Adverse Drug Event Surveillance and Drug Withdrawals in the United States, 1969-2002: The Importance of Reporting Suspected Reactions

Background  The Adverse Event Reporting System is the primary surveillance database used by the Food and Drug Administration for identifying postmarketing drug safety problems.

Methods  We analyzed all reports of suspected adverse drug reactions submitted to the Food and Drug Administration from the inception of the Adverse Event Reporting System database in 1969 through December 2002. We documented drug withdrawals and restricted distribution programs based on safety concerns.

Results  During the 33-year period from 1969 when adverse drug event reporting was initiated through 2002, about 2.3 million case reports of adverse events for the cumulative number of approximately 6000 marketed drugs were entered in the database. Most reports were for female patients. During this period, numerous drug reactions have been identified and added to the product labeling as boxed warnings, warnings, precautions, contraindications, and adverse reactions. More than 75 drugs/drug products have been removed from the market due to safety problems. In addition, 11 drugs have special requirements for prescriptions or have restricted distribution programs. Drugs withdrawn or restricted represent a small proportion (about 1%) of marketed drugs.

Conclusions  The Food and Drug Administration’s Adverse Event Reporting System is the primary surveillance database used for the identification of safety problems of marketed drugs. Despite the limitations of underreporting, differential reporting, and uneven quality, submitted reports often allow the identification of serious adverse events that are added to the product labeling information. In rare instances, additional regulations, up to and including market removal, have been required. We encourage physicians, pharmacists, other health care professionals, and patients to continue to report serious suspected and known adverse drug reactions to manufacturers and the Food and Drug Administration.

A primary mission of the Center for Drug Evaluation and Research of the Food and Drug Administration (FDA) is the review and approval for marketing of safe and effective drugs and certain biologic agents. Before approval, drugs are evaluated for a defined indication in clinical trials of relatively short duration in a relatively small number and in restricted categories of people. Following approval and marketing, when the drug is used in larger populations for longer durations and for unapproved indications, previously unidentified adverse drug eventsoften occur. Some of these events have been serious and frequent enough or have altered the risk-benefit ratio enough to result in the drug’s removal from the market.

A primary source of information used by the FDA for identifying postmarketing drug safety problems is surveillance of voluntarily submitted adverse drug event reports. Reports are sent by health care professionals, consumers, drug manufacturers, and others to the FDA, where they are organized in a database referred to as the Adverse Event Reporting System (AERS), previously known as the Spontaneous Reporting System. Herein, we describe adverse drug event reports submitted to this database from its inception in 1969 through December 2002 and provide information on associated regulatory actions, including drugs removed for safety reasons.

The Office of Drug Safety (formerly the Office of Post-Marketing Drug Risk Assessment) of the FDA receives reports of adverse drug events primarily from physicians and pharmacists who submit them on a standardized form (referred to as the MedWatch form) indirectly through pharmaceutical companies or directly to the FDA. Health practitioners and patients may also report adverse drug events by telephone or by accessing the MedWatch Internet site. Published reports of adverse drug reactions also may be submitted.

Each report is entered into the AERS computerized database using a coding thesaurus of adverse reaction terms used for searching and retrieval purposes. Reports are entered manually; in late 1997, the system was redesigned to begin accepting a growing number of electronic reports submitted by pharmaceutical companies.

The process for identifying adverse events and for initiating regulatory action for drug safety problems is as follows: FDA safety staff (primarily pharmacists) receive AERS reports daily and review them for possible drug causality. After a sufficient number of convincing reports and/or other data (including published studies and case reports) have accumulated implicating the drug with a reaction, pharmacists and epidemiologists present the evidence in writing to the division that reviewed and approved the drug. If the reviewing division agrees that the data are compelling enough to require regulatory action (eg, a change in the product information), it notifies the manufacturer and requests the action. Disagreements between manufacturers and the FDA may be resolved through formal dispute resolution and the use of an Advisory Committee.¹ Title 21 of the Code of Federal Regulations (eg, Sections 2.5, 7.1-7.87, 314.150, and 314.151) describe regulatory actions that can be taken for the removal of a drug for safety reasons.² These include recall procedures (usually for lot specific problems), declaration of the drug as an imminent hazard (invoked only once for phenformin and lactic acidosis), and notification of a judicial hearing.³ In most situations after important safety problems become evident, drugs are voluntarily withdrawn by the market authorization holder.

The AERS data are reviewed by safety staff for duplicate and follow-up reports. Follow-up to obtain additional information occasionally is conducted by manufacturers and rarely by FDA staff (because of limited resources). Technology has enabled linkage of follow-up and initial reports as case reports. For this article, case report counts were obtained, but some duplicate and follow-up reports cannot be identified, so the numbers do not refer to individual persons. Reports are entered continuously. The data in this article refer to those accessed from June to August 2003.

Reports with serious outcomes (death, life-threatening event, hospitalization, disability, congenital malformation, and event that required medical intervention) are required to be submitted by drug manufacturers to the FDA within 15 days of receipt and are priorities for analysis. A separate serious outcome category referred to as “other” that was added in 1997 was not included in our tally of serious reports.

Previously, no more than 5 adverse events per report were entered in the AERS, but in 1997 the system was restructured to allow entry of all adverse events mentioned in the text. Because all adverse events are entered but serious events are not identified as such, relatively nonserious events may be included in lists of events with serious outcomes.

We analyzed reports by year, country, sex, age, and adverse events. We identified drugs and drug products that have been removed from the market or have restricted distribution programs because of safety reasons—many of which were found or confirmed using the AERS database.

During the 33-year period from 1969 when adverse drug event reporting was initiated through 2002, about 2.3 million case reports of adverse events for marketed drugs were entered in the AERS database. Through 1997, there was a continuous increase in the annual number of reports entered. Beginning in 1998, annual changes have been inconsistent (Figure). Compared with 1999, the peak year for receipt and entry of reports, in 2002, reporting declined 21%. This, in part, may be because of the FDA’s recent procedural change to discontinue entry of reports with nonserious outcomes for drugs approved longer than 3 years and to allow market authorization holders waivers for submission of reports with adverse events that have nonserious outcomes and that are already mentioned in the product labeling.

About 1.8 million reports (about 80%) originated from reporters in the United States. The second largest source was from unidentified countries. Assuming that these originated in this country, the proportion of US reports was 87%. The country with the second largest number was France (64 021 reports), followed by the United Kingdom (45 864 reports) and Japan (41 053 reports). No country besides the United States had more than 3% of the total number of reports. Most reports (53%) concerned female patients, 35% referred to male patients, and 12% did not specify sex. Patient age was less well reported than sex, with 30% of reports having unspecified ages. Of the reports, 5% were for patients 14 years or younger, 24% concerned those aged 15 to 44 years, 16% were for those aged 45 through 59 years, and 25% referred to those 60 years and older. Race is not a separate field on the form. For 1969 through 2002, about 60% of reports were from health care professionals, 29% were from consumers, and the remainder were from study and literature reports, company data, and other sources. Overall, 29% of reports had serious outcomes.

Table 1 presents the most frequently reported adverse events for all reports entered in the AERS through 2002. The event with the highest count was drug ineffective, with 151 431 reports. This was followed by dermatitis, headache, nausea, pruritus, pyrexia, dyspnea, dizziness (exclusive of vertigo), vomiting, and urticaria (several of which are indicative of hypersensitivity/allergic reactions). The leading events associated with serious outcomes (Table 2) included pyrexia, with 41 529 reports, followed by dyspnea, vomiting, hypotension, condition aggravated, death, asthenia, pneumonia, convulsions, cardiac arrest, and chest pain.

Since its inception in 1969, the AERS database has provided evidence that certain approved drugs/drug products posed serious safety problems. Some drugs have been removed from the market, based in whole or in part, on adverse event reports. Table 3 provides a list of known drugs/drug products removed from the market, dating from 1964, because of safety reasons.^4-7 For many of these products, safety concerns were identified from spontaneous reports or in clinical trials with verification from spontaneous reports, although good historical data are lacking. Table 4 provides a list of drugs/drug products that were removed from 1978 through mid 2003.^4-7 Considering that, as of fall 2003, an estimated 6000 trade name drugs were in the US marketplace (Pat Muller, RPh, National Drug Data File [NDDF] Plus, First DataBank, written communication, 2003), the proportion of drugs withdrawn from marketing for safety reasons has been small. In the 1960s, there were only 532 suspect (for causality) trade name drugs entered in the AERS when 8 drugs/drug products were removed from distribution. Comparable numbers were 36 drugs/drug products removed of 1655 entered in the AERS during the 1970s, 13 of 2357 in the 1980s, 14 of 4129 in the 1990s, and 6 of 3952 entered thus far in the 2000s (counts from both Tables 3 and 4).

In addition to those removed, 11 drugs have special requirements for prescription or have restricted distribution programs (Table 5). These include clozapine, isotretinoin, fentanyl, trovafloxacin, thalidomide, bosentan, mifepristone, alosetron, sodium oxybate, and dofetilide. Three teratogenic drugs (isotretinoin, acitretin, and thalidomide) have formal pregnancy prevention risk management programs.¹⁰

Besides providing evidence prompting removal and restricted distribution programs of certain drugs, AERS reports also have provided information leading to the addition of boxed warnings, warnings, precautions, contraindications, and adverse reactions to the product information labeling¹¹ and the provision of information to patient package inserts and patient medication guides.¹²

Since 1969, a spontaneous reporting system of suspected adverse drug reactions has provided the FDA with drug surveillance information and has supported regulatory decisions about the safety of the approximately 6000 marketed drugs. Because more than 10 000 separate adverse event terms have been used in connection with these drugs, the AERS database is a rich source of drug safety information.

Although the AERS is a valuable resource, an accumulation of well-documented reports does not by itself indicate a causal relationship between a drug and an adverse event. Rather, the spontaneous reports that make up the AERS database are often considered as early warning signals, hypothesis generating, or preliminary evidence with causality assessments often requiring additional supporting data from clinical trials and/or published studies. However, in certain circumstances, analyses of spontaneous reports can lead to an assessment that a drug and a usually relatively rare outcome are causally related. Criteria for such circumstances often include¹³ a temporal relationship between the drug and a rare (often drug-related) outcome, occurrences of positive dechallenge, occurrences of positive rechallenge, the lack of confounders or other explanations for the outcome, biological plausibility, a dose-response relationship, a consistency between US reports and those from other countries, and consistency with, or extension of, clinical trial data. In some situations, using numbers of adverse events and drug exposure data, it is possible to determine whether the rate of the reported adverse event approaches or exceeds the background rate for the outcome in a comparable age and sex group of the general population.¹⁴ Also, using AERS reports, it is possible to determine if drugs in the same class, but not in other classes, report a particular adverse event, and how the number of reports for the event compares with the numbers associated with other drugs in the AERS database. Data mining of AERS reports also may assist in the identification of adverse event signals.¹⁵

Despite these possibilities for using the AERS data, evaluation of reports is often limited by underreporting, possible differential product reporting, and an uneven quality of submitted reports. However, these limitations have not resulted in disuse of the data. While investigators have reported that active surveillance of a particular adverse drug event, eg, thrombotic thrombocytopenic purpura with clopidogrel,¹⁶ provided more complete, timely, and certain diagnoses than voluntary reporting to the FDA, active surveillance probably would not have been undertaken with clopidogrel had not thrombotic thrombocytopenic purpura been identified previously by spontaneous reporting to be associated with ticlopidine,^17,18 a drug in the same class as clopidogrel. Furthermore, active surveillance of the many drugs and potential adverse reactions by the number and composition of FDA staff is not feasible. However, FDA epidemiologists have sometimes used an extramural cooperative agreement research program^19-21 or have participated in studies that have used other funding mechanisms²² to conduct additional studies of selected adverse drug reactions.

Following risk assessment of a drug’s safety problems using the AERS and other data, strategies to minimize the risk may be enacted. These include dose changes, product labeling changes (a continuum from adverse reactions to boxed warnings), issuance of informational health care professional letters, publication of AERS data in the medical literature,^17,23-44 release of FDA Talk Papers and public health advisories for press reports, inclusion of patient package inserts and medication guides with drug dispensing, written informed patient consent before prescription, restricted drug distribution, and drug removal. In the case of one drug’s removal from marketing, the chronology and timing of regulatory actions in relation to the identification of the adverse event have been described in detail.²³ Although the timing of new boxed warnings was the focus of a recent study⁴⁵ and editorial,⁴⁶ documenting the timing of regulatory actions for the drugs removed was not the focus of this article.

Data in this report indicate that a small proportion of marketed drugs have been removed for safety reasons. Besides removal and disuse of drugs because of market competition, the FDA also relies on risk management strategies to achieve postmarketing drug safety standards. Indeed, effective science-based risk management was announced in 2003 as a major initiative of the agency⁴⁷; however, risk management plans will require evaluation to determine their effectiveness in reducing drug risks.

The continuation of the receipt of adverse drug event reports is critical to the drug safety mission of the FDA. Because the AERS data are an important source for detecting postmarketing drug safety problems, we encourage physicians, pharmacists, other health care professionals, and patients to submit reports of suspected (and known) adverse drug reactions to the drug manufacturer or the FDA. They may be reported to the FDA by completing and submitting (by mail or facsimile) the MedWatch form that is the last page of the Physicians’ Desk Reference, by calling (telephone number: 1-800-FDA-1088), or by accessing the MedWatch Internet site (http://www.fda.gov/medwatch).

In September 2004, Merck & Co, Inc, voluntarily withdrew rofecoxib (Vioxx) from the global market because of an increased risk of cardiovascular events. Two months later, the FDA announced that manufacturers of isotretinoin will obtain registration of prescribers of isotretinoin, dispensing pharmacies, and patients who are prescribed the drug.⁴⁸ The agency also announced the requirement of documentation of a negative pregancy test result before isotretinoin is given to women capable of becoming pregnant. In April 2005, valdecoxib (Bextra) was withdrawn from the market because of serious dermatological conditions and an unfavorable risk vs benefit profile.

In 1994, the FDA established that any over-the-counter drug products (eg, quinine sulfate) for the treatment and/or prevention of nocturnal leg cramps is not generally recognized as safe and effective and is misbranded.⁴⁹ 1998, the FDA established that over-the-counter drug products containing quinine for the treatment and/or prevention of malaria are not generally recognized as safe and are misbranded.⁵⁰

Correspondence: Diane K. Wysowski, PhD, Division of Drug Risk Evaluation, HFD-430, Food and Drug Administration, 5600 Fishers Ln, Parklawn Building, Room 15B-08, Rockville, MD 20857.

Accepted for Publication: June 2, 2004.

Disclaimer: The views expressed herein are those of the authors and do not necessarily represent the official position of the FDA.

Financial Disclosure: None.