Effect of Intravenous or Intraosseous Calcium vs Saline on Return of Spontaneous Circulation in Adults With Out-of-Hospital Cardiac Arrest: A Randomized Clinical Trial

[00:00:00] Thank you very much, Rob. My name is Lars Andersen, and it's a pleasure to be back here at Critical Care Reviews to present the results of the Calcium for Out-of-Hospital Cardiac Arrest Trial, also known as the COCA trial. Before getting started, I just want to mention the authors on the manuscript and some of the members of the steering committee. This, of course, has been a tremendous amount of work for a lot of people, and I just want to thank everyone who has been involved. The trial was primarily funded by a grant from the Novo Nordisk Foundation. This is a nonprofit grand that is part of the Novo Nordisk Foundation. We also received funding from a number of other sources including the hospital and the university. I'm going to hand over the work to Mikael, who was the PhD student on this project.

[00:00:59] Thanks so much, Lars. My name is Mikael Fink Vallentin, and I'm a PhD student and a doctor here at the PreHospital Emergency Medical Services in the Central Denmark region, and I'm happy to present the COCA trial for you. First, I want to start out with some incidence numbers for out-of-hospital cardiac arrests. In the U.S., we know that around 350,000 have an out-of-hospital cardiac arrest each year, and in Denmark, the number is around 5,000. It is a high-mortality condition. Around one out of four regain circulation, and if we look 30 days into after the cardiac arrest, around one in six will survive. [inaudible] a curve of the survival numbers for out-of-hospital cardiac arrests in Denmark. On the x axis, you have the year, ranging from 2001 to 2016, and on the y axis, you have the proportion of survival at 30 days. The black line is the nonshockable rhythms and the red line is the shockable rhythms. You can here see that the increase in survival, the trend that we have seen over the last 15 to 20 years is highly influenced by the shockable rhythms, indicating that early defibrillation and also early CPR are important factors in the increasing survival trend over the last years. One of the reasons that we wanted to test calcium is that it could be one of the interventions that could benefit the nonshockable rhythms that are, that comprises around 80% of the out-of-hospital cardiac arrests. We know that calcium when given intravenously is an inotropic agent. It also has a vasopressor effect, and it may also counter some of the proarrhythmic effects that we know can occur due to hyperkalemia. Hyperkalemia is a condition that we often see in the first blood tests when the patients arrive in the hospital. As far as the evidence goes for calcium for out-of-hospital cardiac arrests, there have been two small, randomized control trials that were conducted by the same author group back in 1985, and one of the trials were in asystole, and one of the trials were in pulseless electrical activity. These trials included a total of 163 patients. So, they were small trial. If you look at the outcome, return of spontaneous circulation at hospital arrival, you can see that we have conducted a rather informal metanalysis. That is the figure you see in the bottom left. In the middle, you have a black line, which indicates a neutral outcome for the ROSC at hospital arrival, and any numbers or figures on the left of the line is in favor of placebo, and any numbers on the right of the line is in favor of calcium. As you can see, the black bar, which is the metanalysis, although there is a very wide confidence into the point estimates that indicate that calcium could be beneficial for return of spontaneous circulation. Going back to calcium, the use during, we know that the use during in-hospital cardiac arrest, and that is the figure you see here, this is a figure we have taken from an article by Ari Moskowitz, and you'll see here on the x axis and the proportion of patients that receive calcium during their resuscitation and in-hospital cardiac arrest, these are U.S. numbers. And as you can see, there is rather high prevalence of calcium administration during in-hospital cardiac arrest, and the trend is that this is increasing and has been increasing over the last 20 years. The dark blue figures are for the nonshockable rhythms, and it indicates that it is especially these patients that are receiving calcium administration in a high number. For out-of-hospital cardiac arrest, we do not know how many patients receive calcium during the resuscitation attempt, but I think a lot of doctors will know that it's something that happens anecdotally. We have this figure from Rob MacSweeney, who conducted this rather nonscientific poll on Twitter, and it asks whether you would give a patient calcium during resuscitation for out-of-hospital cardiac arrest, and 13% answered yes. What do the guidelines say? The guidelines say that routine administration of calcium, they do not recommend routine administration of calcium. They do, however, recommend that you could give calcium if the cardiac arrest was caused by one of three conditions, and the patient is in pulseless electrical activity. The three conditions are hyperkalemia, hypocalcemia, and an overdose of calcium channel blocking drugs. These are very rare causes of cardiac arrest, so the evidence for calcium being beneficial in these conditions is rather scarce. Looking at the American guidelines, we see similar recommendations. They do not recommend routine administration of calcium during either in-hospital or out-of-hospital cardiac arrest, but it may be considered if the cardiac arrest was caused by hyperkalemia or hypermagnesemia. Just to sum up why we did this trial, calcium acts as inotropic and vasopressor agent when given intravenously. Two small trials back from 1985 including U.S. or American patients with out-of-hospital cardiac arrest indicated that calcium could have a beneficial effect as far as return of spontaneous circulation at hospital arrival goes. There is a high and increasing prevalence of calcium administration. At least we know this is the trend for in-hospital cardiac arrest using the U.S. numbers, but we do not know the numbers for out-of-hospital cardiac arrest. Therefore, we did the Calcium for out-of-Hospital Cardiac Arrest Trial, also called the COCA Trial. It was an investigator-initiated trial. We randomized on a 1-to-1 ratio between the intervention and the placebo. We randomized in blocks of, random block of two, four, or six. It was placebo controlled, and it was double blind, and what we mean by double blind is that it was blinded for patients, clinicians, and the investigators ourselves. The Central Denmark region has an elaborate prehospital system. First of all, it is two tiered, meaning that we have 69 ambulances, and we also have 10 physician-manned vehicles for a population of 1.3 million inhabitants. Typically, out of cardiac arrest, both will arrive at some point. These are inclusion criteria. We included adult patients over 18 years with an out-of-hospital cardiac arrest, and the entry into the trial was that the patient should receive at least one dose of adrenaline. We excluded patients with traumatic cardiac arrest, meaning we also defined traumatic cardiac arrest with external asphyxia, drowning, and hanging. Another exclusion criteria was pregnancy or if you had a prior enrollment in the trial. And of course, if you received adrenaline before we could arrive with the trial drug at the patient site, that was also an exclusion criteria. If the physician deemed that there were any clinical indications for calcium that could be, for example, hypokalemia, then the physician can make the decision not to include in the trial based on this exclusion criteria. This is a flow chart of our trial, and I just want to mention again that we included both patients with shockable rhythms and patients with nonshockable rhythms. What was definitive and what was the inclusion criteria for entry into the trial was whether the patient received adrenaline or not. We gave the first dose of trial drug after the first adrenaline dose, and then we gave another dose of the trial drug after the second adrenaline dose, and the patient could receive a maximum of these two doses. If the patient received the intervention, it consisted of each dose consisted of 5 millimoles of calcium chloride given either intravenously or intraosseously. This corresponds to around 200 milligrams of calcium or 735 milligrams of calcium chloride dihydrate. We gave it as a rapid bolus, and it was a 10 milliliter bolus, and our placebo was saline. The trial could be double blind because calcium is clear in color, and it is odorless, which made it possible for the double blind and with the placebo. Our primary outcome was return of spontaneous circulation, and we defined this as sustained return of spontaneous circulation, meaning that the patient had ROSC for at least 20 minutes without a need for chest compressions. Our key secondary outcomes were survival at 30 days and also a favorable neurological outcome at 30 days. This was defined by the modified ranking scale. We had additional outcomes, and these were quality of life, and we also will present adverse events. We had time points of 30 days but also 90 days, six months, and one year, and we were able to follow these patients through telephone interviews and access to their medical records. Our sample size was 672 patients. This was based on the [inaudible] for difference in proportions and an assumption of 18% event rate in the placebo group and 27% in the intervention group. This corresponds to a risk ratio of 1.5 and a risk difference of 9%, and we used an alpha of 5% and a power of 80. The analyses included all patients that met the inclusion criteria and none of the exclusion criteria. In the double-blind trial, this approach is unbiased, and it increases precision. Of the binary outcomes, we calculated risk ratios and risk differences, but we only calculated P values for our primary and secondary outcomes. All these analyses were prespecified in the protocol. I'll now hand over the microphone to Lars, who will present the trial results.

[00:12:58] Thank you very much, Mikael, and now for the results. The trial started in January 2020. We had a preplanned meeting with the independent data monitoring committee in early April. They reviewed data from 383 patients. Shortly thereafter, they had their meeting and returned their report, which recommended that we stop the trial. The trial was stopped on April 16, 2021, and at that time we had 391 patients who could be analyzed. So, in total, the trial ran from January 20, 2020, to April 16, 2021. In this time period, we screened approximately 1,200 patients with out-of-hospital cardiac arrest. A few were excluded because they were children, and then a larger number were excluded because they did not receive epinephrine. This could be either because they had return of spontaneous circulation prior to receipt of epinephrine, or because resuscitation was terminated prior to receiving epinephrine. There were a few patients who met some of our exclusion criteria, primarily people with traumatic crowd arrest, a few that received epinephrine or adrenaline from a nonparticipating unit, and then, notably, only one patient who was excluded because the physician felt there was a clinical indication for calcium. A number of patients were excluded for other more logistical reasons. Some had return of spontaneous circulation or early termination of resuscitation after adrenaline but before the trial drug could be administered. Then, there were some where the clinical team forgot or made a decision that this patient should not be included. This did include some patients where the patient was suspected to have COVID-19, and it therefore was not possible to include the patient. We ended up randomizing and giving the drug to almost 400 patients. A few were excluded because they were traumatic arrest and should not have been included in the first place, leaving a total of 391 patients who were included in the trial and in the analysis. These are the baseline characteristics of the patient population. They are very similar to other out-of-hospital cardiac arrest trials. The average age was 68 years. About 70% were male. Most of the cardiac arrests happened at home, and most had a nonshockable rhythm. The time from the recognition of the cardiac arrest to the first dose of epinephrine or adrenaline was 17 minutes. And right after that, the trial drug was administered, giving a time from recognition of the cardiac arrest to the trial drug of 18 minutes. Most of the patients received the drug through a intraosseous route, and most of those were tibial. There was no loss to follow for any of the outcomes. In this figure, we see ionized calcium measurements in the patients. On the x axis, we have the time from the cardiac arrest, and on the y axis, we have the calcium values in millimoles per liter. Measurement of calcium was not part of the protocol, but those patients who had return of spontaneous circulation, and that's those patients included in this figure, had measurements done as part of their clinical care in the ICU. We therefore plotted each individual calcium value in this figure, and then you can see the reds are the calcium group, and the blue is the placebo group, and the solid line is a regression line with the dotted or dashed line the confidence intervals. As you can see, there was a quite large separation between the two groups. In fact, at the first ionized calcium [inaudible] was approximately 1.4 in the calcium group and 1.2 in the placebo group, and this difference was highly significant and had a very narrow confidence interval, illustrating that the intervention actually increased the calcium values in the blood. And now, to the primary outcomes. Return of spontaneous circulation occurred in 19% of the patients in the calcium group and in 27% in the placebo group, indicating worse outcomes in the calcium group. This difference corresponded to a risk ratio of 0.72 and a risk difference of almost 8. As you can see, the confidence intervals included 1 for the risk ratio, and 0 for the risk difference, and the P value was nonsignificant, being 0.09. The results were very similar for our 30-day outcomes at survival and favorable neurological outcome, which was defined as modified ranking scale of 0 to 3. For survival, approximately 5% survived in the calcium group, and 9% survived in the placebo group. This corresponded again to a risk ratio of 0.57 and a risk difference of approximately -4, again, being nonsignificant. Results were very similar for favorable neurological outcome. When we look at outcomes at 90 days, they were identical for survival. No one died in between these two time points, and they were very similar for favorable neurological outcome, again indicating harm in the calcium group but with wide confidence intervals. Here, we illustrate the Kaplan Meier curve for survival out to 90 days, and of course, as Mikael mentioned, we are also collecting data at six months and one year, but that is still ongoing. We also collected data on quality of life. We did this with the EQ-5D-5L. This is a very commonly used score ranging from 0 to 100, 100 indicating the best quality of life. There are different ways of measuring it. One is the score, as indicated here, which is directly assessed by the patient, and then an indexed value as well. As you can see, the results were similar for the two scores and similar for 90 days and 30 days with higher values in the placebo group compared to the calcium group, indicating a better quality of life than the placebo group. Again, here the confidence intervals were wide and mostly include no difference. We also looked at adverse events, and these were only assessed in patients who had return of spontaneous circulation, and below, we also showed them for those who survived at least 24 hours. The results are very similar for these two groups. Not unsurprisingly, there was much more hypercalcemia in the calcium group. This was expected, and only one patient in the placebo group had an elevated calcium value at arrival to the hospital. Other than that, there were no clear differences in side effects although we did see a numerically higher number of patients with dialysis in the calcium group compared to the placebo group. There was no clear difference in tachyarrhythmias, in ulcers, or in acute pancreatitis. All things that have been at least correlated with high calcium values in the past. We conducted a number of predefined subgroup analysis. You can see the subgroups on the left. It was initial rhythm, time to trial drug, route of administration being either intravenous or intraosseous, whether the cardiac arrest was witnessed, and whether there was bystander CPR. We present the results both on the relative scale, the risk ratio, and the absolute scale and the risk difference. As you can see, according to these subgroups, there was no clear subgroup difference. We also conducted a number of predefined Bayesian analyses to help better interpret the results. The figures illustrate the distribution, the post-theory of distribution based on a noninformative prior. So, we have a noninformative prior on our data, and then we get a post-theory of distribution, and that's what's illustrated in the figure, for return of spontaneous circulation, survival of 30 days, and a favorable neurological outcome at 30 days. And these, of course, looks very similar to the point estimates and confidence intervals from the frequentist approach. What's perhaps more interesting is the probabilities that we have calculated below, and that's one of the nice features of Bayesian analysis, that is you calculate direct probabilities of a given effect. So, if we start with return of spontaneous circulation, we calculate the probability that the risk ratio was above 1, meaning any positive effect of calcium, and that was 4%. Then we calculated what the risk ratio was above 1.2. So, a 20% relative increase in ROSC, and that was effective at 0%, at least much smaller than 1%. And of course, the probability that the effect was similar to or larger than our estimated effect, which was 1.5, is much lower. So, even closer to 0%. When we look at survival at 30 days and favorable neurological outcomes, the results are fairly similar with a relatively low probability of any benefit and an even lower probability of a meaningful or larger benefit corresponding to a risk ratio of 1.2. So, to sum up the results, there was no statistically significant difference in either of the outcomes including return of spontaneous circulation, 30-day survival, or neurological outcome at 30 days. The results were similar at 90 days and for quality of life. However, we found a very low probability of a beneficial effect of calcium during out-of-hospital cardiac arrest and an even lower probability of a beneficial effect that would be considered perhaps clinically significant such as an effect of 20%. So, we were, of course, disappointed by these results. We had hoped and hypothesized that calcium would beneficial. We were happy to see that Rob had conducted another poll of the Twitter square finding that about 20% of the viewers also had hoped or thought that calcium would be beneficial and that only a small percentage of the people taking this poll actually thought it would be harmful. Although our hypothesis was that calcium would be beneficial, we found point estimate that suggested that it could be harmful or at least had no effect. There are some possible explanatory models for why calcium could be harmful. These are, of course, extremely speculative and could not be answered by the data in our trial. One of them could be that intracellular calcium overload can lead to a condition with cardiac hypercontraction, which has sometimes also been called the stone heart theory. This has mostly been shown in animal models. Although calcium is a very simple molecule that is used for many things, it actually has a very complex role in the human body. It is involved in multiple things including enzymes, coagulation cascades and multiple other things. So, it's clearly possible that there are many other potential consequences of a high-calcium during cardiac arrest, and many of these we probably do not understand at this time point. Of course, calcium is not recommended for routine use in cardiac arrest, and some might ask, but what about for the special etiologies such as primarily hyperkalemia? Unfortunately, that's not something we can address in this trial. It was not designed to include patients with hypokalemia. Such a trial would be extremely difficult to conduct, and I doubt it ever will. However, when we looked at our numbers, only three patients actually had high potassium levels at arrival to the hospital. Of course, this is a select group of patients who had return of spontaneous circulation, but only three out of 90 of those patients had very high volumes of potassium above 6.5. So, perhaps this is not such a common condition in out-of-hospital cardiac arrest. Although we were disappointed about the results and had hoped for a positive effect of calcium in this condition, we do think that trial results are still very important, especially given these findings from the in-hospital cardiac arrest setting in the United States where we saw, as Mikael showed, that there has been an increase in calcium used during cardiac arrest and that it's actually quite common with approximately a third of all patients with in-hospital cardiac arrest receiving calcium. This corresponds to almost 90,000 patients each year in the United States receiving calcium during out-of-hospital cardiac arrest. And if calcium is truly harmful or at least not beneficial, we think this is an important finding. There are some strengths of this trial. It was double blind. The time to trial drug was 18 minutes, and that might seem like a long time, but it's actually very favorable or similar to other trials or in fact may be faster than most other trials with a similar setup. There are a very few protocol deviations, and there was very little use of calcium outside the protocol. Only a few patients received calcium outside the protocol. We saw that the intervention resulted in a clinically relevant increase in calcium, which is the intended mechanism. We include return of spontaneous circulation as the primary outcome but do also include a number of the patient-relevant outcomes including favorable neurological outcome and quality of life, and we have no loss to followup. Of course, there are some important limitations. First of all, the most important, the trial was stopped early. When we got the report from the independent data monitoring committee, we did not feel it was [inaudible] justified to continue the trial, and we stopped the trial, which was also the recommendation by the independent data-monitoring committee. Of course, when a trial is stopped early, based on the results either being very positive or very negative, the trial do tend to overestimate the effect. So, of course, our results should be interpreted in that context. We only tested one dose and only one timing of calcium, and of course, we don't know if these results are generalizable to the in-hospital cardiac arrest setting. There are multiple differences between in-hospital and out-of-hospital cardiac arrest, maybe the most important being the timing of advanced life support with trial drugs often administered much earlier, and perhaps also other etiologies in-hospital cardiac arrest. I want to once again thank all the authors on this paper. I also want to thank some international experts, Clifton Calloway, Marcus [inaudible] , and Michael Bonino [phonetic] , who helped us in the early phases of trial design, and then, of course, the independent data monitoring committee, [inaudible] Jerry Nolan and Theresa Olasveengen. Lastly, I want to thank al the clinicians who took part of this trial, hundreds of clinicians including patients day and night for more than a year, and of course, a trial like this could not have happened without them. I want to say a bit thank you to Rob MacSweeney in Critical Case Reviews for hosting this presentation and allowing us to present the trial for such a great audience. Lastly, we are happy to say that these results are now published in JAMA. Thank you.

[00:29:20] Lars, Mikael, congratulations. A huge achievement. Very, very well done and my apologies for the tapping. I had a mute failure on my mute handset unfortunately. First time that's happened, first time for everything, but very, very well done. A fascinating, fascinating trial and fascinating set of result for our panel to get into in a little bit. We will cross now to Gavin Perkins in England for his editorial. Gavin, over to you.

[00:29:51] Great. Thank you. Hopefully, you can now see the slides. Apologies for the moments delay. Firstly, thank you, Rob and the Critical Care Reviews team for the opportunity to provide this editorial and also to Lars, Mikael, and co-investigators, you know, for the opportunity to provide a editorial on such a, you know, a fascinating and important piece of work. In terms of kicking things off, I'd just like to share with you briefly disclosures in relation to conflict of interest. I have no relevant commercial conflicts, but I do have a number of academic or intellectual conflicts as outlined on the slide there. I don't think that any of them are necessarily going to have a significant impact on the views that I share with you over the course of the next 10 minutes or so. So, what I thought I'd do is I thought I'd look at these four things, that I'd do a little bit around setting the scene and looking at some of the historical aspects of calcium use in cardiac arrest. Before summarizing it, again, the key study findings, and perhaps exploring my insights into the study and the potential implications for clinical practice. And of course, we'll also have the opportunity of seeing the paper beforehand. I didn't have the opportunity of seeing the narratives from Lars and Mikael. So, whilst there'll be some overlap on this particular slide, I make no apologies for emphasizing how many hundreds of thousands of people around the world each year sustain an in- and out-of-hospital cardiac arrest, and this is work from ILCOR and international registries that summarizes the survival rates from out-of-hospital cardiac arrest. And you can see, it really is incredibly dismal. There is so much scope for us to look for and search out effective treatments to try to improve status quo for people that sustain cardiac arrest. This slide emphasizes or puts a perspective on the chain of survival, and I think it's an important perspective because whenever you're considering public health interventions for cardiac arrest, thinking about the front end of the chain of survival and the things that the community can do, really provide the greatest scope for improving patient outcomes. But as we saw in the Paramedic2 trial, whilst these proximal parts of the chain of survival are incredibly effective, despite that, a significant number, 70 to 80% of people despite early access, bystander CPR, and systems with public access defibrillation don't achieve a return of spontaneous circulation. The addition of adrenaline in the Paramedic2 trial, it required 112 people to be treated in order to increase the survival at 30 days. So, we are desperate to find effective treatments to try to improve outcomes from cardiac arrest, and I think it's really important to think of that context when thinking, you know, why conduct, you know, research in relation to calcium. The reason is, you know, we're desperate to find effective treatments. In preparing for the editorial, I was interested to go back and look historically about how calcium ended up in resuscitation guidelines and actually working all the way back at least as far as I could find the first evidence for a potential role from calcium came from experiments from Ringer back in 1883, where he actually published a correction on his work where he'd been bathing cardiac muscle cells in what he thought was simply normal saline but subsequently discovered that it was contaminated by inorganic salts, and when looking at the composition of that contamination noticed high levels of calcium within that. And he went on in that work to actually pick out all the different components from the water and to conclude that calcium chloride solution added to saline, sodium bicarbonate, and potassium after the ventricle has lost contractility restores good spontaneous beats, which will continue for a long time. He then goes on to say that whilst calcium salts are necessary for the proper contraction of the heart, if unantagonized by potassium salts, the beats will become so broad, diastolic dilatation would arise, and the beats would fuse together, and the ventricle would be thrown into a state of tetanus, the so-called stone heart that Lars so eloquently referred to in his presentation. So, we had some really very basic laboratory data suggesting that calcium may have a role in improving cardiac contractility. There were then a series of animal experiments that I think I would challenge you to say actually provide evidence of efficacy. The first study there had not control arm. The second study, the effects appear similar between intervention and control, and the third arm, it was noted to be inferior to epinephrine. I think one of the pivotal things that led to the inclusion of calcium was an anecdotal case report of four pediatric cardiac arrests that occurred on the operating table where resuscitation occurred as a consequence of the intracardiac administration of calcium. So, we saw back in the early 1980s the include of calcium chloride in the American Heart Association's guidelines published in the Journal of the American Medical Association. It was listed as one of the top essential drugs for resuscitation, and in those first guidelines recommending its use in electromechanical dissociation 2 to 5 mL of 10% calcium. This was reinforced in the 1980s guideline where it was extended to also include asystole, and then, as Lars and Mikael have already referred to, there were two randomized control trials conducted nearly 30 years ago looking at the effect of calcium in refractory asystole, finding no significant evidence of benefit that the point estimate favoring benefit then similarly in electromechanical dissociation that they noted that the rate of return of spontaneous circulation was high. But again, importantly in this study, no significant difference on the more clinically focused outcomes of survival to discharge. And I guess those negative studies combined with the literature in the time and the critical care lecture led in the mid-1980s, so the American Heart Association going from this being a treatment that they advocated to one that they suggested was withdrawn and not used outside of the context of electrolyte disturbances or calcium channel blocker overdose. And so, it comes as a surprise, really, to then note, and these are the dates again already presented, highlighting a significant number of in-hospital cardiac arrests continue in the U.S. context to receive calcium. And I think taking all that background together, it was reasonable to move forward with the trial calcium for out-of-hospital cardiac arrest and Lars and Mikael have just walked us through the results, no significant difference in the rate of return of spontaneous circulation but a signal potentially towards harm for rate of return of spontaneous circulation. And a similar pattern observed on the other outcomes, survival to 30 days, and survival with a favorable neurological outcome. In terms of thinking about the trial, I've kind of approached it by first thinking about the methodological quality of the trial, and I've used the Cochrane Risk of Bias tool to work our way through that. And it's really to take my hat off to the investigators for conducting a high-quality trial that across of the measures, I think, was assessed at a low risk of bias. How generalizable are the results? I've pulled together the treatment arms, so the interventions arms from the COCO trial, Paramedic2, which was intravenous adrenaline, and the CAAM trial, which was the French airwaves trial, which again is another later intervention in cardiac arrest. And I think you'll probably agree that on the whole the studies, you know, the demographics of participants enrolled into the studies are broadly similar between the different groups. There are some small nuances and difference, but I think it provides some assurance that this is a group like the patients that I treat in my out-of-hospital setting. So, we come onto what I think will be the interesting parts of the discussion both to date but going forwards, is the decision to terminate the trial early. The trial protocol sets out very clearly that there were no predefined stopping rules for the trial, either for efficacy or for futility, and that any decision in relation to safety was left in the hands of the independent data monitoring committee. And it's in that context that I think it's interesting to look at the data as it was presented to the data monitoring committee, and this information is all very clear in the electronic supplementary material. But one can see that at the time the data were presented to the data monitoring committee, if we focus in on the risk difference, it was almost minus 10% with a confidence interval not crossing zero. So there was evidence of a reduced rate of return of spontaneous circulation. Now, as is often the case in between an interim result and then full studies results coming in, there were some small changes in patient numbers, which actually, at least in terms of the P value and the point estimate tipped the study findings from being one that I guess people would regard as a statistically significant result to one where the confidence interval crosses zero. And I think it will be interesting in the group discussion for us to reflect on that. I think Lars, again, has already highlighted in his discussion some of the limitations that there are risks associated with stopping a study early, and these have been broadly classified into two elements, firstly, stopping only tends to overestimate treatment effects, and then, it also has a freezing effect on future research, because if it shows an intervention is either beneficial or harmful, that it's, you know, potentially inhibits or inhibits further research in that field. And I'm just going to use this bottom component of a slide to remind me of something Keith Wheatley, a clinical trialist from Birmingham taught me a number of years ago in terms of interim analysis and data monitoring. So, this was a large cancer trial that the investigators were appointed as members of the independent data monitoring committee. It was comparing five courses of toxic chemo versus four courses of toxic chemo, looking at the impact on long-term survival. And when they did their first interim analysis after 100 patients, you can see the P value is 0.05 and the confidence interval didn't cross 0. The data monitoring committee went back a second time. At this point, the P value had reduced to 0.003, and although the point estimate was very similar, the confidence interval was now shifted more to the left, indicating that the longer course in the therapy was potentially beneficial. What impeded the investigators from stopping the study at this point was quite how dramatic that effect was, and they actually did another interim analysis, as you can see here. Not a lot of change but continued to monitor the effects over time, and it's going to have to come out of there, because to have missed the key component of the slide through the reveal would somewhat undermine the story. But I think you can see hopefully now that in the course of this study that that extreme initial effect as more patients were involved. So, it came back to indicate that there was actually no effect between four and five courses of chemotherapy. And I think, you know, in this context, this would potentially be exposing large numbers of patients to addition rounds of chemotherapy where the conclusion of the trial was that there was no overall effect. But of course, the context for this study is a study in patients who lack capacity. It's not possible to consult them in relation to the risks or benefits, and we must be grounded in the premise of first do no harm, and the rights of the patients must always exceed the rights of society or the benefits of the research. I think as we go into the discussion, it will be really interesting to see how the data monitoring committee or representatives of that, you know, grappled with those challenges. Implications for practice. I think I know the implications for my practice, and that's I didn't use calcium before, and I'm not going to use calcium afterwards. What I think will be fascinating is what the insights are from people that have been strong advocates of calcium, and I think one of the advantages of ILCOR, I've declared my conflict that I cochair ILCOR, is that multidisciplinary, brining the research evidence together to develop consensus on the science and treatment recommendations that consider the research findings in the context of the values and preferences of the various societies and patients that we serve. So, to close, many congratulations to the research team and my thanks to the patients, the families, the clinicians, and the researchers that have produced this really important piece of research for us. Many thanks.

[00:45:02] Gavin, thank you very much, fantastic editorial. An awful lot to consider there. We'll go back over to Denmark now and to Mikael and Lars. I'm sure they'll have some comments to respond to Gavin's editorial there.

[00:45:26] Tyvm, Rob, and thanks Gavin for an excellent editorial. First of all, very interesting history of calcium. Of course, we have looked into this as well, and it's a very sort of back and forth thing where certainly it's great, then it's out, then it's back in, and as you know, it's also used for a number of other indications. We use it often in the ICU and the OR. Also for its inotropic effects. So, I think there's still a lot to learn about calcium. Coming back, of course, to the early stopping of the trial. As I mentioned, this is, of course, a limitation, and it's quite possible that we have overestimated the effect. At least there is an increased risk of that. This was not an, I guess it was an easy decision in one way because we thought the results indicated that it was very unlikely that we would ever reach the intended effect. So, the sample size calculation was based on a rather large effect on return of spontaneous circulation, a 50% relative increase. We chose that large effect because we thought it would be unrealistic that a smaller effect on return of spontaneous circulation would ultimately end up in an improvement in survival. As we saw in the Paramedic2 trial, there was a very large increase in return of spontaneous circulation, about 300%, which corresponded to only a 30% relative increase in survival. So, it was sort of on purpose that we have chosen a rather large effect size. So, I agree with you that it's quite possible that we have overestimated the effect, but I still think it's really unlikely that at least this trial would have ever reached the intended effect size that we had planned for. And therefore, we thought it was a relatively easy decision. The steering committee all agreed, and we agreed with the independent data monitoring committee that this would be the best approach. Of course, had we had the same results in the opposite direction, signaling, a strong signal some might say for benefit, we would never have stopped the trial, and I don't think the independent data monitoring committee would have suggested us to stop the trial either. And that's how sometimes evidence is not equal or balanced. As you said, we are including patients, without their consent, and we are introducing a treatment that is currently not recommended. So, that combined with the results we had made it quite clear that we had to to stop the trial. As you showed, the results changed slightly based on the interim analysis at almost 400 patients, and then the final result, and this, of course, because we ended up including about 10 patients more, and that just shifted the results a little bit. The point estimates were very similar, but the statistical significance had changed, both for neurological outcome and for return of spontaneous circulation. And one last comment about the early stopping. Of course, it was based on the primary outcome, but had that been the only signal in the data, we would probably have continued. But, it was very clear that all the outcomes were in the wrong direction. It was true for survival. It was true for favorable neurological outcome. And even in those who had return of spontaneous circulation and survived, they actually had lower quality of life. So, there were so many different outcomes that sort of pointed in the same direction that we actually quite quickly and quite decisively stopped the trial.

[00:49:08] Lars, thanks very much. Great explanation, and I'm sure we will get into that in a little bit more detail with Victoria and Theresa shortly enough. We're going to turn to my colleague, Chris Nutt now, who is going to join us with questions from the chat function and from Twitter. Chris, over to you.

[00:49:27] Thanks Rob and thanks Lars and Mikael for a really great presentation. Twitter has been interesting when the hashtag is shared with a very well-known soft drink, so it's been interesting to moderate that. The general feeling is that calcium is gone and shouldn't be used in cardiac arrest. And we've already sort of, or you've already sort of talked about it a bit, but they have mentioned about the decision to stop, a few people have mentioned there was no stopping criteria. And if you look into the data, there were some, there was a fairly low event, and there were differences in terms of nonshockable, more nonshockable rhythms in the calcium group and less bystander CPR. Is that correct?

[00:50:30] I do think that if there were any differences they were very small. I can't remember the exact number in my head. I have it here in front of me. There were some slight imbalances, although those were minor. We did conduct an adjusted analysis where we did a regression model including of the strong prognostic factors including age and bystander CPR that did attenuate the results a bit towards the null, towards no difference between the groups, but it was still a signal towards harm. But again, of course nonsignificant results. If I can just comment on the stopping criteria, it's true. We have no predefined stopping criteria for either benefit futility or harm, and that was not an oversight. That was a decision we made up front for multiple reasons. First of all, we didn't want to stop for benefit, because we thought it was important that we could also look at secondary outcomes and adverse events, subgroups, and so on. We didn't want to stop for futility, because it was only a smaller trial. We felt it was important to include the entire group. We also had not predefined any criteria for harm, and I think that's a complex question, whether you should do that or not in these smaller trials. Again, we could have defined something that was based on the primary outcome, but I think that's not the whole story. The primary outcome was only relevant to us because it was related to other more patient-centered outcome such as survival and favorable neurological outcome. So, I thought personally it would have been unfortunate to stop the trial had we only seen a difference in return of spontaneous circulation but not in the other outcomes. So, we decided on a different approach. We said, let's find some really smart people who know something about cardiac arrest, and let's put them together, and that includes Theresa but also two other people who are heavily involved in cardiac arrest research and are also clinicians as well. So, we thought they were better suited to make that complex decision based on the data that was in front of us instead of speculating in hundreds of different scenarios upfront. And I know that's a debatable point of view but that was the approach we decided to do.

[00:52:51] Sure. And so, there are those who have sort of already banned calcium. There's a few people who sort wanted to know about how you selected the dose and given that the calcium group had more hypercalcemia, do you think that investigation of different doses perhaps or lower dose would be useful in the future?

[00:53:20] Yeah, that's always a good question. When you test the medication, the first thing people say is what about a lower or higher dose. I think people have said the same with Gavin's trial, Paramedic2 trial and epinephrine. And we decided on this dose because it's the standard dose in most places in Europe. At least in Denmark it's the dose that is mentioned in the guidelines, if you should use calcium, this is the dose that's recommended in the European guidelines. I know the dose is sometimes different in the U.S. It's actually larger in the U.S. most times and sometimes more than two doses are given. We can only speculate what a different dose would do. I don't know. We decided on this from a pragmatic point of view, but we also decided to give the drug fairly early in the resuscitation. Some might argue that it should only be used for refractory cardiac arrest or long cardiac arrest, but we felt we would, one, never be able to include enough patients because if we had wait until that was 20 minutes into the cardiac arrest, and we would also have a very, very low event rate, making it very difficult to say anything. So, that was also a somewhat pragmatic choice that if we wanted to see an effect, we should probably give it early.

[00:54:30] And on that is there a difference in the timing of the dose if you have a shockable rhythm versus a nonshockable rhythm? Is there more of a delay in the shockable site because of CPR and shocks before the administration of adrenaline?

[00:54:51] Yes, there is a difference. We followed the ALS algorithm recommended by international guidelines. So, the patients were generally given three shocks that did not successfully result in a nonshockable rhythm or a pulse-bearing rhythm. So, and the nonshockable rhythms got adrenaline early in their resuscitation. As I mentioned, adrenaline was the entry into the trial. That was an inclusion criteria, and according to that algorithm, the shockable rhythms will have received both adrenaline and the trial drug later than the nonshockable rhythms.

[00:55:34] And, of course, here the guidelines in Europe is slightly different. We give the adrenaline after the third shock, where as I believe in the U.S. it's after the second, but I don't think that makes a huge difference.

[00:55:48] And there were a few comments on how you give the calcium. So, first of all, the speed of administration. So, there are some suggestions that calcium, when administered rapidly could cause vasodilatation as opposed to their inotropic and vasopressor effect that you're looking for. Would giving the drugs more slowly as per the sodium nitrite trial, do you think that would be something that should be looked at, or do you think that it's a mute point?

[00:56:33] No, I actually think it's a very good point. I think when we give calcium in other conditions in the ICU and in the OR we often give it fairly slowly. I'm not aware of the literature suggesting it should give vasodilatation, but there has been some suggestion that it could be proarrhythmic and give tachyarrhythmias. We thought that was potentially less important during a cardiac arrest and therefore decided again on a fairly pragmatic solution where we decided to give it rapidly, because we think that is how it's actually administered during a cardiac arrest in most settings. And although it might say to give it slowly, I think once you're in that cardiac arrest, once you're giving the drug, you will often give drugs fairly quickly. So, we just follow routine clinical practice for how it would be administered if it was to be administered in the setting of cardiac arrest. And of course, our trial did not test a slow administration, so we can only speculate on whether that would results in any different effect.

[00:57:30] And we had a comment from Jerry Nolan about did you see any difference between the IO and the IV administration of calcium, I think, really was looking at the ionized calcium levels?

[00:57:54] So, we did not look specifically at the ionized calcium values for IV versus IO. That is something we had planned to do. The calcium values were actually not a predefined analysis. We just presented them descriptively to sort of show the reader what was going on, did they actually make a difference. And of course, these are only patients who had return of spontaneous circulation. So, it's a select group. So, I think we should be careful in interpreting too much on those results. That said, we had a very high proportion of patients who received the drug in an IO fashion instead of IV, also higher than the Paramedic trial that Gavin just showed. And I think there has been a shift in Denmark where IO is now being used more and more also as a first-line route of access. As it comes to return of spontaneous circulation, survival, and favorable neurological outcome, we could not see any difference in the effect between those who received the drugs IV or IO.

[00:58:57] Thank you. And timing of calcium, you said, you administered it early. You certainly seem to have got it in earlier than in Gavin's trial. Is there any way that calcium, or any drug can be administered earlier than this with the thought that the earlier you administer the drug you may see more of an outcome benefit or an effect in cardiac arrest? In out-of-hospital cardiac arrest, I don't think it's possible to get it much lower. As I mentioned in my slide, where I was presenting Central Denmark region, as aside, we have quite a comprehensive prehospital system that is two-tiered, and we arrive at the patient's side in around seven to eight minutes, which is actually similar to Gavin's trial. If you wanted to test an hypothesis that the time to drug should have an effect, maybe you should test it in in-hospital cardiac arrest, but then, again, you have different etiologies and yeah.

[01:00:08] Yeah. I completely agree with Mikael. I think it would be very, very difficult to institute ALS and administer drugs earlier than what we did in this trial. Maybe you can cut off a few minutes in a very efficient system, but as I mentioned, our time to drug delivery is very similar or faster than other trials similar to this. The in-hospital setting, where we recently presented the [inaudible] trial, we showed that we could give drugs after eight minutes. So, it is possible to give drugs much earlier in an in-hospital cardiac arrest setting. And of course, you should never say never, right. I mean there could be innovative solutions where you could give drugs earlier. We have seen bystanders that can give shocks. Maybe they can also give drugs IM or something like that. So, I think there's always opportunity for looking at things in a different way. But I think with the current system, it would be very challenging.

[01:01:01] Thanks. Are we going to get that trial in-hospital cardiac arrest in calcium?

[01:01:06] I'm not going to do it. I don't think Mikael is either. I think, we had hoped the calcium would work, and our conclusion is very similar to Gavin's, that we're not going to use it, and I don't think, I don't think there's clinical [inaudible] for another trial. I'm happy if someone else wants to do it, but it won't be us.

[01:01:28] Super. Chris, thank you very much. I think what we better do is move onto the panel discussion. It's time to bring in Victoria, Theresa, and Chris. And Victoria, I'll maybe start with you. I know you had some comments perhaps around the early termination of the trial.

[01:01:46] I did. Thank you, Rob. I just wanted to say, first of all, well done to the trial team. I was thinking, this is such a brave trial to take on, and it looks, it's difficult, you know, and it's difficult to do it well, what you have done. I was impressed, you know, it's complicated when we have to navigate, you know, participants who can't consent, and then just performing the whole trial outside the context of that hospital environment. So, I just was, yeah, very impressed, and [inaudible] you taking this on. It's a very well-conducted trial. I was agreeing with Gavin on his low risk of bias assessment that he used. And I did actually have some statistical things to talk about as well, but maybe we'll save those for later. I wanted to talk about with regards to the early stopping, I was sort of stuck by the fact that, you know, the trial started off. You planned it for 430. It was a DMC review which then upped that sample size to 674. And then actually then there was a decision by the DMC also then to terminate at 397. So, that is an unusual combination of decisions from a DMC. And now, I thought it was interesting when you were saying there were no predefined statistical criteria to help you make a choice about whether to stop for harm or futility or efficacy. And I think if there had been one for futility, you may well have met it. And I was stuck in as well in your protocol that you said that even if it was met for futility, and actually there are important secondary questions to continue to get out, the secondary outcomes and also for subgroups. And then, I wasn't quite sure of the strength of the evidence for the stopping for harm, and I wasn't, so I, if it was a statistical formal rule that you'd use there, I'm just wondering whether you would have actually met it, because we look for very extreme results in order to be able to stop a trial. So, I appreciate sitting on the DMC. So, that's the statistical sort of analysis of it. But that's not taking the clinical context at all, and I think the DMC do face that really difficult decision, because you've got to bank the risk of the patients in the trial and also the risks then to future patients if the trial isn't definitive enough then to change practice. So, I think my question is, have we, you know, is there enough evidence now to change the practice to stop administering calcium, and have you answered your questions about subgroups? And in regard to certain things like the shockable [inaudible] and nonshockable rhythm.

[01:04:39] Yeah. Thank you very much for those comments. I'll try to start with the first one about the sample size, because that was a little confusing process. We tried to explain it in the manuscript. We started out with a sample size calculation of approximately 430 patients. Then we did, the steering committee did a blinded review, where we just looked at the event rate and found that the event rate was much lower in both groups combined. So, we didn't know which was in which group. Then we decided that it was not the monitoring committee that we should increase the sample size. We had originally planned for two interim analysis of 50 patients and 200 patients, but then we increased the sample size, and it was therefore natural to also have another interim analysis, which was also in the interest of the monitoring committee, who had of course seen a signal perhaps already after 200 patients. So, that was sort of the story of that. We had hoped, we had been concerned that we would be underpowered because the event rate was so low, that we increased the sample size. We never reached that sample size, of course, as you know now. And then to the stopping criteria, I think it's important, although the conclusion might be futility or no difference or no benefit perhaps. That was not why we stopped the trial. We stopped it because there was a signal of harm. And I also think that it's true that we normally only stop trials if there's a very strong signal, depending on the stopping criteria, of course, and the alpha spending and how we do it, but that's mostly in the context of benefit. I think most people would not have stopping criteria for harm that let's say would require a P value of 0.01 or 001 that sometimes used for benefit. At least that's not something I have seen come used for new interventions. Of course, it's different if it's comparing two equal interventions, for example, or two interventions that are used or recommended, both of them. So, I think it's actually difficult to have meaningful predefined criteria for stopping for harm while also being cognizant of the alpha spending, the family wise error rate and things like that. And again, this was, it was not an oversight by us. This was a conscious decision that we thought that that would be very challenging and therefore instead had a independent data monitoring committee. I agree when you look at it from a strictly frequentist null hypothesis testing perspective our results were not significant. But if you look it from the Bayesian analysis, especially looking at what would be the probability of these larger effect, more than 20% relative, more than 50% relative, it's almost, it would almost be impossible to find that in this trial. Of course, that doesn't mean that that's the truth, but in this trial, it would be extremely difficult to find those things. So, I don't necessarily think that the final conclusion from this trial should be calcium is definitely harmful. My conclusion is it's definitely not very beneficial.

[01:07:55] Yeah. A great point to make there, Lars. Theresa, you were on the data safety monitoring committee. Perhaps you'd like to comment on this aspect as well.

[01:08:07] Sure. As Lars explained, the trial was planned to stop at about 400 patients. We had had what we thought was our final interim analysis at about 200 patients, and we saw pretty much the same signal that we saw later on. So, not knowing, we were blinded as well, not knowing what was what. We were seeing a signal towards one group being better than the other but not knowing which one was which. So, then when the steering committee decided to increase the number of patients, we thought, well, then we thought well then we should also ask for an additional interim analysis at 400 patients to make sure that we're not putting patients at undue risk. And I think to the question is this evidence strong enough to stop using calcium, I think that's sort of, that gets at the core of, I think within the DMC. I don't think any of us were in systems where calcium is commonly used, so I don't think it was a big part of the discussion to need the evidence to stop a practice. It was more, you know, did we throw out the baby with the backwater, you know, decades ago when calcium was removed from the guidelines. Is this, you know, perhaps an old, new, you know, exciting strategy to improve outcome? But then, you know, with sort of on the grounds that this is not in the guidelines, it's not that commonly used, we felt it was important to not put patients at risk for this potential harmful intervention. And, I guess, concluding as Lars that the likelihood of this being beneficial is extremely low.

[01:10:08] And Chris, I think you were the handling editor at JAMA. So, when this paper came in, and there was this early stopping, how did you deal with that amongst the interim review at JAMA?

[01:10:20] Sure. So, again, thanks for having me and congratulations to the investigators, and that was a great presentation today. We're very fortunate to have this paper come across our desk at JAMA, and you're right, there was a really interesting discussion and a little bit of confusion at the beginning as to sort of what happened with all these steps. The investigators who have joined us today were very transparent about all of the sort of steps, and you'd seen in the supplement how they've shared the data that was looked at by the data monitoring committee. And that was the key for us, is that we could unpack the story, and although things may have been done or could have been done differently, there didn't seem to be any sort of fatal flaws or huge risk of bias in how the data was presented to us. So, this was all, of course, on the backdrop that aside from this question that we're debating, it was an incredibly rigorous trial. The double blind and the, you know, complete follow, the adherence in [inaudible] , this was a fantastic trial, and I think as some of the panels have mentioned, it's very difficult to do this in every hospital environment. Not to mention, to test a drug that's been, as we learned, known about for 140 years and itself hasn't been tested in this population in decades. So, all around, we thought this was going to be of great clinical interest and lead to a fantastic discussion, like we're having today.

[01:12:03] Yeah, we certainly are. Gavin, can I come to you? Do you have a genuine feel for what the actual mechanism of harm was? You've done all the, you look back in history and looked at an awful lot of papers. Do you get a feel, and I'll maybe move onto Lars and Mikael in a moment just to see if there's any mechanistic work ongoing, but why was there a reduced level or rate of return of spontaneous circulation and subsequently worse outcomes with calcium, or what is it that's causing the apparent harm/

[01:12:39] So, I think it's probably factorial. We're focusing in on the primary outcome, which was return of spontaneous circulation. Then, I think it's something acting at the heart or the circulatory system level as opposed to, you know, a brain injury or tissue ischemia or other levels, and I do wonder if that stone heart phenomenon described, you know, may actually be the reason that the lower ROSC rate was seen. But of course, the harm as it were was not solely limited to a reduction in ROSC. It was seen across, you know, the outcomes, why typically follow ROSC. But I think probably reasons that the primary outcome was the way that it was, I'd speculate that it was, you know, related to stone heart.

[01:13:31] And Lars and Mikael, do you have any mechanistic work are you looking at as to why this happened?

[01:13:39] Yeah. So, first of all, I agree with Gavin although of course this is purely speculative, because we were not able to measure anything in this trial that would sort of make us come closer to the truth on this question that would require quite advanced invasive monitoring to sort of measure this in any detail. I just want to mention there are also potential other mechanisms. About 20, 30 years ago, there was a big trial testing calcium blockers for patients with cardiac arrest after cardiac arrest because there was thought that this could help with ischemia reperfusion injury in the brain. That trial did not show any difference between the groups, but it's just pointing to the fact that this is very complex, and you could come up with theories in either direction if you want to, depending on what the [inaudible] textbook you sort of opened. We are looking into some additional studies, of course. One of the first we will look at and perhaps one of the more interesting is patients with pulseless electrical activity and the specific ECG characteristics they have at the time of the cardiac arrest and whether there might be some patients that could benefit for the treatment or where the drug is perhaps even more harmful or has a stronger signal. As Gavin showed in one of those early studies from the '80s, there was a very exploratory subgroup analysis looking at patients with wide QRS complexes or signs of ischemia where there was a larger effect, positive effect of calcium, and perhaps we could also try to identify ECG characteristics that are correlated or associated with hyperkalemia such as wide complexes. So, we're looking into that to see if there is any patient group that might benefit or whether the effect is different. Of course, it's extremely difficult to diagnose these things like hyperkalemia and other electrolyte disturbances in the prehospital setting. So, it's very difficult to sort of target that specific pollution.

[01:15:54] Victoria, this trial used a dual analytic approach using both the Bayesian and the frequentist framework. What do you think of that? Is that something trials should be doing more, or should we be sticking with one approach or the other approach?

[01:16:10] I really liked that. Yeah, it was great to see the secondary Bayesian analysis. And I, I mean you could argue why do you need a frequentist analysis, but people are much more comfortable with, so having a secondary and having that, you know, the way that it was presented as well really helps aid those results in the interpretation, being able to talk about the probability of, you know, superiority and realizing that was very, very low. So, whether we need the dual approach, I don't know. Maybe we need this hybrid approach as well while people get familiar and comfortable with it. But with Bayesian approaches, we do have to be careful because they can be quite, you know, we can end up presenting a lot of results, and that can be a little overwhelming, because we have to make different decisions about the priors or the probability, the thresholds we want to calculate the probabilities for. So, and I thought the team did that really nicely, that they presented the uninformative prior and, you know, were limited on the thresholds that they presented. Yeah, and I thought it was great, really good.

[01:17:21] And Theresa, for the data safety monitoring committee, as someone who is not particularly familiar with how that works, I presume you undertake your own analysis, so you may not have necessarily done a Bayesian analysis in addition to the more familiar frequentist analysis?

[01:17:41] No. We did not perform the analysis. Those were sort of provided to us by a statistician used by the study group.

[01:17:57] So, you would have had the Bayesian, did the interim analysis include the Bayesian data, do you know?

[01:18:04] No, it did not. It had just the standard frequentist statistics.

[01:18:10] Okay, so Lars, this just came at the end then after the decision was made to halt the trial, is that right?

[01:18:18] Yeah, that is correct. The interim, we handed over the data to the chair of the monitoring committee. He then analyzed the data with the help of a statistician. And they used the frequentist approach, based on a charter we had sort of created before. If I can mention just a few things about the Bayesian analysis. I'm a big fan of Bayesian analysis, but I think there was an important thing that the Bayesian analysis does not change the data. The data is the exact same. The evidence is the exact same, whether you analyze it one way or the other. So, it is just a tool of communicating and illustrating things better. It doesn't give you more evidence or give you more confidence in the results. And any early stopping that might be a concern to the frequentist approach and the alpha and the chance of false positive is the exact same for the Bayesian approach and the way we did it. So, I just want to, it's a supplement, but it's not better or worse in that sense, and maybe it's a better way of illustrating the data. The Bayesian analyses were preplanned, but they were actually not included in the first version of the manuscript we sent to JAMA because they were planned for a different manuscript. But I think rightfully, the reviewers and the editors said while this would be a very nice way of illustrating this data, and they were preplanned, as I said, so we decided to include them in this manuscript as well. So, no, they were sort of a little bit on the backburner compared to the frequentist approach.

[01:19:50] So, Chris, I'll bring you in on this. Is this a sign that JAMA are moving towards maybe a more Bayesian presentation of analysis, or is it just for this case, this study, this trial that matched very well.

[01:20:05] Yeah. I think it's a combination. I mean Victoria made a great point is that when presenting these results, there needs to be clarity and not confusion. And that's not always the case when we see these post-hoc Bayesian models. In this case, we asked the group for them. We knew that the investigative team was working on this and had done so in prior papers, and so, rather than split these up, lumping them together seemed to make a more compelling case, particularly given the issues with the early stopping and some of the topics we discussed before. So, I would say that, yeah, JAMA has been taking a number of Bayesian trials with responses after randomization in addition to these secondary analyses.

[01:20:53] Lars, I think you were going to come in there.

[01:20:55] Yeah, so, I wanted to comment on that because I completely agree that it can very quickly get overwhelming with these Bayesian analyses. I presented just very few of them here today, but there's much more in the manuscript, and of course, that's one of the reasons is because you have to define your prior, so you're your prior belief in what the effect would be. So, we used what's called a noninformative prior here, which means we basically come with no evidence or no prior belief about the effect of calcium. That's a simple way of doing it, but it's probably not the most effective or the smartest way. So, in the supplement, we have a number of other analyses using informative priors, both that said, well, I actually believe calcium works. I believe it doesn't work. I believe it's harmful. And then, you can then calculate the posterior probability. So, I think that's, if people are really interested in it, they should try to dig down in the supplement material and look at, from my perspective, as I might be an American clinician who used calcium for many years and actually believes that it works, maybe look at the graphs with a informative prior instead of the one we used. So, I think it can illustrate some different aspects there, but it also makes it quite complex to present the results.

[01:22:10] Which brings us nicely on to the topic of generalizability. Gavin, you presented some lovely data looking at the Paramedic2 population for relatively high-income countries with well-resourced prehospital services. Should this data be generalizable across most of those similar sort of countries and health care systems?

[01:22:38] So, I think in similar health care systems, which in reality are probably going to be the ones that would have the capability to deliver the technology, it is difficult to see how a system could be set up differently that would make these results not valid in the out-of-hospital setting. I think, you know, questions, you know, do remain in the in-hospital, you know, setting. Given these results though, I think you would be brave to initiate an in-hospital trial of calcium. But it remains an unanswered question.

[01:23:21] And Chris, I mean we saw the data earlier on about the 90,000 patients potentially per year in the U.S. being treated with calcium during in-hospital, I think was in-hospital cardiac arrest. Yes, this was out-of-hospital cardiac arrest trial, but what do the people using calcium for in-hospital cardiac arrests now do?

[01:23:45] That's a really interesting point about which I was thinking during the presentation and perhaps even my own practice. To me, it comes down to your probability as you arrive on the scene that this patient is having an arrest related to hyperkalemia and that sort of sliding scale of where we fall, and perhaps we should be having, you know, a higher suspicion of that before administering calcium than perhaps we had been using before.

[01:24:15] Lars, yes?

[01:24:18] Yeah, so, so, of course, I have also been thinking about the implications of these results and especially for those patients where you suspect hyperkalemia. I think the other electrolyte disturbances are more rare or at least not suspected as much. So, I really think it comes down to hyperkalemia, and Mikael and I have of course been talking about it and also trying to look at the evidence for calcium during hyperkalemic cardiac arrest, and there is none. There is no evidence that calcium improves outcomes if you have hyperkalemia and cardiac arrest. We couldn't even find good animal studies suggesting that calcium improves outcomes for cardiac arrest caused by hyperkalemia. I think it's extrapolation from some studies in patients not with cardiac arrest where the administration of calcium can improve some ECG characteristics if you have wide complexes and very high levels of potassium. So, I think it's a very interesting question. Of course, it's recommended in the guidelines. It's a given if you have hyperkalemia, but that is based on expert opinion, anecdotes, case reports, and not hard evidence. So, for me, I'm even questioning whether I should use it if the patient has hyperkalemia or not. Theresa, you were going to come in on this?

[01:25:32] Yeah, no. I just had a comment more as a cardiac arrest researcher than as a member of the DMC, but I think this is another example of sort of doing large trials without having a lot of animal data. And I think perhaps this is a field where we would really benefit from having studied things better in the lab. I think, you know, trying to understand the mechanisms and what might be happening with large doses of calcium to the heart, you know, in various settings, to me that would be something perhaps interesting to explore in a experimental laboratory setting.

[01:26:12] Yes, Lars?

[01:26:14] Yeah, I'm very glad Theresa is saying that because I think that is the only way we can really answer some of these questions specifically for patients with hyperkalemia, for example. I really doubt that anyone will ever be able to pull off a trial including 1000 patients with confirmed hyperkalemia and then randomizing them to calcium placebo. So, although I'm definitely a proponent of doing human trials or think we should do human trials. I think this might be something where it would make sense to do some animal work, and we are actually planning that, to look at hyperkalemia cardiac arrest and then some of the various medications for that, including calcium, but also bicarbonate, for example, because I think it would be very difficult to do in humans.

[01:26:59] Theresa, there is a mindset that once somebody is in cardiac arrest then they can't any worse and these sort of therapies, these unproven therapies, can be reached for, it's justified, although we've seen in this trial that arguably that isn't the case. They can do even worse because some of those will recover. Is this, does this trial put an end to that sort of mindset, that unproven therapies can be used, can be justified in cardiac arrest?

[01:27:30] Sorry, I think I had an internet issue. Can you ear me?

[01:27:33] Oh, yes, can you hear me now?

[01:27:35] Yeah, I'm sorry, I missed your question.

[01:27:37] No problems. So, the question I was making or the point I was raising was that in cardiac arrest there's sometimes a mindset that for clinicians the patient can't get any worse, and therefore, unproven therapies are justified, such as calcium, but we've seen in this trial that it's probably not the case. They probably can do worse, the rate of ROSC was lower numerically at least in those who receive calcium. Does this sort of trial put an end to that sort of mindset that we can't do harm if they're already in cardiac arrest?

[01:28:09] I think that's an important mindset to get rid of because I think there are things that we know help during cardiac arrest. I mean we know that it's important with high quality CPR. We know that the VFs need to be defibrillated, and I think introducing new sort of complex therapies always carry the risk of taking our focus away from what we know works. So, I think it's important to get rid of that notion.

[01:28:37] And Victoria, was there anything else within the methods that you noticed. You mentioned at the start that you wanted to come back to a couple of points.

[01:28:48] One was just about the Bayesian analysis, and I just, I thought that was great to [inaudible] as a secondary analysis for interpretation. And actually, there's one other thing I found a little bit surprising was that the primary analysis was an unadjusted analysis, and given that you, and that uses adjusted analysis for secondary and whether actually the results varied at all with your adjusted analysis, because it's our preferred is that you would adjust the randomization stratification [inaudible] was there any strong prognostic factors you had, because that helps us, you know, have a more precise treatment effect. And yeah, so.

[01:29:25] Yeah, I think that's an excellent question, and it's definitely a choice that can be debated. We only stratified according to the station, so where the ambulance, the physician and ambulance were stationed. So, that was, the stratification was more a logistical thing than it was like a strong prognostic factor. Of course, it's sometimes some people recommend adjusting for prognostic factors to increase [inaudible] , and that's what we did with the sensitivity analysis, we did, where we picked some strong prognostic factors. The reason we didn't do it for the primary analysis is actually a very pragmatic thing. It is that it's very difficult to estimate risk differences in an adjust model when the outcome is rare. And the problem is that you have some assumptions that you would use, for example, in a linear model, and that doesn't work very well when there's few events and the outcome is rare. So, the gain in precision we would have from adjusting we would probably lose based on the model specifications. And we thought it was important to present results both as risk ratios and as risk differences, and of course, had we presented them as [inaudible] ratios, it would have been easier just conducting a [inaudible] and then adjusting for the variables. But it was a little bit more challenging, especially for the rare outcomes such as survival. So, a somewhat pragmatic solution to that where we then did a sensitivity analysis, but only for the risk ratios, if you noted that, and not for the risk difference, because it's difficult to estimate correctly.

[01:31:04] Mikael, you've done the trial, a fascinating trial, congratulations on the publication tonight in JAMA. Can I ask, what would you do differently now if you were to do this again, now that you're at the end of the journey or near the end of the journey, what would you do again if you were to set design this trial up from the start and execute it again.

[01:31:25] Thank you so much, Rob. I'm very fortunate to be a part of Lars [inaudible] cardiac arrest group, and I just feel privileged to be able to present this trial here and be able to present and then publish the trial in JAMA. What I would have done differently --

[01:31:45] He would have picked a different supervisor, but he doesn't want to say that.

[01:31:51] No, I think if you look into the supplemental material in table S1, there were a few confusions around some exclusion criteria that I would have done differently, and we're actually planning a new, planning a new trial at the moment now where we have clearly defined such a thing as traumatic cardiac arrest. I think the way that it was defined in the protocol and the way that we had it communicated was not, maybe not too clear, and I think that's the main thing I would have done differently.

[01:32:33] And I think what we have really learned is more of the logistics around conducting a trial like this. It has been a journey from sort of having to understand how you do trial in this patient population and sort of the whole prehospital system, and then we have been very fortunate to have an organization here on all levels, the leadership as well as the clinicians have really supported the trial and sort of worked hard to include patients, been very positive about the trial, including patients. And we were very, very happy about that. And then, I think we have learned how to do things more efficiently with consent forms and other, like practical things, and that's, most people might not recognize, but that's like 99% of doing a trial. It's the whole logistics and the practical things, and we are definitely better suited for the next one, I think.

[01:33:23] Yeah. And I think we're very fortunate to be able to do this trial in the Central Denmark region where the clinicians have been so fantastic at including patients. We're actually, we're reaching for 40% inclusion rate of those that were meeting the inclusion and none of the exclusion criteria, and we were actually able to include two out of three. So, we were just very happy with the way the clinicians worked.

[01:33:46] Yeah, super. Gavin, you were going to say something.

[01:33:50] Yeah, and again, it was great to hear about the support that you got from your local EMS leadership and clinicians. Have you had the opportunity to share the results with them, and what's been the reaction amongst the sponsor and the leadership? Because I guess it could be stress inducing, you know, for them, and sometimes, you know, we've seen, you know, less thought, you know, reactions when a study is potentially terminated for harm, you know, despite the fact that it's, you know, taking forward science, you know, importantly.

[01:34:35] Yeah. I think frustration is not a thing that I have met among the clinicians regarding that we stop the trial early. We haven't had the opportunity to present the detailed results of the analysis. I hope that a lot of the clinicians are listening at the moment right now, but generally, we have met the support, and I think as Chris said, we aim to be very transparent in our way of communicating, and that also includes the clinicians. So, they were very aware from day one why we stopped the trial due to the signal of harm.

[01:35:19] But I think we, and I think this is part of your question, Gavin, is like we were nervous what the reactions would be if we had introduced a treatment that could be potentially harmful, but I think everyone here is really in tune with why we did this trial, what was the rationale behind it, why did we do it, and as you say, that we think despite it potentially being harmful, that these are important findings that could potentially save other people going forward. So, I think our nervousness surrounding that issue was put to shame, and we've gotten a lot of support from both clinicians and leadership.

[01:35:59] And I think, you know, if I could just come and say nothing, that's one of the reasons why it's so incredibly important that such a high quality trial was delivered, because it's delivered an important finding, and without research and without randomized control trials, we're really left guessing about what the most effective treatments [inaudible] with science and evaluation that we're going to improve patient outcomes. Thank you.

[01:36:27] Yes, well said, Gavin. We are almost out of time, so we'll quickly go around the panel once for some final comments, and Chris, I'll start with you in Pittsburgh.

[01:36:38] Oh sure. Just, we again would like to thank the investigative team for sending their paper our way, and we're very excited to work with them. It was probably our shortest statistical review I've ever received. So, kudos to the authors for their excellent work.

[01:36:56] Victoria.

[01:36:58] I just want to say well done. It was such a joy to read, and just I could see how difficult this trial was, and so, yeah, congratulations on that.

[01:37:07] Theresa in Oslo.

[01:37:09] Yeah, no, I think congratulations to the authors. Extremely important and interesting trial. Thank you.

[01:37:16] Great. And Gavin, a last comment from you.

[01:37:19] Yeah, no, just thank you for commitment to science and your commitment to improving outcomes for patients unfortunate enough to sustain cardiac arrest.

[01:37:28] And Lars and Mikael, final word.

[01:37:31] Yeah, so once again, we, of course, want to say thank you to all the people who are participating this trial both as part of the steering committee but also, of course, the clinicians, and then, thank you very much to the panel for their insightful comments and good questions. And of course, thank you to JAMA for publishing and being very thoughtful and attentive in their review. And lastly, Rob, it's always a pleasure presenting at Critical Care Reviews, and it's a pleasure watching other people present. So, thank you very much for doing this, and we hope that we'll be back some time in the future again.