Identifying Safety Signals
Good pharmacovigilance practice is generally based on the data obtained from adverse reports, also known as case reports. In addition, some information published in medical journals or observed during clinical studies may generate signals for adverse effects of drugs.
Case reports are usually submitted to both the FDA and the sponsor. The FDA recommends that sponsors obtain enough information from each case for appropriate assessment of the event.
A good case report should include such information as description of the event; the drug used including dose, lot number, duration; patient characteristics including demographic information and relevant disease history; and diagnosis of the event and relevant therapeutic measures. If the event has occurred due to a medication error, the information and the environment causing that error must be documented also.
As expressed by the FDA, for any individual case report it is difficult to know with a high level of certainty whether the event was caused by the product. FDA suggests that sponsors should search for similar or related cases from the global adverse event database or other available sources such as the FDA's Adverse Event Reporting System (AERS) or Vaccine Adverse Event Reporting System (VAERS).
In the event that one or more cases suggest a safety signal warranting additional investigation, the FDA recommends that a case series be assembled and clinical information be summarized to characterize the potential safety risk and identify risk factors.
Information on adverse reports is used to develop case series, which usually include an analysis of the clinical manifestation of the event, demographic characteristics, exposure duration, dose and lot number, route of administration, any other medication used, and the presence of co-morbid conditions.
The FDA emphasizes that use of statistical and mathematical tools (data mining) can provide additional information about the existence of an excess of adverse events reported for a product. By applying data-mining techniques to large adverse-event databases, such as AERS or VAERS, it may be possible to identify unusual or unexpected product-event combinations, warranting further investigation.
Data mining is especially useful for assessing patterns and trends. The current methods of data mining generate a score for each event, product, or characteristic. The score quantifies the disproportionality between the observed and expected values for a given product-event combination.
The FDA recognizes that although the above methods will assist sponsors in identifying a safety signal, they might not reveal the actual risk to patients, mainly due to under-reporting. Considering this, a rate higher than the background may be an indication that the true incidence rate is sufficiently high to be of concern.
Examples of safety signals that warrant further investigation include new unlabeled adverse events, an apparent increase in the severity of a labeled event, and occurrence of serious events thought to be rare in the general population.
When, due to the severity of the event or an excess of adverse events, further investigation is needed, it is important to put the signal into context by calculating the rate at which new cases of adverse events occur in the product-exposed population (i.e., the incidence rate). In some pharmacoepidemiologic studies, the numerator (number of new cases) and denominator (number of exposed patients and time of exposure) may be readily available.
But for some others the accurate number of patients exposed to a medical product may not be available. When the accurate number is not available, sponsors may use the estimated number to calculate the incidence rate.
When an estimated number is used, the FDA recommends that sponsors submit a detailed explanation of the rationale for selection of a denominator and the method of estimation. To provide further context for the incidence rate, it is helpful to have an estimate of the background rate for the event in general population.
Investigating a Signal
There are various methods for investigating a safety signal. The signal can be further evaluated by carefully designed nonrandomized observational studies of the product's use in the real world and by randomized trials. Some types of randomized trials have been discussed in the Premarketing Guidance and could be used pre- or post-approval.
The FDA focused on three types of nonrandomized observational studies: pharmacoepidemiologic studies, registries, and surveys.
The FDA emphasized that many other methods are available and encouraged sponsors to consider all and choose the one best suited to the particular signal of interest. A summary of these three methods is presented as follows:
(1) Pharmacoepidemiologic studies can be of various designs, and the results may be used to characterize one or more safety signal. A pharmacoepidemiologic study should have a protocol, control group and test prespecified hypotheses. These studies are used for estimation of relative risk and incidence rate and can be initiated at any time (but usually when a safety signal has been identified after approval).
According to the FDA, for uncommon or delayed adverse events, pharmacoepidemiologic studies may be the only practical choice for evaluation.
For the evaluation of more common events, clinical trials may be more practical. Because pharmacoepidemiologic studies are observational in nature, they may be subject to confounding (adverse events with possible etiologies other than the product of concern) and other bias, which may make the results more difficult to interpret.
Because of wide variation in benefit-risk considerations, it is impossible to specify the point at which a pharmacoepidemiologic study should be initiated, and therefore the decisions should be made on a case-by-case basis.
When a sponsor determines that a pharmacoepidemiologic study is the best method for evaluating a particular signal, it should proceed with an appropriate design. The FDA provided some references for methodologies for pharmacoepidemiologic studies, and encouraged sponsors to communicate with the agency when pharmacoepidemiologic studies are being developed.
(2) The term "registry" in this document has been described as an organized system for the collection, storage, retrieval, analysis, and dissemination of information on individual persons exposed to a specific medical intervention who have a condition that predisposes them to the occurrence of a health-related event.
According to the guidance, creation of registries could help sponsors to evaluate safety signals identified from spontaneous case reports, or other sources, and to evaluate factors affecting the risk of adverse outcome such as dose or duration. This tool is particularly useful for collecting information from multiple sources. The FDA suggests that sponsors develop a protocol to include objectives, literature review, and relevant animal and human data.
(3) Surveys can be used also to assess a safety signal. Like a registry, surveys can be initiated by sponsors at any time. The FDA suggests that sponsors develop a protocol including the objectives and detailed description of the research methods. Sponsors are encouraged to discuss survey development plans with the FDA.