Association Between Data Sources and US Food and Drug Administration Drug Safety Communications

Drug safety communications (DSCs) are the primary tool for the US Food and Drug Administration (FDA) to communicate important new postmarketing safety information to patients and health care professionals.¹ Drug safety communications are issued by the FDA for more than one-quarter of new drug and biologic approvals.² There are known inconsistencies in issuing DSCs across national regulators.³ Less is known about the information source that typically serves as the basis of the initial safety signals that lead to DSCs, such as spontaneous reports aggregated through the FDA’s Adverse Event Reporting System (FAERS),⁴ results from randomized clinical trials (RCTs), or observational studies. Our objective was to describe the sources of the initial safety signals that lead to DSCs and examine their associations with drug characteristics and subsequent label changes.

The study included all DSCs issued between January 29, 2010, when DSCs first became publicly available, and December 31, 2018, identified from the FDA website.⁵ We excluded DSCs for over-the-counter drugs. For each DSC, we determined the information source that served as the basis of the initial safety signal, categorized as FAERS data, RCTs, or other sources, including observational data. We used the Fisher exact test for categorical variables and the Wilcoxon rank sum test for continuous variables to examine associations between information source and initial regulatory approval pathways, time in years between initial approval and DSC posting, characteristics of clinical studies included in the most recent drug label prior to DSC publication, and subsequent safety-related changes to the drug label. We used Spearman rank correlation testing to examine for changes in the number of DSCs issued annually. Analyses were performed using SAS, version 9.4 (SAS Institute). Statistical significance was defined as a 2-sided P < .05.

This study was exempt from the requirement for ethical approval according to the Government of Canada Panel on Research Ethics (http://www.pre.ethics.gc.ca/eng/home.html) because of its exclusive use of publicly available data.

The FDA issued 228 DSCs from 2010 through 2018. The most common medical specialties related to the subject drugs were endocrinology (44 [20%]), infectious diseases (39 [17%]), and neurology (27 [12%]) (Table 1). Of the DSCs issued (N = 228), 15% (35) were in 2010 and 26% (58) in 2011, whereas 4% (10) were issued in 2017 and 5% (11) in 2018 (P = .005).

Among the 228 DSCs, the most frequent information sources that served as the basis of the initial safety signal were FAERS (87 [38%]) and RCTs (81 [36%]). Time in years from initial approval to DSC posting was significantly shorter for DSCs triggered by FAERS or RCTs vs other sources (FAERS: median [interquartile range (IQR)], 10.4 years [2.9-21.6 years]; RCTs: median [IQR], 10.8 years [4.8-17.0 years]; and Other: median [IQR], 20.2 years [11.6-32.2 years]; P < .001) (Table 2). Common subsequent changes to drug labels included additional warnings and precautions (103 [45%]) and boxed warning (42 [18%]).

The leading initial sources of DSCs were the FAERS program and RCTs. Despite known underreporting of safety information to voluntary passively collected pharmacovigilance programs, the large proportion of DSCs initiated by FAERS suggests that the FDA relies on the program for postmarket drug safety monitoring. Similarly, the continued utility of RCTs for identifying new safety issues is at odds with recent legislation promoting the replacement of traditional RCTs with real-world data analyses to help support the approval of supplemental indications or the fulfillment of postapproval study requirements.⁶ Our study appears to show that relatively few DSCs were initiated by real-world data.

This study has limitations. First, the source of the initial safety signal triggering DSCs might not have been clearly stated or represent all safety data reviewed by the FDA. Second, the present study was limited to initial safety signals. Inclusion of secondary or confirmatory safety signals may have represented different information sources.

The FDA Adverse Event Reporting System and RCTs triggered most DSCs and were associated with shorter time from initial approval to FDA issuing DSCs. These data might inform clinicians and regulators of the current state of postmarketing monitoring of toxic effects and assist in guiding future improvement efforts.

Accepted for Publication: June 8, 2019.

Corresponding Author: Daniel Shepshelovich, MD, Medicine A, Beilinson Hospital, Rabin Medical Center, 39 Jabotinsky St, Petah Tikva, 49100, Israel (shepshelovich@yahoo.com).

Published Online: September 3, 2019. doi:10.1001/jamainternmed.2019.3066

Author Contributions: Drs Tau and Shepshelovich had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Tau, Gafter-Gvili, Shepshelovich.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Tau, Shepshelovich.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Shochat, Gafter-Gvili, Amir, Shepshelovich.

Administrative, technical, or material support: Tau.

Supervision: Tau, Gafter-Gvili, Tibau, Shepshelovich.

Conflict of Interest Disclosures: Dr Amir reported personal fees from Genentech/Roche, personal fees from Apobiologix, personal fees from Agendia, and personal fees from Myriad Genetics outside the submitted work. No other disclosures were reported.