Clinical care delivery is increasingly dependent on technologies that have increased in number, variety, and complexity. Cloud technologies have become the backbone, supporting the drive to innovate. Cloud-hosted EHRs can optimize clinical efficiency, enhance organizational security, and ultimately have a positive effect on a health system’s bottom line. With the power of integrated services and technologies like Amazon Web Services’ (AWS) continuous release of new EC2 Instances with more compute power and productivity, cloud transformation is rapidly becoming the industry standard. Health systems looking to strategically transform their operations by doing ‘more with less’ are finding cloud-based transformation as a beneficial new asset.
On today’s episode of In Network’s podcast feature Making Rounds, Nordic Head of Thought Leadership Jerome Pagani, PhD, sits down with cloud solution experts from both Nordic and Amazon Web Services (AWS). From Nordic, Digital Health Cloud Practice Lead Saran Sonaisamy and Principal Architect Brijeet Akula; from AWS, EHR Leader Mike Hegyi and Global Epic Leader Matt Dinger. They chat about cloud infrastructure, the benefits of moving the EHR to the cloud, and the key points any healthcare system should consider when transitioning its IT infrastructure to a cloud-based platform.
Listen here:
In Network's Designing for Health podcast feature is available on all major podcasting platforms, including Apple Podcasts, Amazon Music, Google, iHeart, Pandora, Spotify, and more. Search for 'In Network' and subscribe for updates on future episodes. Like what you hear? Make sure to leave a 5-star rating and write a review to help others find the podcast.
Show notes:
[00:00] Intros
[01:35] Benefits of moving the EHR to the cloud
[05:43] Current advancements in cloud-based EHR capabilities
[11:30] Integration of data from different sources
[15:35] Matt’s personal healthcare journey
[17:55] Real-world use cases for cloud-based EHRs
[25:02] Financial opportunities with the cloud
[28:46] The rapid uptake of cloud adoption
[31:38] Concrete first steps in a cloud journey
[35:01] Outros
Transcript:
Dr. Jerome Pagani: All right. I'm so excited for today's podcast. Today, we have two guests joining us from AWS, Mike Hegyi …
Mike Hegyi: Hey Jerome, thanks for having us.
Dr. Jerome Pagani: … and Matt Dinger.
Matt Dinger: Happy to be here.
Dr. Jerome Pagani: And two of our digital health experts, Saran …
Saran Sonaisamy: Happy to be here, Jerome.
Dr. Jerome Pagani: … and Brijeet
Brijeet Akula: Hey thanks for having me.
Dr. Jerome Pagani: All right. So we're going to dive in today and talk about the cloud and cloud infrastructure. For healthcare, one of the most important investments in cloud infrastructure is around the EHR. So, what do we mean by moving the EHR to the cloud? Specifically, what are the benefits for a health organization?
Mike Hegyi: Yeah, I think the first thing we need to do is really talk about why. If we take a look at the EHR, we're seeing a continued delivery of new features that are actually creating a challenge for some of the IT leadership in creating infrastructure. But if you actually look back on it, these new features are really designed to help improve the physician and the nurse workflows. A lot of times, what we're doing is, is as we get the computer to do more work, the physician, the clinician does not have to do that work. And so, we're seeing an increase in the compute to be able to satisfy that. I've been at this for four decades, and we've seen this cycle increase in frequency to the point that health systems EHR is going to grow 20% year over year. That means it's going to double every 3 to 4 years. The way that organizations try to combat that on-premises is they try to predict what they're going to need in hardware two, three, four, five years from now. And Epic, for example, one of the vendors, does a really good job of giving health organizations some idea of size of what it might be like in two years. So what organizations are doing is they say, okay, on-premises, I need to buy servers, I need to buy storage, and I want to make sure that I can amortize the cost of that over a five-year plan, which is great from a financial perspective. But the thing of it is, after years two, you're starting to run on really old hardware, and you know, the idea of trying to project … It works great if you know, you're getting those releases every 18 to 24 months, but they're getting it every quarter. There's a better way of dealing with this. To be able to use an infrastructure that allows you to deliver just what's needed now, and as you grow, add to it and to be able to completely refresh the hardware … is when the new technology is available, not have to lean on that. The amortization of the costs, so you don't take on that technical deck and then improve availability through a much more reliable infrastructure. Really what you're doing is you're transforming an EHR like Epic from electronic health record system into a lab, into an elastic health record system, one that leverages elastic compute, elastic storage, and elastic networking. That really is what the Epic on AWS solution is geared to do.
Saran Sonaisamy: I totally agree with you, Mike. I mean, the advancements across the stack, that is basically what is making us look into cloud from a hosting of EHR standpoint. Jerome, you know, we had spoken about advancements earlier and this means an enterprise-level advancement not necessarily on a particular compute section but having a need to advance all of it from a tech standpoint. That is a good value addition for our customers in considering hosting EHRs like Epic on the cloud.
Dr. Jerome Pagani: So other industries have been on the cloud journey for quite some time and are a little more advanced than we are in health care. Brijeet, what advancements are coming online now, and how is that resulting in the rapid pace that Mike just mentioned?
Brijeet Akula: Yeah, absolutely. So especially in an EHR-on-cloud workload, one of the key components of your EHR system like Epic is your operational database. It requires a very high performance. There is an incredible amount of IOPS that are needed, read-write operations, because it is typically serving up thousands upon thousands of concurrent connections per second. So, we need to have compute capabilities in your cloud vendor that can meet these demands. So, one of the one of the recent instances of compute instances that have been released, it's able to support one of the higher-level workloads that EHR customers, Epic customers have. So 42 million GRefs, it's a specific specification on the high level of read-write operations that the database performs to be able to keep up with this demand. So now, you know, this is a very exciting time because we have the raw compute power that's able to facilitate this very, very, high workload and keep up with that increasing demand. Then there's also third-party companion apps that we can talk about.
Matt Dinger: Yeah certainly, and, you know, as far as those apps are concerned, making sure that low latency is really what you're seeing across the board, regardless of where the health system is being hosted, is critical for these organizations to be successful. I'll tell you, I did not really think that a human could understand what a microsecond is until you are standing in front of a physician doing their workflow waiting for a smartphone to load. Let me tell you, you feel every microsecond. A lot of the way that this is accomplished is through things like Direct Connect, which creates a connection across the shortest path between you and your resources running in the virtual private cloud. Which results in the lowest latency, the most consistent bandwidth. It's really these dedicated connections that are the best option for mission-critical workloads, like health care. I don't know anything that's as mission-critical as that.
Saran Sonaisamy: The other thing to add is basically how do we automate those establishments? Advancements always happens over the period, but how do we set that up? To Mike and Brijeet’s point, we have something called a landing zone accelerator. How do we establish cloud for our customers first? It is this particular arrangement that is fully automated. That means a customer has to go towards a single click establishment of the entire cloud environment for them to host EHR. This puts up a good, significant, advantage for customers who look at cloud and looking at cloud for a further rapid acceleration for EHR workloads through AWS.
Mike Hegyi: And when we talk about EHR workloads, I think what we really have to realize is that the EHR chart isn't the complete health record. When you think about it, if you’re really trying to deliver the complete health record to it, it's the EHR, it's the PAC system, those imaging system that's used for radiology. There's the document management systems that are used to track, there are coding systems, blood bank interface engines that loop it all together. So when we look at operating an EHR and providing it the performance and the reliability, having them run on the same type of cloud platform that allows you to, if you need to fail over for disaster recovery purposes, you're failing over the comprehensive health record, not just the EHR.
Brijeet Akula: Yeah, that's a really good point. This level of flexibility and agility is really important and really effective in this environment. So especially as your workload spikes or you have, for example, on the weekends, there's just not as much demand in some of these systems. You don't need to have your entire infrastructure running at all time. More importantly, when there is an event, when there is a disaster event, that all of a sudden has a spike in patients in the hospital. Now, this is really the time where your systems are being stressed, right? You have an incredible amount of volume. There's incredible amount of activity happening in your EHR, your PAC systems, and all of these things. So the benefit with this modern architecture and automation is that we have the capability, essentially on a one-click basis, to all of a sudden double, triple, quadruple your infrastructure. All of a sudden, if you just need to increase capacity, it's very much possible, and we're talking about like 30, 45 seconds. All of a sudden, you have doubled your capacity to be able to serve those needs, which is really important, that agility is just amazing.
Dr. Jerome Pagani: So Matt, the EHR isn't the entirety of the health record, I'm reminded that our definition of health is becoming more holistic. We're seeing more types of data and more variety of data, and a greater volume of data being incorporated into the health record. So, things like patient-generated data, stuff from Internet of Medical Things, continuous monitoring data … how does cloud infrastructure support the integration of those data types and the variety of data in a way that on-prem storage can't?
Matt Dinger: Sure. It's the idea and concept of hosting this as a healthcare ecosystem. You know, you're talking about information from the patient in terms of wearables, but you're also talking about information that the health system is using maybe third-party applications or other information that they want to incorporate, whether that is social determinants of health, whether that is weather patterns that can help predict, no-shows or, you know, any number of things that come in. Some of that is incorporated in the elastic health record. Some of it is not. So putting that together in a singular ecosystem allows analytics to come together in kind of one place to be evaluated, allows you to leverage innovations in the cloud like advanced medical imaging, genomics, AI, ML, all of that is coming with this, you know, cloud center that you have built to be able to really take action on that, to empower the clinicians and the other end users to really create the best patient experience possible.
Brijeet Akula: I can expand on that. So especially, you know, the cloud platform enables a lot of this kind of innovation. So we have some real-world use cases where Saran and I worked with some researchers that were very much interested in getting Fitbit data. But the interesting thing here was they wanted the raw telemetrics that are coming from the device, the Fitbit itself. I think it has like 20-some things it picks up your temperature, your heart rate and things like that, your movement. The out-of-the-box experience, it gives you aggregated information. So it'll say something like over the last minute the average heart rate was X, or over the last minute, the movement level was Y. But researchers, as we know, they always want as much of raw data as possible so they can sit there and do their calculations. This is one of the things that cloud enables because when you start, you can certainly do this in an on-premise environment, but as soon as you want to scale from, let's say, five patients to 100 to 1000 patients that are wearing this device and generating this information, you will very quickly start running out of, for example, physical capacity for storage. That's an incredible amount of storage that's coming in. For example, the one solution that we did, it was collecting data at 110 hertz, which means 110 readings per second of these dozen or so metrics coming from the Fitbit. So, when you compound that per second, per a thousand patients, for at least a month worth of data, that's an incredible storage amount. Maybe you may have the data space in your data center today, but six months from now you will very quickly run out. So, the elasticity of cloud is what really enables this, that you can just keep pumping in more data, do your calculations, and then you can tear the environment down.
Matt Dinger: And if I could just put a personal take on this, I am a patient with type one diabetes. I was diagnosed when I was five years old, and it used to be that I had a sheet of paper and I would do a finger stick three or four times a day, and I would write that down. Well, let's be honest, my mom would write that down, and we would go into the physician's office every six months, hand him the paper. He would, in a few minutes, try to analyze that as closely as he could and as carefully as he could and provide care and advice and guidance to me to do that. That's certainly not a knock on the physician. They did the absolute best job that they could. Incredible care given for the technology that was available to them at the time, which was literally a sheet of paper. Now I have a continuous glucose monitor. It goes directly to my cellphone; it integrates with my insulin pump, and it provides a closed loop. Most importantly, that information goes directly to my physician's office. They can take a proactive approach, send me a message through the patient portal, say, Hey, you know, this indicates you're running high between ten and eleven pm in the evening. What do we need to do to improve that? Is that, you know, are you not getting enough insulin at dinner? Is it something that's wrong with your background basal rate? In addition, my husband can take a look at the data on his own phone, send me a message, and say, Hey, you know, what the hell are you doing? Yeah, you know, you need to take care of this. And I kid, but he can configure his own alerts. I contrast that to the experience that I had as a child where my mother quit her job, essentially to sit by the phone because she was terrified that something would happen to me while I was in kindergarten, waiting for something to happen. So, all of this holistically, whether it is the patient experience, their family’s lives and time, or the care that a clinician can provide. All of that is made better by technology and technology that the cloud can provide.
Dr. Jerome Pagani: So that's fantastic. What I'm hearing are some real-world clinical and research applications for the speed, power, and extensibility of cloud infrastructure. Saran, are there other sort of real-world use cases that apply maybe on the back-end or otherwise today?
Saran Sonaisamy: Absolutely. Jerome, I think before even I come to the use cases related to EHR, I wanted to add another point, which is pay per use. I mean, what Brijeet and Matt were talking about is scalability, but another side, is allowing our customers to pay for what they use when they really scale it up, they pay it, and then they take it off, they don't pay it. So that's a key concept from a cloud standpoint. In terms of the EHR related use cases. There are four major solutions that are there in the market, and this is completely evolving, you've seen the advancements recently. The first one to talk about is the production workload. A customer can migrate their production EHR workload to AWS. That's the first evolved solution and then the next one is the disaster recovery. We have to think about disaster recovery in, you know, on-premises, you are thinking about talking about a data center, a secondary data center, and equal and parallel infrastructures. The third option is thinking about hosting the training environments. And we know that a lot of environments that that doesn't work great for EHR applications. So, we can host of all training environments on the cloud. The fourth critical use case is also about cloud read-only. There are advantages in positioning a read-only environment in cloud, which I would ask Mike to just talk about and to call up some advancements.
Matt Hegyi: Thanks Saran. You know, health systems face a growing threat from ransomware that is really designed to make systems inaccessible, disrupt care delivery, and potentially impact patient outcomes. So, when health systems discover ransomware, what they'll do is they'll take their clinical systems offline because they have a dual purpose. Not only do they need to care for their population’s health, but they also need to protect the patient record from unauthorized access. Now, CMS has emergency preparedness requirements that health systems must follow so that they can continue to deliver care. So, when ransomware hits, health systems take their EHRs offline. The thought is that that perhaps they can rely on read-only systems, alternate production systems, disaster recovery systems, but they're almost always on the same network as the ransomware. So you end up taking those offline. So in order to meet those CMS emergency preparedness requirements, what they're having to do is do a lot of work on the phone, faxes, to be able to establish a paper record that sometimes can take two or three days to collect, and really only contains about 10% of that patient record. So the thought is, what if you actually had a replica of your production EHR in an air-gapped location such as AWS in a cloud where you can mirror updates in real-time, but it is on a separate plane, air-gapped from where that ransomware hits. So that if ransomware does hit that healthcare system, they sever that connection that was mirroring the updates, and now they point uninfected devices, and this includes not only PCs from a browser, but also iPads and smartphones – in Epic's world, that's Canto and Haiku – to this read-only system. They can deliver that in minutes as opposed to two or three days. So that we call this cloud read-only, Epic cloud read-only on AWS. The idea is that it allows your staff to continue to deliver patient care, informed patient care, because they do have access to the health record, yet it allows the health system to actually create an immutable copy or backup of that health system. They can satisfy the CMS requirements in minutes versus days. This all reduces clinical and business risk, which can have a positive impact on perhaps your cyber insurance premiums. Now, the greatest thing I like about it, is this is something that can be delivered quickly, generally in about 8 to 12 weeks. This isn't a migration of an existing environment. You keep what you're running the same way that you always have. This is a new environment delivering new value. That has the benefit of keeping your clinical piece resilient, but also is the opportunity to get your IT experts trained up on operating your EHR on our cloud.
Brijeet Akula: That's fantastic. So other angles of the resiliency, just talking more from the back end of the infrastructure that we're able to do with these cloud environments. We've talked a lot about elasticity. It also comes with its own baked-in level of resiliency for all of these services. For example, a server that we're able to spin up in the cloud environment, default out of the box, it comes with a 99.99% SLA, four nines, which equates to about a threshold of about 54 minutes of potential downtime in a given year, which is pretty incredible. This is just the default experience without any further customizations. But typically, especially an Epic approved architecture does recommend, you know, further load balancer, basically adding additional levels of redundancies and things like that. Again, all things that are fairly easy to do on cloud platforms like AWS. So we're able to very, very easily go from four nines to even five or six nines. So to keep things to keep in perspective, five nines, for example, that's about 5 minutes of downtime in a given year period, which is pretty incredible, and it takes very little effort to actually set that up.
Dr. Jerome Pagani: Matt, we heard Saran talk a couple of minutes ago about sort of the pay for what you use, but that's not really the whole story, and I think we hear quite often that the cloud journey can be quite expensive. Is that true?
Matt Dinger: So cost is always a concern for healthcare organizations. We hear that more now than ever. You know, as margins are razor thin, as folks are coming out of the pandemic and dealing with all of the aftermath of that. The reality is it shouldn't be. You know, Mike started off talking about this provisioning and these hardware refresh cycles that we see organizations on, where you are way over-provisioned for a couple of years in that refresh cycle. This really allows you to scale that down to be able to use only what you need. Certainly, there's also a proactive piece to this. Our technologies, our services team looks at the instance types and makes improvements that make that more efficient all the time. This is why we see our customers reporting actually significant savings than they would have had on-prem. Really up to 40 and 50% in some of these circumstances where folks have talked to us. The other piece of this is all of the setup that can go into overall cloud strategy, ensuring that you are leveraging everything available to you when you move to the AWS cloud. So these are things like guardrails, resource tagging, active monitoring, alerts, budgets. So you have the proactive approach with individuals and technologies that are there to look ahead of time and say, Hey, you're trending in this direction. You may want to make some adjustments, as well as in the moment setting up alerts that you are getting in case, you know, someone has left something running or, you know, you are headed down a path that is not necessarily somewhere that you want to go. But then in addition, recognizing where organizations are at, and this does represent a significant investment and a significant lift for organizations, is we do have incentives and discount programs that can be taken advantage of both by the health system themselves, and that system integrators like Nordic have access to that can further reduce the cost of implementation and cloud consumption.
Brijeet Akula: That's really great, Matt. So just adding a couple more points to that. One of the important things here is that these cloud providers, AWS, for example, provides an excellent set of Fin Ops capabilities to really help control your overall spend for AWS. As you mentioned, because there is so much flexibility, it is also very easy to end up with some runaway processes or somebody’s doing some R&D work and they end up creating some very expensive resources that are really great for experimentation. But perhaps they forgot to turn it off. So, these Fin Op tools that come directly out of the box as part of the AWS tool kits enables that, provides the governance, and in some cases, also proactively, we'll shut it down if we need to. So you don't end up racking up a huge bill for your consumption rates.
Dr. Jerome Pagani: We talked a little bit in the beginning about the technologies that have come online that are making health systems consider a cloud journey right now. And healthcare, let's be honest, are slow adopters for technology and have been for a long time. Why is that? Why are we seeing this rapid uptake now?
Mike Hegyi: Yeah, you know, this is something that, thank you for asking that question, and I have a big smile on my face, but you can't see that. But I'm smiling because this is actually a question we're hearing more and more now. I actually love that question because a few years ago, it was, ‘Why would anybody want to run the EHR in the cloud?’ So I’ve been around for a few decades with this, and I think we're really on pace with what we've seen in other technology and innovations in healthcare over that time. An example would be, say, virtualization. Virtualization had a great promise, it was high on that hype cycle, and but in healthcare specifically, it had a slow adoption. Initially, it was why the heck would anybody want to have a virtualized Epic? Then two organizations did it, and then four organizations did it, four turned to eight, eight turns to 16. And then it was a similar question Why aren't more organization using virtualization? But then we take a look at, again, on-premises, you're trying to amortize off that hardware, that infrastructure that you have. So the slow pace of it really came around to, hey, when it was time to do a hardware refresh, that's when you went with virtualization. And I think we're seeing a similar type trend with the journey to the cloud. When I first started, there were three organizations, a year later doubled to six, and now we're at 18 organizations that are running Epic on AWS. And so if you would imagine this, this hockey stick of a graph. Initially, it was on the blade portion of it, but now it's going up that steeper angle where I think that as organizations look to do hardware refreshes. And just as Saran had mentioned, take a look first at alternate production and then migrate things into production. I think we're going to see this frequency of use have running HER on the cloud happen more rapidly.
Dr. Jerome Pagani: That's fantastic, and at Nordic, we talk to our clients a lot about optimum performance and that's really a story around being able to do more with less. And what I'm hearing here is, exactly that the migration to the cloud has both immediate and long-term benefits that are going to help health systems achieve that optimal performance over time. So, we've talked about this on a sort of theoretical level, but what are sort of the concrete steps a client can take who's considering a cloud journey?
Brijeet Akula: Yeah, absolutely. So as soon as a client is starting to get curious about this and wants to explore the capabilities that the cloud can provide to them, specifically on their EHR workloads … one of the first steps is reaching out to system integrators like Nordic, where we would start working, collaborating with them to strategize on a plan on how to do this. We did talk a lot about how cloud is a very easy environment, a lot of things are single clicks, everything expands very easily, but also at the same time, having a good foundational structure and strategy on how to deploy this very powerful tool is very important. So, we would certainly help with that journey. Another important component is to get in touch with Epic, start talking to them about your intentions of exploring the cloud as an option. There's a very specific document that Epic generates for all customers once they have been requested. It's a free service. It's called a cloud specification guide and a hardware sizing guide. This is uniquely created for individual customers where they see exactly where your environment is and what you would need exactly on the cloud platform. So, on AWS would be very specifically customized for that environment, talking about which resources you may need and then SIs like Nordic can then help take that and then build out a plan for your environment. Start looking at what your annual run rates for consumption fees are going to look like, and build out financial plans around that, financial models around that, see what kind of incentives we can get added in there.
Saran Sonaisamy: And to add to that, Jerome, what is most important is also having a very active collaboration with AWS, between Nordic and AWS, that ensures we not only look at what solutions are available, and what is coming up when we build a platform, we build it for the next generation, for the long term. So, using what Brijeet was talking about from the input from Epic would work with AWS, and really working towards the migration playbook. It can be a net new installation for new Epic instances. It can be migration from a data center but work along on the migration playbook. That spans from the assessment to migration and transition to services. And you know, Mike can certainly talk about how we have an active cadence between Nordic and NWS.
Matt Dinger: And I'll jump in there, actually. I just wanted to quickly acknowledge that partnership and that cadence and actually thank you all for that. At Amazon, one of our leadership principles is customer obsession, and it is wonderful to work with an organization like yourselves who understands that this truly is all about the health systems and the patients that they serve. So really appreciate that coming through in our partnership.
Dr. Jerome Pagani: I know vendor management is always a concern for our clients, so, you know, it's nice to hear that this is one of those areas where the customer and the solution comes first. Mike, Matt, Saran and Brijeet, thank you so much for joining us today. It's really a pleasure having you on the pod.
Mike Hegyi: Thank you for having us.
Brijeet Akula: Thank you. Thanks for the time.