Thank you, Pat. I'm very excited to be here, and I've already talked to Pat about a few things that I hope Matt will tease out in the questions, and I'm so excited will be the next step for this ABCDF bundle and make it easy for people to adopt. I'm an intensivist at Columbia. I work in the SICU. I actually just came off service yesterday, and the CTICU. One of the reasons I got involved in this is because I work for the EPIC steering board for critical care, which is new. EPIC doesn't pay me to do that, which is something else I'd like to talk about. And what I want to show you here is a new workflow for integrating clinical decision support that we're working on. It's at that stage of being just good enough to criticize. So today, be as harsh as you want in the small audience. One thing that I and EPIC are super stuck on and I want to draw attention to with my talk is how we set goals, especially for titratable parameters like drips. And the last thing I want to walk out here with is identifying who here is really going to be a content expert, because that's what's going to drive change for these big EMR companies. I finished training in 2019. Our CIO asked me to help with our EPIC implementation, this big billion-dollar project. We did our implementation January 31st of 2020. And three weeks later, had our first COVID patient in New York. We then proceeded to do the exact same thing four other times in the last three years, where we would do these EPIC implementations followed by these huge surges. Our IT team was phenomenal through this. When we set up our vaccination site, our IT team built it and ran it, and we had our analysts there helping with vaccines and showing patients around. But that meant we didn't get to build any of the fancy things that Pat has and Matt have. So now in 2023, I look at where our consortium is, I feel pretty foolish because I'm sitting on the steering board pretending I'm very fancy, but I haven't built out the, like, really incredible tools that other places have. But those tools take five, 10, 15 years to build. So I'm jealous. And what I'm looking for in setting up the steering board is a way to catch up really fast. So I was able to get some really incredible people to serve on the steering board. Adam Saperstein from Hopkins, Bill Hanson from Penn, Sarah Nizula from Finland, Adele Beasley-Marcus from across Central Park at Mount Sinai, Arun from Colorado, and it's been incredible to bring everybody together and to start thinking about how I can catch up. And here's the workflow that I think is going to help me catch up. It's really simple. Essentially, I like to walk into the unit and think about the map of the unit. I'm sure many of you who work in the ICU really have this very tangible, physical sense of your unit and, like, where beds are, how many steps away it is from the nursing station. And just coming off service, I can think of exactly who is in every bed in my unit. And from this, I'd like to know what happened the last 24 hours. I'd like these event notifications to just pop up and show me. Like a RAS mismatch, an agitated patient, somebody going into AKI. Right from that notification, I want to go into some incredible review tools. I was able to build this review tool out with Spectrum Health. It really shows some key shock indicators, some key respiratory distress indicators, a brief narrative summary, the key problems, that scoring system, and to build these out for each organ. Another set of review tools that I think are appealing to anybody running a unit are quality checklists, whether that's SCCM's Choosing Wisely, your own hospital's quality initiatives, ABCDF Bundle, exactly the one Pat showed. But to really start building these out in a way people like to use them, a way where doing them brings you to completion, and where doing those checklists actually drives the orders that gets things done. For instance, not just checking off, we can remove this patient's Foley, but actually putting in the orders to execute that action. And then the very satisfying place where you've completed everything you need to do. Orders are really the key for driving actions in Epic, and I think Cerner as well. Think of it as a transactional database, and orders are really the currency here. One thing we should all care a lot about is orders don't work very well for these titratable medications. I think intuitively all of us recognize this as how we run our units. We set a very clear goal for MAP or RAS, and we are delegating to nurses and respiratory therapists how to titrate to that goal. We should really push our EMR companies to build this workflow so it's fluid and reliable. The key thing that's missing is actually setting those goals. An example just from this week is very simply changing a MAP goal for a patient. Right now if I just say during rounds, let's make this MAP 65, and I just looked around and said, you know, where does that get written down in the EMR? How is that executed? Is it in a cardiology consult note? Is it in a nephrology consult note? My intern wrote it down in a little scribble. It's in a handoff. It's in my note. It is in the notification orders for providers, which is some sort of order. It's text in the drip order, and it's in the literal physiological monitor alarm itself, which is fine. I love that people are writing it down. They're listening to me. They're paying attention. They're going to do it. The problem is if I change my mind, that then has to be updated in all of these places, which is just costly. It causes confusion. It's rife with errors. Once we have those goals, we can drive just really rich contextual decision support. I'm going to share an example. Many of you may recognize BPAs, best practice advisories, decision support for things like this that are often pretty ugly. Just use a reference point of convenience. The last seven days of creatinine in this case, they're interruptive. They occur when you place orders for other things. People just blast through them because they're interruptive. I'd actually like to be very thoughtful and explicit about how we set baselines or goals. When you admit a patient to the ICU or the ER, thinking about what a patient's baseline creatinine is is really important and helps drive decisions downstream. But this is why we need doctors still, despite everything with CHAT GP3. I think you really need a doctor to not just look at all these numbers, but to think about what makes sense. Like, what is the baseline creatinine for this patient? And integrate a lot of complex data to do that. But then explicitly decide. The computer doesn't know all the things you do from talking to a patient. And the key innovation here is to create a framework for making those explicit decisions. To say, Matt Aldrich, he's deciding this is the baseline creatinine for this patient. Once we do that, the next innovation is creating visualizations based off of that. Seeing these are from just Kedago's criteria, the increases in creatinine relative to that baseline. Once you've done that, not only can you visualize it, but you can also drive just very rich contextual decision support. That notification when I walk into my unit and I'm looking at my unit map can tell me with words, can tell me with this visual, this increased two times my baseline. I want to make the cycle even tighter. This is why I'm continuing to work with the steering board, and this is how I want things to work. Again, going back to this, setting an explicit map goal. Once you set that goal, I want to make it easy to visualize. I also want to juxtapose it, and this is specific to critical care. I want to juxtapose it with the amount of life support. I was telling my interns this week, look, when you look around the unit, everybody's map is 65, but some of them are on this much pressure, and some of them are only on this much. If we took that all away, what would their blood pressures be? I think very specific to us is to always be showing these goals and these parameters in the context of the support with them, whether that's a ventilator, whether that's mechanical circulatory support, or whether that's these titratable drips. Once we've made these definitions, we can trigger notifications, both from the parameter itself or from the escalating support. I'll bet we've all had a situation where you weren't notified about a rapidly escalating FIO2 or vasopressors, and it slipped through the cracks, and you were suddenly surprised to learn somebody's norepinephrine had tripled overnight. I'd like to put this together with really embedding orders. I mentioned this a little bit in the quality checklist piece, but when you look at these things together, you're immediately going to take an action after that, whether that's adding another pressor, giving a fluid bolus, just changing the map goal. But I can prime you to take the right action by putting those right into this card. I also think the writing you do and the thinking you do is really valuable, and putting a space for that narrative right there is really important to me and I think makes it more valuable. And finally, I'd like you to be able to share this chunk of information with your colleagues, especially as we start embedding communication tools within the EMR. I'd like you to be able to send this directly as a notification to your team members who it would be relevant to. These are a lot of cool ideas. I can design stuff all day, but when I'm sitting there with Epic developers, the constraint is actually that they need content expertise. Epic has, what, 6,000 employees, but it only has 10 doctors. There's no critical care doctor in the company. And so when I heard about what Pat was doing with the ABCDF bundle, I just immediately recognized that that's the connection that needs to happen between an SCCM group that's willing to stand behind content and say, this is what's important for evaluating something like hypotension or quality metrics for a unit, and stand behind those decisions. Because once you've made those content decisions, it's much easier for Epic or Cerner to actually build. This is just an incredible example of success. One of the ways we can build on that is by really adopting exactly what Pat and her team have built into as many hospital systems as possible to really show Epic that they will get adoption if they keep collaborating and keep partnering with SCCM. And I think we can keep making this even better. I think we can take exactly what a nurse does and make it really easy for them. When you look at a nurse choosing, did they get sedation last 24 hours, we can make more content decisions about exactly what does sedation mean, exactly what medications that is, and that should come from us. You think about mechanical ventilation, we can decide exactly what mechanical ventilation means, come up with that definition, and then only show nurses this set of decisions when it's relevant, when a patient is ventilated. We don't want our teams to be doing extra work when it's not relevant. So I'm happy to talk today about and get feedback on this workflow, this idea of notifications taking you to a review tool. That's the simple aspect of it. A lot of it's more innovative, but the simplest part to start with is just a system of notifications that take you to really well-developed review tools. I want people to go back to their healthcare systems and really hone in on this very specific problem of how do we set goals and titratable drips, and really look at how your hospital system is doing it. I don't think anybody's figured this out, and I think it's going to take real development to fix. And finally, I want to advocate for building on the success of the ABCDF bundle. That partnership is fantastic, and I think we all have an obligation to adopt that into our own systems so that we can show Cerner and Epic that this pathway of partnering between SCCM and the EMR is productive, and people want it, and it drives change. Thank you.

clinical decision support

workflow integration

electronic medical record

titratable parameters

collaborations