Radiology is faced with a market paradox. Even as the radiologist shortage continues, the U.S. market for medical imaging equipment is projected to grow at a compound annual growth rate of 8.1 percent through 2012.1 This growth will further drive remote radiology reading and teleradiology services, which have already gained a strong foothold thanks to continued advancements in Web-based technologies.
Yet, even as Web technology spurs the growth of teleradiology, many radiology practices find themselves in the midst of a new dilemma: namely, the creation of a multi-PACS reading environment.
You would think that the existence of search engines such as
Google with the ability to aggregate an infinite amount of data would imply that sophisticated data management would be ubiquitous. But the medical industry defies this logic. On the contrary, in the PACS arena, radiology departments are struggling to streamline the data needed to complete a patient report. A large part of the challenge is gathering images from disparate PACS.
A practice that reads off a different PACS client for each site they read for is referred to as ‘maniac’ PACS, according to Paul Nagy, Ph.D., associate professor, diagnostic radiology and nuclear medicine, University of Maryland Medical Center. “If you combine information from multiple, separate institutions onto the same PACS, you run into another problem. There is a significant amount of back-end work needed when importing to correct the data, such as matching to the medical record number for relevant priors, just so the radiologist can begin reading and compare them side-by-side,” said Nagy. He estimates the time to curate outside studies could easily run as high as 15-20 minutes to import a study from an outside institution.
This is compounded by the fact that a patient may have relevant priors at different institutions within the same health system, says Paul Chang, M.D., professor and vice-chairman, Radiology Informatics at the University of Chicago Pritzker School of Medicine and medical director, Enterprise Imaging, University of Chicago Hospitals. Frequently in a multi-institution system, there is no single archive or database, and the same patient may be represented by a different medical record number at each site. Often, radiologists navigate through patient data on different workstations and Web clients, each with a different user interface. “Yet the radiologist requires a single presentation for a comprehensive patient view to read and report each study,” said Dr. Chang.
The result is an inefficient workflow. Nagy does not believe that portable media such as CDs are an answer either. “RHIOs have tried to use CDs to exchange and share data, but so often the CDs have interoperability issues. Plus, the embedded viewers on the CD still means the radiologist has to look at the images separately.” A key problem, he notes, is that there is no standard usability across different vendors’ PACS. “This environment forces the radiologist to use different user interfaces, which places a cognitive drain on the clinician,” he said.
Other challenges include identifying disparate medical records for the same patient, noted Dr. Chang. “The technology exists to solve this problem, such as the master patient index,” he added. “Yet, the problem remains because there is no universal patient record identifier.” So there is a human factor for bringing together records for the same patient who may be listed differently, such as John Doe, J Doe and Jonathan Doe. “This dependence on manual association of multiple medical record number identifiers into the master patient index is a significant problem,” Dr. Chang said.
Both Nagy and Dr. Chang point to existing IHE standards, such as the cross enterprise document sharing (XDS) profile for distributing electronic medical records among healthcare enterprises. Even with this profile, Dr. Chang indicates that many vendors have not yet embraced it. “There is no real incentive for current vendors to follow this,” he said.
Making Multi-PACS Work
Even with all the challenges confronting radiology practices reading for multiple sites with different PACS, several groups are finding some success. Yet, the common thread is that each site had to implement their own PACS for an efficient workflow model that enables the radiologist to read and report from a single workstation.
Five years ago, Henry Hollenberg, M.D., and a partner at Total Radiology Solutions (TRS) recognized the solution to reading studies for multiple hospitals with disparate PACS was to bring all the data together. He worked with BRIT Systems and the company’s Roentgen Files PACS to route all studies from 10 small- to medium-sized hospitals into a Linux (UNIX/AIX) server driving BRIT’s Linux Reading Workstation.
At each hospital, the modalities, HIS and RIS all communicate directly with the Linux (UNIX/AIX) server, which he calls the “traffic cop.” This “traffic cop” provides features to ensure uniqueness of the data and labels it so the PACS knows which facility it came from and where to return the radiology report.
In some of the hospitals, the Roentgen Files PACS acts as the primary PACS server and communicates directly with the HIS/RIS. For all others, Dr. Hollenberg says the hospitals use a nighttime reading model, where all images, orders and relevant priors are pushed to the server as a back- end or secondary system.
For redundancy, Dr. Hollenberg has two Linux (UNIX/AIX) servers backing one another. “The back-end part that makes this work is synchronization of studies at the PACS level,” he said.
In an effort to better synchronize studies, Jesse A. Salen, vice president of Sales and Technology for Online Radiology (Riverside, CA), implemented Centricity PACS-IW from GE Healthcare’s Dynamic Imaging Solutions Division. Salen set up a virtual private network (VPN) to each customer site so the remote reading service group could more efficiently read studies from 150 different sites. Their hospital client base ranges from rural 50-beds or less hospitals to Level 1 Trauma Centers.
Although sharing data in a HIPAA-compliant manner is one technical challenge, Salen says a larger issue is the fact that not all data adheres to the DICOM standard. For the 10 percent of the studies that have some proprietary data, Online Radiology uses DICOM routers and gateways to send clean data to its PACS. “The biggest challenge is consolidating and moving the data into our environment, so our radiologists have a single patient view,” Salen noted.
At Renaissance Medical Imaging Associates (RIMA) (La Canada, CA), the IT staff is preparing to install Carestream Health’s new solution called SuperPACS. Andrew Deutsch, M.D., president and CEO, says that in a complex multi-PACS reading environment, it is important to be able to do more than just move images around.
“At this SuperPACS level, we can balance workflow and re-route studies to gain efficiencies,” said Dr. Deutsch. Each site will communicate with RIMA’s SuperPACS to create one global worklist.
TRS also took over the worklist with its PACS. Says Dr. Hollenberg, “This allows us to manage workflow and maintain a high level of productivity by using one user interface.” All five radiologists in the practice read from the same worklist, regardless of where they are located. “In addition, we can QC the studies and patch multiple modality studies into one seamless presentation, including history sheets,” he explained.
Reporting is also an issue. “We have to coordinate reporting, then electronically distribute that final report back to the referring physicians,” said Dr. Deutsch. RIMA reads for five hospitals, owns and operates seven imaging centers and reads for one teleradiology practice.
Yet, Dr. Deutsch cautions that the SuperPACS model is not simply taking data from different PACS and ‘dumping’ it into another. It acts as an intelligent facilitator to communicate with existing PACS and maintain the integrity of the data as well as simultaneously satisfying security and privacy regulations.
At all three reading groups, a scalable, redundant solution is a necessity. Each group practice has a data center that can also be used to provide a disaster recovery solution.
Also common to each site is the ability to track studies as they
move through the system. Dr. Deutsch says a robust worklist that helps divide and balance the imaging load is crucial. Dr. Hollenberg suggests having a champion who understands the front end workflow and the back end technology. “Persistence and finding the right vendor who can make it work is the best bet,” he said. At TRS, the PACS also integrates with his billing company to automate this process as well, providing the full loop from order to charge.
“Don’t change to accommodate the PACS – have the PACS fit into your workflow,” said Salen. “Have a group representative be completely involved in the installation process.”
Political, Economic Challenges
Even though these three sites have found success, many reading groups find the multi-PACS environment hampered by interoperability issues and a lack of data sharing.
Dr. Chang believes the real issue is political and economic, not technical. “There are no real incentives for the hospital and the vendor,” he said. “The technical issues are understandable and not impossible to solve.”
The traditional hospital-based EMR is not the solution either, said Dr. Chang. He uses the analogy of purchasing goods at different stores. “For every purchase we make, we use a single trusted proxy – our bank or other financial institution.” He sees this model – the economic record as a single source – one that healthcare should follow.
“In healthcare, we need a central clearinghouse that acts as a trusted proxy for a person’s health information,” Dr. Chang said. This proxy, according to Dr. Chang, could be the personal health record (PHR). The scenario Dr. Chang envisions is once a patient presents their PHR card to a healthcare provider, permission is granted to access that patient’s records. Technology points and brings together all relevant data so that health data is not stored in one place. “The PHR addresses the political and economic issues that create the problems in a multi-site, multi-PACS environment,” he explained.
He also suggests that third-party payers can work with a PHR provider to reduce waste and streamline patient data collection as an extension of the billing process. The third party payers synchronize with the PHR provider to know where the patient’s data resides.
Dr. Chang added, “We have a good precedence already – it is the model of how we handle our money.”
Brock, Amy. Medical Imaging: Equipment and Related Products. U.S. Market Projections for Medical Imaging Equipment by Segment, 2004-2012. BCC Research, Oct. 2007. Available at www.bccresearch.com/report/HLC020E.html.