Disaster Preparedness and HIEs

Christopher Henkenius is program director of Nebraska Health Information Initiative (NeHII), Omaha, Neb. He describes the HIE as the foundation for multiple functionalities— clinical information from multiple sources, as well as connections to the state, connections to the patients, and ordering and imaging capabilities—that many clinicians in the network rely on. “We’ve got doctors at NeHII who would rather use the HIE as opposed to their own EMR,” he says. “If we don’t have disaster recovery processes in place and business continuity processes in place we are significantly impacting those physicians who use NeHII.”

NeHII also contracts its technical services with Optum, and the agreement requires that the vendor has sufficient backup and recovery procedures, Henkenius says. He notes that the vendor has dual data centers with clustered servers and multiple communications vehicles.

Henkenius calls NeHII a hybrid federated model. The HIE takes a copy of each hospital’s stored data, which are backed up on a regular basis. If a server corrupts, the data fails over to the backup copy, he says. “It’s no different than backing up data at a company or an EMR at a hospital. It’s just backing up the data and making sure you have redundant systems that can access the backup,” he says. Each of NeHII’s hospitals has its own edger server for this purpose, and Optum handles the backup, off-site data storage, and recovery processes.

From a backup and recovery perspective, the HIE needs to be sure its edge servers contain the data from each of the member hospitals, particularly the master patient index, the provider directory, application systems that allow physicians to view the data, a VPN connection that allows the information to flow between the hospitals and the HIE, and applications that run in the background that facilitate the dissemination of the information, he says.

HealthInfoNet, Portland, Me., is a centralized HIE model. It began its demonstration phase in 2008 and has been live since 2010. “Everything is centralized for us, with all of our systems within a couple of data centers,” says Todd Rogow, director of IT and the regional extension center. All of the data is HL-7 based, with close to 400,000 messages per day flowing into its system. Access to the services is provided through an internal portal via a VPN connection.

In Rogow’s view, the centralized model has reduced some of the complexity as one system with many vendor modules. He says the HIE and its member hospitals have a basically one-to-one relationship, where the main concern is to keep the data flowing and making sure the portal service is up. Essentially the HIE’s role is that of an aggregator. If a connection does go down, the messages are not lost, they are simply queued up until the connection is re-established, he says.

Rogow says the NeHII database is large, approaching 1 terabyte in size. Clinicians who enter the portal are able to pull up patient information across multiple unique enterprise IDs for that patient, and clinical information across 15 categories. “They always get the most real-time view, and it’s all done within five seconds,” Rogow says.

At present imaging data is not stored, but that will change. In May HealthInfoNet announced it would pilot the nation’s first statewide image archive, sharable within the HIE. The archive will be cloud-based and stored in three redundant data centers across the country, and will be managed by Dell. It will be separate from the existing data storage infrastructure for non-imaging data, Rogow says. “That solution has disaster recovery as a major component, and is a benefit to our customers, because this becomes a redundant system for them,” he says. The pilot is expected to conclude this fall and expand statewide in 2013. 

Rogow says that the HIE is typically the secondary system to the hospital’s primary EHRs, and has been used as a backup on at least one occasion. A large hospital in the state had an outage, making its EHR go down for a period of time, although the data feeds were still in place to the exchange, allowing the hospital to use the exchange to get data on the patients. “That was a real benefit, and we became part of their disaster recovery plan in that respect,” he says.

He notes that HealthInfoNet works with local vendor (Systems Engineering, Portland) that provides services around security and disaster recovery. It provides incremental backups of its databases and systems, going back six weeks in time. “That is something that is always running and backups are always being generated,” he says. If the need arises, data can be backed hourly, daily or weekly, he says. It keeps a digital copy of data going back a week in its own data center, and falls back to a tape backup for an additional five weeks. The tapes are kept at a secondary site about 100 miles from the primary site.

HIEs as a Basis for Regional Cooperation

All of those interviewed say that disaster preparedness was a factor in the choice of the HIEs architecture. One advantage of HIEs is that they offer a communication infrastructure as well as storage infrastructure that providers can fall back on; and, as Rogow notes, the data-sharing made possible by HIEs presents a unified story of patient data. As the ONC has recognized, those same capabilities make HIEs a powerful tool for sharing data across state lines in the event of a disaster.