An Enterprise Imaging Journey Starts With Defining Your True North, Then Throwing in a Helping of Flexibility, says Hartford HealthCare Chief Clinical Innovation Officer & Chief Medical Informatics Officer Dr. Barry Stein

In a recent interview with healthsystemsCIO, Dr. Barry Stein, Chief Clinical Innovation Officer and Chief Medical Informatics Officer at Hartford HealthCare, shared insights into the organization’s enterprise imaging strategy. Dr. Stein and Anthony discuss:

• • • • How Hartford is streamlining access to patient images across their vast network, and how Epic fits into the process
      • The importance of balancing stakeholder needs
      • Integrating artificial intelligence for personalized care
      • How to continue the evolution of enterprise imaging to include additional data modalities, such as digital pathology and intraoperative video, while maintaining a focus on interoperability and patient-centered care
      • Hartford’s application rationalization process

Watch or listen below; scroll down to read 

LISTEN HERE USING THE PLAYER BELOW OR SUBSCRIBE THROUGH YOUR FAVORITE PODCASTING SERVICE. 

Podcast: Play in new window | Download (Duration: 36:27 — 25.0MB)

Subscribe: Apple Podcasts | Spotify | Android | Pandora | iHeartRadio | Podchaser | Podcast Index | Email | TuneIn | RSS

Anthony: Welcome to healthsystemsCIO’s Interview with Dr. Barry Stein, Chief Clinical Innovation Officer and Chief Medical Informatics Officer. Dr. Stein is also a practicing vascular and interventional radiologist with Hartford HealthCare. Dr. Stein, thanks so much for joining me.

Dr. Stein: Thank you, Anthony. Pleasure to be here.

Anthony: We’re going to talk today about Enterprise Imaging but first, if you want to tell me a little bit about your organization and your role.

Dr. Stein: Hartford HealthCare is the largest integrated healthcare delivery system in Connecticut, over 40,000 employees, 500 locations and it’s a fully integrated healthcare system taking care of patients from birth until the end, across multiple specialties, in fact, all specialties in a very comprehensive way. Our focus is improving access to healthcare, affordability of healthcare, improving equity and excellence, meaning quality and safety. Those are our 4 strategic pillars and the vision at Harford HealthCare is to be the most trusted, more personalized coordinated care.

Anthony: Excellent. As I mentioned, we’re going to talk about Enterprise Imaging which is a term that would be interesting to hear you define.  What does it mean and why are we trying to get there and then what are some of the things that we do that further that goal.

Dr. Stein: Enterprise Imaging is a concept that has been around probably for more than a decade and it really started with PACS. Very quickly many healthcare systems organically developed a hodgepodge of different PAC systems, different ways for clinicians to see images and as healthcare systems grew, it became more and more difficult to seamlessly provide access to the entire imaging history of a patient for the clinician.

In other words, if a patient went to one hospital with one PAC system, and needed to go to another hospital in the same healthcare system, because the different PACS are not connected on one particular platform, it became extraordinarily difficult for clinicians to have a full picture of the imaging history. When the electronic health record came about, most of the electronic health record companies did not have an imaging platform. They certainly had a platform to collect structured data and non-structured narrative data but not imaging. It was a significant element that was left out of the strategy of any electronic health record.

At Harford HealthCare, we have 7 acute care facilities and over 500 ambulatory facilities in over 50-odd imaging centers, and you can imagine how difficult it would be if we didn’t have one enterprise imaging platform. Our patients would have their entire imaging history in the palm of their hands and the radiologist reading the study, even if they are in different groups, would have access to the history of the patient’s prior studies to compare to from any facility within our enterprise.

That really was the goal and the reason why we developed our enterprise imaging platform, to consolidate all imaging on one platform to achieve the goals through three different lenses. The first of which was the patient. We didn’t want the patient to carry CDs to get the images to the next facility. We wanted our patients to have access to their imaging history and reports 24/7, the moment the imaging was acquired and the report was signed out.

For the referring clinician, we didn’t want them to have to search and wonder where imaging was performed in our large integrated healthcare delivery system, we wanted them to have access to the entire imaging set within Epic, which is our electronic health record. When they look at their particular patient and they wanted to see imaging, they click on the image in a unified viewer within Epic and they see the entire history including reports from every facility.

That’s the second lens, the first was patient – the second is referring physician and the third lens is for our radiologist, the interpreting specialist, to have access to the prior studies from all parts of our organization to render high quality reports.

Anthony: Very good. Is one way to think of this moving in stages? I interviewed another individual in a position similar to yours and what he said was a first step is getting every particular specialty down to one PAC system. You could theoretically have cardiology, radiology, a number of different PAC systems within that specialty, like you mentioned at different hospitals.

It sounds like you’ve achieved some really significant milestones and I want to give people a paradigm of some high level ways to move in this direction.

Dr. Stein: I think we did it almost the other way around. We created the vision that I’ve just explained to you. It’s necessary to think about this through three stakeholder lenses – the patient, the referring physician and the radiologist. For us to accomplish that, the ultimate goal would be to have images in a central archive in the neutral archive that is integrated with our Epic healthcare system and start there. That was the vision that folks understood and really wanted.

The second piece was we can’t do it all at once, let’s break it down into different sectors. We started with radiology. The next was cardiology. The next was point of care, ultrasound and imaging. The fourth one was visible light, taking pictures with your camera and your iPhone. Right now we are in the process of doing pathology ultimately with the same goal. That if you click on a patient’s name in the EHR, that loads the unified viewer and you can see radiology, cardiology, visible light images, point of care ultrasound, digital pathology, et cetera.

When it comes to deciding should we have a unified PACs with each ology, we realized that an organization may have different groups that are very much devoted and comfortable within PAC systems, that’s a resistant point that could slow the whole process down. With modern day technology, it’s very feasible to have multiple PAC systems because we want to make sure each stakeholder is comfortable with in getting what they want. If you have 5 radiology groups or three radiology groups and each had a different PAC system and they’re comfortable and good and efficient within their system, that should be okay. What we requested, and everybody agreed, was for the images in their facilities to go to the VNA so then the patient could have access to the images no matter what PAC system is irrelevant to the radiologist or the clinician.

We developed our architecture so we could accommodate different PAC systems. Now, I’m not going to say to you that having different PAC systems is frictionless; it requires a little bit more effort and technology to make sure that the interoperability is smooth. There are ways to do that that are elegant and certainly achievable. In our radiology system we started with five different PAC systems, most consolidated into one but there’s a large, large group that’s on another and we continually work out the small kinks but ultimately, they work off their own PAC system, the radiologist are happy. They see all the images from the rest of the enterprise. Their referring clinician is seeing all the images in Epic and the patient can open up their MyChart and see the reports and images and they don’t have to worry about different systems. We chose that route.

When it came to cardiology, we wanted to have one main system, but the one or two other systems we can accommodate. Point of care ultrasound, the same thing. Digital pathology, we have two pathology groups, we may need two different systems. When it comes to visible light images that you may imagine, people are using different devices to take pictures. We felt that that would be a major barrier to progress and I believe that is one of the key decision points that allowed us to accelerate as quickly as we did, remove as many barriers through each respective stakeholder lens to get to the ultimate goal. One VNA, absorbing all the images from across the enterprise that allowed our patients to get images, our referring physicians to see it in Epic and our radiologist to work in an environment that they’re comfortable or cardiologists, or pathologists.

Anthony: It sounds to me like this is a fundamental decision point that organizations have to make around strategy. What you have described is extremely interesting because from what I’m hearing, your goal was to let people use the PAC system that they are comfortable with, so you have less pushback. It is always difficult to get people to stop using a tool they are comfortable with and we talked about this in all kinds of applications. The concept is called, I’m sure you know application rationalization, going from many to fewer.

Dr. Stein: You hit the nail on the head. I would say the application rationalization is a huge driver in our organization for all the reasons that your audience is pretty familiar with. If the goal to achieve Enterprise Imaging through the stakeholder lenses that I just mentioned is the primary goal, it’s always compromised in a large multi-disciplinary organization. For us to move forward, we have to compromise.

I would say the majority was around application rationalization where we had a large radiology group and we still do. That was very much wedded to their PAC system for other reasons and we worked on the interoperability. I think that for organizations that want to move ahead quickly, pick your battles very, very carefully because that one could really stop it dead in the water. And then who loses out? – The patient, the referring physician. And at least at Hartford HealthCare as I mentioned, the vision is to be trusted for the most personalized coordinated care. That’s our north star. Are we going to allow the friction between PACS systems interfere with achieving that vision? It would be a hard argument to make.

Anthony: The decision points here are fascinating, different than the EMR. The reason Epic is where it is, is because we don’t do this on the EMR side. The industry has decided we will not have multiple EMRs in the health system, we will have one because interoperability I guess is not all. It could be also part of the reason is that data transfer from one EMR to another is not as difficult and onerous as moving images from a well used and old PAC system into another one. The files are massive. It’s a different proposition. I think the data transition is easier with EMR data than it is with PACs data. That’s one reason we may be okay living with more than one PACS is because the archiving and data transition is so onerous. Does that make sense?

Dr. Stein: Yeah. There’s a huge difference between comparing consolidated EMR and having multiple EMRs versus having multiple PAC systems. When you think about the stakeholders and the audience that use the EMR, it’s every clinician in the organization. If you, as a clinician, a referring physician, are working in multiple facilities, ambulatory and acute, you want one system to go into to find all your information. It’s key. It’s not only important from a clinical standpoint but from a revenue cycle standpoint, from a compliance standpoint, from a quality data standpoint, everything.

When we’re talking about PACS, we’re talking about a tiny segment of the clinical population of a large healthcare system, meaning the stakeholders are small percentage of the larger group of clinicians. Most radiology groups, if they’re going to be moving different hospitals, they want to be in an environment that they’re comfortable with. It’s not going to impact the clinician and it’s not going to impact the patient. In fact, it’s going to help him.

I don’t believe the comparison is an equivalent comparison at all. It’s, to me, night and day. The two function differently, the stakeholder – the amount of stakeholders that have to be entertained with a consolidated EMR versus the few on the PACS is what – I don’t even know that the ratio, but very significant.

Anthony: And that’s why the strategies are different.

Dr. Stein: Yeah, the strategies are markedly different. At Hartford HealthCare, we have Epic throughout. For all the reasons that I just mentioned and a whole lot more that everybody else knows. We’re very grateful that we have one EMR because it allows our clinicians to see everything in one system.

Anthony: Can you talk about the integration with Epic? Is it individual PAC systems feeding into Epic or did the PAC systems feed into the VNA that then feeds into Epic? What does it take to have a good solid integration there?

Dr. Stein: It took a little bit of work upfront but now it’s a well-oiled machine. Basically, the fundamental premise is we have a Sectra VNA, a Sectra enterprise imaging platform, the Sectra platform and we worked very hard with Sectra and Epic to integrate Sectra’s unified viewer into Epic. Easy launch and it allows access to any image on the Sectra platform. If the images from any PACS are pushed to the platform and they’re on the platform, you should then be able to see them through the unified viewer in Epic because Sectra and Epic are tightly integrated with the VNA and Epic. Any image that’s on the Enterprise Imaging archive is visible in the viewer, simple as that.

Now, obviously, with some technicalities but that‘s not a major barrier at all. It’s not basically saying ‘okay, you use X PAC system, therefore that system has to then work with Epic to integrate the viewer.’ No, no, no. The first and only step is for us to get images, current and past, onto our VNA. Once they’re there, we can access them through our unified viewer that’s integrated with Epic. No matter what system the specialist is reading from, it’s irrelevant.

Anthony: What’s next for you around this area? What are some future goals, further integrations or your vision of imaging, how you want imaging to work at Hartford HealthCare three years from now?

Dr. Stein: Again, I’ll go back to keeping in mind trusted for personalized coordinated care. Personalized is a word I’ll come to in a moment, and how we’re going to continue to do that and coordinate it. The patient needs to be able to see the images wherever they are in the world, whenever they want to, easily. Number two, our referring physician should have access to anything they order any time and our radiologist and our interpreting specialist, cardiologist, pathologist need to have access to the latest and greatest technology moving forward.

With that being said, we’ve done radiology, we’ve done cardiology, we’ve done point of care ultrasound, we’ve done visible light, we’re doing pathology. What’s in the future? We’re going to continue to work on pathology which is not totally mature, it’s almost 10, 15 years behind radiology, and also understanding and then we would like to consider all the video, intra-operative video, et cetera, also being stored in the same environment so that everything’s accessible in the unified viewer.

That’s the coordinated care. The specialist, the referring physician and the patient are all tightly coupled here. They were given what they want and need, 24/7, contemporaneously, tightly coordinated. How do we make this more personal? Artificial intelligence is going to allow us to personalize care much more accurately and much more granularly. Where does radiology fit into the picture? {Excuse the pun}

Radiology has always been at the tip of the spear in terms of artificial intelligence in the diagnostic capabilities to improve it as well as the quality and safety of the care that we provide – at the tip of the spear, far ahead of almost any other specialty. With that being said, we haven’t unlocked the full potential of artificial intelligence to personalize more care that’s imbedded in every single image that we acquire. For us to unlock their full potential, we need to have our images in one platform, organized and architecturally designed that when we’re developing algorithms. It goes beyond just being able to diagnose an intracranial hemorrhage, a pulmonary embolus, aortic dissection.

It goes far beyond that and at Hartford HealthCare, we have a center for artificial intelligence in healthcare. With the vision and goal of unlocking the full potential of AI for our patients in a safe and trustworthy way. That includes understanding that you can’t do this alone, so we have world class ecosystem partners to help us develop algorithms to predict and prescribe and optimize patient care in the likes of MIT, Oxford University, Google, et cetera.

Number two, we’ve done an enormous amount of research around this domain including most recently using multi-modal data to increase the predictive capability of these algorithms for disease process. It goes beyond just using radiology. You need radiology, you need EKGs, you need structured data, you need unstructured data. If you think about it, we’re using multi-modal data now to improve the personalization of the care we get and radiology is such a fundamental part of that. As we think about enterprise imaging, we have to think about how that connects with the multi-modal data strategy. An organization has to allow algorithm development, deployment, validation, et cetera.

The third component to our center is governance, around when do we allow AI into our organization. The fourth piece is how we integrate this in a very disciplined way in our enterprise, such that our clinicians who are using these algorithms understand what goes into developing them, what it is, what it isn’t, how to recognize a problem when it occurs, how it may change and how do you monitor it. There’s also an enormous amount of education.

The last pillar is what we call our data layer. We have a partnership with Google Cloud platform to develop that data layer. It allows the multi-modal data, organized in a way that we can unlock the potential of AI for our patients in a safe and trustworthy way. With our enterprise imaging platform, with our Epic environment, neither one has the capability to harmonize all that multi-modal data with the capability to develop algorithms, run algorithms, monitor algorithms. We need to figure out an efficient way to connect and transfer information to these multi-modal data platforms from the EHR, from the imaging enterprise platform and that hasn’t been completely worked out. Which is the future, and their future is all about unlocking the power of AI in a safe and trustworthy way, to provide personalized accurate care for our patients.

Anthony: Excellent. I want to get one more question in, Dr. Stein, and then I’m going to let you go. I’m wondering where the push for this comes from, if it comes from the clinicians and the users in the ologies. You’ve got the practitioners and the clinicians, you may have informatics folks embedded in those specialties depending on how the organization is set up, you may or may not, depending on if IT is embedding informatics folks in those disciplines and I’m not sure who they report up to.

Then you’ve got folks like yourself that bridge that gap between the IT world and the clinical world, the two titles you hold and the CMIO title. Then you’ve got people like Joel Vengco, your CIO, who is living firmly in IT. When it comes to enterprise imaging, is there a typical way this push gets started and then what are the keys for both the CIO role and the CMIO role to move this forward? It seems it could go any number of ways.

Dr. Stein: Yes. Great question. I think, historically, because Epic was such an expensive roll out for not only our organization but every organization. Boards have to approve hefty checks and Epic has a very prescriptive way of deploying Epic, it’s almost like a tactical kind of thing. IT did a lot of that over the last decade in most healthcare systems, pushing technology to the clinicians. We all know there’s a huge revolt in the country around technology burden that hasn’t made their lives a whole lot easier a lot of the time. I think most people would agree that at least we got the data and everything in the one place so that it improves quality and safety. But it has increased the physician or clinician burnout very significantly.

When I personally took over 6 years ago, Informatics falls under the clinical affairs department, reports into Ajay Kumar, our Chief Clinical Officer, and we represent and support and advocate appropriately for our clinical partners. Our organization flipped the operating model on its head the following way. We made a compact with our institute leaders. We’ve got 7 institutes, heart and vascular, cancer, neuroscience, behavioral health, bone and joint, kidney. We’ve got 17 clinical councils that standardized clinical care across the enterprise in four different regions.

We basically said the following: clinicians are now going to surface the problems they want solved, they’re going to work with Informatics members that are under Clinical Affairs and are embedded in each institute, each council, each region. You’re going to work together to crystallize your problem statement, what problem are you trying to solve, and then once we’ve done that, we prioritize what’s the most important problem, you’re going to have 50 things or 60, you’re going to have a few and then we work with IT partners like Joel’s team to find the best solutions and in partnership with the clinician champion that’s surfaced to the institute, we work in an iterative way to make sure that the solution solves the problem.

Fundamentally, this isn’t putting a ticket to the loudest voice. You must have a crisp problem statement that is a priority in the organization, that is a consensus, signed off by leadership, worked together with Informatics and then we go to IT who are the experts in the technology to help us find the best technology. That has worked magnificently to get clarity as to what are the priorities and it’s not three or four different ways in. The only way you can get your problems solved by technology in our organization, from a clinical standpoint, is the way I’ve just described. We’ve empowered the clinician with caveats. They’ve got to work on their problem statement. I’m not interested in the solution and they have to prioritize it. If you can’t do that, then expect somebody else to do it for you.

Anthony: That’s fantastic, Dr. Stein, great advice for our listeners. I want to thank you so much for your time.

Dr. Stein: You’re very welcome. It was my pleasure.