Good afternoon. Thanks for being here. My name is Emily Fink, and I am an intern physician at Kaiser Permanente. And I'm here today with Dr. Fazila Asim. Together, we are presenting work that I helped lead with Dr. Asim at UNC School of Medicine, which is my alma mater. And today, we'll be talking about how an interdisciplinary team in the neurocritical care ICU at UNC School of Medicine implemented a seizure risk scoring tool to detect seizures in the ICU and reduce costs and improve value care. Neither of us have financial disclosures. So our objectives today are to start with a literature review and introduce continuous EEG monitoring. We'll discuss the current system and operational practices We'll look at clinical practice guidelines and introduce the two-helps-to-be score. Finally, we'll move into the project specifics that we did at UNC, and we'll conclude with a discussion of our results and next steps. So I'll turn it over to Dr. Asim, who will take us through the literature review. Good afternoon, everyone. I'm Fazila. We'll go through our literature review here. So as we all know, neuromonitoring is essential. It's vital in the ICU. And continuous EEG monitoring is important to, it's very sensitive to detect changes in brain structure and function in general. And it's important to realize that critically ill patients are at increased risk of seizures, and that's seizures and convulsive and non-convulsive seizures. And a lot of times, they're also at risk of prolonged post-ictal effects, sedative effects, and like I said, non-convulsive seizures, which both of them, whether it's convulsive or non-convulsive, have increased mortality and can affect our patient outcomes. So hence, EEG monitoring is important and a lot of times can help us with making meaningful changes in our treatment plans. So these are some of the common conditions, systemic and neurological conditions that are at high risk of seizures. And the incidence and prevalence of seizures in general is wide among these conditions, whether it's ischemic stroke, hemorrhagic stroke, TBIs, CNS infections, brain tumors, and even sepsis-associated encephalopathy in our MICU patients. This study is an interesting study, and it's basically what they were trying to, this was an early study, and they were trying to diagnose or the rate of misdiagnosis of seizures. And so as you can see up here, let me see my pointer. So up here, these were some of the positive seizure semiologies, like motor semiology, and then some of the negative seizure semiologies were up here, for instance, patients with altered mental status or aphasia. And as you could see, the rate of misdiagnosis of seizure was higher among patients with negative seizure semiology. And the difference was statistically significant. What that means, basically what this slide means is for our critically ill patients, we're not very good about clinically telling if they're having seizures, and so monitoring them on the EEG is reasonable. And here's, in terms of time to record the first seizure, comparing comatose to non-comatose patients, a lot of times, and this was an early study as well, but what it highlights is a lot of times our comatose patients require longer duration of monitoring for detecting non-convulsive seizures and seizures in the ICUs. And then we have many societies, international and here with us in the States, American Clinical Neurophysiology Society that have guidelines for EEG monitoring in the ICUs. Many indications, and it's indicated for diagnosing electrographic seizures, for status epilepticus, for spell capturing spells, and then also like patterns of ischemia, monitoring effects of medication, anti-seizure medications, and then we have guidelines for prognosis of seizure and prognosis of after-cardiac arrest as well. So these are, again, some of the same indications that we have consensus for in guidelines, and the duration of monitoring in general is about 24 hours of monitoring, but most of our providers in the guidelines suggest 24 to 48 hours of monitoring in critically ill patients, but there's no optimal duration of monitoring itself. And then that's why a lot of times we have these routine 24 to 48 hours of monitoring in the ICU, and this has increased the number of EEGs in the ICU now, and so it's very difficult. It has increased our demand overall when we don't have the resources in all institutions to monitor patients. And resource limitations, it's not just across small hospitals. There's bigger institutions, tertiary care centers as well. EEG, continuous EEG monitoring is very resource intensive. You have to have the text, the machine, and an epileptologist who is available 24 hours to read the EEG. So it's costly and labor-intensive in general, and but there are studies that shows that actually your diminishing returns with continued EEG duration, so your cost effectiveness overall with time, sorry, your cost effectiveness decreases over time with continued EEG monitoring. So how can we provide value monitoring and monitor patients that are at high risk for longer compared to those who are at low risk? And so the routine, we know that the routine EEG monitoring does not work just because of resource limitation in many institutions, including ours, which is a tertiary level institution. And so we need to appropriately utilize our resources based on individual patient needs. And then we have to have a cost-effective monitoring that does not compromise patient safety. That is, we are not missing seizures overall. And so that's where some of the literature guidelines come in and the first one that helps to be score is a seizure detection scoring system that stratifies patients based on their risk of seizure. And this has been validated in the literature based on about 5,000 EEGs and then was validated, this was in 2017, and then most recently it was validated in about 2,000 patients in 2020 as well. And I will go in through the clinical criteria for calculating the HELPS-to-be score itself. So there are EEG characteristics. And so some of these features are the EEG features. If you have brief ictal rhythmic discharges, you get two points. If you have lateralized periodic discharges, you get a point up here. And so these are the EEG features and then you have clinical features and that is clinical history based on whether the patient had a prior seizure, they came in with a clinical seizure or if they have a history of epilepsy, so they get a point. And that is the score that has been validated in the literature. And this is a graph of the time-dependent seizure risk that is stratified based on the HELPS-to-be score itself. But basically it shows that our duration of monitoring can be lower for those who are at low risk. And then based on our guidelines, for low-risk scores, the duration of monitoring for seizure detection less than 5%, about one hour. If you want to get it to less than 2%, it's about 3.3 hours. Medium risk, on the other hand, up here, you can, it's 12 hours and then 29 hours and high risk, about 24 to 48 hours. And I will invite. to invite Emily back to go through our project here at UNC. Thanks. So we had three primary aims for our project at UNC. We wanted to determine our baseline utilization across all adult ICUs at UNC, and we wanted to prioritize potential waste and determine ways to reduce that waste. We then developed a shared protocol for identifying and triaging patients who needed continuous EEG monitoring as well as those who can come off and when they can come off the EEG safely. And then we developed and implemented the clinical risk scoring system and helped clarify roles between and amongst teams. Our intervention comprised of three phases. We focused on communication, we focused on implementation solely within the NSICU, and then the third phase we brought into other ICUs at UNC, and today we're really proud to present the data across all of the adult ICUs at UNC. So some of the details of our project. The first phase, as I mentioned, we targeted communication and we created a simple daily touch point between the neurointensivist as well as the reading epileptologist, defining a particular time that these two leaders of the clinical team were able to speak about the patients. Then simultaneously we taught all levels of providers from all trainees about the scoring system as well as attending physicians about the scoring system in all disciplines through journal clubs, grand rounds, as well as lecture series. And then finally we implemented this score directly into rounds as well as ensuring consistency across our documentation by creating smart phrases within our EPIC electronic health record. And this, by incorporating, we were able to incorporate the vernacular directly so that we were all on the same page. So our QI project criteria really mirrored that which was used by Strzok et al., who were the folks who, the group that determined the two-helps-to-be score from the beginning. You can see our inclusion-exclusion criteria. We primarily, we included adults who were over age 18 on continuous EG and in the ICU, and importantly, our exclusion criteria are the same as those that were used by Strzok et al., and our interventions spanned across all levels of the care team, from readers to fellows to residents, APPs and attendings. Our measures included what was the average duration in hours of continuous EEG as well as seizure detection rate, and what we did was we pre-intervention, we retrospectively looked at charts, and we calculated what that two-helps-to-be score would have been, and then post-intervention, we, with the actual two-helps-to-be score in the EEG report, we were able to compare pre- and post-cohorts of patients. So these are our results, and what this highlights here is that over 437 patients, we were able to reduce the average number of hours of continuous EEG monitoring, particularly in the low- and the medium-risk scoring group. This second graph shows that the seizure detection rate per risk group, so patients in the high-risk group had the higher incidence of seizures compared to our low- and moderate-risk groups. We didn't appreciate a statistically significant difference in seizure detection rates after application of the two-helps-to-be score. And this graph looks at the duration, the average EEG duration as mapped against the two-helps-to-be score, and the bigger the bubble indicates the number of seizures or approximates the number of seizures that were detected. So you can see that the bigger the bubble is, the greater number of seizures that were detected during that time frame, which correlates with the two-helps-to-be score. So briefly, in our last remaining time, I'll just discuss how we understand the results. And so most of our patients across all ICUs had low- or moderate-seizure risk, and that also correlated with a low incidence of recorded seizures on continuous EEG monitoring. And our two-helps-to-be implementation of the two-helps-to-be risk scoring model identified these low-risk patients, and we did not compromise our detection rates. Next, we plan to look at how often are we potentially reapplying the continuous EEG in the event of removing too early, potentially. The group at UNC is also going to be looking at differences in outcome measures based on utilization of anti-seizure medication, what their Glasgow outcome scale is at discharge. And then, interestingly, we are hoping to see how we can use the two-helps-to-be score as a way to triage patients who are outside of our institution and might be transferred to UNC, and help our colleagues at community hospitals outside of UNC determine whether their patient needs continuous EEG monitoring at our tertiary care center. So I really wanted to thank everybody for being here, and particularly to the Congress, as well as my alma mater. And this is the great group that's continuing to do the good work. I also wanted to thank Kaiser Permanente for sending me here, or allowing me to come here today and not go to wards today.

seizure risk scoring tool

continuous EEG monitoring

patient outcomes

two-helps-to-be score

communication