Hi, my name is Stephanie Cha and I am an Assistant Professor of Anesthesiology in Critical Care Medicine. I specialize in Cardiothoracic Anesthesia and Cardiac Surgical Critical Care, and this panel is about difficult delirium drug decisions, specifically should antipsychotics be used or avoided. I do have two relationships to disclose listed here, neither of which will influence the content of this talk. Here's a broad outline for this talk. We'll discuss delirium, how to define it and the scope of its problem. We'll review some literature looking at antipsychotic use for prevention and or treatment of delirium in ICU patients. How all of this fits in with the 2018 SECM published pain agitation delirium guidelines and some other unique considerations. The problem of delirium is complex. While it is common, with an incidence reported to be as high as 80% of ICU patients, it can be difficult to diagnose due to its many subtypes, the most common of which is probably the image of a hyperactive patient, you know, arms flailing, nurses running to the bedside, providing restraints, needing a medication to calm them. But really hypoactive delirium, which is marked by quietude or withdrawn behavior, is much more common. And in addition, some patients have a mixed subtype where they may have both hyperactive or hypoactive episodes. Regardless, when present, delirium does increase both patient and provider risks for harm and preclude necessary cares, especially physical therapy, rehabilitation, ventilator weaning, et cetera. And we know it is associated with poor long-term outcomes, although the causality is really a mystery. Is delirium just a marker of severe systemic disease or does it actually precipitate some of these poor outcomes? What are some of these poor long-term outcomes? Well, ICU delirium, when present, is associated with a greater risk for six-month mortality, longer hospital length of stay, greater hospital costs, with one report of up to $38 billion per year in elderly patients, and a greater risk for post-ICU cognitive impairment. On the right-hand side of the slide, you can see a famous study by the Vanderbilt Group in 2004, which demonstrated delirium as a predictor of mortality in here. We can see the survival curves of ICU patients with delirium, without delirium, diverging markedly from time of enrollment, with the survivorship of delirious patients about 65% and those of non-delirious patients closer to 85%. The pathophysiology of delirium is multifactorial and poorly characterized, but I thought this schematic showed nicely how many components converge in final pathways that produce what we call acute brain failure. So we know that precipitance of delirium, infection, trauma, surgery, metabolic derangements, hypoxemia, combined with patient risk factors such as pre-existing dementia or cognitive impairment, elderly age, can all produce, among different pathways, dysregulation of our neuroendocrine system, producing acute brain failure or delirium. And I think thinking of it as acute brain failure is very helpful because it highlights a specific organ dysfunction versus an entity that we sometimes mistakenly characterize as a psychiatric process. What do you all do in practice? Well, even though there is no FDA-approved drug for delirium, we know that antipsychotics are frequently used, the most common of which would include first-generation antipsychotics or Haldol, which act on the dopamine pathway and are therefore associated with a number of side effects that encompass extrapyramidal symptoms, akathesias, dystonias, et cetera. But also atypical antipsychotics, also known as second-generation antipsychotics, that act on the serotonergic pathway. And we can see on the right-hand side of this slide a survey from 2011 of ICU pharmacists who reported that the majority, over 85%, believed that delirium should be treated with pharmacotherapy and almost 70% reported two or more meds should be used. And we can see the most common agents used here were haloperidol, catiopine, and lorazepam. And again, just to specify, those atypical antipsychotics from this list include catiopine, olanzapam, and zeprasidone. There was a lot of early interest in investigating the use of antipsychotics for delirium treatment. And while some did potentially demonstrate a reduction in delirium severity, in general, these early studies failed to demonstrate consistent benefits in some of the more meaningful outcome categories that I have listed here, like mortality, hospital length of stay, or delirium-free days. In addition, this group of studies was limited by their small size, generally. Study heterogeneity and how and which antipsychotics were utilized. Again, a limited number of meaningful clinical outcomes, as well as heterogeneity in population studied, and in many cases, language restrictions or barriers. More recently, there was a meta-analysis performed in 2019 of adult inpatients with delirium. And so 5,607 patients were studied among 16 RCTs and 10 observational studies that you can see listed here. And if you can look at the fine print, we'll see that most adult inpatient categories were non-critically ill patients, but there were several that looked at critically ill patients. These groups utilized assessment standards of either the DSM or CAM-ICU confusion assessment method for diagnosing delirium, and at least this meta-analysis looked at common outcomes of interest, including mortality, delirium severity, hospital length of stay, and cognitive function. These schematics come from that meta-analysis. Here we can see three illustrations that correspond with different outcome measures, cognitive functioning, delirium severity, and hospital length of stay. The shaded boxes correspond with no significant difference or insufficient evidence found, and the circles each represent a different study. The larger the circle, the larger the sample size. And so overall, what was determined was that there was no difference or there was insufficient evidence in these outcome measures when comparing Haldol versus placebo, second generation antipsychotics versus placebo, or Haldol versus the second generation antipsychotics across outcome domains, which included, again, mortality, length of stay, cognitive functioning, delirium incidence, and sedation level. This plot illustrates the safety profile. You know, the hypothesis was that the second generation antipsychotics might cause less extrapyramidal symptoms than the first generation antipsychotics, but may be associated with more cardiac effects due to their QTC prolonging properties, which could lead to the dreaded ventricular arrhythmias, particularly torsades. But what we can see here is that in general, when it came to cardiac or neurologic effects, the drugs were relatively safe. I'd like to take a look at one of those trials included in the meta-analysis in slightly more depth, and this was the MIND-USA trial of 2018. This trial was the largest trial of the series, and was a one-to-one-to-one RCT that compared placebo versus ciprocidone, which is a second generation antipsychotic, versus haloperidol. 566 medical and or surgical ICU patients were included who had been diagnosed with delirium by the confusion assessment method, and they were in acute respiratory failure or shock. And they were given drugs titrated at 12-hour intervals for a pre-designated max dose. So for haloperidol, it was 10 milligrams per dose or 20 milligrams per day max, and for ciprocidone, it was 20 milligrams per dose or 40 milligrams per day max. Also important to note that since this is a more recent study, there was great emphasis on clinical practice for incorporation of our ABCDEF bundles, some of these non-pharmacologic interventions in practice. The major takeaways from this trial were that the drugs were ineffective in treating delirium. There were no differences in delirium-free days, coma-free days, ICU or hospital length of stay, ICU remission rates, or 30- or 90-day mortality. In general, the drugs had a good safety profile, and it was unclear if drugs may have benefit at all. But it is important to note that hyperactive patients only made up a small percentage of the cohort, so 11%. And in addition, there perhaps was an effect of additional drugs given or crossover between groups. So for instance, 20% of trial drug was withheld due to patient, family, or medical team refusal, and over 90% of patients received additional medications and sedatives. Well, what about antipsychotics for delirium prevention? This illustration comes from a meta-analysis of 14 RCTs for a total of 4,281 patients. And again, what we see here is plotted delirium incidence and mortality. There was no difference in Haldol-first placebo for mortality, delirium incidence, or sedation. But there may have been some benefit for those second-generation antipsychotics, first placebo, and delirium incidence. But it is important to note that those three RCTs summarized here were all in post-operative patients. So two were cardiac surgical populations, and one was an orthopedic surgery population. While it looks promising for post-operative delirium prevention, we should take into consideration that while the delirium incidence was less, delirium severity and duration was more. In addition, at least one of those studies, there was no blinding performed. So does this fit in with our pain agitation delirium guidelines? In 2018, SCCM put forth a number of recommendations in which they did state that there is no recommendation for routine use of pharmacotherapies for prevention or treatment of delirium or subsyndromal delirium. And again, just to summarize what we've already discussed, while small studies have variably reported a reduction in delirium incidence, there is no statistically significant outcome of meaning demonstrated with the use of these antipsychotics. In addition, their high side effect profile probably tips the balance to greater risk versus benefit of these agents. So antipsychotics aside, do any drugs work? Here are a number of drugs which have also been studied to varying degrees. There was interest in statins due to one study that suggested that abruptly stopping statin use increased delirium incidence. But to date, there really is no evidence that statins, much like antipsychotics, are associated with reduced delirium duration, length of stay, mortality, or ventilator duration. There's a study that looks at IV acetaminophen with the addition of either propofol or dexmedetomidine with some promising results for delirium incidence, so may be useful in prevention of delirium. But again, this is one study. Ketamine and MDA antagonists have also been looked at with interest. There really is insufficient data to make any conclusions, although trial arms that were given ketamine did have sedative sparing effects, so less propofol or less dexmedetomidine additionally given, not surprising. Propofol and dexmedetomidine have been looked at with a lot more interest, especially in the surgical populations for prevention of postoperative delirium. You know, propofol versus placebo, dexmedetomidine versus placebo, propofol and dexmedetomidine versus benzodiazepine, propofol, dexmedetomidine, head to head. And the conclusions of these are really that there is one RCT that supports low-dose nocturnal dexmedetomidine use in ICU patients. That study was terminated early because the dexmedetomidine or trial drug was used up early, and the patients who received it did show reduced ventilator times. In addition, propofol and dexmedetomidine are conditionally recommended over benzodiazepine infusions due to their more easy titratability, shorter time to extubation, but overall quality of evidence is particularly weak. So I would say when considering sedatives in hyperactive patients, we at least know that propofol and dexmedetomidine are probably better than benzodiazepine infusions, and it's reasonable to consider both of them over antipsychotic use. So finally, what about antipsychotic use in special populations? Again, we've discussed hyperactive delirium as a subtype that may benefit from drug therapy. It is a subtype that is very underrepresented in most of the RCTs that have been mentioned. And what about medical versus surgical patients? Again, there are unique pathologies for postoperative delirium that may differ from medical intensive care unit patients, but probably still insufficient data to support the routine use of antipsychotics in these patients. Withdrawal syndromes, there is a very old 1997 consensus statement that does support the use of halperidol as an adjunct for alcohol withdrawal. Not surprisingly, it was found to be inferior to benzodiazepines, but may be useful, again, as an adjunct to control hallucinations. And there really is not much more data beyond this. Another population of interest is patients with sleep disruption or insomnia. I think this is a really interesting population because, you know, speaking from experience, we often use antipsychotics in these patients for their sedative effects. But really, there is no evidence to suggest that, you know, while antipsychotics may target sleep receptors or produce sedation or even increased duration of sleep, that they improve either sleep quality or overall sleep hygiene or cycling in critically ill patients unless those patients have concomitant psychiatric illnesses that require treatment. So at least one meta-analysis suggested that antipsychotics increase slow-wave sleep and total sleep time, but in patients with pre-existing schizophrenia. Another area where there is really no data is use of antipsychotics in weaning high-dose sedatives. So during the COVID pandemic, for instance, we saw a lot of very deep, rascal sedation regimens that eventually needed to be weaned, and we often did so with agents like atypical antipsychotics. But there really is no evidence for this. In fact, what we do know and is more established is that patients who begin antipsychotic use in hospital are likely to continue it unnecessarily in their post-hospital course as well. And therefore, subject to all of the side effect risks that come with them. So in conclusion, we really cannot make any recommendation for the routine use of antipsychotics for prevention or treatment of delirium. We could consider antipsychotics for delirious patients who have the hyperactive subset, potentially those who have post-operative delirium. But if that is the case, then we should also consider other medications and definitely consider non-pharmacologic therapies, and that we should not routinely use antipsychotics for treatment of sleep disruption or withdrawal syndromes as a sole agent. Thank you so much for your time. My email is listed here, sch4hh.edu, for anyone with additional questions. And again, thanks for your interest.

antipsychotic drugs

delirium

ICU patients

diagnosis

non-pharmacologic therapies