Interim Findings

The Collaboration is expanding to enable a broader set of organizations—both within the public and private sectors—to participate in the initiative. The Collaboration is inviting national, state and local organizations, including provider organizations, health plans, health IT vendors, and health information exchange organizations to run the three use cases in a research environment and share their experiences, to enable further assessment and refinement of the methods for both data collection and interpretation of the results.

The following summarizes key interim findings based on the first iteration of testing of the three use cases with Partners HealthCare System and the Regenstrief Institute.

#	Interim Findings	Impact
1	It is feasible to use electronic health information from sources at the community level—including clinical data alone, claims data alone, and clinical and claims data together--to detect adverse events in the context of drug exposure.	National efforts related to drug safety and post-market surveillance can be bolstered by the use of electronic clinical health information residing within the care delivery system—a majority of which is ordinarily found at the local level.
2	It is feasible to use electronic health information at the community level-- including clinical data alone, claims data alone, and clinical and claims data together--to detect designated medical events.	National efforts related to drug safety and post-market surveillance can also be bolstered by the use of electronic clinical health information residing within the care delivery system—ordinarily found at the local level.
3	Experts residing in multiple environments vary in their approaches to definitions related to post-market surveillance due to availability of data, local expertise, and interpretation of the literature. To enable consistency, a distributed model of safety surveillance, involving more than one data source, requires agreed-upon definitions for cohorts, including both inclusion and exclusion criteria.	National efforts related to drug safety and post-market surveillance should leverage experts in epidemiology, safety and informatics, as well as organizations who are actively running surveillance queries and similar research questions, to develop agreed-upon definitions that can be employed in any model involving more than one data source.
4	Analysis of the impact of the use of different types of data (clinical and claims) for detecting known adverse events revealed that the use of claims data alone for such analyses resulted in limitations in identifying events necessary for cohort definition.	While the use of claims data has been effective in identifying drug safety signals, the addition of clinical electronic health information can refine the precision and specificity with which such signals can be evaluated. It is critical that national efforts related to drug safety and post-market surveillance leverage both claims and clinical data sources to support drug safety signal detection and evaluation.
5	Running the research questions in two separate research environments, and the sharing of both the methods and the results, resulted in the refinement and improvement of both methods and interpretation of results.	Reliance on one research environment to conduct post-market surveillance may limit national efforts related to drug safety and post-market surveillance. The level of knowledge in post-market surveillance can be considerably increased and methods can be further refined and improved by supporting or creating an environment where drug safety signal detection and evaluation activities can be run in multiple environments, and both methods and findings can be shared to stimulate improvements in the nation’s ability to detect and respond to drug safety issues.