Talking with Robert Ball, MD, MPH, who is the deputy director of the Office of Surveillance and Epidemiology, CDER.
Ensuring the safety of regulated drug products is a major part of FDA’s mission. The Sentinel Initiative is a national electronic system that monitors the safety of drugs, vaccines, biologics and medical devices on the market. Started in 2008, the program has moved beyond the pilot stage into a system that works with 17 data partners , such as insurance companies, HMOs, and hospitals, with access to 223 million members. Robert Ball, deputy director of CDER’s Office of Surveillance and Epidemiology, discusses Sentinel’s progress and future.
What is the Sentinel Initiative?
In May 2008, the U.S. Department of Health and Human Services announced the launch of the Sentinel Initiative , which directed the FDA to create the Sentinel System, a national electronic system for medical product safety surveillance. The Sentinel System includes the Active Risk Identification and Analysis (ARIA) system , which was mandated by the FDA Amendments Act (FDAAA) of 2007. FDAAA required the creation of a drug safety surveillance system with data on at least 100 million people by 2012. FDA exceeded this mandate and continues to update the database. This system complements existing FDA surveillance capabilities that track adverse events of FDA-regulated drug products.
In addition, the Sentinel Initiative created focused surveillance efforts around vaccine safety using the Postmarket Rapid Immunization Safety Monitoring (PRISM) system , and it supports regulatory review of blood and blood products with its Blood Surveillance Continuous Active surveillance Network (BloodSCAN) . The Sentinel Initiative also contains FDA-Catalyst, which leverages the Sentinel infrastructure, and supplements it with data from interventions or interactions with health plan members and/or providers.
In 2016, FDA activated the Sentinel System and officially integrated it into FDA’s routine drug safety operations. All FDA centers involved with medical products are now actively using the Sentinel System to monitor the safety of drugs, vaccines, biologics, and medical devices.
How is the Sentinel System different from other systems?
The Sentinel System is different from other systems in important ways. First, it uses a distributed data approach in which data partners maintain physical and operational control over electronic data in their existing environments. This is achieved by using a standardized data structure, called the Sentinel Common Data Model . Distributed data networks allow secure access to multiple data sources to achieve far larger sample sizes than could ever be achieved at a single source. More importantly, they allow data to be aggregated securely and in a manner that safeguards patient privacy.
An additional feature of the Sentinel System’s distributed data approach is the use of Routine Querying Tools , which are designed to run against the Sentinel Common Data Model. They allow for rapid queries across all data partners. These reusable programs increase the speed and volume of safety analyses by eliminating the need for customized programming for each query. These tools can perform both simple descriptive analyses and more complex studies.
How does the structure of the Sentinel System protect privacy?
All data are safeguarded by staying within the firewalls of the original data holders. The only data that is transmitted to the Sentinel Operations Center is what is minimally necessary to conduct the safety analysis. Patient privacy is protected by removing all personally identifiable information and using datasets that have been aggregated into groups of people, rather than individuals (e.g., the number of people with asthma aged 18 to 45 years of age who were prescribed albuterol).
How does FDA use the Sentinel System during the review of new drug applications?
FDA conducts a comprehensive assessment of a drug’s safety data before approving it. In this evaluation, reviewers identify serious safety concerns and develop an action plan for safety monitoring after approval. This action plan may include labeling, Risk Evaluation and Mitigation Strategies (REMS), or post-market studies (e.g., phase IV studies). If serious safety concerns prompt the need for post-approval monitoring, FDA assesses whether the Sentinel System’s ARIA system can meet the specific safety monitoring need. If it can, the agency begins planning for a future ARIA study. If ARIA is not sufficient to meet the safety monitoring need, then the agency can require a sponsor to conduct a postmarketing safety study. In this way, Sentinel’s ARIA system is an integral and strategic part of the regulatory planning process to monitor serious safety concerns identified in the review of new drug applications.
What other ways may the Sentinel System be used, and how are results disseminated?
In addition to considering the Sentinel System during the review of new drug applications, FDA also may use it to examine safety questions identified after a drug is approved, along with other sources like the FDA Adverse Events Reporting System.
FDA routinely conducts feasibility assessments to determine when it is possible and appropriate to use the Sentinel System. These feasibility assessments are based on many considerations, including whether Sentinel can capture the drug and health outcome of interest, whether the sample size is sufficient to make an accurate inference, and whether enough additional supporting data (e.g., laboratory or radiologic information) exists to answer the question. The Ongoing Assessments Page shows the drugs and the health outcomes of interest that FDA is currently studying. Although the Sentinel System can answer many important safety questions, it cannot answer all of them.
To ensure transparency, all results from Sentinel System assessments are posted on the Sentinel Initiative website, along with details about how the studies are conducted. These results are also published in peer-reviewed scientific journals to ensure that the methods and results of each FDA study are critically evaluated and can be reproduced by other researchers in the field.
How are the data collected in the Sentinel System? What are the strengths and limitations of claims data?
The data in the Sentinel System are primarily derived from administrative claims databases that healthcare insurers use for reimbursement for diagnoses, procedures, and medications. Unlike clinical trial data, claims data are not derived from customized data collection instruments (e.g., telephone surveys, patient diaries), or from interventions within the medical system (e.g., using pre-defined follow-up visits).
Claims data are useful because they can allow investigators to observe healthcare across different settings. For example, the Sentinel System can follow a patient starting with his or her initial outpatient visit, to the Emergency Department, throughout a hospitalization, when he or she is discharged home, whether a medication is prescribed, and through a follow-up visit one week later. Unlike electronic medical records, which only capture information within a single clinical setting, claims databases produce a longitudinal record that allows for complex safety analyses to be done.
However, claims data have limitations. Only information about reimbursed healthcare is captured. Sentinel does not include information about visits to free healthcare clinics, drug samples given in physician’s offices, use of low-cost generic medications that do not incur an insurance co-payment, and purchase of over-the-counter medications.
Some say Sentinel lacks important data and has not yet found any safety issues. How does the agency respond to this criticism?
There are many ways to think about the value of the Sentinel System. First, it can provide reassuring evidence that a medical product does not appear to be associated with an adverse health outcome, which allows the agency to conclude that the prescribing information adequately describes the risks and benefits. Second, the Sentinel System is only one part of many sources of evidence used to make decisions. Third, the Sentinel System can contribute important evidence for labeling actions, responses to Citizens Petitions or Advisory Committees, and many other actions that are not easily quantified or visible to the public. Fourth, the Sentinel System has only been fully operational for about 18 months. Prior to 2016, the focus of the Mini-Sentinel pilot was on developing the infrastructure and methods that would allow FDA to take full advantage of the Sentinel Distributed Database. All the benefits of this important new source of safety information will not be evident for several years. Lastly, FDA’s current focus is on use of the Sentinel System for evaluating safety signals as required by the law that created the Sentinel System. As capabilities for signal identification in the Sentinel System mature, it will have a greater role in finding drug safety issues to supplement FDA’s already robust safety issue identification efforts.
Who can access Sentinel resources?
The Reagan-Udall Foundation’s Innovation in Medical Evidence Development and Surveillance (IMEDS) program allows public and private entities access to the Sentinel System. In addition, the Sentinel Common Data Model and tools for querying the data are publicly available on the Sentinel Initiative website. FDA has also offered training for the public on the Sentinel System.