Radia Homogenizes Client Base with Homegrown Worklist

The PACS has come a long way since its inception, enabling communications among all imaging stakeholders in a hospital or health system. But what is the large radiology practice to do that works in a heterogeneous environment, with multiple information and dictation systems, and a centralized reading model? Radia, a 75-physician radiology and vascular practice based in Everett, Wash, found itself in just such a predicament. Since launching its teleradiology operation—teleRadia—in 1998, the service has evolved from a way to provide nighttime emergency reads for client hospitals to a thriving service line that provides nighttime coverage, subspecialty reads, and supplemental reading for more than 30 hospitals, clinics, and radiology practices, all of which have their own information systems. The practice’s information system—devised under the guidance of CIO Jack M. Jones and teleRadia medical director Jeffrey Robinson, MD—works remarkably well considering the number of PACS and information systems involved. But in order to broaden its service and meet demand in its marketplace, Radia knew it needed to exact more efficiency from both its information technology and its radiologists. After scouring the market for a product that could provide a single worklist and dictation system for its scores of clients in its dual-PACS environment, Radia earmarked up to $1 million, hired a software engineer to write the code, and designed a new front end for its two PACS. The project, begun one year ago, is representative of the creative approaches radiology practices are taking to customize garden-variety information systems to their unique needs. ImagingBiz.com interviewed Jones and Robinson, to understand the steps taken to streamline Radia’s teleradiology workflow with a custom fix. The Problem Radia’s core business comes from the nine hospitals for which it performs all professional services, supplemented by 4 outpatient joint ventures and a growing number of teleradiology clients. Currently, all images from Radia’s teleradiology clients arrive at a centralized reading room via DICOM. Orders come in via fax and are processed by imaging assistants who match the order with the image. “The way we do it today is that each of the sites either sends us the image directly from their PACS or they will send to us directly from the modality,” Jones explained. “As long as it is in DICOM format, we can take images into our system here, read the cases, do the interpretations and get the results back out to them, and that is the tricky part. Today we are dealing with seven or eight different dictation systems, and that is the real problem we have today.” Because Radia does not interface directly into its clients’ HIS or RIS, teleradiologists must dictate into the client’s dictation system, programmed directly into the client’s reporting system. Another obstruction to workflow is the need for teleradiologists to switch back and forth between the user interfaces of the two PACS used by the practice, one acquired for its teleradiology practice and the other more recently through a joint venture imaging center with a hospital, whose PACS was chosen by the community. That hospital provides a significant amount of Radia’s volume. The dual systems present a significant productivity hurdle. “Right now we are running two independent PACS simultaneously on one workstation,” explained Robinson, a key advisor in the architecture of the new system. “We have AMICAS and Stentor, and as a radiologist, I have to flip back and forth constantly. It’s like reading from two different books at the same time. Getting everything on one worklist would be a dream in terms of efficiency.” The impact on radiologist productivity should not be underestimated. Jones predicts that Radia will achieve a 5% to 10% improvement in radiologist productivity as a result of the project, which it calls Teleradia 2.0. Ultimately, the new system will have one user interface and dictation system for the teleradiologists and, eventually, a paperless process. “The radiologist will have one worklist that will launch one or the other PACS depending on where the case comes from, the system will know that and launch the appropriate viewer,” Jones explained. “We will have one dictation system here, and it’ll be a voice recognition system. We haven’t made a selection on that yet, but it will be one of the big three. When Phase I of the new front end goes live later this month, orders will continue to be faxed, but they will be read automatically by optical character recognition (OCR) software.” Solution Building Blocks Key components of the new system are the optical character recognition (OCR) software, a database for the orders, a database for the images, the engineers writing Radia’s software code, a voice recognition system (yet to be purchased), and the servers to run it all. Jones found his first software engineer from a company doing rapid consolidation in the chemical testing industry. “As a result of this rapid consolidation they had to do a lot of testing documentation, so, much like radiology, it was very paper-laden, and they had to figure out how to get rid of the paper, do some quality checks, and automate the whole thing,” Jones explained. “So it really was a benefit to us to grab this engineer. Then we found out that the project, like all projects, got a little bigger than we thought, and the timeline by which we wanted to get it done by got shorter, so we brought on a second engineer.” Radia is a Microsoft shop, so the software is being written on the .NET platform. Jones also hired a project manager for Teleradia 2.0. “For lack of a better term, we do call it middleware because it’s really a front end to different systems,” Jones explains. “It’s a workflow product for us, and we’re trying not to reinvent the wheel. We work with those vendors that we currently have PACS relationships with, and the good news for us is that those folks have things like APIs (application programming interfaces), so we can get into their systems, do the calls, get the information back, launch the viewers, and do all those things outside of what their system already does. We don’t get access to the proprietary code but we certainly have a way to work with their systems in a way that in the old days you didn’t have the ability to do. Years ago, if I wanted to go into a major HIS, they wouldn’t let you do it, you used the product the way it was sold and that was it. At least now with the open systems kind of thinking, many vendors provide you with ways to get to the database, you can query it, and manipulate the information to build this worklist, like we are doing.” When Phase I of the new front end goes live later this month, orders will continue to be faxed, but they will be read automatically by OCR software. The OCR software will automate order entry, work currently performed by the imaging assistants. “Initially the fax comes in and the computer looks at it and says, There are words, here, here, and here, I recognize this as coming from hospital X,” Robinson explains. “If it’s from hospital X, then the name is here, the medical record number is over here, and the reason for the exam is down here. Then it draws a box around those fields, and says, OK, what are the letters and then it decodes the letters. That kind of software has been around for a long time, but it has not worked very well. It is getting better. Now, our IT guys’ experience is that it is good enough to use.” Added Jones: “Even though we don’t get discrete data elements, our success rate of scanning is quite high right now so that we can take that information, develop discrete data elements, and embed that into the front end system we are building.” Radia plans to roll out the project in three phases. “In the first stage, the computer will do digital voice capture and then call up the hospital dictation system and log in, play the voice file, and log out,” Robinson explained. “But it’s something I have to do myself now. Eventually, in Phase 2 and Phase 3, we will have our own voice recognition software, and then we truly will be dictating in one way, into our own VR. Then, once we have the HL-7 interfaces, we will be able to export a signed authenticated report directly to the hospital.” Jones’ team will begin implementing the interfaces in short order, both for order intake and results delivery. “We will build the interface with the sending site’s HIS or RIS, wherever they want to send the order from, and on our side we will have a database for orders,” Jones explained. “We will also have a database for the images, and we will insure that for each order we receive, there’s the corresponding imaging information available, and then create a worklist for the radiologist to read the case off of.” Robinson estimates that the practice will have many HL-7 interfaces in place within a year’s time. “We’ve been developing for the last year, and we are just in our acceptance testing phase today,” Robinson said. “We’re planning on using a rapid prototyping strategy in which once we have something up and running, every new feature will just roll out as it is developed. We’ve got our whole list of features and priorities, and the developers are going to work their way down from the top.” A Variety of Sophistication Levels But not all clients have the IT sophistication required to implement and maintain these interfaces, which is why the system is designed to work both for the least and the most technically savvy organizations. “With the hugely heterogeneous environment here, there are all levels of technical sophistication,” Robinson explained. “You have to go to the least common denominator to make it work.” Some teleRadia customers will continue to receive results as they always have (voice files), while others will receive an HL-7-formatted text file that can be printed and scanned or imported into their system. Likewise on the front-end, the OCR software will receive outbound orders and scan the incoming requisitions. “We like to differentiate ourselves when we talk to people about our practice,” said Jones. “We like to say we are all things to all people. Unlike a lot of the other teleradiology groups that will come to you with a solution they’ve invested in and it works for them but oftentimes it means you will deploy a second system inside your operations for the sake of that teleradiology company to be able to do what they say they will do. What we say is, ‘We’re not going to change your operations at all. You keep doing it they way you have, and we will find a way to make it work.’” Radia’s investment in teleRadia 2.0 goes beyond a short-term fix for its teleradiology workflow. “When you think about what we do in working with all of those hospitals, we have become de facto our own RHIO, or can, because we are collecting information from patients across multiple hospital sites,” Jones said. “As long as we are building this database for our own workflow needs, we actually believe that long term we can take that information and use it in more of a global way, and actually be able to share that information with our provider community. So we have plans down the road to clearly expand beyond just workflow. Once we have the information, then it is just a matter of how you use the information.” Meanwhile, the new product will enable teleRadia to grow. “What this will enable us to do is go to a client with a package of services,” Robinson explains. “Where, be it a night preliminary contract, a clinic daytime subspecialty contract, or as a partner with a radiology group who just needs help, we can come in and say all we need is an IP address, and we’ll get your studies in and however you want the reports back, if you want them secure-emailed, faxed, or imported directly in your information system, we can do it. “By having standard interfaces—DICOM and HL7—then we can interface with anyone. We are using DICOM and that is working well. Right now, our common interface is a fax machine. And that works up to a point, but there is a ceiling to how big you can grow that way. We have our imaging assistants managing the paper flow and when they get topped out, then we add on another imaging assistant, and pretty soon we have to have someone managing the imaging assistants, coordinating them, and that’s just untenable. So it allows us to grow by giving us a scalable system with a standard interface.” Just how big teleRadia plans to grow is an open-ended question. “How big is a geographic area?” Robinson asked. “Right now, we limit ourselves to the state of Washington, partly from convenience, and partly because our growth has been organic. We have never gone out and marketed our services. So the people who come to us are in the state of Washington. We are looking outside the state in the region, and that I think is reasonable because we can still be a quasi-local group. From Oregon, we’re still local, from Alaska we’re the local group. Montana, we’re local. From Arkansas, we’re not going to be the local group.”