and welcome to today's Journal Club Spotlight on Pharmacy webcast, which is supported by the Society of Critical Care Medicine CPP section. My name is Kelsey Ladd, a clinical pharmacy specialist in the Neuroscience ICU with Jackson Memorial Hospital in Miami, Florida. I will be moderating today's webcast. A recording of this webcast will be available to registered attendees. Log into mysccm.org and navigate to the My Learning tab to access the recording. A few housekeeping items before we get started. There will be a Q&A after each of today's speakers. To submit questions throughout the presentation, type into the question box located on your control panel. You will also have the opportunity to participate in several interactive polls. When you see the poll, simply click the bubble next to your choice. You may also follow and participate in live discussion on Twitter following hashtag SCCM CPPJC and hashtag PharmICU. Please note the disclaimer stating that content to follow is for educational purposes only. And now I'd like to introduce your speakers for today. Each will give a 15-minute presentation followed by a Q&A. Our first presenter today is Lauren Schmidt, a PGY2 Critical Care Pharmacy resident at Eskenazi Health in Indianapolis, Indiana. She will present on an indexinib for factor Xa inhibitor associated acute intracerebral hemorrhage. Our second presenter is Brett Snyder, a PGY2 emergency medicine resident at University of Pittsburgh Medical Center Mercy in Pittsburgh, Pennsylvania. He will present on comparison of nebulized ketamine to intravenous subdissociative dose ketamine for treating acute painful conditions in the emergency department. A prospective randomized double-blind, double-dummy controlled trial. And our third presenter is Madison Roberts, a PGY2 Critical Care Pharmacy resident at Virginia Commonwealth University Health System in Richmond, Virginia. She will present the effect of computerized decision support alerts tailored to intensive care on the administration of high-risk drug combinations and their monitoring. A cluster randomized step wedge trial. And now I'll turn things over to our first presenter, Lauren Schmidt. All right, thank you very much for this opportunity and for all of you attending today. I'm excited to present this highly anticipated trial and share all of the findings with you today. So to start off with some background, andexinib alpha or andexa is a modified recombinant inactive form of human factor Xa. So it basically acts as a direct reversal agent for our factor Xa inhibitors and rapidly reduces the effects of dox. And previous data on andexinib alpha was seen with mainly the anexa A and anexa R trials, and then most recently anexa IV, which showed about 80% hemostatic efficacy and 10% thrombotic events, which is a similar trend that we will see with this trial as well. So prior to anexa I, the study we're presenting today, no studies have compared and dexinet to standard of care in a randomized control trial fashion, specifically to PCC or K-Centra. And before we get into any of the information on the study, we're going to start with our first polling question. So what does your hospital use as their preferred reversal agent for factor Xa inhibitor reversal, either PCC, and dexinet, or something else? And it looks like majority of the participants answered PCC, which I expected. So I'm not surprised by those answers. But we can look into the details of this trial and see if maybe you might want to change your mind. We'll see. All right, so for the trial design, it was an international prospective, multi-center, randomized, open-label study. And it was funded by the manufacturer of andexa, which is AstraZeneca now, but previously was Portola at the beginning of the trial. And the purpose was to assess the safety and efficacy of andexinet compared to usual care in patients with acute intracerebral hemorrhage. And then just to clarify, the type of hemorrhage at first in the first version of the protocol, they were including all intracranial hemorrhages. But to help with consistency with the measurement of hemorrhage volume or hematoma volume, they did narrow that inclusion criteria to only include intracerebral hemorrhages specifically. So I think that was a benefit in the method just to help with that consistency with measurement of hematoma. But we will see that not only intracerebral hemorrhages were included in the final result. And then I did want to include some blinding information just because it was funded by the drug manufacturer. But I would say that there was a lot of proper blinding precautions that were taken into account with the trial design. As for how the patients were split up, they were randomized one-to-one to either receive Andexanet, Alpha, or Usual Care. So the Andexanet group either received high or low dose based on the FDA-approved labeling. So this was standard dosing for Andexanet. And then Usual Care group was a little bit more of a wild card. It was up to physician discretion. So most patients ended up getting PCC, which we'll see later in the results. But this was a very large multi-center study that occurred in multiple different countries. So standard of care might differ in these different sites. So some ended up getting four-factor. Some got three-factor PCC. Others might have gotten some other factor products that are not PCC. So it kind of just depended on the site and the physician. And then as for the results, efficacy analyses were performed in the intention-to-treat population and safety analyses in the extended population, which included all patients. And this population included intracerebral hemorrhage, which I explained earlier was a change that was made with the protocol. And then they had criteria on the exact hematoma volume and timing with the dose of the factor Xa inhibitor, symptom onset, and imaging before they were randomized. And then patients were excluded if they had previously received Andexanet, GCS Lesson 7, NIH-assessed score over 35, and then if they were going to get emergent surgery or had a thrombotic event recently before enrollment. And then moving into the endpoints, the primary outcome was hemostatic efficacy. And this was assessed at 12 hours after randomization. And this was a composite outcome of three major clinical points, which included change in hematoma volume of 20% or less, which was deemed as excellent change in hematoma volume, and then 35% or less, which was deemed as good change in hematoma volume. And then the next criteria was increase in NIH-assessed of less than 7 points, and then no rescue therapy needed. So this included a second dose of Andexanet, a dose of PCC, or emergent surgery within that 12 hours. And then secondary endpoints included this first bullet point basically is stating how well the reversal agent is working, so how it can get the anti-10A level from whatever level it was at to zero within the first two hours of randomization. And then another secondary endpoint was the MRS score or modified Rankin scale, so kind of looking at functional status of the patient. And then safety outcomes included thrombotic events and death at 30 days. And as for our statistical analysis, 900 patients were needed to reach 90% power. This was initially 440 patients, but with a protocol amendment that was changed to 900. The statistical test used for the primary endpoint was the Cochrane-Mantel-Hansel test, and then secondary endpoint used the ANCOVA test. And these were both tests that were appropriate for the data that was being analyzed. For the interim analysis, I did want to go through this a little bit because the trial did stop early after the interim analysis, so just to provide some information on that. At first, they had in the original protocol, they would allow the trial to stop early if a p-value less than 0.001 in favor of indexinate occurred for the primary outcome. But after the first amendment of the protocol, the rule was changed to allow the trial to stop early once they reached 450 patients enrolled, and then with a p-value favoring indexinate of 0.031 or less. So this did seem like a pretty standard interim analysis approach, and it was led by a statistician. So overall, I think that it was a pretty standard way to do the interim analysis, but just realizing that they did stop the trial early at the 450-patient mark and because they were favoring indexinate at that point. So as you may have noticed with the previous comments, they had made many amendments and changes to the study protocol, and I'm not going to read through all of these to you, but there were many, many changes that were made, and these just provide a summary of the major updates that occurred. So major changes I wanted to point out included changes to the inclusion and exclusion criteria. They did change some of the statistical tests that they were going to use, and then they removed some outcomes as well. And I would say this is kind of a limitation of the study, knowing that different, like especially with the inclusion and exclusion criteria, knowing that that had changed midway through the trial and some of the patients ended up being included that were not in the final protocol. It's just something to keep in mind and kind of a larger criticism of the study as we are able to analyze these results, just keeping those in mind. So going into our results, the study population ended up including 530 patients in the extended population, which is where the safety outcomes were analyzed, and then 452 patients in the intention-to-treat population. And then a big thing I wanted to point out with the population is that only 85.5% of the usual care group received PCC. So other patients ended up getting different factor products and some were unknown what they ended up getting. So just keeping that in mind as we are comparing these two groups and knowing that comparing Andexanet to usual care, if you are thinking of PCC as your usual care, this study isn't necessarily comparing one thing to another. It is comparing Andexanet versus a variety of different things. And although it is the majority of PCC, it's just something to keep in mind when you're analyzing the results. And as for the demographics of the study population, there were a few findings I wanted to point out. So most of our patients got Apixaban as the agent that was reversed, which is expected as that's one of the more popular DOACs. And then majority of the patients were intracerebral, which makes sense based on the inclusion criteria change that I discussed. And then another weakness I would say that I wanted to point out was the percentage of patients that had hemorrhage preceded by trauma, especially when we're considering using this trial at level one trauma centers or areas where we're going to be reversing anti-tenant factor inhibitors in the setting of trauma, knowing that a lot of majority of these patients were not trauma-induced hemorrhages. So just keeping that in mind as well. And then as for the GCS and NIHSS scores at baseline, these patients had pretty minor impairment considering their functional and neurologic status with median GCS of 15 and then NIHSS of 9. And then the median time from hospital presentation to receipt of treatment was a little bit longer than I would have expected, considering this is usually considered a stat agent. So it's about two hours for both of the groups. All right, and then let's move into the primary endpoint results. So for hemostatic efficacy, this was where we saw benefit with Andexanet compared to usual care. And this was 67% versus 53.1% in favor of Andexanet. And when they analyzed these results further, they did find that the biggest driver of this statistical difference was the hematoma volume change of 35% or less. So keeping that in mind that it was more of a objective measure, I would say, that drove the statistical significance and knowing that it was the change in volume rather than like an NIHSS or a modified Rankin score or something that's kind of measuring your clinical status of the patient. But they did have a statistical difference in the hemostatic efficacy overall. And then as for the secondary endpoints, they did show reduction in anti-TIN-A activity in favor of Andexanet as well. So it did have better reduction in the anti-TIN-A activity and overall reversal of the DOACs, and that is shown in the graph on the right as well. And then for MRS score, they did have no difference between the two groups for the MRS score between Andexanet and usual care. So when reviewing these primary and secondary analyses, they did show significant differences in more objective measures, but not as much in our clinical outcome. And then as for safety endpoints, this is probably the biggest kicker of the study. So for the thrombotic events, there was a statistically different result in thrombotic events with more thrombotic events in the Andexanet group compared to usual care. And unfortunately, this was not driven by like DVTs or more of our clots that are more benign, but it was your myocardial infarction and ischemic stroke with the biggest differences in the groups. So these are definitely huge safety endpoints to keep in mind when making a decision on which drug to use. So as for the strengths and weaknesses of the study, the strengths would be, it was a very large study population. It was a prospective randomized trial design. There was adequate blinding in place, although it was funded by the drug company. And then it was the first study of its kind to evaluate Andexanet alpha versus usual care in a randomized control fashion. And then for weaknesses, there are many, many authors with financial relationships with the drug company or industry partners of the study. There was a substantial amount of protocol amendments, which makes it a little bit difficult to assess the outcomes of the study and assess the groups that were included. And then the efficacy and safety outcomes were assessed in separate groups. And although this isn't necessarily a weakness of the trial, I think it would have been helpful to see the results in both of the groups so we could at least assess that information. And then the usual care group was not standardized, as I mentioned, which makes it difficult to see what we're truly comparing Andexanet to, especially if you are trying to compare it to PCC, which is what majority of us were using at our health system. So in conclusion, Andexanet has the high hemostatic efficacy balanced with the thrombotic risk. There are many significant changes to the study protocol and the results focus on objective improvement rather than your neurologic and functional status of the patient. And then I would say as my personal conclusion, further studies comparing Andexanet directly to PCC are needed before I could make a strong recommendation in favor of Andexanet. And that leads us to our final polling question, which is pretty much the same question, but based on the results of Annexa I, which reversal agent would be your preferred option for your DOAC reversal? All right, and it looks like everybody's still on board with K-Centra, which I agree with. And then I am happy to answer any questions that you all might have. Thank you, Lauren. So we have a few questions coming in. The first question that we have, so when in your practice, if any, would you choose to use Andexanet Alpha over PCC for intracerebral hemorrhage? That's a good question. I honestly would say there would be very few times that I would choose Andexanet in favor of PCC. In my opinion, I think the data's not quite there yet for any specific populations that I would recommend it in. And then especially considering like the cost of the agent and overall risk versus benefit, I'm not sure if I can actually pick a population where I would specifically use it, unless if we're able to show more, I guess, clinical benefit and clinical outcomes with the patients with like your MRS scores and actual functional neurologic outcomes. Thanks. And if you were to redesign this trial, how would you change either or both the intervention and the outcomes which were utilized and assessed for this trial? Yeah, that's a good question. I would say that I would like to have seen, which I kind of hinted at in the presentation, but I would have liked to see the usual care group more standardized so that we could compare one drug to another drug. And I know that would have probably taken a lot longer time and been more difficult of a study to design, but I think that would have been more helpful, especially considering majority of hospital systems are using PCC as their preferred reversal agent. So I would probably design it in a way that your like Andexanet group would be the same, but the PCC group would also have like standardized dosing and have standard like drug that they're actually getting. And then as for the outcomes, I think that it makes sense, the outcomes that they chose, especially comparing it to NXF-4, they chose pretty similar outcomes to assess with like the hematoma volume change and the 10A activity, which makes sense to assess that, but I think we know that the drug works and it reverses the agent, but I think what we want to see is the functional outcome, so I would probably design it to kind of focus more on like your modified Rankin scale and NIHSS scores rather than those like objective measurements. Awesome. Thank you. And we do have one final question in the chat here. So it's asking for at your institution, which index, so we're assuming it's not on formulary, what would you recommend for patients with a recent history of HIT for reversal? Yeah, that is a good question. I would probably have to look into that. I'm honestly not sure of the answer to that. Yeah, I'm sorry. I'm not sure what I would recommend in that situation. No worries. Thank you so much, Lauren. And now I'll be turning things back to our second presenter, Brett Snyder. Thank you very much. My name is Brett Snyder. I am a PGY-2 emergency medicine resident in Pittsburgh, Pennsylvania. Today I'll be providing an overview of the formula that compared nebulized ketamine to intravenous sub-dissociative dose ketamine for the treatment of acute pain in the emergency. My objectives include that I'll highlight current practices for using ketamine to treat acute pain. I'll also outline the methods and results of this trial and discuss ways that we could hopefully apply the findings to pharmacy practice. Ketamine is an NMDA antagonist, meaning it inhibits excitatory nerve transmission, and that allows it to be an effective analgesic. Ketamine is typically given intravenously, and for analgesia, the dose is 0.1 to 0.3 milligrams per kilogram. For comparison, an anesthetic dose would be closer to 1 milligram per kilogram. And we like to say that the dissociation starts around 0.5 milligrams per kilogram. Other routes of ketamine include intranasal and intramuscular, but the alternative route that was specifically addressed in this study was an inhaled route, such as through a breath actuated nebulizer. Ketamine has lower bioavailability compared to intramuscular or IV. It is about 20 to 40% of IV, but the duration of action is moderate at 20 to 40 minutes. Potential advantages of using nebulized ketamine is that it's less invasive for the patient, could have lower rates of adverse effects, and it could be practical in the emergency department when we do not have IV access. Taking into account that both IV and nebulized routes can provide adequate amounts of ketamine, the proposed hypothesis of this trial was to determine if IV ketamine, abbreviated as IV-SDK, at a dose of 0.3 milligrams per kilogram would provide better analgesia at 30 minutes following administration compared to nebulized ketamine, which is abbreviated as KBAN. The dose for that that was studied was 0.75 milligrams per kilogram. This was a randomized, double-blind, double-dummy superiority trial. It took place from October 2021 to September 2023 in a large community teaching hospital. The population of focus was adults who presented to the emergency department with moderate to severe acute pain. They followed an intention-to-treat study model, and it was determined that at least 67 people were needed in each group in order to meet 80% power. They included patients who were age 18 years or older, presented with acute pain, having an initial pain score of at least 5 on a 0 to 10 scale. The patients also had to be deemed candidates for ketamine based on the attending physician's judgment, and also following the institution's ketamine protocol. The patients also had to be able to verbalize adverse effects. Therefore, people were excluded if they had altermental status or were acutely intoxicated, or if they had unstable vital signs or painful conditions that required more immediate intervention, making them less than ideal candidates for ketamine. The primary endpoint was the difference in pain score between the two groups at 30 minutes after ketamine was administered. The superiority margin was set to be a difference of at least 1.3 between the two groups. Secondary outcomes looked at pain score differences also at 15 minutes, 60 minutes, 90 minutes, and 120 minutes. They also kept track of the need for rescue analgesia. The protocol allowed for IV morphine to be used for rescue analgesia. Other secondary outcomes that were followed included vital signs and the occurrence of adverse events. A total of 221 subjects were reviewed, but after applying the inclusion-exclusion criteria, 150 were included for final review. They allocated 75 patients to each group. There are no significant differences in their baseline characteristics, but to summarize the population, the majority of them were females in their mid to late 40s with baseline pain scores of 8.2, and that was the same baseline pain score for both ketamine groups. And both groups had normal vital signs at baseline. The most common indication for analgesia for both groups was abdominal pain, followed by traumatic musculoskeletal pain in the IV ketamine group, and then flank pain was the second most common indication for the nebulized ketamine group. Looking at the results for the primary outcome, there was no significant difference in the change in pain score in 30 minutes, and therefore the study failed to meet the superiority margin of 1.3. The average difference in pain at 30 minutes between the two groups was 0.23, with the IV ketamine group having the numerically lower pain score. For the secondary outcomes, there were also no significant differences at pain scores at 60, 90, and 120 minutes. The only time there was a significant difference was at 15 minutes, but the authors decided this was too soon to assess pain reduction since both of these medications can take up to 15 minutes to administer. Rescue analgesia was administered in both groups, but it was higher in the nebulized ketamine group. 21 subjects versus 10 received rescue analgesia in each group, respectively. However, to that point, the study protocol was not fully adhered to. There were medications given other than morphine, including acetaminophen, contorolac, and tramadol, and that was more so seen in the nebulized ketamine group, so it's possible that it could have misrepresented the effects of nebulized ketamine on the pain score. From a safety standpoint, both groups had stable vital signs after receiving ketamine, and more adverse events were observed in the IV ketamine group, and specifically, they noticed more sedation, dizziness, restlessness, and feelings of unreality. A per-protocol analysis was performed to address the fact that rescue analgesics were administered outside of the study protocol. This was intended to resolve the possibility that administering these analgesics could have diluted the treatment effect of ketamine. Despite this per-protocol analysis, the results were the same, and showed that there was no significant differences in pain scores, vital signs, and adverse effects. To summarize the authors' conclusions, their main takeaway was that they found no difference between ketamine-administered IV or through nebulization for the short-term treatment of moderate to severe acute pain in adult ED patients. The authors also support that both routes of administration can lead to a clinically important reduction in pain score at 30 minutes. Whenever I was assessing the article, strengths that I found first were the study design. I felt the study design allowed the research question to be appropriately addressed. It was a robust study design, randomized trial, and I also thought within that study design, they looked at an appropriate study population. One-time doses of ketamine are typically given in the emergency department for pain, so focusing on adults specifically in the emergency department was an appropriate way to compare nebulized to IV ketamine for acute pain. Additionally, the study was very relevant to emergency medicine practice, and a mainstay of our practice is we want to provide prompt pain control when someone does have acute pain, and sometimes IV access can be a barrier to this. So showing that the nebulized route is a viable alternative is certainly useful in emergency medicine. Initially, this study had strong external validity. The dose they chose for the nebulized ketamine was based on past research that showed that there was no difference in dosing between 0.75 mg per kg, 1 mg per kg, and 1.5 mg per kg. So if there was no difference, they opted for the lower dose as a way to help mitigate side effects. So I thought that the dosing for nebulized ketamine had strong external validity, along with the dosing for IV ketamine. It lines up with what our institution uses as well. There were some limitations within the study design, though. It was single-center and also had a small sample size, which limits its internal validity. The IV ketamine was consistently administered over 15 minutes, but the nebulized ketamine had varying inhalation times ranging from 5 to 15 minutes. And on top of that, the actual treatment time for the nebulized group was not documented. So it's possible that some of these patients weren't inhaling the full dose, or maybe if they were inhaling the full dose and taking the full time, the researchers were assessing their pain scores too soon. Initially, there was some non-adherence to the protocol and also blinding. As I mentioned, rescue analgesics were administered outside of the study protocol. So my potential theory here is if the nebulized ketamine group was getting more rescue analgesics, this could have falsely lowered the pain score in these patients at a faster rate. So that could have potentially caused a type 2 error, meaning we're not finding a significant difference between two groups, whatever there was one. Also, this was supposed to be a double-blinded study, but IV ketamine in particular can have a side effect of nystagmus. It can happen with nebulized ketamine too, but it is higher incidence in IV. So the research associates were not blinded to this side effect of nystagmus. To summarize this article, nebulized ketamine can serve as a viable alternative to IV ketamine. For the management of acute pain in the emergency department, my conclusion is that its place in therapy may be whenever IV access is unattainable or whenever adverse drug events are a bigger concern. But if you want to achieve analgesia faster pharmacokinetically, it makes sense to prefer the IV route. So that leads us into our pooling questions. The first one I had just out of curiosity is, what route of ketamine is used at your institution for acute pain? The options are nebulized, intravenous, both, or other. Up until I read this study, I had never seen or heard of nebulized ketamine. I've seen us use other intravenous agents like tranexamic acid through nebulization, but using nebulized ketamine for analgesia was something that was new to me. I can relate. I have only ever seen intravenous, but I thought that nebulized ketamine would be a novel approach. And I think it would be also something that's practical in pediatric patients, and there are case reports of that where it's used for them. It might just be more patient-friendly if they have needle phobia or just having trouble getting IV access for that patient population, too. But I would agree, from personal experience, I've only ever seen intravenous ketamine for pain. All right, the next pooling question. I was wondering if continuous ketamine infusions or PCAs for pain are permitted in your institution's emergency department? Oh wow, perfect 50-50 split. I was curious about this question because at my institution we have a policy that does not allow continuous ketamine infusions specifically in the emergency department just because we require more constant monitoring if we are going to use that and we just don't have the nursing staff to do that. So that was something I was just curious if other institutions had the same restrictions. So I'm glad to see that there's a variety of different approaches. But that's all I had. Thank you for listening. I'd be happy to answer any questions you have about this journal article. Thank you, Brett. So our first question, while this study didn't redose patients with ketamine, would you consider offering a second dose of either of these prior to a breakthrough medication like an opioid? And if so, what dose and route of ketamine would you choose? Yes, I can speak to the nebulized version first. I think it would be appropriate to offer them a second dose if we had a more objective measurement that they didn't receive a full first dose. So to elaborate on that a little bit, the same research group did a study comparing three doses of nebulized ketamine, 0.75 milligrams per kilogram, 1 and 1.5 milligrams per kilogram. And what I noticed was that in the smaller doses, patients were able to inhale more of that medication. There was less residual drug left over. So besides that point though, in all three of those doses, there still was residual drug left in the breath actuate nebulizer. So I think for nebulized ketamine, it's inevitable that the patient's not going to get the full dose. Ketamine as itself is not an ideal size for nebulization. It's estimated to be around like 10 to 25 micrometers, which is larger than an ideal aerosolized particle size. So I think it would be appropriate to redose nebulized ketamine. I think I would start maybe at a smaller dose, like 0.5 milligrams per kilogram, but I would also factor in how much residual drug was left over in the breath actuate nebulizer whenever we decide to redose. And I would definitely wait until up to like 30 minutes at least, if not 60 minutes. For IV ketamine, I don't have experience redosing that in the emergency department. The effects can last up to 10 minutes, sometimes a little bit longer than that. So I would at least give it that amount of time before redosing and maybe consider a slower infusion if we are going to redose, just to decrease the chances of side effects. All right. Would you recommend nebulized ketamine over other routes such as intramuscular or intranasal ketamine for acute pain if you don't have IV access? I was debating that because with IM, you're also causing like another small source of pain to the patient. And there's really not a need for that. Whenever we use IM, it's typically designated for agitation because it's just it's not as safe to get IV access in that patient or like unable to cooperate with the patient. So I would, after reading this study, seeing that both groups started out with the same identical pain scores, 8.2. And at 30 minutes, they had very similar pain scores of around 3.6. I would be open to using nebulized ketamine. We're not going to increase the chances of side effects by using nebulized over IV. If anything, it's going to be lesser risk. It's just I would remind myself that they might require more analgesics after the nebulized dose. So that's just like my mindset as emergency medicine informed citizens. I'm always trying to think ahead to the next steps and plan ahead. So if I did recommend nebulized ketamine to someone who didn't have IV access, I would just be prepared to give a rescue analgesic. So how would you redesign this study specific to look at the potential impact on opioid use or morphine male equivalents specifically? Well, so part of this study was they required the emergency medicine attendings and medical residents to attend courses that way everybody clearly understood the ketamine protocols, but built within those ketamine protocols. They did not specifically talk about rules for rescue analgesics, like those were specifically part of the site design, but not part of the institution's protocol. So I would somehow incorporate that into those required education sessions, just reminding them that they're only able to use, if they are going to need rescue analgesics, they should only be using morphine, and if they use other rescue analgesics, then we'll have to exclude these patients from the study, but that's how I would redesign it. All right, that concludes our Q&A session. Thank you, Brett. Now I'd like to introduce our final presenter, Madison Roberts. Hi, everyone. So to round us out, I'm going to be presenting at Patient Safety Center Journal Club. The title is a little bit of a doozy, so if I get through it, we'll be on a roll. But my article is titled, The Effect of Computerized Decision Support Alerts Tailored to Intensive Care on the Administration of High-Risk Drug Combinations and Their Monitoring, and this was a cluster randomized stepped wedge trial. So really the background for this journal club choice is that, you know, in our modern healthcare systems, we very often are utilizing these electronic health records that have within them these integrated clinical decision support systems, or CVSS, and oftentimes these alert systems are giving us an abundance of alerts. And alert fatigue has become a high buzzword concern in a lot of healthcare systems. These clinical support systems often have very low impact alerts that they are firing off, and it can lead to providers ignoring both low and high alerts because they are just so frequently getting inundated with these alerts. So currently it's estimated that around 90% of ICU alerts related to drug-drug interactions are overridden by providers, and within that 90% that are overridden, about 16% are considered clinically or critically important to patient care. So really just to put into perspective that while we may not be overriding all alerts inappropriately, there is definitely a proportion, specifically in the ICU, that we may be avoiding incorrectly. Just to pull some literature, I will say obviously electronic health records are relatively new in this grand scheme of healthcare. So there is emerging data on how we are utilizing these EHRs as well as these clinical support systems that are within them. So Nabavati and colleagues in 2017 tried to evaluate really if these drug-drug warnings were improving outcomes for patients. They found that the majority of the DDIs that were firing off did lead to improved outcomes with provider response to drug-drug interactions. And they also found that alerts were most effective when the provider was given the alert at the time of their decision making. So when they were actually entering the order, the alert firing then versus at a later date. Zapata and colleagues in 2022 tried to evaluate the frequency of drug-drug interaction alert overrides and really to see if the amount of overrides that were occurring was appropriate. And they found that around 50% of drug-drug interactions that were generated by these clinical support systems were overridden in general. I will say this was not specific to an ICU. So it can be argued that maybe in some of our floor patients there are less alerts firing off in general. And then Wong and colleagues in 2018 wanted to evaluate if the alerts that were being overridden were appropriate specifically in an ICU. And they found that four out of five alerts were overridden correctly. However, when you think about the volume of patients that we see in an ICU, one out of five alerts were inappropriately overridden, which is around 20% of patients. Or 20% of alerts, which when thinking about the number of patients could be quite sizable. And they also found that when alerts were overridden inappropriately, the likelihood that an adverse drug event reached the patient was six times more likely. So obviously much of this data is newer and there's not a lot of data on really evaluating ICU alerts and how it is impacting care, which is why I wanted to choose this article. So for my article, their hypothesis was that if you remove low yield drug-drug interaction alerts, you will, by default, increase provider awareness of the high risk drug-drug alerts. So they wanted to remove low yield alerts within an ICU clinical support decision making system. The ICU providers theoretically would then be forced to pay more attention to critically important alerts. And that would overall decrease the number of high risk drug-drug interactions that will actually reach the patient. So this was a cluster of randomized step wedge trial of nine ICUs in the Netherlands over a year long timeframe. All patients that they looked at were in adult ICUs, and to be included, every patient had to receive at least two drug administrations. The ICUs did have to utilize a specific CDSS system called MEM, and that was considered the control group for this study. And then a second clinical decision support system, MEM+, was created as the comparator. And over the 12-month period, they randomized these ICUs to either the MEM or the MEM+, group. So they're just briefly their methods. Any ICU that was not currently utilizing this CDSS actually had to turn it on. And they did give all staff in these ICUs training before the study actually started. MEM+, which was the comparator group, was created by a panel of intensivists and hospital pharmacists. They went through all of the alerts that were clinically considered to be drug-drug interactions by the system. And they worked through them to decide which were considered high risk versus low risk. And if they could not decide on an alert, they gave the ICU staff the choice to either keep it or to remove it. And all of the ICU alerts that they could not make a decision on, the ICUs chose to keep in the alerts. Just to give you an example of an alert that they could not reach a consensus on, one was epinephrine plus a beta blocker. Another was a potassium binder plus potassium supplementation. So just to kind of give you a frame of reference of what some of the alerts that they gave the ICU the decision to keep or to leave out. So for the control group, the MEM group, they had 139 drug-drug alerts, drug-drug interaction alerts in total. Versus the intervention group, which was MEM+, had 86 alerts in total. I just wanted to give a visual of what their design really was with this cluster stepped wedge approach. Because I think it's probably a less common trial design. So throughout these 12 months, ICUs were either started, they all started with the MEM clinical decision support system. And they were randomly placed into the MEM+. ICUs were not aware of when that would happen. And over the course of 12 months, every ICU slowly was placed into the intervention group. Their primary outcome was the number of administered high-risk drug combinations per 1,000 drug administrations per patient. Secondary outcomes were length of stay in the ICU, the number of administration of high-risk drug combinations in total. And the proportion of appropriately monitored high-risk drug combinations. For their statistics, they utilized a generalized mixed effects model. So they had multiple models that they designed to try to evaluate these alerts. Their primary one, M0, had no adjustment. And then they had the M1 model, which adjusted for baseline COPD and admission type, which there were differences between the groups in both of those categories. And then their M2 model was adjusted for sex, age, type of admission, cardiovascular disease, immunodeficiency, and their APACHE-IV scores. For some baseline characteristics, the median age for both groups was mid-60s. There were more males than females, and there was an APACHE-IV score of 51. The majority of patients were admitted for some type of medical reason. And then there was about, the rest of them were either for emergency surgical or for an elective surgical procedure. Looking at baseline and chronic conditions, so chronic renal failure, respiratory failure, CV disease, cirrhosis, malignancy, AIDS, and immunodeficiency were all the same between the two groups. However, as I mentioned previously, there was a higher instance of COPD in the intervention group, which is why they incorporated it into their first model. I will say of note when thinking about our patient populations in an American ICU, I found that the the rates of chronic conditions in these patients were very low. Just thinking about my sticky right now, chronic renal failure is way more prevalent than five to six percent. So just something to consider when thinking about the overall sickness and baseline illness these patients may have had. So in total, they had around 200,000 drug administrations with around 15 administrations per patient per admission. 0.8 high-risk drug combinations per patient were administered, but only about 35% of patients were actually exposed to those high-risk drug combinations. Looking at their primary outcome, so this was just to revisit it, the number of high-risk drug combinations that were administered per 1,000 administrations per patient, kind of a mouthful, but they did find that in the control group there was 35.6 high-risk administrations versus in the intervention group there was 26.2 high-risk administrations. And those p-values were statistically significant regardless of the adjustments they made in their models. I do think it's important to note what exactly they deemed to be a high-risk drug interaction. So their most common high-risk drug-drug interactions were two QT prolonging agents, and then the second one was NSAID plus corticosteroids. Some other ones to point out were NSAIDs plus serotonergic agents, phenytoin plus sulfuric acid, and then these iron preparations plus antacids or carbonate preparations. And I point these second ones out and really the first ones as well just to consider in our own ICU practice what we would consider a high-risk drug-drug interaction. Oftentimes these drug-drug interactions are a risk that we are either willing to take or willing to monitor appropriately. I know I don't very frequently consider NSAIDs plus corticosteroids a huge worry. So just considering the actual risk of these drug-drug interactions in this study compared to what we may be administering in our own ICUs here in the state. For their secondary outcomes, they actually did see a decrease in ICU length of stay in the MIM-plus or the Intervention Group. And they also saw a decrease in the number of administrations of high-risk drug combinations in the Intervention Group. Of note, these percentages that I have here are each based on all three models, which is why there's a range. And then of the seven drug combinations that they deemed to have monitoring strategies, there was actually an increase of appropriate monitoring in the Intervention Group. To briefly touch on what they considered to be appropriate monitoring, the vast majority was either adding an EKG for a patient, adding some type of gastric protection, whether that be a PPI or an H2 blocker, or getting therapeutic drug monitoring as necessary. Looking again at these drug combinations with appropriate monitoring, as I mentioned earlier, there was a 9% improvement in the appropriate monitoring in the Intervention Group. And this was statistically significant. So between groups, there was an improvement in monitoring of these drug combinations. However, it was driven predominantly by QT-prolonging agents or by agents that were able to monitor the patient for longer periods of time. QT-prolonging agents or by agents that were able to get levels on. So thinking from a pharmacy perspective, getting levels for these drugs was the driver of really the difference that they saw here. Some strength of the study, it was a large multi-centered study across multiple ICU environments. It did evaluate patient-centered outcomes that could potentially impact direct patient safety. And in a lot of the articles that look at data analysis for these alert systems, they really don't look at patient-centered outcomes. And even though those patient-centered outcomes were secondary in this study, I did think that was beneficial, especially when thinking about trying to remove alerts from an EHR and really having to justify how that may impact patient care. Some limitations. These interventions were not blinded. The ICU, again, did not know when they were going to be placed into the Intervention Group with less alerts. However, it is obvious, most likely, if you are all of a sudden getting less alerts from a provider standpoint. So it could increase their risk of bias in knowing that they were in the Intervention Group. Again, like I mentioned, many of the drug combinations that were considered high risk are drug combinations that we very commonly utilize in an ICU, and we consider them to oftentimes be an acceptable risk. There was only about 15 administrations of drugs per patient. So when thinking about, again, an ICU in the U.S., I know my patients in an ICU typically have 15 drug administrations per day at least. So likely these patients were potentially less critically ill than some of the patients we care for in our ICUs. And also they didn't actually look at the harm that was related to any of these drug interactions that reached the patient. So it's hard to know what the impact actually was in reducing the number of administrations that reached the patient. They also didn't really disclose the total of alerts that the providers were getting. Obviously, we know the alerts that were possible to receive, but we don't know the total number of alerts that providers were having to kind of wade through to make appropriate decisions. So takeaways from this study. Limiting the number of low-yield alerts may increase provider attention to high-yield alerts in an ICU environment. The number of high-risk drug combinations administered did decrease in the limited alert group. So when providers were faced with less decisions to make, they actually paid higher attention to the decisions that were given to them. And finally, CDSS optimization does require continued research, especially in an ICU environment where we are constantly adding new drugs to formulary, changing how we care for patients, have new guidelines, and providers are really stretched to make decisions that require a lot of intense thought and really require a good level of focus. So now for my polling questions. At your institution, do you feel that alert fatigue or high number of alerts directly impact patient care in the ICU? Yeah, so I agree. I know even as a pharmacist, when we're verifying orders, we are constantly inundated with alerts. At my institution, we have Epic, and I feel like I'm constantly benefit outweighs risk on all the orders I'm verifying. And I do think even in my own practice as a pharmacist, there have been times where I have missed very critical alerts because there were so many alerts that were low yield, and I just automatically was clicking through them. So I definitely think it's something to consider in all patient care, but specifically in an ICU. And then my next polling question, at your institution, do you... Oh, this is the same question. Let's see if I go to the next one. Oh, here we go. At your institution, are you currently working on or discussing to limit the number of low yield alerts in your EHR? Perfect. So definitely a conversation that is going on in the background and maybe you can utilize some of the findings of this study to kind of help make that argument. I know we are currently working on this pretty frequently and trying to limit alerts that have low yield for patients. And I will now take questions. So our first question we have here. What barriers would you anticipate in implementing a clinical decision support system to filter out low yield potential drug-drug interaction alerts in the ICU setting at your institution specifically? Yeah, that's a great question and something I think that we're still trying to figure out. I do think it's hard because low yield does not always mean not clinically applicable. So there's definitely the chance that a drug alert that is considered low yield in the vast majority of patients could be high yield in a specific patient population. So it's really difficult to make umbrella decisions on all patients, especially when you're looking at an EHR that's serving multiple patient populations. And I think that's a lot of the barrier to taking out some of the low yield alerts is that there's always the argument that to a small patient population or a single patient even, that alert could be really impactful. Especially when we have so many learners in the environment, I think it's really difficult for everyone to decide when an alert can really be low yield. Awesome. And then are there any other alerts aside from drug-drug interactions that you encounter at your institution that you believe may also contribute to alert fatigue and may also benefit from ICU specific settings for? Yeah, absolutely. Actually, one that we're currently working on at my institution are pregnancy alerts. We currently get them for all ages and obviously at a certain age that likely is not a high risk for a patient. We also currently get alerts for insulin in patients over the age of 65, which obviously insulin is a high risk medication, but we have diabetes at all ages. And so I know specifically those two alerts have had a lot of conversation recently, just because especially with something like insulin that could potentially have very high yield alerts associated with it. We're kind of stacking these low yield alerts in the order sets. So trying to work through how we can minimize alerts that do not apply to the vast majority of patients in general. Awesome. Thank you so much, Madison. Yeah, thank you guys. Yeah, thanks. That overall concludes our Q&A session for today. So thank you to our presenters today and the audience for attending. Please join us on the third Friday of the month from 2 to 3 p.m. Eastern Standard Time for the Next Journal Club Spotlight on Pharmacy. And that concludes our presentation for today.

Pharmacy

Intensive Care Unit

Andexanet Alpha

Factor Xa Inhibitors

Intracerebral Hemorrhage

Nebulized Ketamine

Intravenous Ketamine

Computerized Decision Support

High-Risk Drug Combinations

Patient Safety

Webcast

Pharmacology

Neuroscience

Analgesia and Sedation

Intracranial Hemorrhage

2024

Professional

Select