Sociodemographic Characteristics of Adverse Event Reporting
Introduction
Postmarketing surveillance is an important way to identify drug-related adverse reactions (side effects) that may not be observable in the clinical trial setting because it is a relatively controlled environment. Compared to participants in clinical trials, real-world patient populations are generally more demographically diverse and are more likely to have additional medical issues and take other medications. In addition, patients often use FDA-approved medications for longer time periods than participants in a pre-approval clinical trial, allowing experts to detect adverse reactions that take a while to develop. Plus, there are a lot more patients in the real-world setting, increasing the chance for rare safety concerns to emerge.
Once a drug is approved, individual patient adverse events are the most common source of information that drug sponsors use to update safety information in drug labeling.1 Recognizing the value of postmarketing surveillance for protecting public health, FDA manages the FDA Adverse Events Reporting System (FAERS), a repository of adverse event reports associated with drugs and biologics. The public can submit reports directly to FAERS through the MedWatch program or to the product manufacturer, who is required to submit the report to FAERS. While FAERS can provide important drug safety insights, its voluntary nature may result in under-reporting. There is also not much information about how sociodemographic factors (such as sex, age, income, race and ethnicity) may influence people’s likelihood of reporting. If different sociodemographic groups are not reporting at similar rates, the surveillance system may not detect safety issues affecting the under-reporting groups. In turn, this could exacerbate the pervasive health care disparities in the U.S.
Methods
For this reason, researchers in CDER’s Office of Surveillance and Epidemiology and other experts sought to examine the sociodemographic characteristics of individuals who submit FAERS reports.2 In this study, the researchers extracted reports that consumers (e.g., patients, caregivers) submitted to FDA through MedWatch. The researchers evaluated this subset of reports because it has detailed information about individuals’ geographic location. The investigators then organized the FAERS reports by the county where each individual lived and aggregated (added) the totals per county.
Using these data, the researchers calculated a reporting rate per 100,000 residents for each county and categorized the reporting rates into quartiles, or four sections. They then compared these reporting rates across counties. Their goal was to see if sociodemographic factors and health care access — age, sex, race, ethnicity, highest level of education completed, English language proficiency, insurance and health status, health resources, income, rural/urban living, and prescription drug use — within a county influenced reporting rates. For example, did counties with a higher proportion of residents living in rural areas have lower adverse event reporting rates than counties with a lower proportion of people living in rural areas (and by how much)?
The investigators performed a sub-analysis of Texas, comparing the counties’ reporting rates with respect to sociodemographic factors. They chose Texas because of its large size and relatively large number of residents with limited English proficiency (defined by the Census American Community Survey as individuals over five years old who reported speaking English less than “well”). The researchers also conducted a nationwide multivariate analysis (where they evaluated multiple factors to find associations) to evaluate all the previously mentioned sociodemographic factors while controlling for the number of dispensed prescriptions per person.
Results
The results showed that reporting rates varied significantly across the U.S., with counties ranging from less than 5.5 reports per 100,000 residents (the lowest quartile) to more than 17.6 reports per 100,000 residents (the highest quartile) over the five-year time span of the study.
Counties in the highest reporting quartile had significantly lower proportions of Hispanic or Latino residents, American Indian or Alaska Native residents, and individuals living in rural areas within the county. Counties in the highest reporting quartile also had higher proportions of White residents, higher household incomes, more primary care physicians and mental health providers, and more prescriptions dispensed per resident. When the researchers excluded counties with fewer than 20,000 residents, these patterns were consistent. Another trend also emerged: the highest reporting counties had a lower proportion of Black or African American residents.
Regarding the Texas sub-analysis, counties along the U.S.-Mexico border were commonly in the lowest reporting counties. Relative to the rest of the state, these counties generally had the highest percentages of Hispanic or Latino individuals, highest percentage of people with fair/poor health, lowest percentage of residents with English proficiency, and lower household incomes. The northernmost counties of Texas, located between New Mexico and Oklahoma, were also generally among the lower reporting counties. These counties typically had higher proportions of residents living in rural areas and residents without health insurance.
The researchers found trends in the nationwide multivariate analysis as well. Counties with higher reporting rates had greater proportions of females, individuals aged 65 years and older, mental health providers, and individuals without health insurance (an inconsistent finding with the Texas sub-analysis results). They also had higher household incomes. Counties with lower reporting rates had a higher percentage of individuals aged 18 years or younger, American Indian or Alaska Native residents, individuals with limited English proficiency, and those living in rural areas. The number of dispensed prescriptions per county resident had no association with reporting.
Discussion
This research shows that sociodemographic factors are associated with different levels of adverse event reporting. Specifically, the study highlighted that adverse events in some populations may be under-reported. This may affect the ability of FAERS to detect certain safety signals in specific populations. These findings are concerning because many of the under-reporting populations are vulnerable and already subject to health care disparities. These populations may either be unaware of the surveillance system or unsure of how to use it, which may be particularly true for people with limited English proficiency and people without access to health care professionals to inform them about the system (which include individuals living in counties with high rural populations and low household incomes, which also tend to have fewer health care professionals per resident).
This study showed that adverse event reporting data, while valuable, may not fully capture the experiences of the entire population using the drug. It is important that we continue to help populations whose adverse events are under-reported understand the value of the surveillance system. CDER encourages all individuals who experience drug-related adverse events to report them to MedWatch to help the agency detect safety signals and, ultimately, help improve medication safety. This webpage has information for health care professionals, patients, and consumers, including information and a reporting form in Spanish.
Other Resources
- Reporting Serious Problems to FDA
- MedWatch Online Voluntary Reporting Form
- Questions and Answers on FDA’s Adverse Event Reporting System
- MedWatch Tips and Tools
- Postmarket Drug Safety Information for Patients and Providers
- 1Croteau D, Pinnow E, Wu E, et al. Sources of Evidence Triggering and Supporting Safety-Related Labeling Changes: A 10-Year Longitudinal Assessment of 22 New Molecular Entities Approved in 2008 by the US Food and Drug Administration. Drug Saf. 2022.Feb.doi: 10.1007/s40264-021-01142-3.
- 2Munoz M, Dal Pan G, Wei YJ, et al. Sociodemographic Characteristics of Adverse Event Reporting in the USA: An Ecologic Study. Drug Saf.2024. Feb: doi.org/10.1007/s40264-024-01397-6.