Dig deep with data analytics to improve imaging quality

2014 01 23 14 51 33 773 Data Stream 200

Sure, your radiology department tracks statistics such as report turnaround time and imaging volume. But is that enough? Not if you want to flourish in a future that emphasizes value over volume, according to a presentation at the recent New York Medical Imaging Informatics Symposium (NYMIIS) in New York City.

By using analytics software that can also analyze text in radiology reports, imaging practices can track a wide range of metrics and uncover opportunities for meaningful improvement in quality, said Dr. William Boonn, an adjunct assistant professor of radiology at the University of Pennsylvania, as well as president and CEO of imaging analytics software developer Montage Healthcare Solutions. That will be critical to avoid commoditization and keep a leg up on the competition.

Radiology practices face many challenges today, including declining reimbursement, more regulations, and increased competition from local groups or national teleradiology providers. Data analytics can help you not just survive, but thrive, Boonn said.

"There are opportunities for practices that will take quality seriously and use it to succeed," he said during his during his NYMIIS 2015 talk.

Not smart enough

Unfortunately, many existing analytics systems are dumb when it comes to radiology, Boonn said. They can analyze basic metadata such as time stamps and provide information on turnaround times, imaging volume, and relative value units (RVUs), but those figures don't fully account for the value of the radiologist across the imaging chain.

"Most analytics systems ignore the radiology report," he said. "They ignore the report because we can't do simple calculations on a report. We need to be able to understand what's contained within that report."

Analytics software that incorporates text analysis via natural language processing technology can also yield information such as quality metrics, outcomes analysis, and medicolegal risk evaluation, as well as data that can be used to optimize revenue.

These types of analytics algorithms can also be invaluable for identifying inappropriate imaging utilization, such as excessive studies for pulmonary embolism in the emergency room, according to Boonn.

"You need to be able to analyze this and understand it in order to succeed in today's environment," he said.

Such algorithms can also pinpoint how many radiology recommendations for imaging are overdue, ensure that critical results communication has been documented, and highlight dictation errors in reports that could lead to significant medicolegal risk, Boonn said. In addition, data analysis can help reduce length of stay by identifying actionable recommendations for interventional procedures.

Analytics and Imaging 3.0

The American College of Radiology's Imaging 3.0 initiative offers a road map for transitioning radiology from a model based on volume to one based on value, Boonn said. Data mining and business intelligence is central to that process.

"In order to succeed in today's environment, you need to be able to measure how you're doing," Boonn said.

Studies in the literature have highlighted the large percentage of patients who do not receive the follow-up imaging studies recommended by radiologists in their reports. In a poster presented at NYMIIS 2015, researchers from Albert Einstein College of Medicine and the Hospital of the University of Pennsylvania also reported that 60% of emergency room patients and 37% of patients overall did not receive follow-up imaging within one year.

"This is a big problem, and what we need to be able to do is find all of the cases where follow-up was recommended, and then find all the cases that are overdue," Boonn said. "You need to be able to track these and find what the outcomes are of these follow-up recommendations."

Analytics can also assess the quality of follow-up recommendations, which can vary significantly between radiologists.

Quality of care is important to the U.S. government. The consequences of not participating in its Physician Quality Reporting System (PQRS) for Medicare can be severe: For a 50-person radiology group, the potential PQRS penalty can be $132,700, according to a 2013 study published in the Journal of the American College of Radiology, Boonn said.

Gender, laterality errors

Sometimes a patient's gender can be incorrectly identified in radiology reports, and these types of errors are particularly problematic now that patients increasingly have access to their reports, according to Boonn.

"They read every word of their report," he said. "[If their report has a gender error], that patient's confidence in me as a radiologist, in that whole radiology practice, and in that whole health system has decreased significantly because of these stupid little errors."

Analytics can determine the frequency of these gender errors in reports and who is making them, enabling corrective action to be taken. This could involve presenting the radiologists with data showing how their error rates compare to those of colleagues, according to Boonn. The same principle applies with tackling laterality errors, i.e., cases in which the left and right sides of the body are misidentified in the radiology report.

What's more, analytics can improve confidence in radiology reports by searching for "hedge" phrases used by radiologists. Those who frequently use such phrases can then be coached to reduce this practice.

Analytics can also be useful for critical results and the problem of missing notifications. Failure to communicate is the second most common cause of medical malpractice litigation in the U.S., so it's important to have the ability to extract critical results and make sure that these results were, in fact, communicated to the referring physician, Boonn said.

"You need to be able to do text analysis to understand what's going on here ... and [understand] where these problems are occurring," he said.

Volume still matters

While these types of quality metrics are critical, volume analytics is also important. Instead of relying on one overall number (i.e., a radiology report turnaround time of 45 minutes), Boonn recommends a more granular analysis that measures data by the hour of the day. This level of detail can yield valuable insights and uncover opportunities to improve operations, such as staggering lunch hours for staff to make sure there's no slowdown at noon, he said.

Analytics can also be used to track outcomes from radiology interpretations by correlating them with pathology results, he said. Getting this type of feedback can help reduce errors in future cases.

Radiology is at a crossroads right now, with a past based on volume and a future based on value, Boonn said.

"My belief is that if you use data analytics -- and you use it successfully -- you can help bridge that gap and go from volume to value, and go from Imaging 2.0 to Imaging 3.0," he said.

Page 1 of 603
Next Page