Building a Better Mousetrap: Developing Models for Cross-enterprise Image Sharing
When researchers at Wake Forest University Baptist Medical Center in Winston-Salem, N.C., undertook phase one of a two-part National Institutes of Health/National Institute of Biomedical Imaging and Bioengineering (NIH/NIBIB) research project focused on cross-enterprise image sharing, they uncovered a nasty, universal truth. “No one likes CDs—this includes physicians, patients and staff,” says Yaorong Ge, PhD, assistant professor of biomedical engineering.

The primary cross-enterprise image sharing mechanism—the lowly CD—consumes an inordinate amount of physician and staff time, often doesn’t include the necessary viewing tools, presents security risks and fails to provide an adequate framework for historical comparison in many cases. Health Information Exchange (HIE)-based image sharing offers a partial solution, but isn’t nationally scalable. HIEs also are plagued by growing privacy concerns, with some facilities foregoing or reconsidering participation.

Federal agencies have recognized the universal nature and magnitude of the problem and are investing in a pair of research projects designed to craft a next-generation solution. NIH/NIBIB awarded Wake Forest Baptist a $2 million, two-year grant to investigate the image sharing challenge and develop a patient-controlled, scalable solution geared to address the needs of smaller and rural providers. NIBIB also granted RSNA $4.7 million to spearhead a two-year pilot project and develop an internet-based network for sharing of patient images in partnership with five academic medical centers —Mount Sinai School of Medicine in New York City, the Mayo Clinic in Rochester, Minn., the University of California, San Francisco, the University of Chicago and the University of Maryland in Baltimore.

The status quo: Failure points abound

But how problematic are CDs? According to a January survey published in the Journal of American College of Radiology, more than half of sites reported that up to 10 percent of patients underwent repeat studies because of media problems. However, in many cases, the physician needs prior images to adequately evaluate the patient, which means the patient must be rescheduled. Both scenarios disrupt productivity for the physician and patient.

The process may be flawed even when the CD is fully loaded with a complete image set. That’s because the image viewer forces the physician into an unfamiliar workflow, or the tools needed for adequate comparison may be unusable or unavailable. “The end results range from lost productivity derived from trying to view images or, more importantly, potential errors associated with nonstandard viewers,” explains J. Jeffrey Carr, MD, professor of radiologic science at Wake Forest Baptist. One of Carr’s orthopedic colleagues estimates that it takes up to 30 minutes per CD to view images on outside CDs if one accounts for all the steps involved. (The IHE has proposed a Basic Image Review profile to standardize the operations of viewers to reduce the challenges, but this is not yet widely implemented.)

Although developing HIE image-sharing mechanisms offer a partial fix, they do not address all clinical scenarios as patient care often crosses HIE boundaries. “One of the biggest challenges [of cross-enterprise image exchange] is observing proper protection of patient confidentiality,” explains Bradley Erickson, MD, PhD, co-director of the radiology informatics laboratory at Mayo Clinic. HIEs bypass the challenge to a degree with a Business Association Agreement (BAA), which allows fairly free exchange of information among groups of organizations, such as members of an HIE.

However, as Erickson points out, it isn’t feasible to allow a nearly infinite number of BAAs. “BAAs make sense in high volume image exchange situations, but there is a limit to HIE scalability.” What’s more, according to Ponemon Institute’s Benchmark Study on Patient Privacy and Data Security, published in November 2010, HIEs expose healthcare providers to a host of privacy concerns such as unauthorized access, violation of data breach laws and potential identity theft.

Finally, in such organization-mediated solutions, providers do not know which images will be needed. For example, a person with a twisted ankle and negative x-rays is unlikely to require image sharing. However, a prior chest CT or brain MRI performed at another facility in the clinical settings of a potential stroke could alter patient management if available.

Simplified sharing

Wake Forest’s Personally Controlled Access Registry (PCARE) model bypasses the massive financial and IT investments required to build and maintain regional image sharing mechanisms. In the PCARE model, patients control authorization for every sharing event, which addresses privacy regulations. “We’re confident that large medical centers can develop a solution that works for them, but other imaging stakeholders such as public health clinics, rural hospitals and outpatient imaging facilities generally don’t have the same IT infrastructure or support,” explains Carr.

Wake Forest Baptist’s Personally Controlled Access Registry (PCARE)
A PCARE Image Sharing Network consists of One PCARE Master Server and many PCARE Facility Servers—one for each healthcare facility.

PCARE acts as a broker for permission and transport of images. How does it work? After a provider joins the network, a card reader is placed at patient check-in. A patient swipes a PCARE identification card and authorizes sharing and transfer of specific image sets to providers who may require them. Prior to a cardiac study, the patient might grant future access to a primary care physician and cardiologist. This patient-mediated authorization addresses Health Insurance Portability and Protection Act (HIPAA) requirements.

Edge servers connected to the card readers and PACS transfer patient-authorized images to the PACS of the receiving provider. Thus, problems associated with navigating an unfamiliar viewer are avoided. Ge and colleagues are in the process of testing the integration pieces of the system; they are focusing on standards-based solutions and using the Integrating the Healthcare Enterprise Cross-enterprise Document Sharing (IHE XDS-i) profile to ensure smooth transfer of images between facilities.

“Some of the major problems with regional image sharing are related to the Master Patient Index,” explains Carr. PCARE actively solicits patient participation to overcome integration challenges. In PCARE workflow, patient registration doubles as a cross check. When a patient swipes a PCARE card at any site, the patient’s medical record number at that organization is linked to the patient’s master PCARE identification number and any name variants are added to the account. After a patient authorizes image transfer, the system replaces the originating site’s medical record number with the recipient’s. This enhances the ability of the healthcare team to rapidly compare the prior imaging studies at the outside facility to any future imaging studies.

Like the PCARE project, the RSNA Cross-enterprise Image Sharing Initiative relies on patient control of imaging data. The RSNA project leverages existing commercial personal health record (PHR) software and provides a variety of patient-directed image sharing mechanisms. Imaging data and reports are transmitted from the originating hospital to a temporary store. At a later time, the patient logs into his PHR with an identification provided by the hospital, which allows him to pull the examination from the temporary store to the PHR. After that:

  • The patient can log into the PHR and direct images to a specific hospital or provider;
  • The patient can pull up the PHR at a hospital or provider and share access; or
  • By summer 2011, a planned initial deployment of software will allow hospital-to-hospital sharing from PHRs in emergency situations.

Receiving providers can view images within the PHR viewer, which typically suffice for straightforward image viewing functions, Erickson says. More complex image viewing processes require import into the PACS or 3D viewer. In these cases, the hospital has to determine how to handle foreign identifiers. The IHE Import Reconciliation Workflow provides a mechanism to facilitate consistent patient and procedure exchange while maintaining data integrity.

The five RSNA subcontractors expect to begin sharing with selected local hospitals imaging by April.  

Both PCARE and the RSNA pilot are works in progress with a few remaining hurdles. “We want to move reports as well as images. DICOM is highly interoperable, but HL7 isn’t,” shares Erickson. Another challenge is determining physicians’ responsibilities once reports are accepted.

Ultimately, the writing is on the wall. Standards-based image sharing solutions will improve current mechanisms and should deliver an array of benefits—reduced costs, accelerated patient care, fewer duplicate studies and improved physician productivity and satisfaction.