Mammography Matters: The importance of expertise—and optimized IT

That old saying “jack of all trades, master of none” can be applied to medical imaging. Even the bright minds that make up the radiology workforce need focused training and years of specific experience in order to understand all the intricacies of the profession, and there’s one area in particular where specialized expertise is particularly valuable: mammography.

Optimal patient care demands that radiologists with extensive experience in reading mammography take on interpretation of those studies, and that may require a re-evaluation of workflows and IT strategy to make sure the most experienced in the field are the ones reading the images.

Researching performance

A number of published papers support this notion. Back in 2013, for example, researchers in Australia set out to look for markers of good performance in mammography. Mohammad A. Rawashdeh, MSc, of the University of Sydney, and colleagues published a paper in Radiology showing that the number of mammographic readings per year was positively related to performance, as was the number of years a radiologist read at least 5,000 scans.

More recently, this idea was put to the test by researchers here in the U.S. Tracy Onega, PhD, a principal investigator at the Norris Cotton Cancer Center at Dartmouth-Hitchcock Medical Center, and colleagues wanted to test whether high-volume facilities had better screening outcomes, so they looked at 116 facilities participating in the U.S. Breast Cancer Surveillance Consortium.

More than 3 million screening mammograms from 2000 to 2009 were included in the study, and Onega et al. found that facilities with a total volume of 5,000 to 10,000 mammography reads were 32 percent more likely to diagnose invasive tumors with good prognoses (i.e., cancers detected at an early stage) than facilities that only interpreted 1,000 to 2,000 cases per year.

Onega and colleagues published their findings this year in the Journal of Medical Screening. In a statement released by the researchers, Onega noted that surpassing the threshold of 2,000 annual interpretations should be easier in the era of digital mammography, where small facilities could send cases to be interpreted by larger facilities.

“[W]e’ve now laid the foundation to study the volume benchmarks in a more granular way before this could translate directly into policy and practice changes,” Onega said.

Technological assist

In order to optimize accuracy and detection rates, facilities will need to make sure their workflows and technologies are aligned to get images to the right people within a practice or, in many cases, to outside groups that specialize in mammography. Steve Deaton, vice president of Viztek, a Konica Minolta company, says he hears this all the time from radiology groups.

“In all cases I think radiologists recognize that it having a radiologist who focuses on mammography versus asking a generalist to also read mammograms and other breast images will increase detection rates and accuracy,” says Deaton.

For many facilities having a specialist means outsourcing reads, and that means rethinking workflows. An outside group that specializes in mammography will have their own workstations, voice recognition and other systems. Reports will have to be sent back to the originating site via HL7, email or perhaps fax.

“You go from having reports built natively in the facility’s system to now having reports being built in this third party system and transmitted back,” says Deaton. “There has to be accountability so it’s known whether a study has been transmitted and how long the site has been waiting.”

Expediting results in mammography reading can be just as critical as the accuracy of the read, adds Deaton, who notes that the wait between exam and results or between follow-up scans can be excruciating for patients. Patient-centered care now means having diagnostic mammography interpretations available instantly, so that if a subsequent breast MR or ultrasound exam is warranted, it can be taken care of before the patient leaves.

There are a number of technological capabilities that are essential to this process, which Viztek designed its Exa platform to include. Deaton explains that at the top of the list are zero-footprint viewers and server-side rendering.

On the viewer, having zero-footprint means simplicity as there’s nothing to install. This makes communication with outside groups that much easier, according to Deaton. “An existing radiology group that’s remote with a different vendor’s PACS doesn’t have to worry about butting heads with software because all they need is a web browser.”

Server-side architecture is the concept that DICOM datasets will never have to be pushed down to the workstation. This eliminates the need to transfer the incredibly large files involved in mammography and solves a number issues with bandwidth speed that would otherwise be a bottleneck.

Here, again, mammography is a special case, since reviewing priors is more important in this context than almost any other, and transferring all those previous studies can be quite a burden if IT is not optimized. The importance of priors was underscored at the 2015 annual meeting of the Radiological Society of North America (RSNA), where researchers from the University of California, San Francisco, discussed their findings when they looked at recall rates by the number of priors reviewed.

Speaking at the RSNA meeting, Jessica Hayward, MD, explained that her team did a retrospective search of more than 46,000 screening mammograms in their institution’s database, and found having access to two or more prior exams reduces the odds of recalling that patient by 14 percent.

Deaton says Viztek, with its new Exa Mammo Viewer, aims to make viewing these priors as easy as possible. “Mammography is not necessarily just what you see on one film, it’s what you see changing from one year to the next,” explains Deaton. “Something might not look like a problem if you just looked at the current one. But if you saw two prior years’ worth of films, and you saw that it increased in size or changed geometry, that’s important to know. And so all that information has to be readily available to the interpreting radiologist.”

Business intelligence and analytics dashboards provide another piece of the technological puzzle for sites looking to optimize operations. Long a staple of manufacturing, radiology has really embraced analytics in recent years as the call to maximize value grows louder.

In the context of outsourcing mammography reads, such dashboards could provide invaluable information on the efficiency of partner sites. Facilities could measure turnaround speeds achieved by outside sites, sending more studies to those who read the fastest. This type of technology offers more granularity than the basic number of studies seen in billing information, meaning more intelligent allocation of resources.

And while efficient operations can help a practice achieve its own business goals, in the end, making sure that the most experienced physicians are reading mammography improves patient care, which is the ultimate goal.