Late last month, Yale New Haven Health System (YNHHS) hosted its first annual National Symposium on Value Innovation at Yale and health system leaders shared their lessons learned as part of ongoing effort toward improving quality and lowering cost across the system.
The event was a collaboration with YNHHS, the Yale School of Management, the Center for Outcomes Research & Evaluation (CORE), the Yale Center for Biomedical and Interventional Technology (CBIT) and health IT vendors Strata Decision Technology and PeraHealth.
During that event, Lisa Stump, interim CIO at YNHHS, specifically addressed how the health system is using data and health IT to drive value to patients.
“We want to deliver the right data to the right people at the right time, in the right format, to drive outcomes,” she said. “We are thoughtful about the way we deploy technology and avoid implementing new technology and dashboards simply for tech itself.”
The health system has deployed a number of health IT initiatives such as complementing its electronic health record (EHR) system with advanced decision support tools, the development of a tele-ICU program at Yale New Haven Hospital, which has led to improved clinical outcomes, and using technology tools to increase patient awareness of research trials, which has increased enrollment in clinical trials, Stump said.
YNHHS is a 2,130-bed health system that operates three hospitals in Connecticut, including Yale New Haven Hospital, a 1,541-bed tertiary medical center. In early 2012, YNHHS implemented a number of key multiyear strategic initiatives aimed at improving overall value.
Stump leads the health system’s technology strategy as interim CIO, and previously associate CIO for two years. She joined YNHHS in 2008, first serving as administrative director of clinical informatics and then leading the health system’s multi-year implementation of Epic’s clinical and revenue cycle application across YNHHS hospitals and ambulatory practices, North East Medical Group physicians’ offices and Yale School of Medicine faculty practices. Stump’s educational background is in pharmacy administration and she led pharmacy services at Yale New Haven Hospital before moving over to health IT at YNHHS.
Stump spoke with Healthcare Informatics Assistant Editor Heather Landi and drilled down into the health system’s health IT strategy as it continues its journey to value-based care.
How is Yale New Haven Health System using data and technology to drive value?
We have focused a lot of our strategy on building a core platform, so a single electronic medical record (EMR), which is the main system of record for the clinical data, and then we are supporting that with an enterprise clinical data warehouse, and for that we are using Epic’s Cogito platform. And then we’ve complimented that with a Hadoop environment [data processing platform for big data analytics] to allow us to manage the big data and prepare ourselves for the genomics wave. We have a lot of data and now it’s about turning data into meaningful information for people and delivering it in a way that’s meaningful, and that’s probably one of our biggest challenges. We’re working with a variety of platforms to create the dashboards and analytics tools and we’re really looking to be predictive in our analytics. We want to start to use the data to understand the trends that are there and predict future events before they happen, so identify patients who are at risk for declining health in the next 30 days, or to identify patients who are more at risk for orthopedic conditions, so that we can be advocating health interventions and certainly preparing for what might be a more invasive intervention for those patients.
Interoperability also is a big focus of that. And I often say, we’re not trying to compete on the data, we are willing to share the data, and we also encourage others to share data with us. We working with our vendor partners, whether it’s EMR or medical device vendors, to ensure that we are getting the right data exchange and trying to turn that into meaningful information.
When did this effort begin?
The journey began in about 2010-2011, when we recognized as a health system that we needed to be on a single EMR solution. So, all of our network hospitals, our physician groups and the Yale medical group faculty physicians, all agreed that we needed to be on a common platform that would allow us to start to accumulate the same data in the same way in a consistent manner. And about the same time, we started developing the clinical data warehouse. I will tell you, the first few years of that journey we were really focused on getting the EMR platform in. Then it took us a little bit of time to understand all the data that we had. And, so it’s probably been about 18 months that we’ve really centralized on what we call our joint data and analytics team.
And one of the remarkable things is that we had hundreds of people across this health system—in all kinds of departments, clinical quality departments, nursing services, physician’s services—all culling data from the medical record, doing something with it, sending it to external registries; and we said, “We need to do this in a better way.” So we centralized that team, put it under the direction of our CMIO, and really stared to work on standardized tools, data definitions, simple things such as “How will you define a missed appointment?” Believe it or not, people define something that simple differently, so we needed to come up with those standards and from there we could create dashboards that allow us to understand ourselves over time and allow us to compare clinic A with practice B and identify our areas needing improvement and our best performers so we can learn from each other.
As you said, turning data into meaningful information is often the biggest challenge. Are you making headway in that area?
Yes, I think we really are. We’ve had some recent successes again creating those predictive tools. We have the ability to now predict readmission for patients based on characteristics that we gather naturally during their hospital stay. So we understand a patient’s home living arrangements, we know their physical vital signs and other metrics, and we’re able to create algorithms, based on that data, that the patient has a 30 percent chance of readmission and we can appropriately put interventions in place to try to avoid that unnecessary re-hospitalization.
We’ve been able to predict patients who have a terminal illness and are likely to expire in the next 30, 60 or 90 days. Again, so we can be proactive in offering palliative care services to avoid an expensive, uncomfortable admission for a patient. It’s an interesting time. We’re starting to see that having all of that data, it raises the sense of responsibility about it. With a population of patients or an individual patient, if there is a 90 percent chance of readmission, I intervene. But what if that patient has an 88 percent chance, do I not intervene? How do you start to draw those lines? So, we’re starting to hear clinicians feel that sense of, now that I have the data, there’s this added sense of responsibility for me to act on the data.
You mentioned that part of this health IT initiative is preparing for genomics. How are you moving forward on that work?
We are capturing the data, so where a patient’s genome has been sequenced, we have the ability to pull in that data, and we are excited to start to look at the opportunities. I’m a pharmacist by training and some of what is available to us in your human genome is the way that you metabolize medications, for example. Right now, I can’t take your pulse and figure out if you are going to quickly metabolize this drug or not. Narcotics are a good example: if you are a rapid metabolizer of narcotics, you need higher dose. By giving you the average dose, we are delaying the time that you will feel pain relief. On the flip side, if you are a low metabolizer, even a normal dose could be a toxicity for you and cause an overdose.
So the ability to use that human genome data when physicians are prescribing that drug, to say that this is not a good medication choice because we understand the phenotype and the genotype of this patient, that brings incredible opportunities to avoid adverse events. And, ideally, we would be pushing the information as an alert. So much like when you order a medication today and you receive an alert that the patient is allergic, now, we would get notification that the patient is a rapid metabolizer and a recommendation for another drug. Some organizations are ahead of us in that regard and are starting to do that, and we are looking forward to being able to build on that too.
What are the other health IT tools that you are leveraging?
We are using tools that help us, for example, pull social media data to bring in information about our patients satisfaction scores. I think there’s an unlimited potential to start correlating data from our employee engagement surveys and our physician engagement surveys to help us understand patient satisfaction. We can think intuitively that unhappy physicians and unhappy employees probably make for unhappy patients, but showing that, I think, allows us to drill down and intervene. We’re using technology for remote monitoring and remote surveillance and taking advantage of the fact that most of the world’s population is carrying a smartphone that’s tracking steps. If you’re using the health application on your phone, then we can start to cull all that data and correlate it, and there’s a lot of analytic strategy there.
What have been some of the lessons learned throughout this initiative?
I do think that it really is about engaging with the clinical and business stakeholders to understand what are the questions that you’re trying to answer. We often get into the trap with people coming to us and saying, “I need a report that shows me x, y and z. I want these columns and these rows.” And, then when we produce that report, invariably they say, “Well, that’s what I asked for but that is not what I need.” So I think investing the time up front to really understand what’s the question, what’s the problem, allows us to be much more thoughtful and produce more quality results in a more timely fashion. I think once you have all the data, it’s easy to get into that trap of creating a lot of stuff and that is what overwhelms people. So making sure it’s targeted at the real need, I think, is key.
What are the biggest challenge facing healthcare IT leaders right now?
I think it’s the pace. There is a tendency to look at technology as the magic bullet that’s going to solve all of our problems. We have this complicated problem or complicated patient population, so let’s buy an app. And you constantly have to say, no, let’s look at the problem that we’re trying to solve. An app is not going to solve the problem unless you address the fact that person A never talks to person B. As an example, a colleague once brought me a list and said, “We’re trying to solve this problem with telemetry and I need you to build me an order set.” Or, “We’re trying to solve this problem, I need you to build me a dashboard.” I said, “Wait a minute, don’t bring me a solution. Tell me what you’re problem is and let’s talk about what you need.” And when you get under the hood on that, an order set was not going to solve the problem and we needed to look at other things.
It’s difficult as everyone, with good intent, my colleagues across the health system and I, are out at conferences just like this talking to vendors, thinking that they’ve got a solution. But the more we cloud that mix, with all of those interoperability challenges, the more it can result in overload for our clinicians. If I can create a solution in one app, isn’t that better than having to click between five different apps to navigate what I need? So I do think it’s the pace and we’re all trying to keep up with that.