S4E6: Bridging Contextual Data & Operations (ft. Paola Ballester, Johns Hopkins All Children's Hospital)

Jordan Cooper   1:14
OK. Appreciate that. So jumping right in here, I understand you've been working on clinical documentation and building your physician advisor program to support documentation optimization review. You're shifting away from your reliance on external vendors and leveraging internal expertise to build education feedback loops with which.

Paola Ballester   1:21
Yeah. You got it.

Jordan Cooper   1:35
External vendors can't match. Can you tell us more about what you're doing with that project to move towards a more proactive utilization of data approach?

Paola Ballester   1:45
Yeah. So for us, when we looked at opportunities, right, there are cost opportunities. And when you look at the expansion of our physician advisor program, you know, even just on the surface, we were able to leverage internal resources rather than using external vendors for a cost saving right off the bat. But for us, we really looked at it as an opportunity to build our internal program because the benefits of having an internal program really supersede just the financial margins. So we have shifted to now having the opportunity to improve education, improve feedback. Improve communication and training within our CDI teams, our coding teams, leveraging these two experts that we have that we're already working for our institution, Dr. Jesse Huang, Dr. Mayuri Jindal. And so we're super excited because we're seeing the return on that investment very.

Jordan Cooper   2:40
Mhm.

Paola Ballester   2:44
Quickly.

Jordan Cooper   2:45
Right. So I mean from the perspective of moving from reactive to proactive data management and using the use case of this documentation review, can you describe how you've been moving from, how you've been moving towards proactively using this documentation review and what the impact has?

Paola Ballester   2:52
You got it.

Jordan Cooper   3:04
And what sort of data sources are you looking at and how are you aggregating them and normalizing them? And what's the operational impact of taking this more proactive approach?

Paola Ballester   3:15
So there's a there are a lot of different use cases for how we're doing it. I think one that really tells the story is how we're using it in our length of stay. So length of stay is something that many health systems, many healthcare executives are really focused on, right. It seems to be a benchmark not only for kind of quality of care.

Jordan Cooper   3:24
Mhm.

Paola Ballester   3:35
But also you know your margins and telling the story of how you are able to maximize your revenue and and and keep your doors open and your lights on. But when you really peel back the layers of understanding length of stay management, a number in isolation of just what is your average length of stay does not actually tell the story.
And so first of all, you have to understand your data to understand which metrics really peel back and really get to telling the story. Because if you just look at an unadjusted length of stay, you're really looking at your patient mix. You're not really looking at are they staying as long as they should be staying. So first of all, you have to understand. And what data is available. So for us looking at our observed to expected length of stay helps us get a better understanding of not only that risk adjustment to to account for the patient complexity, but are we performing where we should be performing and looking at that data further for opportunities.

Jordan Cooper   4:17
OK.

Paola Ballester   4:34
So when you look at the healthcare system and how it operates as a whole, all of your quality metrics, all of your data at the end of the day really in healthcare come from your documentation when it comes to those quality metrics like length of stay, etcetera. You know what's your case mix index, what's your severity of illness, what's the risk? Of mortality, et cetera. So if you don't invest in strengthening that clinical documentation on the front line, you are missing the opportunity to accurately tell your story. So by investing in having internal experts that can now not only give us feedback in real time to say, hey. We as clinicians, as physicians see opportunities in the documentation where we can leverage compliant methods like queries to help us clarify the documentation in the medical record because the the goal at the end of the day is really it's not up coding, it's not down coding, that's absolutely.

Jordan Cooper   5:28
Hmm.

Paola Ballester   5:34
Not. The goal, the goal is to capture an accurate and complete medical record. So when we do that, we're able to see that our expected length of stay is typically more optimized because we're telling the story of the complexity and why that patient may be needed to stay for four days rather than three days or two days and so. Because we've strengthen strengthen that internal program, we and we are really investing in improving our feedback loops to our frontline clinical teams to give them concrete examples from the data. What are your most queried topics? What are the topics that are generating the most impact on severity of?

Jordan Cooper   6:10
Mhm.

Paola Ballester   6:14
Illness, risk of mortality, when those clinical indicators are present, we've had some phenomenal conversations with our frontline clinical teams to say, hey, we're seeing these topics come up. Let's talk about the clinical indicators so that we're all on the same page and we can also provide that feedback to our CDE specialists and to our coders. So there's really a lot of impact throughout the system and data that really stem from that quality documentation and investing and increasing the accuracy and the thorough and completeness of that documentation.

Jordan Cooper   6:49
I love how you're getting very specific about the differences between observed and expected length of stay and the many different implications of understanding the severity, the risk, mortality, the case mix. You mentioned something interesting. You said that internal experts through I think documentation optimization review are able to give real time.

Paola Ballester   6:56
Correct. Mhm.

Jordan Cooper   7:09
Feedback. Obviously there has, I'm assuming that that that involves some sort of automation, right? Because I'm interested, can you speak about how some how providers are receiving and providing feedback in real or near real time?

Paola Ballester   7:10
Mhm. Right. And so the this process specifically is, you know, heavily regulated and there are a lot of compliance elements to providing feedback. So for example, our physician advisors can't just reach out to a doc and say, hey, you, you know you if you say this.

Jordan Cooper   7:33
Mhm.

Paola Ballester   7:44
You know then you can increase that's absolutely non-compliant and we would never want to risk that or do that. But we can leverage compliant queries to say the physicians have reviewed this chart, they see opportunities. Hey, could you clarify you know if these if these additional diagnoses. Were present or not, and obviously there are a lot of very structured and templates that we use to ensure that the queries are compliant. But that's one way that we can help clarify the medical record while maintaining A compliant process and ensuring that we are optimizing that medical.

Jordan Cooper   8:20
Mm-hmm.

Paola Ballester   8:23
Doctor. Another example is we leverage on the utilization management side and the case management side. We leverage data from our concurrent coding that our CDI specialists and our coders do. So the patient's actively in the hospital and they are already working towards code assignment. So we leverage that data to say, hey, what is the expected length of stay for this patient concurrently while they're still in the hospital. So that the utilization management and the case management teams, we huddle then every week we review potential and active outliers based on that concurrent data so that we can.

Jordan Cooper   9:03
Mhm.

Paola Ballester   9:04
Actively mitigate and say, hey, what are the barriers to safe discharge? What are the barriers to progression of care? What can we do while the patient is still here to help them safely move through the system towards discharge or the next level of care, rather than only getting that feedback after they've already been discharged? And we can't do.

Jordan Cooper   9:19
The.

Paola Ballester   9:24
Do anything to intervene. So as we're reviewing case, reviewing our long length of stay outliers with our case managers and our utilization review nurses, we can also take that step back to say, hey, this patient is super sick, right? They're they're intubated, they're ventilated, they're still on drips, they're, you know, they're nowhere near. Or medical stability, let alone discharge. So then we can give feedback to the coders and the CDI specialists to say, hey, let's take a second look at this chart because somehow they're falling to our length of say outlier report when they're actually really sick and we think that the maybe the coding is has not been optimized yet. Can we take another look? And see if maybe they have not captured those additional diagnoses that are telling the story of why it's completely appropriate for that patient to still be in the hospital. So there's a lot of bidirectionality.

Jordan Cooper   10:12
So. I think a lot of our listeners are hearing this story and are wondering how you measure success of being more proactive in your data approach, specifically with regards to length of stay. And I know that you're working on, as I mentioned at the beginning of this episode, patient flow and developing processes that do safe discharges before.

Paola Ballester   10:23
Yeah. Yeah.

Jordan Cooper   10:34
Or noon alleviating your bottlenecks and getting more people from the Uh and or ICU's transferred into bed. So can you talk about what your metrics are for success? We've moved from reactive to being more proactive within the length of stay use case. What are you doing about it and how are you measuring that?

Paola Ballester   10:35
Yes. Yeah. So we could go in a lot of different directions there, but I think one of the key outputs that we have been tracking, not only looking at our observed to expected length of stay and seeing if we are moving towards maintaining an optimal ratio, right with something like length of stay, if you drive it down too far, you're risk. Increasing your readmissions, right? The goal is a safe discharge when the patient is ready and they have had an optimal length of stay. Too short can have just as many drawbacks as too long, right? So we're looking at that. We're looking at there are other tools that we leverage using Fizz data. From CHA and that helps us with benchmarking against like peers, other academic freestanding children's hospitals. So that work that data story is apples to apples and not apples to oranges. Because even within Pediatrics, which is inherently different to adult medicine, we want to make sure that that data that we're using is comparing to our. Our peers. So we use that. We look at their also metrics like excess days that Fizz reports out and allows us to look at our opportunities. How many days above and beyond that expected length of stay are our patients staying? What pockets, what diagnosis groups are they falling in so that we can really drill down further?

Jordan Cooper   11:54
Mhm.

Paola Ballester   12:13
To see where our opportunities kind of process wise and operationally lie. But I think you know when it comes to kind of that discharge in that patient flow, we're using a different lens and we kind of took a step back when in in the height of COVID when a lot of the community pediatrics.

Jordan Cooper   12:18
Mhm.

Paola Ballester   12:33
Units shut down around us, we had a sudden kind of pressure to our system where we actually were starting to have to decline patients because we didn't either have the bed or the nursing or the capability to take care of them and that was something new for our system. So we took a step back and we looked at our patient flow and how do we, you know, optimize because the answer is not always right, more beds. So how do we optimize the beds that we have and the flow within our system so that we alleviate those bottlenecks so that when we have post-ops coming out of the ORS and we have patients that need a lower level of.Care transferring out of our IC US or patients coming up from our EC that we have beds available for them. So we use data to take a step back and look at. We mapped out our average admission time so that we could see were there any trends, any peaks, valleys throughout the day and we did the same with our discharge times. And what we saw was a very, very persistent pattern that there was a huge peak of discharges. And I think this is probably common to many healthcare systems very late in the day. So the end of the day shift like between 4:00 to 6:00 PM, we would see a huge spike in discharges when the admissions had been relatively stable. Throughout the day. And so as we peeled back the the layers of understanding why was that peak happening, what can we do to safely redistribute, flatten the curve a little bit and and safely redistribute those discharges throughout the day to more evenly match the admissions. That's when we started to use data. Really. Understand where the process was not optimized, right? Rather than asking physicians and nurses work harder, do more. It's how can we rethink the system and how it's set up so that we can optimize for earlier safe discharge planning and the patients that are ready in the morning can go in the morning rather than waiting for the. Late afternoon. So that's where we really started to tell that story and understand our opportunities for improvement.

Jordan Cooper   14:36
Have you been able to measure the impact on patient satisfaction with HEDIS measures regarding allowing patients to be discharged when they're ready? And also what kind of how have you been evaluating provider adoption versus pushback to changing their workflows?

Paola Ballester   14:51
Right. Yeah. Change management is also, it's such a huge part of anything that we do, right. We have not seen any negative impact in patient satisfaction. First of all, I think that's important. We have found that telling the story and engaging not only clinicians and our physicians. And our resident physicians who are a big percentage of our workforce and in our particular healthcare system. But it was critical to engage not only nursing counterparts and our colleagues and nursing leadership for us to understand how these late discharges impact them.

Jordan Cooper   15:24
3.

Paola Ballester   15:30
All the way to speaking with our dietitians, our clinical pharmacists, our environmental services staff, we realize by this process and working through the data and really understanding how everybody's individual workflows all impact and what the downstream effects are. For example, I practice as a pediatric. Hospitalist. So I I have myself gone to the bedside to discharge a patient and EVS is in there cleaning the room. So I'll tell them, hey, we're actually, I'm about to discharge this patient so that you could do one terminal clean rather than doing a daily clean and then coming back two hours later to do a terminal clean. That helps their efficiency.

Jordan Cooper   15:58
Mhm.

Paola Ballester   16:10
Tremendously. So if they're turning the rooms over more efficiently, that's another critical element in alleviating that patient flow bottleneck. So we engage them in the process so that they could understand and better plan for what are our anticipated discharges in the morning versus the afternoon or the evening, so they can also map out. About their own workflows for nursing, it was critical because they also because of the work that we've done with really trying to prioritize safe discharges by noon in clinically ready patients, they've been able to adjust their staffing because now it's become ingrained in part of our culture to discuss.

Jordan Cooper   16:41
Mm-hmm.

Paola Ballester   16:48
When these potential discharged by noon who they are the day prior so that when the and we also communicate overnight with any updates on those anticipated early discharges so that the nursing leadership team can also adjust assignments so that one nurse isn't getting you know a four patient assignment and they're all going to be discharged. In the morning because we know that will be a system barrier to safely effectuate all four of those discharges. So it's really been powerful for our team to learn the impact of what I used to think was my individual decision as a clinician, but how that impacts all the other teams so that we really got that buy-in from everyone.

Jordan Cooper   17:24
1.

Paola Ballester   17:28
This is important information, not just for our patients, but for other members of our team so that we can all work together more efficiently to provide that really high quality care.

Jordan Cooper   17:39
I appreciate you going into depth about the contextual implications of making these data-driven decisions. There are real human implications on every decision you make. As we approach the end of this podcast episode, I'd like to ask a final question.

Paola Ballester   17:49
Absolutely.

Jordan Cooper   17:54
You know you're well into this journey of moving from reactive to proactive data-driven decision making. What advice would you give to yourself a few years ago, earlier in this process?

Paola Ballester   18:09
I think that what we often see and what I often see as it relates to data and operations is there is a tendency to want to take a data element and then go solve a problem with that data. But just like as a clinician when I am meeting a patient they. Presenting to me with a problem and a concern. I don't order a lab panel, a standard lab panel for every patient, and then based on their lab results, go find the diagnosis, right? You have to use context. You have to understand the patient's history. What are their presenting symptoms? What is their physical exam? And based on that information, find the right data to help you make that diagnosis. And I think the same applies with data and operations. You can't just find, OK, here's length of stay. Our average length of stay is 7 days and therefore we need to reduce it. That's not how it works. You've got to. To understand a is it the right data and what do you do with that data so that you can make impact? Because there's a tendency to over rely on data in isolation. When you really need clinical context, you need to understand what is the right data, who am I benchmarking against?

Jordan Cooper   19:24
3.

Paola Ballester   19:28
So much context to data and then you find the right data to help you tell the story of your problem and then you go fix it, right? So I think that's the key for me is is the finding the right data in the right context to tell your story and help you solve a problem rather than just blindly.

Jordan Cooper   19:38
OK.

Paola Ballester   19:48
grabbing a piece of data and then trying to solve a problem with it.

Jordan Cooper   19:51
I I love, Paula, your emphasis on the data narrative and on the context. I think hearing our conversation today, I'm hearing the emphasis on data context for data-driven operational interventions. You can't just take things in isolation. You can't take teams in isolation. You can't remain.

Paola Ballester   20:05
You got it.

Jordan Cooper   20:11
Remain siloed within the healthcare delivery system, but in order to understand you need a comprehensive, holistically approach any organization or any system before recommending an intervention and take various data sources collectively to inform your decisions. Does that sound about right?

Paola Ballester   20:28
Absolutely couldn't agree more.

Jordan Cooper stopped transcription