CIAO DATE: 3/00

The Safety of Pharmaceuticals: Monitoring and Regulation

Lucien Leape, Richard Platt, Hugh Tilson, Janet Woodcock, Michael Cohen, Susan Ellenberg and Eleanor Vogt

Conference Series
July 1999

The American Enterprise Institute for Public Policy Research

The withdrawal of several medications from the market in recent months has coincided with the publication of a stream of articles on drug safety in prominent journals. These developments have caused policy makers, pharmaceutical firms, physicians, and the Food and Drug Administration to look especially closely at drug safety and to consider the following questions: With the increased pace of drug approvals, is sufficient attention being paid to drug safety? Are markets and regulators doing a good job of monitoring safety? Is there more to be done?

AEI held a conference on March 26, 1999, to hear the views of scholars, the medical community, and the FDA on the major issues of drug safety. Resident Scholar John E. Calfee organized and moderated the two panels. An edited summary of the presentations follows.

 

Panel 1: Pharmaceutical Safety in Perspective

Lucien Leape
Harvard School of Public Health

We have high standards for drug approval, and they have resulted in a remarkable array of highly effective drugs that are quite safe when used according to instructions. The safety of those medications when they are improperly used is a very different issue, however, and it requires different methods for control. We refer to an ill effect from a drug when it is used properly as an adverse drug reaction, whereas we refer to an error in use as a medication error. Problems due to error in use are preventable, and it is those cases that we need to address.

Historically, both the FDA and the pharmaceutical industry, as well as all of us in medicine, have taken the approach that safety is the responsibility of the individuals involved in prescribing and dispensing drugs. We expect doctors, nurses, and pharmacists not to make mistakes. We select good people, we train them highly, and we hold them to high standards. We are just beginning to realize we have to do a lot more than that. We have to design systems that make it hard for people to make a mistake.

In a multiple–step process, people are more likely to make a mistake than in a simple process. It is therefore important to standardize everything as much as possible. That does not mean there should be no room for variation; nor does it mean one cannot take into account individual differences. It means one should standardize those things that can be standardized. I have in mind four particular areas: names, labels, packaging, and products. In each area, we can take certain measures right now, we should start moving on others fairly soon, and there are far–off issues that we ought to address in five years or so.

We clearly need to have one name for each drug. Not two, not five–one. It is crazy to have a proprietary name and a generic name. I know there are lots of reasons for the generic names. Most of those reasons are of interest to pharmacists and scientists and have nothing to do with the practice of medicine. With two names, there is already greater opportunity for error. The reasons for generic names are understandable, but they are not as important as patient safety.

We should institute human factors testing for all proposed new names of drugs. That is the process of checking to see whether a process is safe and hard to do perform incorrectly. Certain companies have already started doing human factors testing. The FDA should not approve any drug name that has not gone through extensive testing for name confusion.

Next, we ought to eliminate misleading derivative names within a couple of years. If a company has an effective product with a well–recognized name, it may be used on a totally different product, and that is a setup for error.

Every product should be bar–coded. We ought to bar code each product in the hospital to ensure it is the right drug for the right patient at the right time. Each drug should have its own distinct label. We still lack basic uniformities for the size, direction, and color of the print and for the organization and content of the information provided. Those elements of presentation should be standardized.

We should have one pill design per drug, so that all the ibuprofen look s alike, and so on. These are simple steps, but they are not trivial.

Richard Platt
Harvard Medical School and Harvard Pilgrim Health

I believe that the drugs we have today are safer than they have ever been and that this is largely due to the increasingly effective programs that the FDA uses both before and after drugs are marketed. Nonetheless, there are now probably more problems with adverse reactions to drugs than there ever have been. I say this partly on the basis of the literature that I read and partly just from looking at the experiences of my own family. During the last two years, both of my parents have sustained life–threatening adverse reactions to drug treatments. My father developed profound hypothyroidism as a consequence of a drug used to treat a heart condition, and my mother developed a life–threatening infection following use of a technique to prevent bleeding after coronary angiography.

Their experiences typify four important problems that we have yet to deal with satisfactorily. The first is that our population contains a rapidly growing segment of people who are living to a much older age than used to be common. The elderly are inherently at greater risk for adverse reactions than young people are. Second, the elderly are much more likely to take drugs in combination, and those combinations have the potential to exacerbate each drug’s adverse reactions. The pace at which new drugs are being introduced is increasing rapidly, and that has many positive effects–we are a better society because of them. Yet there are many more opportunities for adverse reactions to enter our treatment system, and we have less time to understand new drugs. Third, to the best of my knowledge, neither of my parents’ experiences was reported to the FDA. So there was no way that their problems could enter the public consideration of reactions from drugs. Finally, drugs are widely used in populations that are not well represented in the formal testing that is conducted, which means that we are missing many of the tools that we ought to have to decide whether a drug is useful in a specific therapeutic situation.

I subscribe entirely to the view that drugs are not inherently safe or unsafe; our job is to balance the risks and benefits. To do so we need three kinds of additional information. First, we need information about drug reactions that were not identified before approval. Next, we need information about reactions that are not unique to these drugs–that is, reactions that occur in increased frequencies even in the absence of drug administration. Finally, we need information about how drugs perform in special populations–the elderly, children, pregnant women, racial and ethnic groups–each of which might respond differently to specific drugs.

We can address some of these needs by using a relatively new resource that has become available: large automated systems for linking medical records. Insurance companies have developed them to help with payment for care, and medical–care delivery systems have developed them to provide care more efficiently. Although they were not designed as research tools, they turn out to be very useful for epidemiological research and health services research, including research on adverse drug reactions.

The FDA’s entire budget for these kinds of studies is less than $2 million a year. That’s about enough to support one good–sized study in the existing system. I believe that, to make the system useful, we need at least a tenfold increase in the resources that are either available to the FDA or that the FDA can direct. The research community could certainly make thoughtful use of $25–50 million to learn a great deal more about how to make drugs safer. That is a reasonable expenditure considering the importance of the problem, the costs of the alternatives that could possibly address these problems, and the amount of money that our society spends on pharmaceuticals.

Hugh Tilson
University of North Carolina

The protection of the public against unseen hazards of drugs lies in the hands of doctors, nurses, pharmacists, and community workers, all of whom need better training, information, and communication. Learning needs to be a continuous process for physicians, nurses, and pharmacists, but we do not have the local infrastructure to enable that important public health need. Most local health departments are languishing for lack of resources, particularly physicians who are trained or certified in preventive medicine and who understand these issues.

I ran a program for monitoring drug safety within a pharmaceutical company that received reports from the field on adverse drug reactions. We tend to think of drug safety protection as something that the FDA does and does well. The FDA is a remarkable enterprise, well–constituted and well–focused on the fundamental principles that the Institute of Medicine lays out for a public health agency. But they act in partnership. More than 80 percent of the reports on adverse drug reactions come to the attention of the FDA because physicians reported them or mentioned them in a contact with a pharmaceutical company. In turn, the pharmaceutical companies get 80 percent of their reports through contacts with practicing physicians, nurses, pharmacists, and health care workers. We in the pharmaceutical industry have to know as much as we can, as fast as we can, and as accurately as we can about these reports in order to know whether they constitute a new problem, provide better information about an old problem, or just add more information. We transmitted information to the FDA and became partners with them to ensure that communication of this information improved.

I second what Dr. Platt said and suggest that the use of population–based monitoring ought to be fundamental in drug safety considerations. Let me suggest a name to remember: Large Automated Multipurpose Population–Based Databases. LAMPs show the way to improve drug safety and to help the people who wish to help us understand population epidemiology if they are to do the job they need to do.

One of the great experiences in working in the pharmaceutical industry was working with my industry partners under the auspices of PhRMA—Pharmaceutical Research and Manufacturers of America. We learned best practices, and we advanced them. We met with the FDA on a regular, collegial basis to learn from each other, respecting each other’s roles but recognizing that public health benefits greatly from this kind of partnership.

We are working to automate medical records and databases and to institute proper privacy protections for patients without hampering our ability to learn from them. (An example is the large, general–practice research database in the United Kingdom–which is supported, in part, by the FDA. That allows us to learn about the experiences of three million people with the push of a button by an informed, properly trained physician or scientist.) Those challenges are not just domestic; they are of international concern. Drug safety knows no national bounds. We have partners from abroad who are interested in these issues as well, and we learn from each other new ways to improve medical practice.

Janet Woodcock
Food and Drug Administration

The standards for evaluation of pharmaceuticals have become stricter, mainly because the science of drug evaluation has advanced and because we know more about how to study drugs before they are approved. The FDA has put out guidance documents requesting that drugs to be used by the elderly be studied in the elderly before they are approved. Previously, the elderly were excluded from trials but would receive a drug once it was approved. Sometimes unexpected problems have arisen because the elderly typically take more medicines and are generally frailer than younger people. We also now require that gender differences be studied, because clinical trials had tended to exclude women, often for societal reasons.

Another way to look at the success or failure of the pre–market review program is to look at outcomes. There has been a lot of discussion about withdrawing drugs from the market. The rate of drug withdrawal, if you look at when the drug was approved and at the number of drugs the FDA is approving, is actually lower now than in the past. Some of the drugs withdrawn were approved decades ago. If you look at the rate of changes to labels of drugs, and in particular where serious reactions have to be added to labels because they are discovered in a post–marketing period, those rates are also lower than they were in the past. That doesn’t mean we couldn’t do better, and it doesn’t mean there won’t be problems. No drug is going to be without risks. We all have to face the risks and try to manage them. But I think the pre–market side of managing drug risk is doing better.

Now, moving to the post–market side, what about injury and death from drugs? In 1995, a journal article estimated that the annual costs of such adverse events were in the tens of billions of dollars. In 1998, another journal article estimated that about 100,000 deaths annually in hospitals were attributable to adverse reactions to pharmaceuticals. That is an ongoing problem and is a question of medical practice. We have wonderful medicines that provide opportunities to help people. But there is also opportunity for great harm. For pharmaceuticals, though, I do not think you could find an expert who would say that the risks outweigh the benefits and that therefore we should not have drugs. Almost everyone believes that drugs produce a massive public health benefit. And so the question is how do we optimize that benefit and minimize the risk?

Known side–effects and medication errors are not the only sources of preventable adverse events. Defects in product quality are also a source. But the pharmaceutical industry and the FDA actually do a very good job on product quality. Americans do not have to worry about their pills falling apart or being the wrong pills, the way people in other countries do. In my opinion, the greatest problem in this area is the lack of adequate and reliable data. Many drug–related injuries are caused by drugs that have been marketed a long time. That is not the kind of problem that our spontaneous reporting system is set up to deal with properly.

A cardinal principle of medicine is that before you preach, you have to make the diagnosis. That should apply to public health policy as well. We need to evaluate risks in a systemic context. The FDA needs to be a part of that context, as do pharmaceutical companies, health–care delivery systems, physicians, and even patients.

How do we manage risk? We can all do better, but we ought to know what we are doing and what problem we are trying to address. Societal awareness about the issue of quality is increasing, and I think this is a very hopeful time for dealing with the safety of medicines.

 

Panel 2: Monitoring Pharmaceutical Safety

Michael Cohen
Institute for Safe Medication Practices

For surveillance of medication errors at the post–marketing stage, our organization favors a program of voluntary reporting by practitioners. In our opinion, an effective system would act before multiple deaths are reported. If a practitioner told us about the potential for user error with two labels that looked alike or names that sounded alike, that information would immediately appear in our publications.

The basic problem for our organization is not that we do not know what to do but that the information we retrieve from practitioners has not been properly utilized. We have had situations where numerous deaths have occurred because of a medication error that was product–related but where no action was taken. The good news is that the FDA has developed the new Office of Post–Marketing Drug Surveillance, and we have seen more rapid action than in the past.

The goal, however, should be to get information about potential errors and then react to them before accidents actually happen. One reason we favor voluntary reporting is that it is unlikely that serious errors or potential errors will be reported in a mandatory program. An effective prevention strategy could be developed by requiring pharmaceutical companies to undertake systematic analysis of the potential for user error and by having practitioners look at new drug names, packaging, and labeling before the products are marketed. A mandatory reporting program would not generate those types of information. The contextual information that we need from the practitioners is absolutely the most important information we can get about user error. It sometimes takes us an hour or two of discussion with practitioners who have been involved with these errors to truly understand what actually happened.

We spend a lot of time on information related to labeling and packaging–trying to get samples of the doctor’s handwriting and so on. I am not sure that such information would always be available in a practitioner reporting program. Proper funding is needed to enhance the voluntary practitioner reporting effort. We need some incorporation of human factors experts. We need more physician participation. We have only one physician on our staff to overlook the reports as they come in, although U.S. Pharmacopeia–we are independent from U.S. Pharmacopeia–also has some medical input.

We need the failure analysis, the label standards, and the technology that Dr. Leape talked about. In particular, we need physicians to be able to enter orders into portable terminals, so that orders are printed and never appear in handwriting. Medicine is the only profession that still continues to handwrite communications rather than communicate through electronics. That information needs to be transmitted directly to a computer system, and then automated dispensing would take over. Bar–coded technology is needed for drug administration. It would prevent most of the errors that we see, even with the labeling and naming problems.

We need to do something about gaining access to data reported to the FDA through its MedWatch program. The purpose would be to get that information back out to the reporters. To increase reports of medication errors, the most important thing we can do is to increase the feedback that the reporters get. When they know that actions are being taken, they will increase their voluntary reports, and much more information will become public than would ever be published under a mandatory program.

Susan Ellenberg
Food and Drug Administration

At the FDA, safety is not examined only in the six– to twelve–month period when a product is being assessed or enters the market. It is looked at from the very moment the product is presented to the FDA for possible delivery to humans. Companies do pre–clinical studies, animal studies, and laboratory studies. They must submit the laboratory and animal data to the FDA before starting a Phase I study with human subjects. For both the company and the FDA, the question is whether those data indicate that the product can be safely administered to humans at an effective dose. The Phase I data will be analyzed both by the company and the FDA before Phase II, which starts to look at efficacy.

When all of those data are put together, the company is ready to do the definitive Phase III studies. If the drug is approved for marketing on the basis of Phase III studies, reports of serious events not observed before marketing have to be sent to the FDA as soon as the manufacturer finds out about these events.

Those are the good things. What are the problems? Well, the first problem everybody talks about is underreporting of adverse events. There is almost surely substantial underreporting and also inaccurate reporting. Details may be noted incorrectly, illegibly, incomprehensibly, or incompletely. The volume of reports is so vast that not every one of the hundreds of thousands of reports can be followed up. Furthermore, we have duplicate reporting. Sometimes people report directly to the FDA and also to the company, and sometimes multiple people will report a single event. That happens with more serious events because people are more concerned about them. It can be hard to sort out this duplication. If reports omit somebody’s middle name or misstate the age, for example, it will not always be clear that the event has already been recorded.

What happens when problems are discovered? As others have noted, we will never be able to identify all possible problems before marketing if we want to make new drugs available. So we expect to discover new things after the drug is on the market. Sometimes that simply leads to a labeling change. In some cases we get reports of problems that we had not observed, ones not important enough to cause great concern but important enough to publicize and to add to the list of observed risks on the label. Sometimes we learn about populations that should not get the drug, and we can add a contraindication. When something more dangerous is observed, a warning label can be added to the product–a black box that the physicians will see right away. Once a particular concern is raised, the FDA will work with its MedWatch partners to send out “Dear Doctor” letters, so that physicians in all the relevant disciplines will be individually and proactively notified about the problems. We can also issue press releases. Of course, the most drastic measure is to take a product off the market.

Safety assessments are performed at all stages of product development, not just in the final stage before approval, and not just after marketing. We think our post–marketing programs work. We do find new problems and inform the public. We think they can be made better, and we work very hard to do that both in the technical area, by means of database development, and in organizational issues.

Eleanor Vogt
National Patient Safety Foundation at the American Medical Association

If there is one message that I want to convey, it is that we know more now than we ever did, and that is good. Despite the doom and gloom in the media, we are crossing an evolutionary jumping–off point in our understanding of what we do.

The issue at stake here is whether we can devise a system that is accountable, that serves the public advocate who wants zero risk, yet at the same time serves the critically ill patient who has explored all other possible remedies and therapies and is willing to absorb and sustain a high level of risk. Do we have a system that allows for all of that? A related question is how to maximize the benefit. We make judgments based on our information and our values. We use decision–making skills, we educate, and we communicate. How can we do those things more effectively?

Those are the concerns and the questions that the National Patient Safety Foundation is asking. We want to learn from error and to prevent error. We provide a forum, a safe place to talk about mistakes and to move beyond blame. The people in our organization who address human factors come from other high–risk industries like aviation and nuclear power, and they remind us that blame is a social construct that we ascribe in hindsight. We often use it to catch the culprit, and we often use it, regrettably, to close the books. I hope that a future forum at AEI will address the inability to get access to the information needed for learning. We cannot improve the system if we cannot learn. We have to begin asking this question not only about our health care system, but also about our legal system: How is it helping or hindering our ability to get the information needed for learning?

There is also a problem of lack of communication among health professionals and institutions. The primary–care physician sends a patient to a specialist, and then perhaps the patient goes back to the primary–care physician and then off to the pharmacy–which is usually outside the system–and then to maybe a nurse and then to a community health professional. That process involves institutions from hospitals to nursing homes to clinics, but each typically operates in isolation.

Our organization is developing a proposal to address these problems: Where are the responsibilities? Who is accountable? Where are the holes in the system? Where are the vulnerabilities? Where are the strengths? What is it that we are already doing right? We are bringing together all the players in the pharmaceutical safety chain, starting with company research and development, and going through the regulatory agency, the prescribing physician, the pharmacist, and the end–user–the patient.

I see some very good signs. We are beginning to ask the right questions and to take the initiative with safety. That is new for us. We are starting to look at incentives to think prospectively in the adoption of best practices. I highly recommend the work of Dr. Leape and Don Berwick, and also the work of David Classen at Latter Day Saints Hospital, who has taken the factors that we have talked about and turned them into a proactive disease management program. With the use of antibiotics, for example, the program has eliminated 70 percent of the adverse drug events.

Janet Woodcock
Food and Drug Administration

Our post–marketing surveillance system, which looks primarily for rare, serious adverse events, is functioning well for that purpose. And through the Council of International Organizations for the Medical Sciences and the International Conference on Harmonization, we have harmonized with other countries’ surveillance systems.

We are working very hard on building our new computer system. We believe that, when we get electronic submissions from people, we will be able to provide better data to interested parties who want to search our databases. We cannot provide access to the databases now because they are not available electronically and because it is not practical to try to provide huge paper records to people. So the new system has a lot of promise in helping everyone.

But we all need to be aware that the FDA’s current adverse–event surveillance systems were not designed to deal with medication errors. The FDA approves drugs found to be safe and effective for the population. The drugs are then given over to the medical community and to the health–care delivery system to use safely, with adequate information. If an unexpected problem develops, the FDA deals with that. But the FDA and the whole medical system around the delivery of pharmaceuticals are not set up to deal with prescription errors. That is an important issue that is surfacing.

We are moving from a convention under which the consequences of misuse, off–label use, suboptimal use, and misprescribing were generally dealt with by the prescriber, and now we are moving to a more systematic approach. But as our system becomes more complex, more drugs are available, and there are more opportunities for problems. We have to decide what entity or collaboration of entities should be in charge of addressing them. The FDA has a leading role, and we can be very effective in solving packaging problems, drug strength problems, obscure parts of labels, or name confusion. We can do something about those sorts of things, and clearly we have a traditional role.

The FDA has acted in the past to manage the risk of drugs we considered particularly risky. For instance, we have mandated that the firm producing a new drug do certain testing–for example, blood testing to monitor for development of anemia –to ensure that problems are not developing. We have sometimes made agreements that the pharmaceutical would be restricted to a certain group of practitioners.

The further public policy issues that arise for the FDA are these: We could raise the bar on approval standards, and we could also get more data on pharmaceutical uses. Our role in managing medical practice problems, however, is not well defined. We need to explore the kinds of problems where the FDA should take the lead, which would be a departure from the past.