Robert Temple is the director of the Office of Medical Policy of the Food and Drug Administration's Center for Drug Evaluation and Research and the acting director of the Office of Drug Evaluation I (ODE-I). He has been with the FDA since 1972. Here, he discusses the difficulty of obtaining data from clinical trials in children -- and in conducting those trials in the first place; the relationship between the FDA and pharmaceutical companies; and the challenges in enforcing laws against promoting off-label uses of prescription drugs. This is the edited transcript of an interview conducted on July 19, 2007.
- Interview Highlights
- Pros and cons of black-box warnings on antidepressants
- Interpreting the term "off-label use"
- Why it's difficult to gather data in children
- Newer medications aren't necessarily better than the old ones
- Weighing the risks in off-label use of antipsychotics
- How the FDA monitors marketing of drugs
- The FDA: Partner or regulator of Pharma?
At this moment, how do you decide what [medication] needs to be studied? Because you're the one saying, "We need X studies done."
What is studied usually is the same thing the drugs [are approved] for in-house, but not absolutely always. If there's a pediatric disease that's different, we could ask for that. The company, as you can imagine, offers some view about what we ought to ask them to do, and then we decide whether that's the way to do it.
We pay particular attention to exactly what kind of study would do it. ... For example, in depression, we feel that no study other than a placebo-controlled study is informative, so we insist on those. Not everybody likes to do placebo-controlled trials in children. They worry about how long they're off therapy. They worry about the fact that children don't really give consent; their parents have to consent for them.
So there are always problems. However, as you know, there have been quite a large number of pediatric depression studies. And there's lots of studies in ADHD, attention deficit [hyperactivity] disorder, things like that, primarily a disease of children. ...
Where do you feel the knowledge is at this point? From what I gather, we're just chipping at an iceberg here. We know very little about children, medication and the way it impacts the brain.
Well, the long-term consequences of treatment and other things, like effects on growth and development, are very hard to ascertain, because to really get an answer, you'd have to do a very long-term placebo-controlled trial, and no one would ever let you do that, because you'd have to leave people untreated.
What you do there is you look at short-term studies which reveal that there is some growth inhibition with many of these drugs, something to plainly keep in mind. I don't think anybody has found long-term developmental abnormalities, but it's something you worry about. So you watch children. ... [T]hat's, in some ways, the best you can do. When you can't do a long-term placebo, it's very hard to get a precise answer.
Also, you don't quite know what you're looking for. But the general guidance people would give is that with children who are changing, you should check every now and then and see whether they still need the drug. A lot of times they will. ...
I hear a lot of criticism of a system that prizes innovation, wants to get a lot of drugs out to the market at the cost of safety, so the FDA has way too many people approving drugs, and very few people taking care of safety issues. ...
Well, that really depends on how you count. We don't think that's an accurate description. There is a group of people who specifically monitor what's called the Adverse Event Reporting System, AERS. But that's not the only source of information about safety.
The ordinary reviewers, who are the same ones who would review the effectiveness data, are also looking at these events; they're reading the literature. And there's many, many sources of data. The problem is that the Adverse Reaction [sic] Reporting System collects large numbers of what are called spontaneous reports.
Well, that may not be how some long-term consequence shows up. It might not be recognized as that. Or -- and this is a bigger problem -- the events that you're worried about may occur in the population ordinarily. What you'd be worried about is an increase in the rate of those events. Very hard to pick up from spontaneous reports. So we think about but don't always have the right idea of how to do them. We think about long-term follow-up studies. And we have some capacity to do epidemiologic studies. We're hoping for more capacity to do those. But there's a continuing interest in safety after drugs are marketed by a lot of people.
Give me a sense of the resources you have to look at safety. How many people are on the beat?
I can't tell you how big the Office of [Surveillance] and Epidemiology [OSE, formerly the Office of Drug Safety] is. I'm sure it's in the hundreds. But let me give you an example from the antidepressant story, OK? As everybody knows, when antidepressant therapy is initiated, there's an increased rate of what we call suicidality. That means either thinking about committing suicide or actually making some attempt that looks sincere. Now, where did that come from?
Well, the discovery that this was true came from review by the ordinary review division that looks at studies. When we looked at the trials that were done for Paxil -- paroxetine -- in depression, it turned out, if you looked at these suicidality events and looked at them properly, in at least some studies, they were increased compared to placebo.
We then, after noticing that, asked all the people who had done all the trials of depression to look within their database in a particular way. We told them how to do it, and see if you could find suicidality in those trials. Now, there was some controversy about this. But we thought each of those suicidality events ought to be looked at by experts in this. So we contracted with the Columbia University Department of Psychiatry to look at those, because there's some interpretation needed of each case.
We eventually found that if you look at all the drugs together -- there are variations from one to the next -- there was about a twofold excess of the rate of this suicidality. Now, that was done jointly by people within OSE. ...
... [T]hat's the major finding we're talking about. It didn't really come from adverse event reports, although there have always been adverse event reports of suicide in people who are depressed. It's just that they're hard to interpret. But these were controlled trials. And we pulled all the data and reached a very important conclusion. Then, as you know, we asked the people to look at their adult trials to do exactly the same analysis.
We recently asked companies to point out that the same phenomenon that was seen in adolescents and children is seen in young adults and, strangely, not seen in older adults. So the analysis is shared by a lot of people. And believe me, a lot of people participated in that analysis. ...
Well, people use the term differently. When we approve a drug for depression, for example, we don't put the age of [the person using it] in there. We say it's for depression. So it doesn't specifically say, "Don't use in children," but it doesn't say, "Do use in children."
Maybe in the other part of the labeling, there will be something that says, "Safety and effectiveness haven't been evaluated in children." I've always thought when you use a drug in a child, whether that's off-label use has always seemed to me something of a debatable point. The more clear term is, if you approve a drug for depression and someone uses it in bipolar disease or schizophrenia or something like that, that's off-label, because we haven't seen studies that caused us to be willing to put that into the label as a use. ...
So my dividing line is when you do something the label says you really shouldn't do or fail to do something that the label says you should do, that, to me, is off-label use. If the label is silent, it's a somewhat more ambiguous term.
There's no question that people have thought the widespread use in children is sort of close to an off-label use, but it seems not so clear whether that's off-label use. But what is true is, we didn't have any data on use in children. So we all believe that you needed more data. ...
Off-label use: How many psychiatric medications have been approved for pediatric use?
I can't answer that. I told you there's only one that's approved for depression. I think one of them also has a claim for obsessive compulsive disease. There have been no antipsychotic drugs approved for schizophrenia in children or for bipolar disease in children. But watch the newspapers shortly, because that's about to change.
Editor's Note: In August, 2007, the FDA approved the use of the atypical antipsychotic Risperdal for use in treating children with schizophrenia and bipolar disorder.
The drugs for attention deficit disorder, of course, are all for children. They were all developed initially in children, and only later got the adult claim. But not too many [medications have been approved for use in children]. I can't give you a complete count, but relatively few.
Well, there's a number of reasons. First of all, most people believe that you shouldn't start studying children until you have a good idea of what the drug does in adults. So you want your adequate database.
You don't want to expose children to an undertested drug. So almost always, unless it's pediatric disease, studies in them almost always begin well after the adult studies. That's number one.
Number two, until the Best Pharmaceuticals for Children Act, there was very, very little incentive for companies to do it. Something bad happens to a child, there's hell to pay, and they're nervous about it. There were many disincentives. Those were substantially overcome by the really important financial incentive of getting more exclusivity or [a longer] period of exclusivity [the extension of a patent against generic drugs]. Since then, there were lots of studies in children. Now, why the depression studies in children didn't work very well, I don't know the answer for that. I just don't know. There are three or four favorable studies of fluoxetine, Prozac, but the others, not so clearly. ...
And for ADHD, there's lots of studies in children. All of the drugs are properly studied.
It's ironic in the sense that these medications are still used in children every day. Isn't it sort of backward logic in that we're doing an experiment on kids without gathering the information?
Well, in the case of the antidepressants, people did try to gather the information. ... They didn't succeed in showing the drugs worked. There are studies of all of the most popular, recently approved drugs in children, usually two studies, because we told companies that unless they had two studies in depression, we wouldn't approve it because it's been so hard to get good data in children.
But there are 20-plus studies altogether in depression and obsessive compulsive disease, and there have been a few that showed an effect. It isn't only Prozac. But no one has had two, and I don't think people have a good idea why it seems to be so much harder to show an effect in children.
I think most psychiatrists think it probably does work in children, or at least in some children, but they know perfectly well the trials haven't really showed it. Some of them have leaned in the right direction, but they haven't really done it, so --
Maybe you're asking, why do they keep using it anyway? I think their answer is, what do you do with a depressed child, leave them alone or try to treat? I don't have to worry about that, because I'm not a child psychiatrist.
But the irony of the Best Pharmaceuticals for Children Act is that in a sense, it showed that most of the psychiatric medications used in adults don't work very well in children at all.
That wouldn't be true for attention deficit disorder, where they seem to work. And the main other place they looked at was depression. ...
First of all, trials of the drugs we know work in adults in depression fail about half the time, about half of all trials. This has been true for years and years, even while trials have gotten bigger.
About half of the clinical studies in adults now, of antidepressants -- and these are the ones we know work -- can't tell a drug from placebo. One of the reasons is, when you take some [people] and put them in a depression trial, there's a lot of improvement in the placebo group. It's a warm, nurturing environment; people are talking to them. And of course, depression is a cyclical illness, so as you get them, they're on the way to getting better. But for whatever the reason, the trials often can't show much. That's plainly part of the trouble in children. ...
In adults there's another kind of trial that you have to do in order to get a claim that you maintain effect over time. It's called a randomized withdrawal study. You take people who are on therapy and doing well, and then you randomly assign them to groups in which the drug is continued just the way it was, or discontinued, and they go onto placebo.
So it's a randomized withdrawal study. That's an ethical thing to do if the doctor thought it was time to see if the person would be OK without it. Anyway, those trials, oddly enough, always win, or almost always win.
There's a number of reasons. One, what you're waiting to see there is when the disease comes back, so there's no warm, nurturing environment, because there's nothing to nurture and be warm about. Second thing, of course, is these are people who are doing pretty well on the drug, so it's selected for a population that you think it's going to work in. What's important about this is that the real overall benefit of drugs probably doesn't come from their ability to treat the acute episode, although that's a value. It's preventing recurrence.
We have not been able to get any trials like that in children, because no one wants to take a drug away from a kid who's doing well on it. You can understand their ethical concerns. So some of the information that would be most helpful, we don't know how to get.
There are questions that have been asked about why clinical trials always compare a drug against a placebo. Why not compare it against an old drug?
It's not that that might not be interesting. And in fact, [in depression], many trials do have, in addition to a placebo, an act of drug comparison.
... And in atypical antipsychotics?
Well, there aren't any drugs that are established yet in children, so there's nothing to compare it to. This goes to a complicated problem that I have been interested in for more than 25 years. What you're describing is what's called a non-inferiority study.
You compare a drug that you know works, but you have to know exactly how well it works, and with a new drug -- and you say, "If I can show that the difference between these drugs is smaller than a certain amount, I'll conclude the new drug works, too." Now, the trouble is, if I told you that a drug you think works like an antidepressant fails to work about half the time, [and] now I run a trial, and I include this old drug that's an effective drug that works only about half the time, and I can't see any difference between the old drug and the new drug, what have I learned? Was this a study that could have told something from nothing, or is there a study that couldn't have distinguished an active drug from placebo?
That's the trouble with this design. You have to absolutely, positively know what the effect of the drug you think you understand was in this study. And without a placebo, there's no way to know. ... So in depression, it's not an informative study.
The irony of it as well is that in depression, for instance, Prozac seems to be the best drug, according to many clinical trials that compare many different antidepressants. And for antipsychotics, it seems like the best drug may even be Haldol, according to the CATIE [Clinical Antipsychotic Trials in Intervention Effectiveness] study that just came out. So are we, in embracing all these additional new medications, approving medications that are less helpful than old medications?
That's a good question. First of all, no one has ever said that the new antidepressants are more effective than the old antidepressants, than the tricyclics. No one has ever showed that they are, and there's no reason to believe they are. They are attractive to people because they don't have certain side effects that the older ones have. And that's largely true for the antipsychotics. We've never concluded, never said -- with one exception -- that the newer ones are more effective than the older ones.
A lot of studies in adults now of the newer antipsychotics had a Haldol control group, had a haloperidol control group, and they didn't beat it. It wasn't better. There, too, though, there were certain side effects of some of the older drugs that people were worried about: the twisted positions and all those things. The CATIE study suggests that at least some of the older drugs may not be so problematical when it comes to that. But the results of that study are complicated.
One drug within the study -- I'm not sure these are public, so I think I can't say -- but one drug within that study looked like it worked a lot better. Many fewer people dropped out for lack of effectiveness on one of the drugs, although they dropped out for toxicity more. That study looked for its major measure of how good the drug is by whether you could stay on the drug for a long time. But that's a complicated question, because you stay on the drug if it works and if you tolerate it. You can drop off because it didn't work or because you didn't tolerate it. ...
I need to know more what the discussion is. There's no question that people on some of them have gotten very bad diabetic reactions and things like that. What you weigh is the overall value of the drug against what its toxicity is.
Let me give you a very clear example. The most effective antipsychotic drug is called Clozapine. It's been clearly shown that it works when people fail on other therapy.
It's a drug that has a known potentially fatal risk. It causes something called agranular cytosis [a potentially fatal drop in white blood cell count] in somewhere over 1 percent of all people. We have a program in which you have to get blood drawn before you can get it, but that doesn't prevent it completely. And there still is a rate of agranular cytosis, and there have been some deaths; there's no question about it. Everyone who takes the drug is aware of this; they're told about it. And clearly, the community and we, too, believe that it's worth it for people who -- I mean, it's labeled for people who failed on other therapy. If you're schizophrenic and failed on other therapy, you're in very bad shape.
You always have to weigh these things, and you have to see whether you can improve the performance. A good example is terfenidine, the first non-sedating antihistamine. If you take it with certain drugs that inhibit its metabolism, you can get a potentially fatal arrhythmia called torsades de pointes.
When we discovered that, we told everybody as vigorously as we could, "Don't take it with these other substances," and the rate of taking it with these other substances fell. There are various estimates, but at least in a study that I participated in, it was something like 90 percent. On the other hand, there were still some deaths -- not a lot, but were still some deaths from an antihistamine. And when a drug came along that didn't have that liability, and that was just as effective, the drug went away, because there was no reason to use it anymore.
I'd say most people, when they're dealing with psychiatric drugs, really would be scared if they only had one drug in the class or only two drugs in class, because the reactions to them are perceived as being different. It's very hard to validate and document that perfectly. But each one has a different array of side effects, probably somewhat different array of effects, although that's hard to prove. And as long as it seems within the range, you keep it. You also look to see whether you can prevent the unacceptable toxicity.
But the world is full of all that. When you take aspirin, a certain number of people bleed into their heads, and you know that. You know it, but it's worth it anyway, because you prevent a large number of strokes and heart attacks of other kinds. But you've got to give it to the right people.
There are a lot of studies for ADHD medications and for antidepressants. But there is a dearth of studies for antipsychotics, atypical medications. And why isn't the FDA insisting that this change? There's so many kids out there taking these medications.
We have been encouraging them, and there are a lot of studies for use of those drugs in acute bipolar illness, and they are moving it to children.
For schizophrenia, they were studied in adults. But in, say, bipolar disease, it's only fairly recently that these drugs have been studied even in adults. So children -- that's the way it is. They're going to follow it.
And these are drugs with considerable toxicity, so you want to be pretty sure that it's likely to work before you go giving it to a lot of children. I think that's the sort of ethical standard people would observe. So it should follow decent data in adults.
But a lot of kids are getting this.
Well, because people don't know what to do, so they give it a try. Basically, what they do is, when they hear the reports in adults, they use it in children even before the data come. That's a problem. Maybe a better thing would be to start using them in children right from the beginning. But again, these are not benign drugs. Using them in schizophrenia, that's one thing. Using them in bipolar or -- it's scary, and people are reluctant to do it. But they are happening. As I said, you will see in an announcement fairly shortly.
They have unpleasant side effects. ... Some of the drugs -- there's some debate about whether they're all like this, or whether it's only some of them -- ... cause significant weight gain. And some of the same ones can cause decreased glucose tolerance. Those are bad things for a long-term drug, and you worry about those. They cause ... extraparametal disorders, where you get twisted and things like that. ... [T]hey all seem to have some potential for causing something called tardive dyskenisia, which is a late-developing distortion of position that can be permanent. Again, these are serious businesses. You don't use them lightly.
So it makes sense to me that people were cautious in moving them into children. Now, ironically, what your question raises is the irony that says actually studying something in children is formidable and is plagued by ethical problems, but you can just use the drugs in people, and nobody worries about it.
That, unfortunately, is the nature of the game. We put different standards on doing an experiment in people versus treating them as best you think you can. You're given considerable discretion if what the intent is [is] to do the best for the child. People step back and look more closely when you're doing an experiment.
That seems very troubling.
Well, people who value clinical studies have remarked on this in the past, the disparity between doing whatever you want if you're treating the patient and being very cautious. I would say our attitude is that it's usually possible to do a trial, and we urge people to do it. But we can't tell the IRB [Institutional Review Board] that it's OK if they think it's not.
For example, these randomized withdrawal studies that I mentioned, which are immensely valuable to show long-term benefit, it's very hard to get them done, very hard, because you're taking a drug away from people who seem to be benefiting from it. They don't like that.
You're also in charge of making sure that pharmaceutical companies don't advertise and promote off-label use of medications, correct? What are sales reps doing in child psychiatrists' offices if so few drugs are approved for use in kids?
We don't know. We don't have tape recording; we don't have a spy group. What we see is what they print, and that's what we know about. So when we see a print ad or something that implies use in children or use for diseases that the drug isn't for, we can jump on that. They're plainly not allowed, in the privacy of an office, to promote these off-label uses. It's not legal. But we have very little capacity to discover what they're doing. It's very hard. There isn't any record of it. We don't know. Doctors very infrequently report to us that it's going on. If they did, we could look into it. But that's not a common thing to be told.
Why not ban it?
It's already banned. They're not allowed to do it. What you're talking about is enforcement.
Yeah. But what are they doing in that doctor's office to begin with, if they're there representing AstraZeneca and Seroquel, and those are not approved medications in childhood?
Why not ban sales reps from going to pediatric offices? ... Our ability to ban [such] behavior is, if anything, less than it has been in the past. There's a lot of consideration of what we can say about limiting commercial speech. I have great doubts that we would be able to do something like ban them. It would clearly take an act of Congress to do that. There's no way we could do that without that. It's not within our authority. We don't have any rule that allows us to say where a manufacturer's representative can go.
They're already not allowed to promote uses that are not on the label. They're clearly not allowed to do that. But then the question is, can we find out whether they've done it?
... There have been, mostly by local or states' attorneys general, there have been attempts to go after companies that have systematically undermined their labeling by promoting off-label uses of gabapentin and things like that. We have very little capacity to know whether that happens.
... I know people are worried about that. But we don't notice whether they're giving us money or not. We regulate them anyway. We haven't changed the number of applications we turn down. ... [I]n the day-to-day of it, you don't really notice where the money comes from. ...
"Partner" is not the word I use. We are the regulator. We have certain common interests. We both have an interest in having a study be a good study. We're not happy if a study is stupidly designed so that it can't show the truth. We want good studies, too. We don't want to find out, and you'd think most companies don't want to find out, that it does something harmful that you should have detected beforehand.
... We're the regulator, and we try to help them do the right studies, study the right side effects, things like that. So "partner" isn't the word I use. ... It has been used, though.
I hear a lot of people saying you guys aren't bad; you're just totally underfunded, and [there are] too few people to do a huge job. Look at all the advertising out there, all the marketing, all the medications that come out, all the adverse events. ...
Well, we've said there are a couple of areas where we've made it very clear we'd like to be able to do more. We need more people to look at the adverse reactions that come in. There's something like, I don't know, 400,000 a year, and we need to be able to look at them. ...
We also plainly would like access to more databases and the capacity to do studies within those. We've said that repeatedly. We have some hopes that we will have more money to be able to do those things. We also hope to have more people -- we have a lot more than we had 20 years ago -- to review the data that comes in in support of an application. My sense is we're not desperate there, but we can do more. And there are new techniques for doing things.
It's worth remembering that in what many people think are the good old days, we had 70 medical reviewers, and not all of them were anywhere near as qualified as the people we have now. We have well in the several hundreds now. But our demand for what we expect them to review -- if you've read a current review and compared it with one from 20 years ago, it's a staggering increase. What we expect people to do in analyzing the adverse reactions that are reported is night and day compared to what we used to do.
So we've increased the demands on people, and that means you probably need more people. We also clearly want to be able to look more, both in the drug review part and in the surveillance and epidemiology part, at the reactions that come in, the new studies that are published and things like that.
And it isn't only looking at them. ... You then have to think long and hard about what they mean, what the best remedy for a problem is; whether you change the labeling, introduce certain restrictions and things like that. We're much more conscious of the range of activities that might be in order. ...
We thought it was a good enough idea to do it. That was at least nominally my decision, made along with others. Remember, the black box was never intended to say, "Don't use these drugs in children." It didn't say that. It happened to mention that they weren't approved for use in children. [What] we wanted to be sure [of], for both adults and children, is that people pay attention to a newly diagnosed ... person [recently started on medication to combat depression], because there were bad things that could happen during that early period.
This was not entirely a surprise. The label has said for many years that there's a belief that sometimes someone who's so depressed and immobilized that they can't do anything can suddenly become able to do something, like take his own life, when he starts to get better.
So there had always been that theory. But [what] we now had with the Paxil and then the subsequent studies was pretty clear evidence that that was more likely to occur in the people who were treated. And the main remedy is to watch for it. We never thought you shouldn't use the drugs because of that. And again, that's because one of the main benefits is the continued use of the drug to prevent recurrence, which hadn't been studied, but is very important.
So we recognized that some people might read the black box and be scared off and not treat people. We were conscious of the fact that by most studies, the rate of adolescent suicide appeared to be declining as these drugs had come along. We were nervous about that, but we felt we had to tell people anyway.
And as you know, there were critics who said, "You're going to hurt people this way," and they now believe that the black box may have done that. It may have discouraged use. We find it hard to believe that a child psychiatrist would be discouraged from using the drug by the black box. It just tells them how to do it. But it certainly is possible that people not as sure of themselves -- you know, general practitioners who aren't as comfortable in making the diagnoses -- may have backed off.
And it's not even out of the question that that's had consequences. But you have to tell people this, so we didn't think there was a real choice. We thought you had to tell people.
I think it was February 2004, when there were those [black-box] public hearings. Can you take me to that room and tell me what the climate of that room was?
... We presented our results. The main controversy there was whether we should present a conclusion that we'd reached already or tell the people there that we thought we had to go to the Columbia University group to look at the cases. That's probably the only controversial part of that meeting. We didn't let Dr. [Andrew] Mosholder present his conclusions. We thought -- and he was fully aware of this -- we thought that these needed a blinded reading. And he couldn't do a blinded reading anymore, because he knew which case was on [the] drug and which case was not.
The Columbia thing we arranged, which was only going to take a few months, we thought was necessary because it was a big step, because if we do this and we're not doing it right, we might discourage use of the drugs; it might do harm. We thought it was very important that we be as right as we could. That was the major area of controversy.
We didn't ask the committee to vote on whether we were there yet or not. They, in their spoken responses, seemed very comfortable with taking these cases to the Columbia University Department of Psychiatry. Again, we didn't ask them to vote on it.
But I think they mostly agreed that if it turns out to be true, you have to tell people. Of course, we told people about it even before we put it in the label and made them to med guides.
And the end result is that a very well-studied medication that is by most accounts incredibly safe has a black-box warning that is very severe. ... Most people say it's well worded, but it has caused reluctance to use this medication --
... You're right.
-- [but] that the very dangerous drugs out there, like atypicals, that are being used in kids don't have this black-box warning. That strikes me as ironic.
The atypicals have a very strong warning against something we know they shouldn't be used for, which is demented elderly people. But our trouble is when you have fundamentally an absence of data, which is largely the case for children, it's hard to write a box, unless you knew that something terrible was going on. But we don't know that, either. We tend to be able to do things when we have data. You want a black box to be pretty solidly based, and we thought this box was very solidly based. We think that the evidence is very strong that there is an increased risk of suicidality.
Again, this doesn't mean it crossed the person's mind that he might commit suicide. The person was absolutely thinking of killing himself -- that's what that means -- and made a gesture that looked serious. Now, that doesn't mean hitting yourself on the head; it means finding a rope, stuff like that.
So those are serious. People need to know them. It's true, the drugs are safe in many respects. But [the instances of increased suicidality are] not a good thing, and we thought people needed to know about it. And we were very worried that it would discourage legitimate use, very worried.
The new version of the box that we've asked people to implement reminds people that suicide occurs as a result of depression, too. We certainly don't want people not to use the drugs, but we want them to pay attention and have caregivers know that they should pay attention. They should listen for signs that the person's getting worse, because you might find something terrible if you don't pay attention. ...
People worry that the FDA is too close to the pharmaceutical industry; that, for instance, information about Paxil causing suicidal ideation -- the knowledge that it may have caused suicidal[ity to increase], it took many months for that information to reach child psychiatrists and to reach the public in general.
I don't think that's actually so. I don't remember the details of the announcements anymore or the dates of them, but we had a fairly early announcement about Paxil, even before we had the data on the rest of them. No, we didn't put it in the label yet; we didn't put a black box; we didn't have the meeting yet. But we put out an advisory that mentioned that it was there. That was all we felt we could do at that point.
Also remember, the data were not overwhelming. I think -- I'm not sure I remember this right -- there were three studies. Two of them, you hardly saw anything. Most of the excess came from one of them. You always wonder, when that happens, whether you've got a true bill.
That's been characteristic of a number of these. For Prozac, the first three studies that we had, which were the basis for approving it for depression, didn't show any excess at all. It really looked different from the others. And then the NIMH [National Institute of Mental Health] study showed it worked, but it also showed a lot. Well, it's not so easy to know what to do when you have data like that.
What we found most persuasive -- and that's what led to the box --was the analysis of all of the studies, which showed a pretty good consistency and high level of statistical probability.
How does it work, though? The pharmaceutical industry gives you the data, but it's proprietary information. So at what point can you reveal, if you're concerned, and not violate ... that agreement between the pharmaceutical industry and yourselves?
When we're worried about something, when we think people need to be warned, then we reveal the data. We're obliged to do that, and that's not a problem. It's not proprietary anymore. We can do that. You have to ask one of our lawyers to say exactly what [is] OK. But that's what we do. So as soon as we had these Paxil findings, ... we didn't make them public in detail until we thought it was ready, but we told people about the finding.
And why not make all the data public?
That's a question that I'm not the person to answer. My own bias is that it all should be public once a drug is marketed. I'd say there's a fair movement in that direction in Congress, and even in the industry. Many companies within the industry are making their detailed data available on their own Web sites. I think things are moving in that direction.
My bias is that studies in humans ought to be available to the public, at least for a drug that's available.
What keeps you up at night?
(Laughs.) Well, you know, you're always afraid you're going to make some kind of mistake. But I don't have trouble sleeping. If I thought, for example, that there would be impediments to doing the right thing, that would drive me crazy.
But despite what people may think, despite some examples that they're upset about, like [the emergency contraceptive] Plan B, we are free to do what we think is right. There's always debates about what the right thing to do is. I'm not telling [you] everybody agrees, but my experience since 1972 -- that's a long time -- is that what people here want is to get the right answer. There's lots of debates about what the right answer is and what the best way to do it is.
I didn't know this was going to be the case when I came here. What were my going-in assumptions about federal agencies? They weren't uniformly favorable, that's for sure. But what I've always found in my experience is that people want to get the right answer. They bring their own prejudices, bias and all those things to it, but that's still what they want to do.
And that's true whether you're me; it's true whether you're [drug safety researcher and former whistleblower] David Graham. We don't always agree, but I think everybody's trying to get the right answer. So I don't lose a lot of sleep. I think that's what we want to do.
When it comes to kids, John March, who is [a professor of psychiatry and chief of child and adolescent psychiatry] at Duke University, he says: "We're using these medications. We don't know how they work, if they work or at what cost, and it amounts to a huge experiment with the lives of American kids."
Well, we're talking about two different things. FDA is responsible for labeling, for data, for getting the right studies done. We know what we know, and we know what we don't know. I can't tell you whether the drugs for say, depression, other than Prozac work, OK? I don't know. They don't have evidence that they do. My gut says it seems surprising if they wouldn't. But that's what we know; that's what the labeling says.
What someone out in practice is supposed to do with the current state of knowledge is not for me to say, and it would be very hard for me to say. It should be remembered, there's a million things we don't know. How low should blood pressure go? We don't know. We know epidemiologically that from a very low blood pressure, any increase is bad for you. Does that mean everybody's blood pressure should be 90 over 60? We don't know. It's very hard to study, and it's really important. Millions of strokes depend on the answer to that, and we don't know.
There's a million things we don't know, partly because they're hard to find, partly because they're not in anybody's commercial interest to find them out. ... A lot of us think there ought to be more attention by the government, by NIH [National Institutes of Health], by somebody, to finding more of these questions and getting the answers. But there's always going to be a fair number of unanswered questions. You're always on your own on some of these things.
What we can do is look at the data we have, try to represent it as best we can. But I don't know whether you should use Paxil for children or not, and I can't know. My own personal bias is, I look at the label and try to see what's known, try to stick to it.
But you're the FDA. People look at you to figure out what is the right thing to do for my child.
But when there's no data, we can't tell them. What we're in charge of, in some ways, is the evidence, asking the right questions within our capacity under the law to ask them. But that doesn't mean in the absence of data, we're any smarter than anybody else. You know, many of us aren't practicing and I can't tell this person whether Prozac is not -- I mean, if it were me, I'd say, "Why don't you try Prozac first, because that's been shown to work."
But there may be some reasons they don't want to. You know, when you stop it, it doesn't go away for a long time, ... or they may have some other reason. The kid is depressed; what should they do in the absence of data? Doctors face that all the time, and we can't give them an answer all the time.
Other countries are far more cautious about medicating kids. Britain is sort of famous for having ruled out all these meds, antidepressants, and it caused a flurry of concern. Do you think that culturally America is less cautious than other countries about using medications? And what role does the FDA have in this?
Well, I'm not sure how good the data are. In fact, the various warnings about antidepressants, as far as we can tell, that the British use were based on the data we caused to be generated. I'm not absolutely positive of that. They said something and we said something at almost exactly the same time.
And their labeling isn't more cautious than ours on this. They've used the [word] "contraindicated" for some of them, but they don't really mean it. It says, I forget the exact language, but it's contraindicated unless you really need to [use it]. What does that mean?
Whether they're really more cautious than we are or not, I don't know. My impression is that we use a lot more therapy for ADHD than other countries do, so there may be some bias in that direction.
But my short answer is, we don't make that decision; that's left for the medical community. And that's appropriate for the medical community. Why there are these differences, I don't know.
But you do make the decision about what advertising is allowed. For instance, in England, there are no ads for pharmaceuticals right before the [nightly] news comes on. And there were no direct-to-consumer ads. Why are we so accepting of all this advertising?
The short answer is, direct-to-consumer ads are legal in this country, and that's the way it is. Whether they're a good thing or a bad thing is debated all the time.
I have mixed feelings about this. When advertising encourages people to get their lipids checked, that's a good thing. I wish there were more ads devoted to getting people to check and control their blood pressure, but there aren't too many of those. I don't think it's bad to remind people that depression is a bad disease. You can get different answers on how important it is to have ads for sleep aids and things like that.
... It's remarkable that when people complain about ads to us, they hardly ever give us an example of an ad they hate. And I think what it is, that people just don't like the idea that products are promoted directly to consumers.
How you feel about that depends on how you think and whether you think doctors can stand up to pressure. It always amazes me that some doctors think that if somebody asked them for a drug, they have to give it to them. I mean, I haven't been in practice for a long time, but that wouldn't be true for me. Someone asked for a drug, and I didn't think it was right, or I thought there was a cheaper drug or a generic to give, I would give them what I thought was necessary, not what they wanted, because that's my job.
Nonetheless, the studies we've done suggest that doctors can be very heavily influenced, and maybe even influenced to do something they didn't think was such a great idea. Well, that's not a good thing.
The short answer, again, is that direct-to-consumer advertising is legal in the United States, [so] we try to make it as balanced as we can. I'm positive people don't think we always succeed. But we try hard.