Statement of Diana Zuckerman, PhD, President of the National Research Center for Women & Families at a House of Representatives Briefing on Medical Devices

Statement of Diana Zuckerman, PhD

President, National Research Center for Women & Families

May 17, 2011

I am glad to have the opportunity to speak at this briefing on Baby Boomers and medical devices.  I am president of the National Research Center for Women & Families, which is a think tank that uses scientific and medical research to develop strategies to improve the health of adults and children.

My perspective is as a researcher trained in epidemiology and public health at Yale Medical School, who was on the faculty of Vassar and Yale and conducted longitudinal research at Harvard, and is currently a fellow at the University of Pennsylvania Center for Bioethics.  I am also a former Congressional staffer for the House subcommittee that has oversight jurisdiction over the FDA, and for the Senate Veterans Affairs Committee.

And, I also have a hip joint and will speak very briefly about my perspective as a patient of a device that was not subject to clinical trials before being sold in the U.S.

What are the implications of dangerous medical devices for Health Care and Medicare?

Today you have heard about the potential risks and long-term costs of one type of medical device, surgical mesh.  You’ve heard from a patient and a surgeon about how widely used mesh is and how terrible the complications can be.

I will put that in context.  Most of us assume that the FDA requires all medical products to be tested on human beings in clinical trials before they can be sold.  As you have heard today, that isn’t true for surgical mesh.  The FDA tells us it also isn’t true for more than 90% of other medical devices, including many implanted devices.

To determine how many of these devices are harming patients, our Center recently conducted a study that was published in the prestigious peer-reviewed journal, the Archives of Internal Medicine.

We studied all the recalls between 2005 and 2009 that the FDA designated as the highest risk because they could cause “serious health problems or death.”  Our study found that more than 3 out of 4 of those high-risk recalls were not approved by the FDA’s pre-market approval (PMA) process, but rather were cleared through the much less stringent 510(k) process or were exempt from any review because they were considered so low-risk.

At a recent hearing of the Senate Aging Committee, the GAO explained that the FDA has been ignoring the law when they clear high-risk devices through the 510(k) process.

I will explain how that harms patients.

There are 3 essential safeguards missing from the 510(k) process that are in PMAs and are also in the FDA approval process for prescription drugs.  The major safeguards that are missing are as follows:

1.           There are no clinical trials-no scientific testing on human patients prior to allowing the device to be sold.

2.           Pre-market inspections are not required to make sure the device is made correctly, appropriately sterilized, and so on.

3.           Post-market clinical trials or epidemiological studies are not required as a condition of approval.  So, in other words, the devices are not studied in clinical trials before they can be sold, and the FDA doesn’t require scientific post-market studies as a condition of approval.

Defenders of the status quo have said that what’s important about the medical device approval process in general and the 510(k) process in particular is that less than 1% of device applications are later subject to a high-risk recall.

That positive view of the process might make sense from a business point of view, but not from the public health or policy points of view.

Americans are dying and being harmed because devices that they relied on were not tested in patients or inspected prior to sale.

What Are the Public Health Implications of High-Risk Recalls?

The detrimental impact of these defective medical devices on patients and the public health is substantial and could be reduced with relatively small changes in the 510(k) process and how it is implemented.

It is substantial because from 2005 through 2009, the 113 highest-risk device recalls involved 112.6 million recalled products.  In the first six months of 2010, more than 437 million devices that had been cleared through the 510(k) process were subject to high-risk recalls-1.4 devices for every man, woman, and child in the U.S.

There were only 113 high-risk recalls from 2005-2009, but the GAO tells us that there were 2,905 moderate-risk recalls during those same years. Some of those “moderate-risk” recalls resulted in a need for inpatient surgery and lengthy hospital stays and rehabilitation.  For example, the 158 recalls of hip, knee, and ankle implants from 2005 through 2009 were not included in my study, because the FDA did not consider them high risk-recalls.  A recalled hip, for example, often requires revision surgery much sooner than expected, and that costs an average of $35,000 and results in a 3-5 day hospital stay, at least 6 weeks walking with crutches or a walker or cane, four weeks where the patient is not allowed to drive, and several weeks or months of rehab or physical therapy.  Despite the high cost and debilitating impact on patients, which could be even more devastating to elderly patients who lack someone at home to care for them, the FDA did not consider any of the hip and knee recalls from 2005 through 2009 to be high-risk.

The bottom line is that even “moderate-risk” recalled devices can sometimes result in death during surgery, and certainly add billions to Medicare costs when they result in additional surgery and hospitalizations from the complications of defective devices.

And, more important, the standards for recalling a device are very high.  Despite well-documented problems with surgical mesh, it is still on the market and is costing Medicare and health care millions – perhaps billions – of dollars.


What Are the Policy Implications of High-Risk Recalls?

As a scientist and as a logical person, I believe that if a device can kill you, it is not low-risk or moderate-risk. I’m not talking about lightning striking out of the blue in a device that seemed absolutely safe.  I’m talking about predictable failures, such as an implant that deteriorates inside the human body or a diagnostic test that is inaccurate. Those predictable life-threatening problems could be reduced if implanted devices, life-saving devices, and life-sustaining devices were subject to the PMA process instead of the 510(k) process.

In our published study, we questioned why the FDA would determine that a device can cause death or serious harm when it fails, then not consider that same device high-risk when categorizing it for review.  Although it might be possible for a moderate-risk device to result in a high-risk recall for a completely unpredictable reason, by definition, the vast majority of moderate-risk devices should not have the potential to cause such harm.  An  implant that breaks or deteriorates inside the human body, or a diagnostic test that is inaccurate are predictable problems that should be considered when the FDA determines whether or not those devices are high-risk.  Therefore, the fact that in our study 71% of high-risk recalls were cleared through the 510(k) process and an additional 7% were never even subject to FDA review indicates inconsistency in the FDA’s policies regarding the classifications of risk for review compared to recall criteria.

It doesn’t make sense that standards for even the most innocuous drug-for constipation or stomach aches for example-are more rigorous than for life-saving medical devices.

Reported Deaths and Serious Adverse Reactions

We don’t know how many people die every year from unsafe medical devices.  Experts agree that adverse reaction reports, despite their many shortcomings, represent a fraction of the actual harm.  The FDA reported 116,086 device-related injuries and 2,830 deaths in just one year (2006).  Our analysis of subsequent years indicates much higher statistics.  There were 4,556 reported deaths in 2009.  These deaths and the hundreds of thousands of serious complications reported on the FDA web site every year are just the tip of the iceberg, because experts tell us that most doctors don’t report them.

How many of these reported deaths and injuries were caused by devices that were not subject to clinical trials or pre-market inspections, because they were cleared through the 510(k) process?  That information is not available.

The Cost of Not Approving Medical Devices Quickly

Industry warns that clinical trials and other requirements of the PMA process take more time and cost more than the 510(k) process, stifling life-saving innovations.  However, there are no available data on whether lives have been lost because of delays in getting devices to market in the U.S.  In fact, by definition, any dramatically innovative medical device should be submitted through the PMA process, not the 510(k) process.  The 510(k) process is intended for more incremental changes.  Without clinical trials, it is impossible to determine whether those changes save lives.

Moreover, since Medicare requires that medical devices be proven beneficial in clinical trials, devices cleared through the 510(k) process will not necessarily be reimbursed by Medicare or insurance companies until clinical trials are completed.  That means that the delay caused by PMA criteria such as clinical trials often does not mean a corresponding delay in the widespread use of a device, compared to the 510(k) process.  Thus, Medicare’s stricter standards can help keep 510(k) devices off the market that were never studied in clinical trials, but Medicare’s decision to cover a device would not provide the PMA safeguards of pre-market inspections or post-market studies of the product’s use in the general population.

Pharmaceutical companies do not question the need for clinical trials or pre-market inspections, and the higher standards the FDA requires of those companies have not interfered with pharmaceutical companies’ high profitability.  There is no public health reason for heart valves and other high-risk devices to receive less scrutiny than prescription drugs for relatively minor health problems such as constipation; in fact, the reverse would be more logical.  If that is burdensome to smaller device companies, there is no proof that this burden outweighs public health considerations.


How Safe is Safe Enough?

Is the process for medical devices safe enough as is?  Devices are common.  Those of us who wear contact lenses or hearing aids, or have a replacement hip or knee, or had LASIK or an injectible wrinkle filler, or use test strips for diabetes rely on medical devices every day.

Lives could be saved and patients would spend less time in the hospital if the FDA implemented the law as required.  Billions of Medicare dollars could be saved.

The 510(k) process may be acceptable for devices that are truly low- or moderate-risk, but not for implanted devices or those that diagnose or treat potentially deadly diseases.

Our Center’s Policy Recommendations

The 510(k) process relies on bioengineering testing and other tests, rather than clinical trials.  Although clinical data such as subjective reports are sometimes included, well-designed clinical trials are almost never part of the 510(k) criteria.

If clinical trials were conducted, they would often catch errors of design or manufacturing before the product was sold, such as the example of glucose test strips, described below. Even if a defective product were cleared for market without either clinical trials or pre-market inspections, requiring post-market studies as a condition of approval would allow problems to be caught more quickly than through reporting of adverse events outside a study environment.

I recommend that, as the law requires, implants and other high-risk devices always be subjected to the most rigorous review, the PMA process.  In addition, the definition of a device that is high-risk enough to be subject to the PMA process should include all devices that can cause death when they fail.  How can a low- or moderate-risk device create a predictable life-threatening situation?  Glucose test strips used for diabetics are an excellent example: a seemingly simple product cleared through the 510(k) process.  When Abbott glucose test strips were found to be very inaccurate in 2010, the FDA pointed out that when test strips falsely conclude that glucose is considerably higher or lower than it really is, the results can be fatal.  That is why the FDA issued a high-risk recall of those strips in 2010, resulting in a recall of 359 million products.


American patients are dying and undergoing additional surgeries and hospitalizations that would have been avoided if their medical devices had been adequately studied or inspected before the FDA allowed their sale.  Device problems would be caught much earlier if post-market surveillance was supported by adequate technological reporting systems, pre-market clinical trials, or post-market epidemiological research or clinical trials.

Medical devices are more ubiquitous than most people realize.  Millions of Americans rely on the safety and effectiveness of medical devices every day.

The FDA approval process will never be perfect and so there will always be unsafe medical products, but small changes in the 510(k) process can make medical devices much safer.  Lives could be saved and patients would spend less time in the hospital if the FDA implemented the law as required by using the PMA process for all devices that are potentially dangerous in predictable ways.  Billions of Medicare dollars could be saved as well.  The 510(k) process, or a variation of it, may be acceptable for medical devices that are truly low- or moderate-risk, but not for implanted medical devices or devices used to diagnose or treat potentially deadly diseases.

This number is based on the FDA’s Medical and Radiation Emitting Device Recalls database, which separately counts each model of a device that was recalled.  That calculation is consistent with the examples in this paragraph, although it results in a higher number than GAO, which combined separate recalls for models of the same device.