BB&T Washington Editor
Donna-Bea Tillman, PhD, is the director of the Office of Device Evaluation (ODE) at FDA's Center for Devices and Radiological Health, a post that gives her a great insight into device approvals, which have been the subject of seemingly endless controversy of late.
Tillman started at FDA in 1994 as a reviewer of the branch dealing with devices for obstetrical and gynecological use, then served as the branch chief of the Pacing and Electrophysiology Devices branch. Tillman's career track took her to the deputy directorship of the Division of Cardiovascular Devices.
Her resume also includes a stint at the Consumer Product Safety Commission.
Tillman earned an undergraduate degree in engineering from Tulane University, and a PhD in biomedical engineering from Johns Hopkins University.
BB&T: According to the annual report by ODE for fiscal 2006 and 2007, the average number of days needed to review PMAs and panel-track PMAs has remained more or less flat since 2001, but that doesn't address the number of reviewer hours required. Is there a substantial change in the number of hours required to review devices in that time?
Tillman: It's hard to generalize too much, but our sense is that PMAs are getting more complex. We don't have any data on the number of people are on a [typical] review team. In terms of the the numbers [of PMAs] we get in a particular year and what they are, it's really quite variable. That's one of the reasons we renegotiated the user fee program, to have it based on registration listings because that's a somewhat more fixed rate. The number of PMAs is more variable from year to year. But it's our sense that devices are getting more complicated and the issues are getting harder to resolve.
BB&T: What does that complexity do to your personnel requirements? Does ODE end up needing more PhDs to get the job done?
Tillman: Most of my review staff does not have PhDs. Generally when we hire reviewers, we're hiring engineers and scientists directly out of undergraduate school. My sense is that the majority of my staff have a bachelor's degree — maybe 50% to 60%. Another 20% have a master's and maybe 10% have PhDs. So PhDs are more rare.
BB&T: Do you typically have a difficult time hiring PhDs or can you get them when you need them?
Tillman: We can get them when we need them. When we go out and do hiring – for example, we're doing quite a bit of hiring right now, we got some additional positions at the center from new user-fee money — we do one of two things: we go out and look for engineers fresh out of school we can train to be reviewers. A lot of times, we're looking for biomedical engineers or electrical or mechanical engineers. It depends on what the review branch is looking for. But other times, we have a specific expertise that we need, for instance, an expertise in materials science or fluid dynamics or one of those topics. When we want that kind of particular kind of specialized expertise, we go out and look for people with advanced degrees such as masters and PhDs.
Generally, we're pretty much able to hire what we need on the science side. Where we run into trouble with is the clinical side. We struggle to find clinical specialists. For example, we've been trying to find a plastic surgeon for I can't tell you how long. It's very hard to bring people in (partly because of salary differentials).
We've struggled for many years to find a neurosurgeon. Through our MDFP program, we have a relationship with Johns Hopkins (Baltimore), where we bring in neurosurgeons who are in their post-doctoral fellowships and they work for us half-time for a year and we rotate them on a yearly basis. I think we're on the fourth year of that program and that's been a tremendously effective way for us to get a specialty that we ordinarily would not have a chance to hire.
So the biggest problem we have is finding people with a specific clinical expertise willing to come in and work with the government.
BB&T: Are there any areas of specialty for which a nurse practitioner's skills would suffice?
Tillman: Yes, most certainly. We definitely hire nurses and there are some areas, frankly, where nurses are better than doctors because they spend more time with the patients. With some of the devices, in the respiratory therapy side of things for instance, these are the people who are managing the patients on a day-to-day basis. As a matter of fact, we have been trying to hire more nurses than in the past.
BB&T: A year or two back, there were some rumblings that a sizable number of people at FDA are either eligible for retirement or early retirement over the next five years. Do you see that having an impact on your office?
Tillman: People have been worrying about that for 10 years, an ongoing concern about the brain drain. I don't see that as being a problem in my office. My sense — and I don't have any specific numbers — my staff is more on the younger side than on the older side. We obviously have some people who have been here for a while who are obviously enormously valuable, and when they do decide to retire, we're going to miss them greatly. But my sense is that my staff is not heavily weighted toward those who are going to retire in the next couple of years.
BB&T: Most device makers would argue that it's getting tougher to get through the FDA process. As for PMAs, it does seem that the amount of clinical data required is getting greater and the post-market review requirements are also rising. Also, there have been several flaps in the media suggesting that some reviewers are not happy with how things are done at CDRH. Is there something of a culture shift among device reviewers?
Tillman: The standard for PMAs for approval hasn't changed. It's still a reasonable assurance of safety and effectiveness. I think that it's variable. When we come up with new technologies and new indications we haven't seen before — where we don't have a lot of historical experience to rely on — I think there is an expectation in those situations that a sponsor needs to provide a little more data. Some of what people may be perceiving is due to the fact that there's a lot of innovation out there. The kinds of data we need for a first-of-a-kind product may be more than we need for the tenth coronary stent.
I know there is a perception that we get easier, then harder, and maybe there's a political tie to that depending on the administration. I would say that we try to learn from our past, so where we make decisions on devices that go out on the market and we observe how they perform and maybe there are issues or questions that get raised, we try to tie what we learned back into the next iteration. Sometimes we may find there were questions we weren't asking before that maybe we should have been asking and now we should be asking for a little more data based on what we've learned.
We try to always balance this level playing field that industry is always interested in against our primary mission, which is public health protection and promotion. So we're trying to make sure we're consistent, but sometimes the science changes and so the bar has to change, too.
BB&T: The Sprint Fidelis lead story seems to revolve around the notion that a slimmer lead triggers less occlusion around the lead, but the Fidelis, which was a PMA, was subjected to bench testing but no in-human testing prior to approval. Can you talk a little about the way bench testing of medical devices works?
Tillman: [Declining to answer the Fidelis lead question directly] I think that the way the 510(k) program is designed to work, the first thing we ask companies to do for a new device is bench testing and pre-clinical testing and if that testing can address all the issues, then that's what they provide. If there are any remaining issues, they have to animal studies and if there are any issues from that, they have to do a clinical study. Bench and pre-clinical testing is sort of the core of the device evaluation process.
BB&T: With regard to leads for implantable devices such as implantable cardioverter defibrillators Is there a feeling that bench testing ought to be accompanied by at least animal testing before being approved in any case?
Tillman: Pacing leads — a lot of our leads — are PMA products. Some are 510(k) products. The question you're getting at — and we've had a couple of discussions with the press on this issue — is that of when do you need clinical data and when is bench testing enough. Our position is there are some questions that are better answered on the bench than in a clinical study.
When you have a clinical study, first of all, you're limited in terms of how big it can be. You're not going to be able to do a 10,000-patient clinical study out to 10 years, it's just not feasible. So it's hard to look at device failure modes from clinical studies, so you do bench testing, where you can take a device and stress it and see where it fails; where you can take a heart valve and take it out to four hundred million cycles and see how it performs over its expected lifetime.
A lot of times, bench testing is a better way to get at device performance than a clinical study. I think our approach has always been that we look at the particular device, we look at the issues and the questions and the risks, and based on a consideration of that, we work with the company to decide on an appropriate set of tests to be done.
BB&T: There are times when stories of controversial product clearance or approval episodes burst into view in a way that seems to suggest that a reviewer may seem to have come to a mindset about an application that isn't terribly objective. Has FDA ever reassigned an application because of something like that?
Tillman: Certainly we expect our staff to do reviews based on the appropriate scientific and regulatory considerations and not on their personal views of the device, and we do everything we can to make sure that's how a review proceeds. If the management felt a reviewer was unable to be objective and make a decision based on science and regulatory policy, then it would be within that manager's right and responsibility to assign it to another reviewer. I'm not personally aware of an instance in which that occurred but cannot say it hasn't occurred.
BB&T: The GAO report didn't really seem to have anything particularly damning to say about the 510(k) program other than to note the need to assign the remaining 20 or so pre-amendment devices. What's involved in that sort of effort?
Tillman: There were a fairly large number of pre-amendment devices (which originally numbered in excess of 100) and of those there are only these 20-something left. Like many organizations, we have competing priorities, and the ones we did first were the ones with the biggest public health impact and the ones there were the most concerns about. What's left are some things we certainly need to deal with but once again I think it's a competing priority issue.
What we are doing — and it shouldn't be too much longer — the first step is to put out section 515 requests for information about these remaining pre-amendment class III devices so we can make a decision about what we're going to do with them. Once we do that, there are a couple of different options. We can call for PMAs for ones that we think that special controls are not adequate to mitigate the risks, we can down-classify them, and there may be some of these pre-amendment class III devices that aren't being made anymore. In those situations — we call them "fallen angels" — we would call for PMAs but there won't be any because nobody is making them anymore.
What we're trying to do right now is to look at the remaining and to get additional information through the 515 process so we can make decisions about what we want to do. Reclassification can be a fairly lengthy process. Part of it depends on whether we need [an advisory] panel meeting. If they're down-classified [into class II], there will be a special control guidance document that we will prepare and there will be a draft guidance and a proposed rule that will go out for public comment. Then we'll have to issue a final guidance document and a final rule. Realistically, it's not going to happen within the next year. We are committed to making decisions on all those devices.
BB&T: Can you give an example of some of the devices that are on the list?
Tillman: Single-chamber, non-rate responsive pacemakers, pulse generators. When companies started coming in with dual-chamber, rate-responsive units, we found them not substantially equivalent and made them into PMAs. Pacemakers are on there, pacemaker leads are still on there, even though the products ... are not made anymore. Another device that's on there that receives a lot of attention are AEDs, automated external defibrillators, are pre-amendment class III devices. They were actually found substantially equivalent to arrhythmia detectors and alarms. That's how we cleared the first AED. We down-classified arrhythmia detectors and alarms into class II with special controls a couple of years ago, but we retained the AEDs in class III because we weren't certain yet as to how we would write special controls for AEDs, so we left them [in class III] pending more information.
BB&T: The 510(k) review process has generated a mountain of controversy of late. Has the process become unwieldy?
Tillman: The philosophy of how we do reviews and how we deal with differences of scientific opinion hasn't changed. We encourage a very rigorous discussion and review process that involves a lot of opinions and a lot of different perspectives. One of the things is that there is not always a right or a wrong answer. Two experts in the same area can look at the same information and come to very different conclusions. Part of what we always deal with is that when we make decisions, it's the agency making a decision, but it's based on an evaluation done by an individual or group of individuals. They all have their own scientific perspectives and ways of looking at the world. I think people need to realize that science is not black and white. Sometimes it's very difficult trying to figure out how to balance out the various scientific perspectives.
BB&T: You obviously have a number of reviewers from various scientific specialties who look at some of these PMAs, but it's not clear from the outside precisely how they go about reviewing a product. Do they all look at the application, then get together and vote? If not, what is the process like?
Tillman: A PMA comes in and goes to a particular review branch based on what it is, so lets take an example. Let's say it's a hip-resurfacing system. It comes in and goes to the orthopedic joint branch, goes to the branch chief and that branch chief will decide who the lead reviewer is going to be. The lead reviewer is one of our lead scientists. In contrast to reviews for drugs, our clinicians are never our lead reviewers. It's always one of our scientists or engineers, someone who has been here for a while. We don't expect them to know everything about the submission because they have their particular area of expertise.
Let's say the lead reviewer is a biomedical engineer who knows something about mechanical engineering and failure modes. So they're going to serve the lead reviewer function, which is a coordinating function. They're going to assign consults, and so they send it over to statistics, and statistics picks a biostatistician to review it. They'll send it over to the Office of Science and Engineering Laboratories maybe for a consult on a particular area. There will be a clinical review, so it will be assigned to one of our medical officers here in ODE, and the lead reviewer may also do a specific sub-review.
The requests for consults go out and each reviewer writes a memo that represents their assessment of the data. Most of the time, what comes back the first time around is a memo with questions. Those questions are evaluated by the lead reviewer to ensure that they are appropriate. "Appropriateness" in this context means that the information is needed to make a decision and that they're not just nice-to-know questions.
The leader puts the letter together and it goes out to the company, and they come back with the additional information and it gets sent out to the consulting review. This goes on maybe three or four times. Some consulting reviewers make recommendations about approvability and some don't.
It's the lead reviewer's job to consolidate this information and make a recommendation based on everything, so if there's a major issue that has not been addressed, the lead reviewer has to either say that it's not an issue that needs to be addressed or they need to make their recommendation based on that.
Ultimately, it's the lead reviewer who owns the recommendation. They don't have final say because the person whose name is on the letter has the final say, which is either the division director or me. Their job is to present the team's recommendation. They don't have the ability to veto someone.
That's why you have to a fairly senior person to be lead reviewer. You have to know enough about statistics and these other areas that you can assess the input you're getting from your consulting reviewers and take all that information in and make a recommendation. Frequently in our studies we don't always get 100% of the data in a perfect world we'd like to see. We have to make decisions based on a reasonable assurance. The lead reviewer's job is to look at what a consulting reviewer is saying and figure out whether the company has provided that reasonable assurance.
BB&T: Do you have any feeling as to whether as to when the scale of the size of the device shrinks that a certain inherent uncertainty might go up? For instance, material on a nanoscale might introduce levels of uncertainty that is not seen in larger-scale items, even if only because of the unknown.
Tillman: Certainly nanotechnology is a very challenging technology area and you have a center and working groups working to figure out what are some of the additional questions and issues. There are new questions that will have to be addressed.
BB&T: Is it your feeling that anyone who wants to get something that works on a nanoscale through the PMA process that they might want to hedge their bets by getting a little more data generated before filing their PMA?
Tillman: Well, no. I think what's important is any area where you have new technology and new indications is come and talk to us first. It's not always that you need more, maybe it's just different. Just to make sure that the company understands what all the questions are that they're going to need to address.
It's not a volume question. It's not like the more complicated your device is the more the submission needs to weigh. It's just that these more complex new technologies provide more questions and more questions mean more data.
BB&T: Speaking of the size of an application, there are those in industry who complain that the "fast track" description of a 510(k) doesn't jibe with the reality at all. They say that yeah, for some 510(k)s that aren't that complicated that the paperwork does not exactly add up to a tome, but they have told me that some of these 510(k) applications are as voluminous as your average PMA.
Tillman: I think that's a really common misconception out there in the press that this notion that the 510(k) is a fast-track process. People need to realize that the 510(k) program was designed to be a risk-based way to account for the fact that there's a wide variation of risk to medical devices, that we need to have an approach that balances public health protection with public health promotion, and it doesn't make sense to ask for the same amount of data for a stethoscope as it does for a drug-eluting stent.
The 510(k) was designed to be a risk-based program that enables us to ask the companies to provide the appropriate amount of data to allow us to make the decision. There are certainly some 510(k)s for which the amount of data needed is fairly small, because they are low risk devices, they're fairly simple, they're well understood and there are a lot of standards out there the company has used. But there are also 510(k)s that are very complex.
The other common misconception that's out there is that the 510(k) is a sort of administrative review. It's not. It's a data-based review process. Companies have to submit data. Most 510(k)s go to market based on bench and pre-clinical testing, but some of that testing is fairly extensive.
BB&T: Do you have any sense that the notion of substantial equivalence has proven somewhat elastic? Has it changed in your time at ODE?
Tillman: I think that substantial equivalence has logically changed. In 1976, when you got the Medical Device Amendments and for the first couple of years [after that], substantial equivalence was a new device compared to a pre-Amendments device. Now, if you look at the chain, you have 510(k) devices that are equivalent to something that's equivalent to this and equivalent to this. As more time elapses, the substantial equivalence tree gets bigger and broader because there are simply more devices out there to serve as predicates. I don't think the process has changed — it's just all the devices out there.