Radiologists have spent decades stockpiling digital images and reports into electronic databases, but extracting useable information from those systems is no simple task.
Big data. It’s one of the latest catchphrases in health care and other data-rich industries. Generally, big data is defined as data sets that are so large and complex, they cannot be managed with traditional processing tools.
In radiology, the idiom often refers to the sheer volume and size of the reports and images that radiologists generate and collect in their databases. While discussion is swirling about the potential benefits of big data, industries like radiology are challenged with figuring out how to better organize this information to improve efficiency and outcomes.
Radiologists began accumulating electronic data in the 1980s, when the first PACS was developed. The archiving system was revolutionary at the time and dramatically changed radiologists’ workflow. Instead of producing images on film and reports on paper, radiologists began generating digital products that were easily stockpiled in electronic databases. The addition of RIS and EMRs has since made it possible to store and manipulate even more information electronically. But after decades of feeding information into these and other databases, many radiologists are finding it difficult to extract relevant data in a way that is advantageous to patient care.
It’s a problem that extends beyond radiology, as a patient’s medical record is often fragmented across many different databases even within a single institution. As a result, radiologists and other physicians must access multiple information systems to find specific data elements that may be critical to providing optimal patient care, says James H. Thrall, MD, FACR, chair emeritus of the department of radiology at Massachusetts General Hospital. “We have dozens of information systems,” Thrall says. “There’s one for the blood bank, there’s one for the biochemistry lab, the hematology lab, the microbiology lab, radiology, histopathology, immunopathology, the operating room, and so on. It’s all electronic, but when you sit down at the computer, you have to log onto each one of those databases separately to see if there’s information about a particular patient.”
In today’s fast-paced health care enterprises, radiologists and other physicians don’t have time to search multiple databases to retrieve patient records. But without access to a full medical history, the chance of providing the highest-quality care possible diminishes significantly, says Gary J. Wendt, MD, vice chair of informatics, professor of radiology, and enterprise director of medical imaging at the University of Wisconsin-Madison in Madison, Wis. Therefore, he says, tools must be developed to aggregate the data and make it available across multiple platforms. “If you don’t organize and present the data in a usable fashion, people won’t use it,” Wendt says. “Obviously, that’s a problem, and it’s something we’re going to have to address.”
Thrall knows from personal experience what can happen when radiologists don’t have all of the information they need to treat a patient. His brother-in-law underwent multiple imaging procedures with the same group of radiologists, none of whom, presumably, had taken the time to review the others’ reports on the case. It wasn’t until Thrall himself examined all of the images and reports that he realized the team had failed to properly diagnose his brother-in-law. “When I had all of the information together, I realized he had cancer,” Thrall explains. “I don’t want to criticize those radiologists, but they were probably just way too busy or they thought they were too busy to review all of the information. I think this happens quite frequently.”
Without immediate access to all of the pertinent information within a patient’s medical record, missed and improper diagnoses will continue to occur, says Safwan S. Halabi, MD, director of imaging informatics at Henry Ford Health System in Detroit. It’s impossible to provide comprehensive, accountable care based on only one data point at a time, he says. “There’s a saying that goes, ‘garbage in, garbage out,’” Halabi notes. “If the information presented to the physician or other health provider is incomplete, it is assumed that the downstream care, whether it is treatment or diagnostic testing, will also be incomplete.”
While PACS technology was groundbreaking when it was introduced, some radiologists say that as a standalone system it is now obsolete. After realizing that he had spent most of his career concentrating on how to get information into PACS and other databases and relatively little time considering how to get it out, Thrall concluded that such systems provide limited value. “PACS and RIS are dumb systems from the standpoint that all they really allow you to do is collect data. They do not help you manage it or use it,” Thrall says. “You can go into the PACS and bring a case up on a monitor, but the system does not integrate the image data with the rest of the patient’s medical records.”
"If you don’t organize and present the data in a usable fashion, people won’t use it. Obviously, that’s a problem, and it’s something we’re going to have to address." — Gary J. Wendt, MD
Massachusetts General Hospital has developed its own data-mining tools to extract data from PACS and other databases and display that data in a comprehensive format. Known as QPID, LEXIMER, and RaceTrack, the tools have since gone to market. Other institutions have integrated similar products into their PACS. Michael P. Recht, MD, the Louis Marx Professor and chair of the department of radiology at NYU Langone Medical Center in New York, says PACS have become like smartphones, and data-mining tools are like the apps that enhance the phones’ performance. “We use a commercially available PACS system that we like very much, but then we put another product on top of it,” Recht says. “It’s that product that I like to say is the brains. It takes the data from our PACS, RIS, and EMR, and puts it all together to make sure we have all of the relevant clinical information before we interpret our images. It is also where all of our metrics come from.” The tool mines patient history information as well as such metrics as department volume, patient time in the department, and billing data in real time — allowing radiologists to make adjustments as needed to improve patient care, Recht says.
Wendt thinks it’s time for a nextgeneration archiving system that not only has data-mining capabilities but also works across platforms, including desktop computers, tablets, and smartphones. “That’s going to require a sort of PACS 4.0,” Wendt says. “It would be a system that can aggregate data and pull from multiple sources and then present that data in a clinically relevant manner for the problem at hand.” Thrall agrees and says radiologists should ask the vendor community for new PACS, RIS, and EMR features. “They should ask the vendors to install data-retrieval programs and data-analytics programs,” he says. “If the companies hear that people want these kinds of capabilities, then they’ll begin to develop them and add them to the systems.”
Empowering Big Data
Institutions that have already begun using data-mining and aggregation tools have seen significant increases in efficiency. Thrall says Massachusetts General Hospital began considering data mining approximately 10 years ago. Now, thanks to the programs it has developed, radiologists can hit just one button and obtain as many as 75 data elements about a patient at a time. Those data elements may include the patient’s allergies, hemoglobin levels, renal function, medications, and other details that can help radiologists provide better patient care. Although a formal study has not been conducted, Thrall says the tools have undoubtedly streamlined radiologists’ workflow. “Once you’ve used these tools, the benefits are so obvious that you’re not going to do a scientific study,” he says.
The possibilities go beyond day-to-day operations, says Recht, who envisions using advanced data-processing tools to manage and share big data nationally so that radiologists can conduct more comprehensive research studies. The results of those studies could then be used to demonstrate the value that radiology brings to health care and further advance ACR Imaging 3.0™. “We need to come up with really large databases so we can use big data to make statements about how we impact patient care and patient outcomes,” Recht says. “Then when people start sitting down and talking about bundled care and reimbursements and the value chain of medicine, we really have some significant arguments to say why radiology is so important.”
But ultimately, radiologists say, the real value of these tools is improved patient care. When radiologists and other physicians can view complete patient histories, the number of mistakes and repeated tests decreases, while the chance for faster and more accurate diagnoses increases, Thrall says. “People expect you to spend the time and effort to learn enough about the patient to make a correct diagnosis, but if you’re faced with 50 CT scans and the referring physician is providing very little historical information, what are you going to do?” Thrall asks. “With these data-retrieval tools, you can get the information you need.”
Therefore, he says, it’s imperative that these tools be developed and implemented to ensure that all of that big data isn’t going to waste. “It was a huge accomplishment to develop the first generation of PACS. Just developing the systems that could store the amount of data that we generate was a monumental accomplishment,” Thrall says. “But now that we have all of that data in there, let’s use these data-mining techniques to help doctors take care of patients better.”
By Jenny Jones is a freelance writer.