The FDA’s Sentinel System, coordinated by the Harvard Pilgrim Health Care Institute, brings together many of the nation’s largest insurers and health systems to create a large distributed database with access to primarily claims data. The Sentinel System was developed to complement existing FDA surveillance capabilities that track adverse events reported after the use of FDA regulated products by allowing the FDA to proactively assess the safety of these products. The initial steps to creating the Sentinel System involved a pilot, Mini-Sentinel, to test the feasibility of and develop the scientific approaches needed. With the increasing availability of electronic healthcare data, there are opportunities to leverage the Sentinel Infrastructure for other purposes, including creating efficiencies in clinical trial conduct.

Between October of 2012 and December of 2013, FDA, Harvard Pilgrim Health Care Institute and CTTI formed a multi-stakeholder project team to assess the potential for conducting randomized trials in the Mini-Sentinel environment, specifically the capability and willingness of participating Mini-Sentinel data partners to collaborate in randomized trials. This project identified potential opportunities to utilize the database for clinical trial operations, such as protocol development, trial implementation, assessing outcomes, and identifying and enrolling trial participants. Detailed results from the project “Uses of Healthcare Electronic Data to Support Clinical Trials (with Mini-Sentinel)” are described in the findings section below and in the Developing Approaches to Conducting Randomized Trials Using the Mini-Sentinel Distributed Database report and webinar.

Building on the findings of this project, work is now underway to initiate an interventional trial as a proof-of-concept study conducted utilizing the Sentinel Infrastructure as a part of the FDA-Catalyst program. The FDA-Catalyst program seeks to provide FDA the ability to use the Sentinel Infrastructure to access existing data and complement this data with either interactions or interventions with patients or providers to answer questions about medical products performance. This is in contrast to the Sentinel System, which only utilizes the Sentinel Infrastructure to access existing data to inform questions about medical products performance (i.e., observational studies).

IMPACT AFib is a proposed randomized trial to test the effect of education on initiation of anticoagulant use in patients with atrial fibrillation meeting guidelines for stroke prevention who have not received such therapy in the preceding year. In addition, the trial will assess the outcomes associated with the treatment (i.e., rate of stroke-related hospitalizations). This trial was selected for several reasons:

1. Increased stroke risk among patients with atrial fibrillation is an important public health issue that can be reduced by use of oral anticoagulants, when appropriate.
2. The electronic data needed for this trial includes inpatient and outpatient claims data for clinician encounters, laboratory tests and pharmacy dispensing. As such, this trial will test the integration of these different data and their utility to researchers.
3. The size of the trial will appropriately test the feasibility of using of claims data from patients enrolled in different health plans and across all 50 states.

Uses of Electronic Healthcare Data to Support Clinical Trials (with Mini-Sentinel)” (2012-2014)

Preparatory work and support implementing IMPACT-AFib (2015-ongoing)

• To understand the feasibility of using electronic healthcare data to facilitate the conduct of clinical trials and create efficiencies
• To plan and execute an interventional randomized clinical trial to assess the impact of electronic healthcare data to facilitate the conduct of clinical trials and create efficiencies
• To create reusable, generalizable methods that could be leveraged for other interventional trials utilizing electronic healthcare data

To demonstrate that the use of the electronic healthcare data, such as that contained within the Sentinel Infrastructure can reduce costs and save time during the conduct of clinical trials.

• Convene a collaborative workgroup to evaluate the capability and willingness of Mini-Sentinel data partners to use the Sentinel Infrastructure to participate in randomized trials (completed)
• Convene a collaborative workgroup to produce an executable trial protocol, including statistical analysis plan; interventional materials; infrastructure (e.g., dedicated suite of programming tools) to support conduct of trial; and a detailed implementation workplan and timeline for activities.
  • CTTI’s social science experts assisted with the development of the intervention materials through outreach with representative samples of physicians and patients
• Convene a collaborative workgroup to implement and evaluate the impact of the IMPACT-AFib trial using the Sentinel Infrastructure.

Findings from the first project defined the characteristics of a trial that would make the Sentinel Infrastructure a favorable platform for facilitating trial implementation. These characteristics include a broad study population, commonly available clinical outcomes, and easily applied interventions, among others. In addition, the project identified that data partners indicated various factors as important to a decision to participate in a trial using the Sentinel Infrastructure, including staff capacity, topic of research, and ethical considerations, among others. As the report describes, data partners were optimistic about using the Sentinel Infrastructure to facilitate patient recruitment and follow-up of participants in randomized clinical trials and have considerable and relevant experience to use the Sentinel Infrastructure to conduct randomized trials.

The report further considered potential use cases that could demonstrate the use of the Sentinel Infrastructure in the conduct of a clinical trial, including the IMPACT-AFib trial. Work is underway to finalize the materials needed to implement this trial. However, initial feasibility assessments of the data within the Sentinel Infrastructure indicate that the data partners have a sufficient number of eligible patients to power a trial to meet the study objectives, namely whether an intervention can increase the proportion of patients started on oral anticoagulation for stroke prevention and reduce the rate of stroke related hospitalizations.

Watch this space for updated findings as the project progresses.

• Mini-Sentinel: http://www.mini-sentinel.org/
• http://www.fda.gov/Safety/FDAsSentinelInitiative/ucm2007250.htm