FDA Behind The Curve In Monitoring Safety of Approved Drugs

Over the past decade or so, there have been at least 20 prescription drugs removed from the market, including several cases of high-profile blockbuster drugs that were found to be harmful only after millions of patients had taken them. Vioxx, the pain reliever sold by Merck is one example; taken by an estimated 20 million Americans, it increased the risk of heart attack and stroke in some patients. The company ended up paying out a $4.85 billion settlement after the drug was pulled from the market in 2004. The diet drug Meridia, also linked to a higher risk of heart attack and stroke, was on the market for over 12 years before it was withdrawn last year. According to the watchdog group Public Citizen, drugs taken off the market since 1993 were sold for an average 4.1 years before being pulled.

The takeaway from these high-profile cases was that industry-sponsored drug trials, whose results are used to inform the Food and Drug Administration’s approval process, may miss serious side-effects that only become apparent once a medication enters the market and is used by millions of people. The consensus was that the FDA needed to expand its role in monitoring the safety of prescription drugs and had to do a better job of detecting these problems, especially in newly approved medications.

In 2007, Congress passed the FDA Amendments Act mandating the agency to establish an “active surveillance system for monitoring drugs” that collects electronic data from patient health records, insurance databases, medical registries and a range of other “information holders.” Four years later, the FDA’s still nascent Sentinel Initiative—the agency’s response to that mandate—remains in an early stage of development. In 2010, the FDA awarded Harvard Pilgrim Healthcare $72 million over four years to develop a “mini-Sentinel” surveillance system that will function as a kind of demonstration project; collecting drug safety data from the insurer’s electronic health records and other databases. But a widespread rollout is still years away.

Instead, the agency still depends on MedWatch, an out-dated and ineffectual system that relies on voluntary reporting of adverse events by physicians (sometimes in handwritten, barely legible notes), patients and according to Cardiology Today; “most pointedly, the MedWatch system relies primarily on drug manufacturers for the majority of collection, evaluation, and reporting of data from postmarketing studies of their own products.”

According to a recent article on the site Xconomy, “Only about 500,000 reports are sent to the FDA each year, about one-tenth of the estimated number of actual bad reactions. And once reports get entered, they are littered with misspellings, misclassifications, incomplete entries, and incompatible file formats that make it extremely difficult to search the database.” This has stymied efforts by pharmaceutical researchers, physicians, or consumers concerned about serious side effects to get reliable, up-to-date safety information about drugs currently on the market.

Take the case of zolpidem tartrate, the trade name for Ambien, a sleeping pill (available as a generic) used by tens of millions of people worldwide. Keith Hoffman, vice president for scientific affairs at a new start-up called AdverseEvent.com told Reporting on Health’s William Heisel that the FDA “has over 200,000 different raw drug names in its database,” including 442 variations on Ambien. Hoffman continues; “That includes generics, non-U.S. brand names, and every form of misspelling you can imagine. The problem is that there are adverse event reports associated with each one of those 442 different names. So, the only way to get a complete view of side effect problems with Ambien using the FDA’s data is to know all of those 442 names and add together the side effect reports and outcomes for each of those. It’s not something that can realistically be done manually.”

AdverseEvent.com is capitalizing on the shortcomings of MedWatch and the FDA’s seeming inability to effectively monitor the safety of drugs already on the market. The start-up has developed a “proprietary 17-step data refinement process” that allows it to sort through disparate serious drug side-effect data and come up with a clearer picture of risk for individual drugs. The company recently conducted an analysis of case reports of birth defects submitted to the FDA for 18 common epilepsy drugs used during pregnancy. Their findings were surprising; researchers discovered that some of the top epilepsy drugs deemed less risky during pregnancy were actually just as likely to cause birth defects as those relegated to a “riskier” category.

In 2007, along with Sentinel, the FDA began another initiative to ensure that drugs currently on the market are safe. If the agency suspects that a certain drug might have safety risks, they can now require manufacturers to develop Risk Evaluation and Mitigation Strategies (REMS) to “ensure that the benefits of a drug or biological product outweigh its risks” in the post-marketing period.

These strategies include one or more of three components. First, REMS may include a Medication Guide; an FDA-approved handout that provides information about safety risks that doctors or pharmacists are required to “distribute with medications posing a serious and significant public health concern.” Second, companies may be asked to devise a Communication Plan; letters to doctors or medical societies that outline the risks associated with the use of a particular drug and in some cases, recommend lab tests or other monitoring to detect adverse effects. Third, some REMS may involve Elements to Assure Safe Use, such as requiring providers or pharmacies to receive training and certification in how best to dispense a particular drug, or in some cases, the REMS could require patients to enroll in a registry. This is the case for patients using Accutane, an acne drug that is known to cause birth defects.

Recently, there has been controversy over REMS for strong opioids like Oxycontin that are increasingly prone to abuse and diversion into the drug trade. Critics of the FDA’s strategy in monitoring the use of these drugs feel that doctors and pharmacists need more education and tighter certification, stricter prescribing rules and other risk mitigation strategies to prevent abuse, while advocates worry that some of these REMS are burdensome for doctors and pharmacists, and keep patients from accessing needed drugs.

In the latest issue of the Journal of the American Medical Association (JAMA), researchers take the view that the REMS system is not functioning as well as it could be in communicating risk. The authors point out that half of the 139 drugs currently identified by FDA as posing “a serious and significant health concern” rely only on medication guides—safety and risk information that is often stapled to the package when a person picks up their prescription at a pharmacy or, less frequently, is handed to a patient by a prescriber—to communicate risk.

As the JAMA authors write, “Although some patients may receive, read, and synthesize the information contained in Medication Guides, there are a variety of reasons that many do not, and Medication Guides remain part of a fragmented and disorganized system of consumer medication information that leaves many patients inadequately informed.”

In the end, it seems that despite being given increased authority to oversee the post-market safety of prescription drugs, the FDA is still behind the curve. The collection of adverse event reports remains scattershot and technically obsolete—even as hospitals, large physician practices, insurers and other providers develop sophisticated databases to record health information. Instead, the drug agency continues to rely on voluntary—sometimes handwritten—reports from doctors and patients; or perhaps more worrisome, on data provided by the very companies that sell the drugs being monitored. Strategies to lower the rate of complications and serious side-effects associated with drugs currently on the market but known to present risks also seem insufficient.

Last month, a House Appropriations panel seemed eager to rectify this situation. The panel recommended that the FDA work on developing “independent postmarket surveillance of drugs approved for market” and to submit a report by March 31, 2012, “outlining the process needed for creating an independent office within the agency focused on post-market evaluation, with the controls and separation of duties necessary for making unbiased decisions.”

This sounds like a good idea. But, true to form, the House panel did not recommend any new funding for the agency to jumpstart its Sentinel Initiative—a project that would achieve just this kind of independent surveillance system. In fact, although the Obama administration asked for an 11.6% funding increase for the FDA back in March, the House and Senate both recommended virtually no change—and even a slight decrease—in funding for the agency in their various versions of the federal budget. New user fees collected from drug and medical device makers are expected to bring in an additional $1 billion or so to the agency; but most of this is earmarked for hiring more personnel to speed up drug review and approval.

The FDA will likely remain understaffed and too often, under siege by drug makers, patient advocates and Congress alike. A start-up like AdverseEvents.com might be able to carve out a niche for itself in providing more accurate risk information about drugs currently on the market for those who want to pay for it. But it’s no replacement for a nationwide surveillance system; an adverse-effect monitoring system that could prevent the next Vioxx debacle long before it becomes a widespread public health problem.

6 thoughts on “FDA Behind The Curve In Monitoring Safety of Approved Drugs

  1. Keith Hoffman is 100% incorrect. The FDA AERS system has an automated dictionary that recognizes each of the 442 names associated with zolpidem (as well as alternative names for other drugs) and as the same drug for statistical purposes. This is not a recent innovation; it has been true for years. Hoffman is not a disinterested party and you shouldn’t take what he says at face value.

  2. Maggie – thank you for an excellent article, one of the best I’ve seen recently online.
    Dr. Stone – your comment is interesting. In almost two years of working on this project and talking to innumerable drug safety professionals, yours is the first mention I’ve heard of the FDA doing something similar to what we’ve done with our RxFilter process. Since we’re not FDA insiders I can’t refute your statement. But, I have to wonder why the FDA (if they have the ability to standardize and normalize that data as we have) wouldn’t choose to make that cleaned-up data available instead of the unstructured data. In the current download or FOIA format, it serves the public and the healthcare professionals who need it very little good.
    And to be clear, we are absolutely not a disinterested party. We are a for-profit company that sees both an economic and public health benefit to making this data more widely available. We make the data we have available online and searchable at no charge to users. We charge users for the analytical and alerting tools we’ve developed to better analyze and mine the data.
    I welcome your further questions and comments and can be reached directly at brian@adverseevents.com.
    Brian Overstreet
    President, AdverseEvents

  3. Marc, Brian
    Thank you for your comments (Brian, fyi I wrote this article, not Maggie Mahar). There seems to be some inconsistency here between how the FDA compiles data internally and what is made public to those trying to access it, as Brian Overstreet and his colleagues at AdverseEvent.com are suggesting. I think I was clear that AdverseEvent is a for-profit start-up and has piqued interest in its services because of a perceived lack of information on drug side effects. Perhaps the two of you have connected off-line about this area; I’m interested in hearing more…

  4. Interesting post. Thanks for the info on the House Appropriations committee asking for an FDA report on setting up a separate drug safety office.
    Preapproval trials are usually powered for efficacy, not for safety. If an excess of adverse events occurs in the preapproval trials, it can be difficult to know if the effect is real or a statistical blip, especially when there are dozens of different types of adverse events.
    In the case of small risks and common diseases, it is difficult to know based on adverse event reports whether there is an increased risk or not. For one thing, there is no denominator or control group. For example, an older person having a heart attack or stroke while taking Vioxx did not usually raise suspicions because heart attacks and strokes are very common in older people. Sometimes it takes a randomized controlled trial to show such an increase in risk. As you know, the excess of heart attacks from Vioxx was first shown in a Merck-sponsored RCT comparing Vioxx with naproxen. The excess of heart attacks in the Vioxx group raised suspicions but Merck argued that naproxen had a protective effect. What should have happened then was for the FDA to require a large RCT with cardiovascular endpoints conducted by independent investigators. Instead, several years went by until the risk became obvious in another trial that Merck was conducting to try to get a new indication.
    In the case of Meridia, the European drug regulators ordered the sponsor to conduct such a trial and the results of that trial caused the drug to be pulled from the market in both Europe and the U.S.
    Another thing that needs to happen is for companies to make patient-level data available to independent investigators. This is being done in the case of a Medtronic product called Infuse. See this commentary by Harlan Krumholz and Joseph Ross for details.
    Krumholz, et al., A Model for Dissemination and Independent Analysis of Industry Data. JAMA.