Post‐market drug safety evidence sources: an analysis of FDA drug safety communications

For over 40 years, spontaneous reporting of adverse events has been the cornerstone of Food and Drug Administration (FDA)’s post-approval drug safety monitoring system.1 Over time, clinical trials conducted after approval, as well as observational studies based on actual patient experiences, have also been the source of important drug safety evidence and have informed safety-related regulatory decision making (e.g. aprotinin and mortality;2–4 ESAs and mortality).5–7 Currently, the three primary postmarket drug safety evidence sources include clinical trials, spontaneous reports, and observational studies; data from pharmacodynamic and pharmacokinetic studies, registries, and other sources may also generate safety issues. Recent examples of drug safety issues, e.g. ADHD drugs8–10 and rosiglitazone11–13 and serious cardiovascular adverse events, delineate the fact that multiple evidence sources may contribute to the identification and evaluation of drug safety issues. In addition, a recent publication discussed the relationship between safety issues generated from spontaneous reports and FDA regulatory actions.14 The contribution of different evidence sources to drug safety issues has not been quantified. Understanding this is important because it (i) facilitates the evaluation of how various types of evidence contribute to drug safety issues, (ii) enables the characterization of the relative value of individual evidence sources, and (iii) may inform the integration of emerging data sources such as the Sentinel system into drug safety surveillance.15,16 We examined FDA drug safety communications and determined the source of safety issues and the type of regulatory decision that resulted from these issues. FDA drug safety communications issued from 2007 to 2009 were identified.17 Communication types included early communications, communications about ongoing safety reviews, healthcare professional sheets, and public health advisories. Duplicate communications were removed; related communications (updates or follow up to the original communication) were counted with the original communication. Other communications that were not related to new clinical drug safety information, such as product quality, were excluded. Evidence sources from which each safety issue was derived were identified and characterized by compiling these sources cited in each Drug Safety Communication and following up with relevant FDA/CDER staff as needed to confirm the sources that contributed to the identification of a safety issue. More than one evidence source could contribute to a safety issue. An analysis of the contribution of each source based on regulatory action type was conducted based on the information provided in the drug safety communication and by consulting with CDER staff. The following types of safety-related regulatory decisions were considered: (i) open (not yet concluded), (ii) no action taken other than continued monitoring, (iii) product withdrawal, (iv) new medication guide (MedGuide) requirement or modification, (v) product labeling change, (vi) other (change of indication, recall, warning for medication error). Of the 116 drug safety communications listed on FDA’s website, a total of 69 unique drug safety issues were identified (Table 1). Nine duplicates were removed, and 24 follow-up communications were included with the original communication. Other excluded communications were those communicated before 2007 (7), communications pertaining to emergency use announcements (4) and environmental or quality issues (2), and one communication concerning an unapproved drug. When including all drug safety communications based on issues identified by one or multiple evidence sources from 2007 to 2009, 39/69 (57%) of the communications were based on spontaneous reports; 17/69 (25%) from clinical trials; 9/69 (13%) from observational studies; 4/69 (6%) from systematic reviews, pooled analyses, or meta-analyses; and 6/69 (9%) from other evidence sources (Table 1). With respect to regulatory actions related to these drug safety communications, 2/5 (40%) of the withdrawals, 7/11 (64%) of MedGuide, 19/33 (58%) of the product label changes, and 6/7 (86%) of other safety-related regulatory actionswere based on spontaneous reports (Table 1).