 percent). Of the latter, only 0.3 percent were required to provide clinical data.13, 14, 190 Supplemental applications are submitted by manufacturers to the FDA for changes to a device (such as new or upgraded models) that could affect effectiveness or safety—or when the manufacturer proposes a change in the label to include new diseases or conditions for which the device is intended.

  Even when devices did undergo clinical testing, a 2009 study of seventy-eight applications for high-risk cardiovascular devices found that only 27 percent of studies (33 of 123) were randomized, only 14 percent were blinded, 88 percent of the primary end points were surrogate markers, roughly half of the studies had no control group, and nearly a third of those that did were retrospective—a form of review that is generally not as scientifically sound as a randomized controlled trial.15

  A study published in JAMA, the Journal of the American Medical Association, found that from 2008 through 2012, the FDA cleared for market approximately four hundred implanted medical devices considered of moderate to high risk—without requiring clinical testing.191 This means that manufacturers have been able to sell everything from filters inserted into the inferior vena cava to certain cardiac stents, total hip implants, and surgical mesh without any clinical trials to ensure safety and efficacy.

  Clearance of high-risk implanted devices through the 510(k) exemption, which has occurred in 21 percent of cases, is particularly dangerous because it is based on two potentially false assumptions—first that the predicate device is safe and effective and second that the manufacturer’s claim of “substantial equivalence” ensures that the new device is at least equally so. In truth, devices cleared through the 510(k) pathway were simply grandfathered in without any clinical trials, so there is no guarantee that either the predicate device or its follow-on device is safe or effective. In 1996, the US Supreme Court concluded that “since the 510(k) process is focused on equivalence, not safety, substantial equivalence determinations provide little protection to the public.”192

  Since 2009, the FDA has reclassified almost all previously unclassified devices—a move intended to ensure that high-risk devices will be labeled as such and subjected to the agency’s premarket approval process, requiring “scientifically valid” evidence. However, only a tiny fraction of such devices are subjected to clinical studies, and the FDA’s definition of “scientifically valid” evidence includes “partially controlled” trials, the use of historical controls, case reports, observational data, and surrogate end points—the sort of data most experts understand to be often misleading.123 Vinay Prasad, an oncologist and a nationally recognized expert on evidence-based medicine, said of the FDA’s approval process, “[It] is so bad [that] some physicians I know privately call the device market the Wild West, where a company can still ‘strike gold’ and where it really doesn’t matter if the product works to make patients feel better or live longer.”193

  Yet the Supreme Court ruling on preemption presupposes that the approval process is bulletproof and provides sufficient protection to the public. If that were the case, the FDA would not have to recall about eleven hundred devices annually. A number of recalls are “class 1” recalls, which, according to the FDA, means there is “a reasonable probability” that the recalled device could “cause serious adverse health consequences or death.”

  Individuals implanted with recalled devices face the difficult choice of undergoing surgical removal, with all its risks, or waiting it out with the worry that something might go wrong.

  As devices become increasingly complex and invasive, the number of class 1 recalls has been rising year after year: in 2003, there were eight class 1 recalls; that number rose to 176 in 2013.194 The number of the FDA’s supposedly “stringent” processes of premarket approval, intended to ensure the safety of devices before they go on the market, has declined over the years, while the use of other pathways to approval or clearance has increased—pathways that require little if any clinical testing. This has occurred at the same time that implanted devices have become more complex by several orders of magnitude, with computer-driven, Wi-Fi-enabled devices programmed to electrically stimulate the heart and brain and devices that combine biologic products and drugs with high-tech gadgetry.

  In 2001, the agency approved seventy-one devices via the PMA pathway; in 2005, the number dropped to forty-three. By 2009, only twenty devices were approved via the PMA pathway. During the same time frame, PMA supplement approvals rose, increasing from 641 in 2001 to 1,394 in 2009. So how did all these high-tech devices get on the market? In addition to the many 510(k) clearances, manufacturers can apply for a supplement approval when they alter an already approved device in a manner they say does not alter safety or effectiveness. There is also a so-called real-time supplement application pathway, which allows manufacturers to make a phone call to the agency to review the changes for approval, after which the agency will fax approval to the manufacturer.195 In most instances these alternative pathways do not require clinical trials.

  Of course the device approval process upon which the Supreme Court rested its preemption ruling is itself a thin reed, because only a tiny fraction of the highest-risk implanted cardiac devices that won approval under the PMA process were subjected to two randomized blinded trials.15 And because only 16 percent of high-risk devices go through the original PMA process, the overwhelming majority of high-risk devices have never undergone a single controlled clinical trial much less the putative standard for drug approval: two randomized controlled clinical trials.15, 190

  The VNS device is a poster child for the problems that arise when the public is forced to rely on the FDA’s premarket approval process as an assurance of safety and efficacy. Even though the VNS device underwent two randomized (partially) controlled clinical trials prior to approval, the FDA’s lax oversight both during and after the approval process is evident. The company was let off the hook regarding the possibility that the device causes deaths, which couldn’t be definitively determined because Cyberonics failed to include a comparison group of patients treated with medicines only. Nor did the FDA demand such a comparison group.

  In 1997, during premarket deliberations, when Steven Piantadosi, one of the FDA advisers, expressed concern about the seventeen deaths that occurred among the first one thousand patients—a number that was particularly alarming in part because patients had been implanted for only a few years at most—a Cyberonics representative assured the panel that the deaths were the result of sudden unexplained death in epilepsy (SUDEP). In other words, according to the company, epilepsy, the underlying condition, not the VNS device, caused the deaths. But Piantadosi continued to express concern, saying, “I’m still a little worried about the death rates that we are seeing…Should we be concerned by that?”133

  FDA reviewer Ann Costello responded to Piantadosi by citing a study that showed a somewhat higher mortality rate from SUDEP among patients about to have brain surgery to treat particularly severe seizure disorders. W. Allen Hauser, a member of Cyberonics’ scientific advisory board, was also quick to acquit the device, saying, “I don
’t think that the sudden death is an issue specific to the device. It’s a specific issue in terms of people with bad epilepsy.”

  There were reasons to doubt Cyberonics’ claim of safety. The company used data from “worst-case-scenario” patients—those who were about to undergo brain surgery as a “last-hope option” to treat intractable seizures. But those patients often have well-defined risk factors that increase their chances of dying early, such as multiple grand mal seizures; and serious developmental problems. The VNS device was not being tested for use on people who experienced grand mal seizures: instead it was tested on individuals with partial seizures, which pose a lesser threat of early death.196

  But something else made a comparison of Cyberonics’ test subjects to “worst-case-scenario” patients unfair: all Cyberonics’ potential test subjects were screened and excluded from clinical trials if they had serious conditions known to increase mortality, such as diabetes and heart conditions—or a serious seizure disorder causing two or more episodes of status epilepticus in a year. In other words, Cyberonics study patients as a group (who comprised a portion of the first one thousand implanted patients) were not as likely to be in the same high-risk categories as patients about to undergo brain surgery.

  The only true way to know whether the VNS device was saving lives or killing people would have been to enroll a third group of test subjects in a medically managed group.