One of the biggest research changes many health systems and universities are facing involves building an enterprise data warehouse that integrates data from various institutions. Its potential has grown in recent years with the growth of electronic health records.

Increasingly, hospitals and other institutions have the ability to share data outside of their own networks. This opens up data-driven research potential, such as personalized medicine, but also poses more problems for IRBs.

“What we’re doing here is complicated and could be an IRB nightmare,” says Michael Ames, MBI, associate director of a data warehouse called Health Data Compass at the University of Colorado, Denver. The university is building an enterprise data warehouse that integrates data from the University of Colorado Health System and area hospitals. All of the institutions work with the Colorado Multiple Institutional Review Board (COMIRB).

“Our objectives are to facilitate a wide range of research types,” Ames says. “We’re concerned with HIPAA, basic science, clinical research — a broad range of things that could present an ongoing nightmare of need to our IRB.”

Fortunately, developers of the enterprise data warehouse have worked to anticipate IRB issues and concerns. Their goal is to reduce IRB burden through development of a protocol that acknowledges how Compass will be used. For instance, the protocol outlines waiver of authorization and other nuanced issues like the appropriate use of prisoner data, he says.

IRB leadership has provided guidance on how to craft the protocol and avoid pitfalls, Ames says.

“The more we can answer questions upfront and clearly delineate these in our protocol, the less the IRB will have to field and deal with those kinds of questions [during research involving the data warehouse],” Ames says.

“The protocol will minimize the way new work is brought to IRBs,” he adds. “This broad protocol highlights areas for which they do not need IRB involvement, reducing the amount of time it takes people to approach the IRB for Compass-related activities.”

Researchers can use de-identified data sets without a requirement for any additional IRB input, for instance.

The protocol will delineate all of Compass’ research purposes: “Here is a list of research activities that could be fully covered under this protocol; here are activities that will require you to go to the IRB,” Ames offers as an example.

The very large database will support research activities that will be covered under the protocol and not require additional work by the IRB, Ames says. “We have policies we’re implementing to try to steer people to that solution whenever possible.”

Developing an enterprise data warehouse is a complicated project because of the different structures of independent entities involved — even when they share a headquarters, as is the case with the collaboration between the University of Colorado and two hospital systems, Ames says.

“We have not collaborated very well, historically, in sharing data for research purposes,” he explains. “This is a project five to 10 years in the making.”

The consortium was begun with the goal of funding a single central data warehouse to support clinical research, molecular discovery, personalized medicine, and operational excellence, Ames adds.

“We will be integrating the fully-identifiable medical record data from both hospital systems,” he says. “We are taking millions of patient records and integrating them in this warehouse.”

Labs are conducting molecular testing for cancer and other diseases with the goal of optimizing treatment paths. The objective of Compass is to greatly expand knowledge of treatments. The data can be linked back to patients’ medical records, making it possible for researchers to discover and create personalized medical treatments, Ames says.

Without the collaboration and data warehouse, the information necessary to make these medical breakthroughs is siloed and unavailable to researchers, he notes.

“Some lab is holding on to years of these very detailed analyses of patients’ genetic profiles, and hospitals are amassing records about other tests from that patient, but those tests live in silos,” Ames says. “Compass is providing the infrastructure that will combine those together, a crucial intersection of data to allow researchers to make new discoveries about how a person’s genetic profile can influence treatment.”

The alternative to a consortium like Compass is for individual researchers or research programs to establish their own data repositories. Each data repository would require some IRB involvement, creating more work for IRBs and research organizations, Ames notes.

“As a consortium we are going to do this one time, and we’re going to do this really well and really thoroughly, leveraging the work we’ve already done,” he says. “In the end, we’re reducing the number of these kinds of resources that have to be created, and it reduces the amount of effort needed by an IRB.”