Exciting things have been happening these days at Boston Children’s Hospital, where clinician leaders have been moving forward to implement advanced solutions to support the integration of data from patient monitoring devices into real-time care management in the intensive care unit (ICU) and cardiac intensive care unit (CICU) care areas.
The effort at Boston Children’s was led until recently by Peter C. Laussen, M.D., and Melvin C. Almodovar, M.D. Dr. Laussen, who spent 20 years at Boston Children’s, the last 10 of them as chief of cardiac critical care, in September 2012 moved to Toronto to become chief of the department of critical care medicine at that Canadian city’s Hospital for Sick Children. Dr. Almodovar, medical director of the CICU at Boston Children’s, continues in his leadership role there.
Together, Drs. Laussen and Almodovar brought in programmers and designers from the Burlington, Mass.-based Arcadia Solutions. Through their collaborative partnership, the people at Arcadia Solutions and Boston Children’s have developed a combination of solutions called T3, which stands for “Tracking,” Trajectory, and Trigger,” and helps critical care physicians and nurses to track all information coming out of patients’ bedside monitors; calculates trends in the data and displays a patient’s real-time trajectory; and triggers actions to avert crises and to make the most efficient use of the resources available in the critical care units.
The partnership with Arcadia has also led to a variety of additional capabilities, including a “stability index algorithm,” a tool that generates a numeric index to indicate the extent to which a patient’s physiologic measurements deviate from clinician-chosen parameters; complete data visualization or morbidity and mortality reviews; a historical data repository, to provide researchers with access to all the historical data for large-scale analysis; and integration with additional demographic information, and with intermittent (aperiodic) blood pressure data.
Given that physicians can have more than 10 devices at each patient’s bedside whose data they need to track and analyze, the implications of the development of such solutions are considerable. In order to provide insights into all of this, Drs. Almodovar and Laussen spoke recently with HCI Editor-in-Chief Mark Hagland. Below are excerpts from that interview.
Can you share with our readers a bit of the history of this co-development? You were both involved in the development and implementation of the T3 solution, correct?
Melvin C. Almodovar, M.D.: Yes, that’s correct. T3 was conceived three-and-a-half years ago, and it started through a partnership with Arcadia.
Peter C. Laussen, M.D.: How this all began is that, nearly four years ago now, Mel and I agreed that our biggest problem in the ICU was being able to integrate these massive amounts of physiologic data streams, being able to integrate and interpret the data, as well as analyzing it and being predictive in how we manage patients. So our first step was to try to find a partner externally that would be able to support our vision. Arcadia Solutions had approached us to develop some kind of collaboration; and that approach had come about through circumstance, when Mel met the company’s COO at the time at a function, and that gentleman came in and met with Mel and me. So we shared our vision with them, and Arcadia went away for a little while, and then they came back and said they’d like to explore this further.
We were fortunate at Boston Children’s, in that I had an endowed chair, and was able to use funding from that endowment to hire Arcadia initially to develop a sort of proof of concept. Arcadia thought about the problem and came back to us with possible solutions; and we realized at that point that a solution was achievable, and that we could capture the data and store and analyze it, in a much more meaningful way. So we began working together, while funding the work on a fee-for-service basis for their involvement. This was a new area for them as well, so they hired some very key personnel, and for the next three, four years, we’ve worked extremely closely with a wonderful T3 team of program developers and managers to where we have an end product that’s been in place since June of last year [2012] in the 29-bed cardiac ICU and in the 30-bed medical/surgical ICU at Boston Children’s.
And now you’ve begun using it at the Hospital for Sick Children as well?
Laussen: Yes, we’re now using it in our 42-bed ICU here as well.
Let’s drill down a bit on the capabilities of the solution.
Laussen: First of all, the name T3 was chosen to represent what the tool does: it’s a tracking, trajectory, and trigger tool, a web-based software platform that captures continuous physiologic data feeds from the bedside through an HL7 feed. The data is then captured and stored on a separate server, which then enables us to do various calculations and to perform integration with the data, and then the data is fed back to a platform that allows us to look at the data in a different way. It’s sort of like using Google Maps or iTunes playlist; we can continuously look at different pieces of data, and really understand relationships between critical physiologic data at the bedside; that’s the tracking component of it.
The trajectory element helps us to understand the response to treatment, and whether the patient is following the trajectory we had hoped. That’s where the Stability Index Algorithm comes in: it’s a composite score of key physiologic variables that indicate whether a patient is progressing as well as expected. And then both the tracking and trajectory elements then trigger a response. If the patient is not progressing as you expect, or you are seeing some problems via the data, it will trigger a response, to either change the management, to communicate with somebody, to huddle at the bedside; but it gives everybody the same platform, so we’re all looking at the same data, not trying to interpret isolated pieces of data.
Almodovar: And a key point is that we’re talking about the continuous collecting of physiologic data. There are various electronic platforms; and we certainly have a robust electronic patient record; this is different. The data that’s entered into the EHR is entered intermittently, usually hourly in ICUs, less often than that elsewhere, and usually is manually entered and interpreted by the person entering the data.
This is a more automated process, then?
Yes, fundamentally, the system listens in and captures and routes the stream of data; it’s listening in on the automated data stream, and capturing it. We’re talking about volumes of information. And it’s also the type of information. Our focus has been on intensive care environments, where physiology is what we’re focused on, in terms of understanding the patients and their function. So we’re talking about a high-volume process, with an emphasis on physiology.
Laussen: This amount of data has not been available to us before.
Is this the first time anyone’s tried to do this in this way?
Laussen: Yes. The monitoring systems we have don’t allow us to capture and store continuous data streams; data appears and is then lost. So we’re unable to look back at key data; and all the data streams are vendor-specific and don’t talk to each other and have different time stamps on them. So the ability to be able to continuously capture all these different types of data is enormous for us; it allows us not only to capture the data in real time, but also to analyze it.
Laussen: And the other piece about that is that we’ve always made assumptions as to what we think are the right variables to capture or use at the bedside to tell us how well a patient is progressing; there’s been no randomized clinical trials to tell us which variables are better or more important; and we’ve never been able to use these types of data to make continuous calculations. So we’re actually using the data that’s integrated and interactive, so it’s giving us new information and insights. And we’re talking about potentially between 20 and 30 or more information types that we can capture.
Tell us a bit about the process so far.
Almodovar: As Peter mentioned, we first went live with T3 last June; and there certainly has been experience with the system; we then released an enhanced version this past January. We’ve had growing clinical experience over the past nine months with T3. Meanwhile, part of our focus during its development was on two areas in particular, the user interface and the stability index. As part of that effort, we collected data on a population of newborns with one form of congenital heart disease, a variant of hypoplastic left heart syndrome. That represents one of the more complex forms of congenital heart disease that someone can be born with; it requires surgical treatment during the newborn period. And Dr. Laussen and I have treated many babies with that condition. And our focus in developing T3 and the stability index was to collect data specifically on that patient population.
That data collection started in October 2010, and with a sample of 38 patients, we focused on 29 patients, and were able to learn a bit about the tendencies of that population, and by applying analytics with respect to the stability index, have been able to look at the predictability of the stability index, with respect to one outcome in particular, and that’s length of stay. In short, we’ve been able to show value with respect to predicting one outcome in particular, and that is length of stay among the population of newborns with a variant form of hypoplastic left heart syndrome.
Laussen: What Mel’s touched on is very important here. T3 can help us to reduce length of stay among children with congenital heart disease; that’s something we’ve been doing in Boston, and will be doing in Toronto. But we’re also using T3 as a way to capture information continuously as part of a clinical practice guideline or algorithm. The plan is that we’ll treat a certain subset of patients a certain way, capture that information as we follow that pathway, trying to be consistent in the ways in which we utilize resources and trying to limit variability in that practice; and the T3 will help us in analyzing a continuous variable, rather than assigning patients to a particular algorithm and managing them in a certain way, and learning later what we’re doing with them. So T3 helps us to do dynamic learning, in real time. That will then allow us to be more strategic in our care management, more predictive, to be more proactive, and to reduce variability in our care management.
In addition, this solution allows us to actually share information between institutions, so that we’ll be developing a data warehouse platform that will allow hospitals in Boston, Toronto, and other cities, to share and pool information, so that we can learn from each other and begin to reduce practice variability between and among institutions, as well as internally. That’s one of our major goals.
In other words, this initiative represents technology-facilitated progress towards evidence-based care?
Laussen: Absolutely. One great advantage was that, because they had come from other industries, the team from Arcadia brought rich experience from other areas, and weren’t bound up by a narrow view of healthcare. So we just ran with it. Mel and I would meet with them almost on a weekly basis; we would do rapid cycles of testing and decision-making. And within 18 months of starting the project, we were collecting all of the data from the critical care spaces in Boston onto a separate server.
What would you say the biggest lessons have been learned, to date, from this co-development experience?
Laussen: I would say there are three. First, taking a vision that you have to improve the care you provide at the bedside, requires external partnership; it can’t be done internally. Second, you need to look at external viewpoints, so you won’t be constrained by your own experience—and then you need to be willing to listen to alternative perspectives, including around how to display the data. The third is that this is expensive. We were fortunate in that we could initially fund the development of T3 through my endowment at Boston Children’s; and then the chiefs of cardiology and anesthesiology continued to support us to fund the development of this over four years.
Almodovar: One lesson has to do with the development and implementation of a system like this in our institution. We certainly were able to leverage some of the resources of our hospital’s IT department; but at the same time, it was important for us to work somewhat independently of those resources, because of the complexity involved in a development process like this one. We have information presented to us in various formats; the question is, what are we going to do with it? And how will we use it to improve medicine? Because of my experience with T3, I get frequent inquiries from my colleagues about this. So the possibilities are there for us in healthcare to learn new approaches.