Background: Approval of new treatments is usually based on evidence from randomized clinical trials (RCTs) that provide information about benefits and risks. Randomized trials are generally considered the best way to measure these benefits and risks; however, these trials usually do not provide enough data for fully informed treatment decisions. For example, patients are often observed in trials only for a short period of time. Outcomes that occur after the end of the trial are not captured.

One way to enhance the information generated from randomized trials is to link the trial data with insurance claims data reflecting participants’ routine encounters with the healthcare system. These administrative data are a valuable source of information for researchers because they enable the evaluation of medical treatments and subsequent health outcomes, even if these occur outside of a randomized trial.

In this study, data from patients who participate in a randomized trial will be linked to insurance claims data. By making this linkage, we will be able to investigate, for example, whether treatment reduces hospitalizations even after the end of a trial, or whether treatment causes long-term adverse events that are not captured in the trial.

This project will serve as a model that demonstrates the feasibility and value of these types of linkages for generating outcomes that are important for patients. We envision that such data linkage will become a routine part of future RCTs.

Study Aims: The study aims to do the following:

• Assess the added value of linking RCT and claims data.
• Develop methods to determine sources of discrepant results between randomized clinical trials and claims-based observational studies.
• Develop an approach to expand randomized clinical trial results to broader patient populations.

Study Description: We will use data from an eight-week randomized trial evaluating a novel drug for heart failure. The trial is currently under way and is expected to end in Summer 2016. During the last follow-up visit, patients will be asked if they consent for linking their trial data to their insurance claims data.

We will assess the added value of linked data by evaluating the complementary role of information available from the trial and claims data. We will examine the extent to which the impact of treatment on short-term trial outcomes translates to long-term patient-centered outcomes identifiable in the patients’ claims data, and we will examine the concordance between outcomes that are captured in both the trial and claims data (e.g., heart failure hospitalizations, myocardial infarctions, strokes). The sources of discrepancies between trials and claims-based studies will also be investigated. In addition, we will estimate the extent to which treatment effect varies based on patients’ characteristics and will develop an approach to expand the results of randomized clinical trial to broader patient populations.