John C. Spinosa, MD, PhD
Scripps La Jolla (Calif.) Memorial Hospital is a community hospital that is part of the larger Scripps system of four hospitals on five campuses in the greater San Diego area. Laboratory Diagnostics Medical Group, a group of six pathologists, provides the anatomic pathology services at Scripps La Jolla, or SLJ. The pathologists in the group had expressed interest in digital pathology and, as such, were invited by Aperio to participate in a pilot study on the use of digital pathology for tumor board applications.
Conceptually, the value of digital pathology for tumor boards seemed clear; in fact, the technology was being used for that purpose in other institutions. However, its use had not been studied in a rigorous way in a community hospital. Accordingly, the purpose of the pilot was to test and measure the effectiveness of digital pathology as applied to tumor boards in a community hospital in furthering the following goals:
- Streamline the preparation for tumor boards, making it less time-consuming, more efficient, and less cumbersome and stressful.
- Improve the experience of the tumor board. Enable the real-time flexibility to show any region of a slide, not just those photographed beforehand. Provide better information by showing higher-quality images.
- Improve the satisfaction of all parties—pathologists, other clinicians, and staff—associated with the tumor boards.
- Identify potential enhancements to digital pathology hardware or software that could improve the applicability of digital pathology to tumor boards.
A 17-week study was conducted in summer and fall 2007. Eight weeks beginning in July 2007 were dedicated to collecting data on the existing process; the pilot of the new technology ran for nine weeks, concluding on Dec. 13, 2007.
Scripps La Jolla holds a breast tumor board and a general tumor board every Thursday morning, typically covering six to eight cases. Pathology is typically notified of the patient list on Tuesday afternoon. The administrator in pathology then pulls all relevant case information, including historical information, if any. The pathologist reviews the reports and identifies which slides he or she would like to see. The slides are pulled manually and given to the pathologist, who reviews them and chooses the ones to photograph. At SLJ, a digital five-megapixel camera and an Apple Powerbook with camera-specific digital imaging software are used to take, save, name, and then present the digital photographs. For a typical case, the pathologist takes five photos, for a total average of 31.9 photos per week. One pathologist acknowledged that if the image acquisition process were easier, he would choose more.
Pathologists estimate that they spend an average of two hours and six minutes weekly preparing for tumor board, though there is considerable variability in the time required. During the eight weeks of pre-pilot tracking, the time spent in preparation ranged from 15 minutes to three hours and 40 minutes. Pathologists handle the additional workload of preparing for tumor boards in different ways: Some come in early, others stay late the previous day.
This approach that had been in place presented several problems. The technology used to take, save, and name the digital photos was a source of dissatisfaction. In fact, taking and naming digital photographs were among the lowest-ranked activities associated with tumor boards (Preparing for tumor board). On a scale of one to five, where one is laborious and inefficient and five is smooth and efficient, the processes of taking digital photos and of saving digital photos were each rated 2.25. The ease of using the software was rated 2.50; the ease of using the hardware only slightly better, 2.75. Specifically cited as challenging were getting images into focus, having to readjust exposure and white balance for each change in lens power, labeling images, and being interrupted while photographing.
The pathologists had to take the digital photographs themselves as the last step in the process. This meant that, for them, any delay in the process translated into the photography needing to be done at the last minute. Accordingly, the process often felt rushed and chaotic. For the administrator, pulling the needed slides in a timely way was a recurring challenge.
As a result of these problems, the preparation process was seen as important but often burdensome and time-consuming for the pathologists and administrator. The overall experience of preparing for a tumor board was rated 3.0.
Other issues raised by the pathologists and administrator were the frequent inability to access material for cases from outside institutions, the difficulty of preparing photographs for last-minute add-on cases (or the inability to do so), and the wasted time spent photographing cases that would get bumped from the tumor board agenda.
At the tumor board itself, the pathologists presented digital photos from their laptops. The process had several shortcomings. First and foremost for the pathologists was image quality, particularly image quality of low-power images. The quality of low-power images received the lowest ranking of any variable in the process, 2.13 (one=very poor, five=very good) (Image quality, ease of preparation, and more). The quality of high-power images was rated higher than that of low-power images but still only at 3.00. Pathologists acknowledged that image quality may have been sufficient for many of the clinicians attending tumor boards and estimated that the clinicians would rate image quality 3.4 and 3.5 for high- and low-power images, respectively. Their estimates were close: For the nine clinicians interviewed at the start of the pilot, the clinicians ranked image quality 3.3 and 3.6, respectively.
Lack of flexibility was another issue the pathologists cited. With the current process, pathologists have to anticipate which questions will arise at tumor board and prioritize their photo-taking accordingly. At tumor board, however, questions do arise that might have been addressed by a different photograph but, for a variety of reasons, that particular photo was not taken. One pathologist estimated that in about 20 percent of cases he wished he had taken another or a different photograph. The clinicians had different opinions on this point. Several said it was often (every tumor board or every other tumor board) that they wished pathologists had taken additional or different pictures. Others did not consider this to be an issue.
The nine-week pilot of the new digital pathology system went live on Sept. 24, 2007. An Aperio CS Digital Pathology System, including Spectrum software, was installed in September. A decision was made to keep the system standalone (rather than on the Scripps network) for the purposes of the pilot. Aperio provided on-site training for the administrative staff in the pathology department. Each pathologist received 20 to 30 minutes of training by Aperio in advance of his or her scheduled tumor board.
The pilot was successful in meeting its goals. First, the process of preparing for tumor boards became more efficient. Pathologists saved an average of one hour each week in preparation for tumor board, cutting their time spent in half. Preparation became smoother, less laborious, and less chaotic for the pathologists and the staff. Second, the experience of the tumor board improved. Pathologists were better able to demonstrate their findings and respond to ad hoc questions at tumor board.
Third, the satisfaction of all parties improved. Pathologists, pathology staff, and nonpathologist clinicians who attended the tumor boards preferred the new digital pathology system over the previous approach. Pathologists shared ideas about how to enhance digital pathology hardware and software to make it even more useful for tumor board application.
The overall impact of digital pathology was positive. In its 2007 Year in Review report, the Scripps Cancer Center concluded, “Digital slides are a marked improvement in the presentation of pathology at tumor boards and increase the educational value of this activity. The context of the patients’ pathology is much better presented and understood.” Pathologists agreed; their comments included these remarks: “Digital is vastly superior ... conveys much more information,” “An excellent tool for these meetings,” and “I don’t want to go back to photos.”
The overall impact on the preparation process was also positive. As shown in Overall experience, ratings of the overall preparation experience improved significantly, from 3.00 to 4.67. The preparation process was less stressful and chaotic and the hardware and software were easier to use. When asked to rate the overall impact of using digital pathology for preparing for the meeting (one=made things substantially worse and five=improved things substantially), the pathologists’ average score was 4.67.
For the pathologists, streamlining the process to make it more time-efficient was of critical importance. Before the pilot, pathologists spent an average of two hours, six minutes preparing for meetings. During the pilot, they spent an average of 50 minutes preparing for meetings. Those 50 minutes were split almost equally between reviewing a case and choosing slides to be scanned (26 minutes) and reviewing slides once they had been scanned (24 minutes). The net time savings of 76 minutes was close to the pathologists’ average estimated time savings of one hour. Given that the reduction in time served to reduce the preparation time by about half, this was a significant finding. Several comments from pathologists relate to the impact on preparation time: “Digital pathology made my job much more efficient,” “This saved me time,” and “Less time was required with technical/ photography preparation, so more time was available to familiarize myself with the content of the path report.” Part of the efficiency derived from the fact that a single scanned slide could convey more information than a single photographed slide. Accordingly, the number of slides photographed fell slightly from 18.3 pre-pilot to 16.6 scanned during the pilot.
It was possible, though, that the new technology would simply serve to shift work from the pathologists to the administrator. While this might be financially attractive, it was also seen as a possible source of new workflow bottlenecks. However, the administrator’s ratings of the system were high. When asked to rate the overall impact of using digital pathology for preparing for the meeting (one=substantially worse, five=substantially improved), her weekly ratings averaged 4.70. She wrote, “On the whole I love the scanner!”
What was the impact on the presentation at tumor board? The use of digital slides for the presentation was overwhelmingly successful. The pathologists were asked to rate satisfaction with digital slides for the presentation; every score for each of the nine weeks was a 5.0. One pathologist wrote, “Having a whole slide allows for a much better experience than a few photos of select areas.”
As noted earlier, image quality had been rated poorly before the pilot. As shown in Table 2, scores on image quality and other key variables improved, with image quality improving the most. Pathologists noted that “low-power images are one of the best things about the system ... a whole slide image is not possible with our digital camera,” and that digital slides “more accurately reflect material we use to make diagnoses than photos.”
One variable to be tested by the pilot was the extent to which having slides led to a better experience. As seen in Table 2, ease of presentation improved with digital slides, as did the ability of the pathologist to adjust to unforeseen changes during the presentation (change in order of patients to be discussed, unexpected questions). One pathologist wrote, “I was able to present in a much more fluid manner.” The pathologists’ satisfaction with their contributions to the meeting increased significantly as well. One pathologist wrote, “I was able to review some areas I would not have photographed because I was not aware of all the clinical issues.”
During the course of the pilot, the Scripps Cancer Center added a question about digital pathology to its standard tumor board evaluation form. The question asked the participants—surgeons, oncologists, radiologists, and other clinicians—to give feedback on digitizing slides for pathology by assigning one of four ratings: excellent, better, same, or not so good. The feedback was very positive: 82 percent excellent; 15 percent better; three percent the same, and zero percent not so good (n=96). Clinicians particularly liked the ability to see the whole slide image overview at the same time that they could see the detail. One surgeon said, “The system gave a better overview of the pathology and was able to show the physical relationship of areas ... zooming in and out was a better way than having different photos at different magnifications—with photos it takes time to reorient.”
Other issues arose in conjunction with tumor boards. Last-minute add-on cases were often impossible to photograph in time for tumor board. While this situation did not arise frequently during the pilot, there was one case that was possible to present by a quick scanning of the slide which, under the previous approach, would not have been possible. Shifting the scanning task to nonpathology staff facilitates such last-minute requirements. Making the scanned slides available on the Scripps network would also facilitate making last-minute cases available for the meeting.
In addition, pathology time was wasted when cases prepared for tumor board were not presented and a different pathologist had to review the case (and take the photographs) in the following weeks. During the pilot, the slides scanned to support the case were able to be used the following week, thus saving pathologist and administrative time.
Finally, SLJ frequently found it difficult or impossible to access material from outside cases that were to be presented at tumor board. During the pilot, this was not addressed. However, routinely scanning and archiving slides from outside cases will provide access to such slides at SLJ even after the original slides are returned to their referring institutions.
As expected, there was a learning curve for pathologists and administration. However, the pathologists in particular were able to learn quickly; none commented that the learning period was overly burdensome.
SLJ is now using the Aperio system for all of its tumor boards and multidisciplinary conferences. In addition, plans are underway to use the system for multi-facility virtual tumor boards.
Dr. Spinosa is pathologist and chief of staff, Scripps La Jolla (Calif.) Memorial Hospital.