A Business Case for Common Sense

For me, the return on investment of digital pathology is about more than just the business use case. It’s about how the system is used – how it truly meets the use case. Much of the time, people justify the need for a new system on paper but then don’t use it to the full extent of its capabilities. The idea is attractive, but the technology itself doesn’t fit into the overall workflow. People say, “Buy digital, start slow, and scale up,” but when it’s disruptive and poorly integrated, that’s hard to do – and then the scanner ends up gathering dust in a corner! That’s what I think differentiates good digital pathology setups from bad: how well they’re used. A bad system is purchased for novelty’s sake, or to meet a niche need; a good one facilitates faster, better, scalable work throughout the laboratory.

Versatile vendors

Labs need to view their vendor-client relationship as a true partnership when embarking on their digital pathology journey. Digital pathology is always evolving, so you’re going to need to work closely with your vendor to help you adopt and customize your system – and if you don’t have that kind of relationship, you’ve made a bad choice. That’s especially true because, with new technologies, you get what you pay for. You obviously have to set your budget wisely – but if you buy something inferior, it might not perform the way you want. My personal key features are “plug and play” simplicity and interoperability; I don’t want to be locked into a specific file format or application, because no one product can do everything, and you need freedom and versatility to grow.

It always comes back to integration. There’s a lot of vaporware out there that vendors have promised will integrate with other systems, or will have high uptime and low scan failure rates. But you only truly know what those numbers will look like when you flip the switch on your own system. Of course, you can’t buy a system “on spec” just to see how well it will perform – so instead, you have to speak to existing users and find out what they think. That can be difficult if you’re an early adopter like we are at the University of Pittsburgh Medical Center. There aren’t too many labs that have extensive experience with new systems before we do!

What is “fully digital?”

I have a problem with the “fully digital” concept. Ideally, if you’re making a commitment to digitization, then fully digital is the way to go – but I think that’s a misnomer. Most labs aren’t really fully digital; they’re mostly digital, because they exclude certain use cases like cytology or hematology. That means they still have microscopes, and the minute you keep microscopes around, you have a hybrid workflow, and it’s naïve to think you can avoid the need for extra personnel or extra work. But it’s also naïve to think you can go fully digital, because not every case is appropriate for scanning – and because there’ll be times when you need to troubleshoot, or when the digital system is down and you still need to deliver care.

But I think that actually helps with the transition. Initially, pathologists need to learn to trust the new system, so they run it side-by-side with the old. They don’t give up the microscope completely because it’s a reassurance. And I think that’s exactly how we should be practising. We shouldn’t be cowboys! We shouldn’t be too gung-ho about diving headfirst into digital; it’s going to take time, and it’s wise to have a backup. I just think we need to be aware that when we say “fully digital,” what we’re actually aiming for is “almost fully digital.”

A fundamental shift

Over the last 10 or 15 years, I’ve noticed a shift from hardware to software to content. In the beginning, there was an emphasis on improving scanners and extending their capabilities; later, people began focusing on image management and sharing software. But in the last five years or so, there’s been an upsurge of interest in algorithms, apps and analytics. That’s shifted the return on investment. In the beginning, when we were talking about hardware, everyone wanted to scan their slides. Then, when we started to think about software, it was worth going digital because the ability to store and share images meant we could introduce efficiency and explore telepathology. Now we’re looking into what we can do with apps and algorithms that a microscope could never do – things like image analysis or content-based image retrieval.

My analogy would be the cellular phone. In the beginning, people worried about which device was the best. Then, they worried about which operating system was the best. Now they don’t really care about either – what they care about is the apps they use, and I think digital pathology has followed a similar path.

Weighing worth

I think telepathology is still the area where digital pathology delivers the best return on investment. It allows you, as a pathologist, to do three things:

1. Balance your workload, so that you can be more efficient. Labs want to get their results out quickly and cheaply.
2. Provide coverage and care where you couldn’t before, so that you can extend your reach. Covering frozen sections at remote hospitals allows surgeons to operate there, whereas previously the patient might have had to travel.
3. Centralize your services, so that you don’t duplicate costly services in multiple locations. For that matter, you don’t need to duplicate pathologists, either; instead of having expert generalists at every hospital, you can subspecialize, sending each pathologist the cases in which they excel and reaping the best return on investment.

Equally, some things are not worth the expense right now. For instance, I no longer think it’s worth scanning every slide you encounter. It doesn’t make financial sense if you have ready access to the physical slides; it’s only necessary when you don’t. I also don’t think labs should be investing in algorithms and apps that aren’t FDA-approved. Vendors offer us these things, but they don’t take them through regulatory approval first, which means we can’t base patients’ treatments on the data they provide, so I don’t think they offer a good business case for a digital transition. Before we can use those apps in a clinical environment, someone has to spend the dollars needed to gain FDA approval.

The other thing that ends up being both frustrating and expensive is that digital pathology systems tend to be standalone – so they don’t work well with existing technology. It places great demands on IT staff to learn, integrate and maintain this “special” system, which results in high indirect costs. It’s like owning a Ferrari – you buy it because it’s beautiful and runs well, but when it’s time for a service, you realize how expensive it is to maintain. Indirect costs make up the majority of digital pathology expenses; you have to consider who’s going to maintain the equipment, who’s going to scan the slides, where you’re going to store your images, how you’re going to change your facility and network setup, and even what peripherals you’ll need in addition to the scanner.

But those investments don’t go to waste, because there are significant sources of potential savings and income as well:

       1. Insourcing new business – whether it’s offering second opinions via telepathology, acting as a reference lab, or digitizing immunostains for clients; all of that can bring in new business use cases for your lab.
       2. Hiring fewer pathologists in areas experiencing staff shortages. Multi-hospital systems can replace some pathologists with scanners, so that instead of keeping consultants in labs with low-volume workloads, you can install a scanner and send images to consultants at central locations.
       3. Reducing inefficiencies in the lab. Our time and motion studies found that pathologists can waste more than 13 percent of their time on administrative duties (1), with travel and archived slide retrieval on top of that. Digital pathology eliminates a lot of those inefficiencies and increases the time pathologists can spend working.

To come up with a hypothetical dollar amount, we did a novel study looking at the top 12 misdiagnosed cancers in our healthcare system (2). Then we looked at the potential cost of making a misdiagnosis – not just in terms of litigation, although we included that, but in terms of over-, under- and delayed treatment as well. We compared that to the cost of buying a digital pathology system and rolling it out over five years, so that every hospital was connected digitally and cases could be distributed to subspecialists for faster, more accurate diagnosis. When we worked out the net difference, our system saved approximately US$18 million! Based on that calculation, our administration gave us the green light to roll out digital pathology – and in five years’ time, we’ll know whether or not we were right!

Four for the future

I see four things in digital pathology’s near future.

First, FDA approval is going to have a major impact. It’s going to encourage widespread adoption of digital pathology, but only then will people realize that this is just the beginning. Once we have approved devices, we’ll need approved software, approved apps, and so on – an endless crusade.

Then, I think we’re going to see more and more new vendors entering the market. I’m talking about companies like Facebook and Google, because they already have the technological chops – they just need to apply them to healthcare. Once they see that we’ve overcome the major hurdles, they’ll likely start applying their tools. Hopefully, that will drive down the price of digital platforms, because the emphasis won’t be on the hardware; it will be on the applications, which brings me to the second thing that will feature in the next generation of pathology. When this happens, we’ll finally have the opportunity to see some real “killer apps.” Killer apps sell platforms – take the Atari system, for instance; it wasn’t very good, but when Space Invaders came out, everyone ran out to buy an Atari. Digital pathology needs the same thing, and I think we’ll see it happen when some other, bigger players join the field.

Next, I think the industry will be forced to standardize. As we adopt digital platforms and integrate with more and bigger systems, we’ll have to fit into enterprise imaging initiatives. I can’t say which standard format will be adopted, but there will definitely need to be some standardization – and I think it’s long overdue.

Finally, I’m concerned that venture capitalists investing in digital pathology may think twice about continuing to do so. Many companies are already in their second round of funding, and the things they were promised – a fantastic market with millions of users and a billion-dollar industry like radiology’s – hasn’t really come to pass. So when those companies go after another round of investment, the funders may be reluctant if the promise of digital pathology hasn’t materialized.

So what can we do? We can use technology appropriately, plan ahead before buying, ensure that we have flexibility, and pay careful attention to what does and doesn’t make sense. If we approach digital transition sensibly and for the right reasons, there’s a lot to be gained.

Liron Pantanowitz is Professor of Pathology and Biomedical Informatics and Director of Pathology Informatics, University of Pittsburgh Medical Center, Pittsburgh, USA.

Read more:

The Doubting Dollar
By Luke Perkocha

A Roadmap to the Future
By Alexi Baidoshvili

Part of a Larger Whole
By Marcial García Rojo

Techniques in Transition
Michael Schubert interviews David Snead