Tuesday, September 26, 2017

Lorazepam, Haloperidol, and Delirium

by Drew Rosielle

JAMA Internal Medicine has published a double-blind, randomized, placebo-controlled trial of adding lorazepam to haloperidol in patients with advanced cancer and agitated delirium. (We had a heads up about this trial because it was presented at ASCO earlier this year.) If there ever was a sort of consensus in HPM about how we should be treating delirium, my sense is that it’s been shattered by the recent RCT of low-dose haloperidol vs risperidone for delirium in Australian palliative care unit patients, showing those drugs worsened delirium symptoms. So, it seems like we should all see what we can learn from this newly published investigation.

The authors note that to be best of their knowledge there has never been a RCT involving a benzodiazepine compared with placebo for delirium. The one kind of famous (if you are a delirium geek) trial which looked at benzos, which was the trial I was directed to when I asked why not benzos for delirium when I was training, involving a 3-way comparison of lorazepam, haloperidol, and chlorpromazine for delirium in hospitalized patients with AIDS, and registered 6 (!) patients in the delirium arm (lorazepam patients did worse). It had no placebo arm.

In fact, there is a lack of high quality drug trials in the delirium world which involve genuine placebo arms, ie, an arm in which there was no active pharmacologic treatment. I have wondered if we’ve made a huge mistake by doing trials which assumed haloperidol was the standard of care, without robust data to actually support that, and so have just done comparison trials of haloperidol with other agents, or like this study, a benzo or placebo added to haloperidol, when then underlying question, does haloperidol or other antipsychotics actually help (compared with placebo) remains open. See for instance this recent review for a nice summary of what’s out there (noting that this even predated the damning low-dose haloperidol/risperidone/placebo trial): it’s not convincing, it’s not the sort of thing you’d read and say the book is closed on this question, we can no longer have equipoise about comparing antipsychotics to placebo in delirium trials.

I was going to get to this point later in the blog post, but I realize I’m already there, so I’ll say it now: delirium is an international health crisis, it is real, it can be devastating (lead to permanent cognitive changes), leads to far worse outcomes for our patients (longer hospital stays, not being discharged home), costs billions of dollars, sucks shit for the patients and families going through it, and we don’t have a real inkling about actual, effective drug treatment for it. There are some inklings about chemoprophylaxis, but nothing definitive. Multidimensional prevention programs seem good, I like those, I’m 100% for those, but we need a lot more. If we can do a bloody RCT of simvastatin for chemoprophylaxis of delirium, surely we can do large, multicenter, patient-diverse (dementia, surgical patients, ICU patients, advanced cancer patients, dying patients), high quality, placebo-controlled trials of a variety of drugs, drug-classes – especially testing the ones people are actually using, and dosing strategies to see if there are any effective disease-modifying agents out there!

Which is why I’m just delighted our friends at MD Anderson did this study as it was well-done, although small, and adds to our understanding.

The subjects (N~60) were patients at MD Anderson’s palliative care unit; all had advanced cancer, and all had agitated delirium (RASS score of 2 or more; it looks like they changed the protocol to 1 or more mid-study – there’s a lengthy supplement involving the protocol changes). All had received haloperidol as a primary treatment for delirium – it looks like they used a protocol of 2 mg IV q4h scheduled + 2 mg IV q1h as needed. The patients were followed with q2 hourly RASS assessments and then received 3mg IV lorazepam or placebo if they continued to have an agitated RASS (mean RASS score prior to being the study drug was 1.6). Importantly, the patients also received a 2 mg prn dose of haloperidol as well regardless of which group they were assigned too. Of note, all the patients studied had delirium for at least 2 days prior to enrollment – these were patients with persistent delirium who didn’t get better quickly after routine interventions. After enrollment the median observation time was 6.4 h (after a median of 6.4 h, the patients had an agitation episode leading to being given the study drug). Median haloperidol doses prior to receiving the study drug were 5-7 mg in the prior 24h.

The primary outcome was RASS score 8 h after the single dose of study drug.

The results are fascinating as hell. To begin with, lorazepam looks really, really good when you are looking simply at the primary outcome of reduction in the RASS score. The RASS score went down rapidly as you’d expect – it was markedly lower in the lorazepam group by half an hour – and remained markedly lower than the placebo group at hour 8: -2.5 for lorazepam, -0.5 for the placebo group (all this was very statistically significant by all the usual tests). Of note, -2 on the RASS is light sedation (briefly awakens with eye contact to voice, but less than 10 seconds); -3 is movement or eye opening to voice but no eye contact.

Ie, these patients were pretty sedated, and much more so than the haloperidol/placebo-only patients.

One may also observe that the placebo patients having a RASS of -0.5 at 8 h means the median state of the placebo patients at 8 hours was something between ‘alert & calm’ and ‘mildly drowsy.’

Got that? While the RASS was a lot lower in lorazepam, the placebo patients’ median RASS could be considered, in fact, a really good outcome, and arguably, a better outcome, than the lorazepam group.

Of course, the reality is much more complicated than that (most delirious patients just don’t go back to normal and stay there), but it’s a good reminder that when using something like the RASS as an outcome, lower is not necessarily better: 0 is normal, and for many patients would be the most desirable outcome. (Notably, the authors address all this explicitly in their discussion.)

Figure 2b in the study is probably more helpful in understanding what happened to these patients than the actual primary outcome graph. What you can see is that at 8 h, in the lorazepam group, basically half the patients were deeply sedated (RASS -3 or less), and half were mildly sedated. In the placebo group, about a third were agitated, and the most of the rest in the mildly to moderately sedated range. If you want, you can actually see what happened to individual patients in the supplement online.

In the secondary findings, they note that many more nurses and caregivers in the lorazepam group than the placebo group judged the patient to be comfortable after the study drug was administered (~80 vs ~30%). The lorazepam group also had fewer ongoing doses of rescue medicines also. Median survival for both groups was ~70 h. I.e., you can essentially understand this to be a study of patients with terminal delirium.

What does all this mean?

It means that lorazepam effectively and rapidly sedates people, better than haloperidol, at the doses studied.

We’ve discussed this on the blog a little before, but this study helps us think about delirium and delirium outcomes better - what outcome we are actually aiming for in these patients. As I implied in the opening discussion of this blog post, I myself am at a point at which I do not consider there to be any active, disease-modifying drug treatment for delirium that I can get behind. By disease-modifying I mean drugs which would return patients to a more normal state of consciousness (i.e. push people closer to 0 on the RASS), and/or reduce the duration of delirium, and/or its long-term adverse outcomes. I think there is hope for antipsychotics, especially used in higher doses than the Australian study, but I don’t think there are any available data which convince me these should be a routine part of our care for delirious patients. We need well-powered, meticulously designed, placebo-controlled, and multi-institutional studies of haloperidol/other antipsychotics.

Until then, it’s just hope, and I honestly don’t know what to do.

I want to be clear, though – with the above I am talking about disease-modifying treatments for delirium. We clearly have, however, rapid & effective ways for reducing the distressing behaviors of agitation by sedating patients.

And I think it’s important that we keep in mind there is a difference between these two goals (sedation vs disease-modification). For our patients near the end of life sedation is often appropriate & acceptable. For some of our patients and their families it is in fact desirable - as this study showed, caregivers as a whole really preferred the moderately to deeply sedated state lorazepam gave these agitated, dying patients. It’s what I would want for myself, or my close loved ones, when close to death.

Sedation however is just not a ‘treatment for delirium,’ in the way that I used to hope low-dose haloperidol was.

Lorazepam has had a bad rap for ‘delirium’ historically. All of us have seen patients become agitated after getting it. Undoubtedly it is a common cause of delirium in hospitalized patients. While I don’t have any data to support this claim, I think much of the bad rap comes from patients who were given small doses of lorazepam, anxiolytic doses, leading to confusion and disinhibition. They weren’t given sedating doses of lorazepam, which is not a hard thing to do, and quite predictably sedates most people. To me, what this study does is help clarify that lorazepam very much does have a role in agitated delirium in patients near the end of life, when the immediate therapeutic goal is sedation. When we do it, however, we should do it right, and use the 3 mg dose like they did in this study, after of course clarifying prognosis and treatment goals with appropriate surrogates.

Pallimed | Blogger Template adapted from Mash2 by Bloggermint