By all indications, most image interpretation in the U.S. is still done on film. This continues years after leading-edge radiologists began adopting and touting the benefits of filmless strategies for managing images.

But that is changing, sometimes rapidly, and the pace at which soft copy is adopted is expected to increase. We convened a group of radiologists and administrators, all of them familiar with both film and soft-copy interpretation, to discuss how their lives changed when their institutions adopted filmless imaging. Our hope was to give radiologists who are still using film some idea of what to expect.

Our panel consisted of six radiologists and two administrators, representing both private practice and academic settings. Their experience in soft-copy reading ranged from just months to several years. They differed on many points, including strategies for training radiologists in soft-copy reading and whether soft copy increases or decreases contact with referring clinicians. They all agreed on one point: soft copy is better and none would be willing to go back to film.

The radiologists on the panel were Dr. David Channin, chief of imaging informatics at Northwestern University in Chicago, Dr. Lucy Glenn, chief of radiology at Virginia Mason Medical Center in Seattle, Dr. David Hirschorn, a fourth-year resident at New Jersey Medical School, Dr. Jonathan Levy, a private practice radiologist in Scottsdale, AZ, Dr. Leonard Swischuk, chair of radiology at the University of Texas Medical Branch in Galveston, and Dr. Bruce Reiner, director of radiology research at the VA Maryland Health Care System. Also on the panel were John Eusek, administrative director of radiology at Virginia Mason, and Rick Perez, senior radiology administrator at New York Presbyterian Hospital. The session was moderated by Philip Drew, Ph.D., a consultant in PACS and radiology technology issues.

Drew: How does the practice of radiology change when you go from conventional film to a PACS? When you go from film to soft copy? Does it make a difference in how long it takes you to complete a study? Does it change the way you arrange your workday? Does it change other things, like the task of interpretation?

Glenn: We were probably among the earliest adopters of PACS, and the biggest change for us was getting used to something different. Radiologists are used to doing things in the most efficient way, and in these times all of us are very productive. So to take a work process that is efficient and change it completely involves a huge transition for some. But once we got over that initial hump, we were better able to look at a large number of images and use different windows. I think CT is infinitely better on a PACS. MR is a little more difficult, just because there are so many different series that you are looking at. Spines are also a little more difficult to get used to.

Drew: The fact that you can use different windows would slow you down, I should think, because you've got to do more things, look at more images.

Glenn: Yes and no. If you are looking at hundreds of images, it's easy to get used to picking out abnormalities as you're scrolling through, and you soon become efficient. I'm sure people who have had PACS for a while know that. The learning process took a couple of months, at least for our group.

Levy: PACS adoption is a two-step process. When you start, you have a bunch of old images on film and you have new material that is soft copy, and it's a struggle to deal with the two different displays. After about six months, when you have a backlog of soft copy, it becomes a lot easier than reading from films. We have more than 20 radiologists now, and we thought there would be some late adopters, but it's turned out that we have none. Every single radiologist likes the PACS better.

Hirschorn: It may be unfair to do a direct case-by-case comparison of how long it takes to read a case on PACS versus on film. What about the times when the tech didn't film it right, and you have to find the tech and say, "Do it right"? Can you take that into account? Or, the worse alternative, where you just throw up your hands and say, "Oh forget it. I'll just look at what

I have here and settle for that." PACS lets you do what you need to do, to look at information the way you want to see it, because you know what's important. Will it take you more time to read a case? Maybe or maybe not, but I can say one thing for sure: The amount of time spent looking at things you want to see is definitely greater with PACS than with film.

WORKFLOW AND CLINICIAN INTERACTION

Reiner: We've done a number of studies at the Baltimore VA looking not only at radiologists' productivity, but at the way they practice. One thing we found interesting is that when we do time/motion studies from the time a radiologist pulls up the image to when he or she completes the dictation, there is an 8% to 15% time savings-8% with CR and 15% with cross-sectional imaging studies such as CT. But a number of other factors add to the efficiency. Radiologists are not interrupted nearly as much. Technologists become more efficient, not having to interact as often with radiologists. And clinicians don't come to the department, because they can access images and reports anywhere throughout the enterprise.

We found that radiologists don't often vary the time from case to case. They'll take certain shortcuts. If you give radiologists an unlimited archive to compare, they will basically keep the number of comparison studies to a minimum in general radiography, because they have an idea when the interpretation time should end and they need to move to the next case. They know what their productivity level should be, and they make adjustments subconsciously. It will be interesting to watch the practice patterns, not only in terms of efficiency, but in the kind of image presentations radiologists use.

Levy: About the consultative process with referring physicians: In our experience, we spend significantly more time consulting with them since we've gotten PACS. That is because in many hospitals and in a lot of private practices, clinicians go to the file room, get the films, look at them, and then read the report, if it's available. If they don't understand what they're looking at, they go the radiologist. But now we've eliminated the file room, so they come into the reading room and say, "Will you show me the films on Mrs. Smith?" Actually, it's been terrific-it's bonded us to our referring physicians. We spend a lot more time with them, not less. Clinicians can get the images up on the floors, but they still come down to talk to us and get our opinion. We value that.

Swischuk: Our results have been mixed. I would say the consultations are cleaner. I do pediatric radiology, and we had set up a hotline-we're also on Talk Station, so we have the report out on IDXrad within minutes. This being the case, I wanted to go ahead with what I thought was a modern way of doing things and get the report online quickly. I expected clinicians not to call me. That lasted about two weeks; then they said, "No, we want you to call us." So we call immediately-outpatient clinics and ER-after we read the film. We get a faxed copy of the consult that alerts us to the fact that it's on the system. We read it immediately and give them a call, whether it's normal or not. The communication is swift and useful.

There are certain specialties that will contact radiologists regularly and specialties that will contact them only when they're in trouble. The ones who did it before still do it, and those who didn't do it before probably do it even less, because we're Web-based and they can pull the images anywhere they want. They're used to looking at them themselves. So I would say it hasn't changed things a whole lot.

Hirschorn: I would agree. We've had a similar experience at the New Jersey Medical School. I don't think the number of true consultations has really changed, there are just fewer residents running around saying, "Where are my films?"

HOW IMPORTANT IS THE REPORT?

Drew: People have said that with PACS and with images being available to referring physicians, radiologists will be cut out of the picture. Referring physicians will look at the images themselves and not bother with the report at all. Is that a real worry? Does that happen?

Reiner: There is a dichotomy in private practice versus academic medicine that stratifies the user groups in terms of how they want to view images, how they want to do consultations with radiologists, and, quite honestly, what their daily schedules are like. A lot of academic practitioners incorporate this educational time into their daily schedule, whereas private practitioners, obviously, need to see a lot more patients-throughput is more important.

Levy: I can give you an example of an increase in utilization: our emergency room. We have two; one is very busy and the other is moderately busy. To get films to the radiologists, the ER doc has to give them to a volunteer, who then walks them down to us. We look at them, call up the ER doc, and tell them what we think. That led to a small number of consultations during the day, and we would find stuff every day that they had missed. Now they call and say, "Pull up Mrs. Jones; tell me what you think." This happens 15 times a day. We love it because it's important that they understand who's better at reading ER films, and now they all do understand.

Hirschorn: You mentioned the ER. That's one of the places where we've experienced the biggest change to our residents who are stuck in that ER box. Now the ER can see exactly when a study was completed, and they can see the images. They say, "Oh, I see an image, so you should be able to give me an interpretation right now." It is killing our residents, who can no longer hide behind, "Well, I don't have the films here."

Glenn: That's a very critical piece, that we're going from a batch and queue process to a continuous flow process. Basically, you are doing everything in real-time.

MORE RESPECT FOR RADIOLOGY

Perez: I've watched the referring physicians gain a totally new respect for radiologists. In the early days, we made sure that we had nine months available before turning over the soft-copy read. But one day we said, "You know, on Monday morning when you come in, there will be no film." It actually was very apparent: The multiviewers disappeared, the workstations appeared, and everything else changed. On the day that we turned everything on, the clinicians had the same reaction: "I've got the image, how come I don't have a report?" So the flow became much quicker.

There is still the rounding that is done on a daily basis by the different services, because they still want that interaction. There's also a more consultative attitude, "What should I do next?" Radiology becomes totally accessible. Utilization has probably jumped because the information is readily available. The referring physicians have helped propagate PACS, because they want the information, and the faster we can deliver it, the more effect it has on length of stay.

Levy: Another thing about workflow is that we have a number of radiologists who read films at different levels of efficiency. But because we have two hospitals on our PACS, you can pull up a work list of the unread exams at each hospital. It has increased our throughput, because people are competing with each other. If you look at your work list and it has 14 cases on it and the other hospital's has seven, you say, "I'm getting behind, maybe I'd better read some more stuff."

We can now doctor-chase. While I've spent a career reading CTs, MRs, and nuclear medicine cases, the percentage of my time that's spent doing things I'm a specialist in has gone down enormously. We now have a neuroradiologist reading, and I don't even have to read some of the trauma CTs any more. The neuroradiologist can be on the other campus or in an office and read the neuro cases. It's an improvement in care, in private practice.

DECENTRALIZATION

Drew: That's one of the putative advantages of PACS, that you can read from many locations. Do you read from many locations? Is that practical?

Hirschorn: Yes, we do. We are starting to branch out, and my chair's feeling is that in order to deal with this consultation issue, the radiologists must get out there in the hospital. He wants one stationed by the ICU. He wants to get the radiologists out into the clinical areas so they're more visible, and that is happening.

Reiner: There is definitely a trend toward decentralizing the radiologists, particularly in a larger, integrated facility. The radiologists can no longer be trapped in the radiology department. You can't justify that. So you go to high-volume critical areas like an ICU, an emergency room. The expectations of radiologists are going to constantly increase.

When CNN came along and offered news 24/7, other channels rushed to beat one another for the news story, because the audience won't wait until the morning news the next day. The same thing is happening with radiology. People want 24-hour service. We're fighting it, but ultimately, more is going to be expected of us. We're going to have to find ways to deal with workflow and efficiency and around-the-clock staffing.

Glenn: In terms of the way PACS has changed the workflow, we're going much more to an anatomic base, as opposed to a modality base, for our stations. I would also say that in terms of moving people around, the critical point is making sure that everything gets done in a consistent manner so the clinician doesn't wander around or call 20 million places trying to figure out who read a study. That's key, that once you set up a reading station, you maintain it in a consistent fashion.

Swischuk: Let me talk about decentralization, because in our place we have a relatively tight campus, so we were able to cut out areas for musculoskeletal reading, chest, etc.; the intensive care unit and the regular chest are all in one area. I put both PACS and voice recognition in together, and once you have that, you can place your radiologists wherever you want. If you do real-time readings, you can place them anywhere and encourage telephone consultation for important cases. That is something everybody will titrate, depending on how far-flung they are in their operations.

Levy: We have a small office, and the hospital wanted to have a small area where films were done, and we couldn't justify putting a radiologist there for productivity reasons. Now we don't have to worry about that so much; we can keep the radiologists in the reading room at the hospital and bring the images in.

Hirschorn: One of the things that's hindered us in being able to decentralize is actually the lack of a good HIS. We went with a HIS a few years ago and now the order entry part of it is dead. We have to go with a new vendor and that's held us back, because the only source of information on the requisition-why this study was done-is still on paper.

Even when you go filmless, if you don't have the clinical history, which we know is essential to interpreting a case, it can really hold you back, in terms of how far you can distribute your radiologists.

Perez: We've gone service line. There are different service lines and our radiologists are geared toward them. Neurosurgery is probably the best example. They're on the same floor as our inpatients and the scanners, so they're close by. MR is not, just because of MR siting issues. But clinicians need to be able to walk to the radiologist. I call it the triangle: the patient, the doctor's office, and the radiologist. It's very efficient. They come in with a problem and it gets protocoled, it gets done, and they get an answer.

You see that with cardio and vascular because they're close by. Radiology is still centralized to a large extent. Radiology could be almost anyplace. I had to give access to the radiology reading rooms to nonradiologists, and it's been amazing that they actually have different logons so that they can get in and have slightly higher privileges than the regular house officers.

TRAINING AND TRANSITION

Drew: We've talked a lot about how practice for a radiologist changes with PACS. If you wanted to advise somebody who was going to install a PACS about the kind of training that should be given to radiologists in anticipation of the PACS, what would you do? What subjects would you try to teach them, and how would you go about it?

Levy: You don't have to know anything about computers to read off a PACS. One of the things that is poor about the way vendors bring these systems in is that they find a guy who has just finished his residency, or the most adept computer nerd in the department, and they have a "train the trainer" system. They train somebody who is very busy and say to him, "Go out and train your colleagues." They should pick the least computer-literate person and spend some time training that person, because somebody who is computer-literate picks these things up intuitively. This "train the trainer" stuff is poor.

Glenn: The vendor is not going to be there 24/7, so you have to have someone who is going to be an onsite expert. You want to have a cadre of people who can train other people and bring them along, a group of early adapters who are going to be your core group and who will then teach the rest of the group.

Channin: Those are good comments, but the "train the trainer" is a tried and tested model-it's the Pokemon model, right? It's the only model the vendors can support, because they don't have an infinite number of applications people.

In retrospect, we should have spent more time training the technical and administrative staff on the relationships of the information systems so that when they make mistakes-and they will make mistakes-they understand whom they have to call. Which downstream system is affected first? Which is the most critical? They can make mistakes in any number of places in the process, and it's important to train them that there are different repercussions.

Levy: We all look at these systems from our own particular point of view, and I try to look at them from the country radiologist's point of view. If you think that the IS department in a 100-bed community hospital is somebody who can teach people how to use computers, you're sadly mistaken. In fact, the IS team in a 100-bed community hospital is some guy who is classified to run a computer. If you have an IS person come in and teach that person how to teach me-well, that guy hasn't talked to a human being in a couple of years, and he's not real good at talking to people, which is why he's in IS.

Perez: The department needs to take a big role in training. The administrators, managers, supervisors, everybody needs to take a big role. Training is ongoing. It does not end, because if you're at an academic medical center, you've got a new staff in every July. Whenever you bring on a new partner, there needs to be training.

Drew: Is it the case that preliminary training is unnecessary, because people won't bother?

Perez: It is not unnecessary, but time needs to be given to the radiologists to do it. Where it falls apart is that the "train the trainers" need to be there. On the day you go live, there need to be people dedicated to helping the doctors in the reading room get through the tough things. Sometimes you'll find little glitches or problems. You know that when you go live with any system, the first two weeks are probably hell. Anybody who tells you, "We're going to plop it in Monday morning at 8 o'clock and things are going to fly"-forget about it. Change your vendor right away, because it's not going to be that easy.

Eusek: You can train all you want, but once you flip that switch you're going to find that there are a lot of things that have to be dealt with. Training the radiologists how to use the workstations is probably the most minor problem that we see. It's getting the front end, making sure your modality work lists are correct, that the technologists know how to use them. And then, on the other hand, we've been talking about how efficient the radiologists are and how we have all these tools to make them efficient, but actually, when you get right down to the bottom point, if you don't have the report out in 20 or 30 minutes, or an hour at most, it really doesn't matter. All of this workstation business means nothing; you've got to have that report out in a timely manner.

Levy: The product is the report, not the image.

Hirschorn: Don't bother training them until the system is there, because they won't remember; they won't care enough. Tell them they can't get a logon until they come for training, which is a big incentive.

For us, training is constant. We always have new people who need to be trained on the system. We've found that the best way is to say, "Here's the trainer; you set up a time that is convenient for you." That seems to work out very well for us.

Swischuk: You do have to babysit when you install these things. The vendors never spent enough time at our facility, and they sent people who worked the keyboard too quickly for anybody to understand what was going on. After that, we had to transition over to IS. We have a cadre of people in IS who were basically trained by the vendor, and they are the ones who are called when anybody in the institution is in trouble.

Hirschorn: I had one other comment to make about something we talked about earlier: The fact that all these technologies of PACS, voice recognition, etc., are enabling technologies that are supposed to boost our productivity. To some degree, though, it's "feeding the beast." We had multislice CT before we had PACS and it was just ridiculous-you had pages and pages of film to hang. So we got PACS to deal with it. But you know what? Now we can do even more CTAs and even more images, because you can take as many images as you want. Is that good? Is that bad? It's probably good, but it's not so clear.

The other main point is that this has enabled us to gain the internationalization of our reading staff. This is something that we are actively pursuing, as a result of PACS. We're hiring radiologists from countries across the world to be on staff, for night coverage and for cleaning up what's left over from the day.

Levy: For radiologists, it's vitally important to look the referring doctor in the eye, and if I had somebody who was living in Afghanistan read a report, I would be very nervous about having that doctor's imprimatur on the report. We think that the most important part of radiology is consultation, and we want the consultant on board.

Perez: I definitely believe that PACS is here to stay and it's going to get better. It's got to be much more intuitive so that the systems are easy to deal with, and the hanging protocols need to become a little simpler.

---

Sponsored by an educational grant from the Eastman Kodak Company
Part of a series about the future of medical imaging