Adverse Drug Reaction Reports Review Practices at the Food and Drug Administration

ADVERSE DRUG REACTION REPORTS REVIEW PRACTICES AT THE FOOD AND DRUG ADMINISTRATION

Serious unlabeled ADR reports submitted by compa-nies, serious ADR reports (labeled or unlabeled) submitted directly to the FDA by health professionals or consumers and reports of selected ‘important’ medical events are electronically transferred to the computer ‘in-box’ of one of approximately 25 safety evaluators, who review them on a daily basis. The safety evaluators are primarily trained clinical phar-macists who are assigned to cover specific groups or classes of drugs or therapeutic biologics. Over time, they acquire in-depth familiarity with the products they monitor.

In reviewing these case reports, the primary focus is placed on identifying previously unrecognized serious ADRs. When such a report is identified, a computer search is made of the entire AERS database for reports of similar cases with the drug in question. These cases are reviewed for clinical content and complete-ness. If important information is missing or supporting medical records are needed for some cases, then the safety evaluator may contact the reporter, usually a health professional, to obtain the needed data. This is a time-consuming but essential process, especially when faced with a serious ADR associated with a widely used medicine. In parallel with these activi-ties, a literature review is performed, national drug usage data are obtained and, frequently, an epidemi-ologist within the office conducts an investigation of background incidence rates and risk factors for the clinical event of interest. For example, a case series of pancreatitis in association with the use of a particular drug product might be supplemented by incidence data from a population-based, randomized survey conducted by the US National Center for Health Statistics.

After a case series is assembled and follow-up completed, it is analyzed for drug relatedness. Several factors are important to this assessment. Temporal association describes the relationship between drug exposure and event. If the adverse effect preceded the drug exposure, then the drug cannot have caused the effect. If the reaction resolves with the withdrawal of the drug, then the ‘dechallenge’ is positive; if the reac-tion reoccurs with the re-initiation of the drug, then the ‘rechallenge’ is positive. Dechallenge is often cited as the evidence of drug relatedness. However, the lack of resolution (negative dechallenge) should not be viewed as evidence against an association. Many adverse effects, once initiated, follow a course of their own. This is especially apparent with certain blood dyscrasias, serious skin reactions and acute liver fail-ure. Positive rechallenge has traditionally been cited as strong evidence of drug association. Our experi-ence suggests that the absence of reoccurrence should not be taken as evidence against the association. For most recognized and serious ADRs, rechallenge is not intentionally performed.

The timing of onset of the ADR after the begin-ning of drug use may provide clues as to possible mechanisms (short latency: anaphylaxis; long latency: cirrhosis). It is also important to note if other explana-tions for the adverse effect are present such as under-lying disease states or other medications. A profound hypotensive episode shortly before the development of acute liver failure may be the causative factor rather than the drug the patient was taking. Alter-natively, the natural course of the patient’s medi-cal condition(s) may be associated with the event of interest. Additionally, other medications, herbal or dietary supplements taken by the patient may be linked to the ADR. Disease states and/or other drugs may therefore cloud or confound the relationship between a particular drug and event, complicating the assessment of case reports. Finally, clinical and laboratory features of the ADR and its progressive unfolding may also provide information that distin-guishes it from underlying or other disease processes (Meyboom et al., 1997). For example, myopathy is a recognized consequence of HIV infection but can also result from zidovudine, used in the treatment of HIV/AIDS. Zidovudine-induced myopathy was found to be caused by damaged muscle mitochondria, distin-guishable from HIV myopathy based on the pres-ence of ‘ragged-red’ fibers in biopsy specimens from affected patients (Dalakas et al., 1990).

The safety evaluator usually stratifies the cases into those with more complete information in which other potential explanations are absent or extremely remote, cases with incomplete information and cases with other risk factors or potential explanations for the adverse event. The case material is evaluated in combination with drug usage data, epidemiologic information and the published literature. In general, a signal results if there are higher quality, unconfounded cases plus supporting cases with less complete infor-mation or confounding factors present. There is no ‘threshold’ number of cases required to indicate the significance of a potential signal; medical judgment is used in each situation. For example, in 2004, the FDA advised healthcare professionals about a new warning for atomoxetine, a drug approved for atten-tion deficit hyperactivity disorder (ADHD) in adults and children. Following receipt of only two reports (a teenager and an adult) in patients who had been treated with atomoxetine for several months, the label-ing was updated with a bolded warning about the potential for severe liver injury. On the contrary, it took over 300 cases of serious cardiac arrhythmias and about 80 deaths before cisapride was withdrawn from the market.

An analysis of the safety issue is presented to the medical reviewing division responsible for ongoing regulation of the drug. A decision is then made about whether the signal is strong enough to warrant a regu-latory action such as changes in product labeling, further study, issuance of a public health advisory, restriction of use or market withdrawal.