ICU Management of COVID-19 (March 2020)

We’re going to go back now to David Griffith. And I don’t know if David heard me the last time introducing him. So I’m going to introduce him again. Because he’s another friend and colleague here and in Edinburgh, a great ally of the college here, a consultant and anaesthesia and critical care in the Royal Infirmary of Edinburgh. And as I said earlier, a programme director for an MSc in critical care that we launched collaboratively with the University of Edinburgh just last year. So David, hopefully, we’ve got you online this time. And I look forward to your talk.

Thanks, Andy.

So hopefully, you’ve got audio this time. So thanks very much for the introduction. Also, thanks very much for the honour of talking this evening on COVID-19. I have to take questions at the end, if you submit those on Slido. If you want to challenge anything that I’ve said or anything to add, please do that. If you want to contact me afterwards, my email address and Twitter handle are there, as well.

This is just a conflict of interest slide. So this talk is an attempt to describe some of the experiences of treating patients with COVID-19 in a critical care setting. And I’ll cover some of the data on demographics, severity of illness, organ failure, and some of the treatments that have been required for this patient group. I’m going to talk then about the ICU management of COVID-19. And this will really focus on ARDS, actually. Because that’s the condition that most commonly brings these patients into ICU. We’ve had discussions already about different modalities of ventilation.

And part of the remit for this talk was to delve into the evidence base behind things like non-invasive ventilation, high-flow nasal oxygen, ECMO, prone ventilation that Gregor was just discussing there, and also some of the adjuncts, like corticosteroids and neuromuscular blockers. I’m not going to talk about local planning, or ICU capacity, or personal protective equipment. These are important issues which have already been covered. And similarly, I’m not going to cover trials of specific new antivirals, immunomodulators, or drugs that could potentially modify the acute inflammatory response. Again, these are all important areas. But I think they’ll be covered in future talks and seminars. So this is the sort of early Chinese experience.

So this data is from the whole of China up to 11th of February, 2020, and shows that 80% of patients had mild disease, 14% had more severe disease, and 5% had what’s being described as critical disease. And I suppose these are the patients that this talk is about. These patients had respiratory failure, septic shock, and/or other multiple organ failure. This is data from Wuhan. This data has already been presented by Adam this evening, but broadly speaking, 80% non-severe, 20% severe, and the severity in this study was defined by the ATS Infectious Disease Society of America guidelines. And of the total population studied in here, 5% of those required intensive care. And of the severe cases, 20%, a fifth, required ICU.

This is another study from Wuhan of 138 patients who were hospitalised with COVID-19 pneumonia, admitted between January the 1st and January the 28th of this year. And of those, a quarter required admission to an intensive care unit. And 4.3%– six patients had died by February 3, so very shortly after admission. This is data from another population in Wuhan. The Wuhan Jin Yin-tan Hospital is a designated SARS CoV-2 hospital. All the patients admitted to this hospital have been transferred from other sites with COVID-19, except the health care workers who became ill whilst treating these patients. So they are also included in these figures. And there were 210 patients in this study with SARS CoV-2 pneumonia.

And the study really focused on the 52 patients, or the 25% of patients that were critically ill. And in this case, critical illness was defined by hypoxemia or the need for mechanical ventilation. We’re becoming quite familiar with the demographics of critically ill patients with COVID-19. So the mean age was 59.7 years. But I think it’s interesting to look at the histogram of the age distributions. And note that the vast majority of these patients were actually under 70 years of age. Again, as per other articles that have been recently published, there’s a predominance for male sex in these patients. So 2/3 of patients admitted to ICU were male, and just a third were female.

Comorbidities– or the most important comorbidities in this case series were diabetes, cardiac, and cerebrovascular disease, which again, seems to be quite a common theme. The overall mortality in these ICU patients was really, really high. And I don’t know if that reflects what we’re going to see here, or whether that’s unique to the Wuhan experience. But the mortality rate in these patients was 61.5%. And you can see that quite a lot of those patients, 63.5%, were treated with high-flow nasal oxygen, which is not what we’re doing here, as Gregor was just saying. 42% received mechanical ventilation. But of those that received mechanical ventilation, there was a really high mortality rate.

So 86% of patients who received mechanical ventilation died in this study. And you can see that 56% of them received non-invasive ventilation. And I’m not sure if the numbers of crossover between the NIV and the mechanical ventilation group. But there’s a fairly high mortality also in those that required non-invasive ventilation. So if you need ventilation, based on this, the mortality seems to be quite high. And what about other organ failures? Well, kidney injury seems to occur in about a third of patients. Cardiac injury, I’m not entirely sure what that is. But arrhythmias and, I suppose, myocarditis was seen in about 23% of the patients, and liver dysfunction in 29%.

There have been attempts to try and determine what the risk factors for poor outcome, in this case, mortality in patients admitted to hospital with COVID-19 pneumonia. This data, I think, at the moment, has to be taken with extreme caution, because it’s actually a very small sample with quite a small number of deaths. So the variables that are put into this multivariable logistic regression model were limited to the five that you see in front of us. So there is obviously unmeasured confounding in this study. But what they pulled out was the age, severity of illness, and elevated D-dimer seemed to be important predictors of death in this particular population. Helpfully, we do have a little bit of UK data.

And Gregor presented some of this before. So as of the 20th of March, this is ICNARC case mix programme data, which describes 225 admissions to England, Wales, and Northern Ireland. And for about 199 of those admissions, which accounts for 196 patients, there is data for the first 24 hours of ICU. 16 of those patients, or 8%, have died so far. And 17 of those patients– so another 8%– have been discharged alive. And there are 163 patients still in critical care as of the 20th of March. Obviously, those numbers are a week out of date now. Interestingly, of those admitted to critical care, at least 132 patients received mechanical ventilation in the first 24 hours.

There is missing data for 21 of those patients. So the 67% is probably, possibly, an underestimate. I’m sorry if you can’t read that very well. But this presents the demographics of the UK or the England, Northern Ireland, and Wales patients. And you can see that, again, 2/3 are male and a third are female. And the mean age is 60, so very similar to what was seen in China. 90% of patients admitted to ICU require some form of respiratory support. And 80% need some form of cardiovascular support. You can see that the requirement for renal support is much less, just over 10%.

On the bottom left of that slide, you can see what we know already from the media, that there’s a high predominance of cases in London and surrounding areas at the moment. So what is ARDS, and why is that relevant? Well, most of the patients that are admitted to critical care seem to have this. Although I suspect we might find out that this is not classic ARDS, or it might be different to what we’re used to.

That said, the similarities between other viral pneumonias and ARDS, and during previous pandemics and the early experience of this pandemic would suggest that it’s reasonable for us to extrapolate our management of hypoxic respiratory failure in these patients, according to what we understand about acute respiratory distress syndrome. So I’m just going to spend a little bit of time just revisiting the definition of that. So ARDS is defined by the Berlin criteria, which were agreed by consensus in 2012 and published in Intensive Care Medicine. You need to have an acute onset of hypoxia of less than 7 days from a known insult. And the severity of the ARDS is determined by the degree of hypoxia.

You need bilateral infiltrates on chest X-ray or CT. And any opacities or hypoxia can’t be fully explained by effusion, lobar/lung collapse, or nodules. And I suppose what’s important to say here is that this is quite a common thing that we manage in intensive care. So 10% of our ICU admissions are ARDS. And 23% of all ventilator patients are ARDS. So this is helpfully something that we are quite used to managing. The challenges are related to the sheer numbers and the issues around infection control and transmission. So what do you see in ARDS? You get diffuse alveor damage, which involves the rapid development of capillary congestion, atelectasis, intra-alveolar haemorrhage, and alveolar edoema.

And then at a later stage in the disease, two to five to seven days afterwards, you get hyaline membrane formation, epithelial cells hyperplasia, interstitial edoema. So all of these things produce impairments in the gas exchange of the lung, which cause V/Q mismatches and hypoxemia. And for this sort of very relatively recent paper in the New England Journal of Medicine describes the classic exudative, proliferative, and fibrotic phase. Now, ARDS can be triggered by intrapulmonary triggers or extrapulmonary events. In the case of COVID-19, it’s a direct insult by the viral pathogen, which causes the release of pro-inflammatory mediators by the macrophages. And these attract other inflammatory cells, such as monocytes and neutrophils.

And these release inflammatory mediators, which basically damage the lung and cause it to become leaky, et cetera. So basically, you get a direct injury to the lung, followed by a destructive immune response. I’ve been asked to talk about ventilators, because ventilators are very topical at the moment. And this is in the press today. The government sorted 10,000 ventilators from Dyson. And I suppose one of the reasons to ask me to speak about ventilators is that they might look inside to be quite complicated things. But as many people are showing at the moment, it turns out, it’s quite easy to make a ventilator. You just stick a bag in a box and put some air into it.

So it’s probably not the ventilators that we need to be focusing on, but what we do with them, and how we use ventilators to manage patients with ARDS. So I’m going to talk through some of that over the next few slides. So excuse my terrible diagram. This is not my forte. But if anyone wants to redraw that for me and send it to me, I’d be very grateful. But basically, what I was trying to show is that in a lung with ARDS, you get dependent areas of the lung which are fluid-filled, and under the weight of the heavy edematous lung, basically collapse.

So that the alveoli in the dependent part of the lung are not aerated during a positive pressure ventilation. But those that are in the non-dependent areas are aerated. And this causes overdistension of the non-dependent areas or volutrauma. And in those airways which are now lacking surfactant to keep them splinted open, you’ve got repeated closure or collapse and reopening. And that causes an atelectrauma. And both of these things cause epithelial injury to the lung. And in doing so, allows the translocation of substances, such as bacteria cytokines from the alveolar surface into the circulation, which creates an inflammatory response, and worsens lung injury. And that’s known as biotrauma.

So I suppose the message from this slide really is that ventilation is required in these patients, but it’s fair to say that any amount of ventilation is actually bad for the lungs. And the ICU management really is about limiting the damage that we cause to the patient’s lungs as the patient recovers. So to try and simplify this a little bit, I think we can think of ventilation and ARDS– or the goals of ventilation as basically allowing adequate gas exchange whilst avoiding these injuries that I’ve described. And I think you can think of that in four ways.

Open lung strategies basically are designed to splint open the airways to prevent this opening and closing and atelectrauma, avoiding distending the aerated lungs, things like low tidal volumes. And then I suppose the ultimate would be to rest the lung or not ventilate it at all. So that would be things like extracorporeal membrane oxygenation, ECMO. And I think it’s worth having a discussion about whether it’s worth trying to avoid ventilation in these patients. I’ll present some of the evidence around the non-invasive high-flow oxygen in subsequent slides. So here’s a picture of a sponge. And the analogy is that the lung in ARDS is like a wet sponge. This is a dry sponge, I’m afraid.

So you just have to imagine it full of water. But basically, if you fill a sponge with water, you know that the bottom of the sponge becomes all heavy. And those nice perforations in the sponge will become filled with water. And they’ll actually collapse under the weight of the lung, or the sponge above. And positive end expiratory pressure allows you to apply pressure at the end of the expiratory cycle. And basically, the aim of that is to try and splint open the dependent areas of the lung during ventilation. And it seems to work, actually. So this is an individual patient meta analysis of over 2,000 ARDS patients from three large trials.

And the high PEEP group had a PEEP of 15 centimetres of water on day 1, versus a low PEEP group of nine centimetres of water on day 1. And the ICU mortality was significantly lower if you had a high PEEP. And so that’s considered the standard of care. Another way that we’ve tried to splint airways open is using something called high frequency oscillation ventilation or HFOV. Basically, when I started my intensive care training, these things were in the units all the time. They were like big speakers that made a lot of noise. And they delivered ultra-low tidal volumes with a relatively high mean airway pressure. And the aim of that was to try and prevent this opening and closing atelectrauma.

There were two large trials which reported in 2013– the OSCILLATE trial and the OSCAR trial. The OSCILLATE trial stopped early due to excessive mortality in the HFOV group. And the OSCAR trial, which completed, had a mortality difference which was equivalent in both intervention and control arms. So following the publication of these two papers, HFOV for ARDS really became a thing of the past. But it can be helpful in some situations if hypoxia is a problem.

Lung protective ventilation– so this is the big landmark trial published by the ARDSNet investigators in NJEM in 2000, that showed that ventilating patients with a lower tidal volume– so six mls per kilo ideal body weight and a plateau pressure of less than 30, reduced mortality. The reference to IKEA is that I heard that IKEA had sent lots of tape measures to hospitals to try and to allow them to measure ulnar radial distance, which is a way of estimating ideal body weight. So we talked about prone ventilation. It’s worth thinking about why that could potentially be a benefit. And it comes down to this unequal distribution of ventilation and perfusion in the lung.

So in ARDS, if you’re lying on your back, as in the top panel there on the right from Gattinoni’s paper you can see that although we know that the disease affects the whole lung– anterior and posterior. But you’ve actually got loss of aerated lung in the dorsal aspect there. That’s also where you’ve got most of your cardiopulmonary output too– your cardiac output too. So it’s not surprising that you get blood being passed through the lungs which is not being oxygenated. And therefore, you remain hypoxic. Now, if you turn the patient over onto their front, basically, that has two effects.

One is to allow the dorsal chest to lift in preference to the ventral chest wall, which allows ventilation of the relatively larger ventral regions of the lungs. And it also allows increased aeration of these dorsal lung regions. So these two effects almost universally improve oxygenation, and helpfully. Next slide, please– and so the improvement of oxygenation is probably mostly to do with recruitment of previously collapsed lung regions. It has been hypothesised that the diversion of blood flow from gasless regions to aerated regions may also contribute. But I think that may have been debunked. So this is a large trial in Guérin et al 2013 in NJEM. And this is where the 16 hours comes from.

So in this trial, patients had a period of 12 to 24 hours of stabilisation. They were then turned prone for at least 16 hours. And then they were assessed, then they were turned back again and assessed in the supine position. And basically, if they didn’t meet criteria for stopping, i.e. there was an improvement in oxygenation that exceeded the threshold set, then they were turned back again onto their tummies. And that treatment seemed to be associated with a reduction in mortality. So improved ventilation in ARDS has very strong evidence base. And I suppose the concern might be that turning people prone may carry all sorts of other potential side effects, like dislodgement of tubes and cardiovascular instability.

But in actual fact, the only difference in adverse events between the two study groups here was in cardiac arrest. And actually, it was the supine group that had the greater instance of cardiac arrest. So it seems to be safe, as well.

So revisiting ECMO, Gregor eloquently described how ECMO works. So we’ll move onto the next slide. So this has been studied reasonably extensively. So the CESAR trial in 2009 was a randomised control trial where the intervention group were transferred to a specialist ECMO centre, which was Glenfield, I think. And those patients that were transferred had a better outcome than those that were– that slide is wrong. Just to be aware, if you’re looking at this at a later date. And important to note, though, that only 76% of these patients received ECMO. And the control group had unstandardized treatment. So they remained in their host institution.

So there was a feeling that this really represented a trial of referral to a specialist respiratory centre, and not a trial of ECMO, per se. But the EOLIA trial published in 2018, basically tried to answer the question again. But this trial was actually stopped early after 75% of recruitment. And it didn’t show any difference between the intervention and the outcome group. Now, this has been reanalyze by Bayesian re-analysis, not a forte of mine. But the suggestion is that it’s highly probable, on this analysis, ECMO reduces mortality. And in a subsequent cohort study during the H1N1 influenza outbreak, there did seem to be, in a cohort study, a low mortality in patients receiving ECMO.

So ECMO remains a useful rescue strategy. And the parameters within which we’re using that, at the moment in the UK, have been described by the previous speaker. We discussed non-invasive ventilation. I think we probably have to be a bit cautious with this one. And the reason people worry about this is that there is quite a high failure rate, and possibly increased mortality in viral pneumonia. So these are two studies. One in nearly 2000 patients with influenza infection requiring ventilation. And in this study, 56.8% of them actually failed their NIV and required invasive ventilation. And that failure was associated with higher mortality. And a Canadian study during H1N1 influenza showed an 85% failure rate. Now, these are just two individual studies.

And I’d be real interested to discuss with Adam, some of his data, which seemed to suggest slightly different outcomes. And maybe look at what the differences between these two populations being studied are. I’m sure he knows more about it than me. But I think one of the other concerns is issues relating to transmission to health care workers, as well, which has been reported in another study. We’ve probably all seen pictures coming out of Italy this week of patients in NIV hoods. I suppose the question is, well, could hoods be better? Could they result in better outcomes? And there is one small single-centre study of ARDS comparing hoods to face masks.

And possibly, there is a signal that a hood is associated with a lower rate of subsequent intubation than a facemask NIV. But really, that was a very small study. And I think we need to wait for more evidence. High flow nasal oxygen– so there’s one big study on this– big study, 310 patients. And the primary outcome was intubation rate. And in this study, high flow oxygen was compared to non-invasive ventilation, which was compared to standard care. I should say, high flow nasal oxygen– for those that are unfamiliar– delivers oxygen at very, very high flow rates, and probably provides very high concentrations of oxygen, but also probably provides a little bit of positive end expiratory pressure.

And so under normal circumstances, we do find it very effective for managing patients with hypoxic respiratory failure. And we’re using it increasingly. This study didn’t find a statistical difference in the subsequent rate of intubation with these three techniques. But it did seem to find a potential mortality benefit, although it wasn’t its primary outcome. But I think as Gregor said, it is probably an aerosol generating procedure. So if you’re providing high flow oxygen, your patient needs to be in a negative pressure area anyway. And I think you need to be very careful about observing these patients for subsequent deterioration. I think there’s a feeling that you might be delaying the inevitable.

So neuromuscular blockers– I think we are using these in Edinburgh for some patients.

The trials of this intervention show conflicting results. There was a systematic review published in 2013 of three trials which showed that the use of neuromuscular blockers was associated with lower hospital mortality without increasing the risk of intensive care units of acquired weakness, which is one of the concerns of using them. The potential benefits are that you have a patient that synchronises better with the ventilation. And therefore, there’s a lower risk of secondary damage to the lungs or ventilator-associated lung injury. The Papazian trial of 2010 showed an improved 90-day survival in the deep sedation group. But the subsequent trial published last year in the New England Journal didn’t show any difference in 90-day mortality.

Although this may reflect the difference now in the way we manage sedation in intensive care, in that the patients who required muscle relaxation were obviously deeply sedated or anaesthetised. Whereas, those that weren’t were probably very lightly sedated. So the evidence behind neuromuscular blockers is probably not as strong as it used to be. And finally, the renal support and fluid management are one of the staples of management of ARDS. And I think this is, again, one of the ARDSNet studies from 2016 that compared two treatment groups. One where patients had their fluid balance very, very strictly controlled– diuretics and renal replacement therapy, if needed. And their cumulative seven-day management was effectively even or just minus 136 mls.

Versus a liberal group, where the patients had a cumulative positive balance of nearly seven litres. And the conservative group had improved lung function, shorter duration mechanical ventilation. But interestingly, no effect on 60-day mortality. But based on the reduction in ventilator days in requirement for critical care, it seems to be a sensible intervention. And finally, corticosteroids were just mentioned in one of the other talks, and during the questions, as well. I think in ARDS, the evidence here is really quite low– or of low quality. And much of the trials which looked at this, pre-dated all the studies on lung protective ventilation.

And more recently, in some of the viral– looking at their use in other viral pneumonitis, they were associated with some significant complications, such as psychosis, diabetes, avascular necrosis, and also, delayed clearance of viral RNA. So there are reasons to be cautious about using corticosteroids. But I think, fair to say, that there might be some groups which would benefit. And I think identifying those is key. And I think we’ll just go to the last slide, if that’s all right.

I’m sorry, that was quite a lot to cover in one talk. But the take-home from this is that COVID-19 patients admitted to ICU, we think they’ll generally have ARDS, which is a common ICU condition. There may be a place for high flow nasal oxygen. But the majority will probably require mechanical ventilation. NIV has quite a high failure rate. And its use in viral pneumonia is associated with increased mortality, from my reading. And I think the mainstay of treatment is lung protective ventilation, conservative fluid management strategy, prone ventilation, and in refractory cases, a small number, referral for ECMO. So I just wanted to say, about the MSc critical care, we’ve suddenly had to suspend our operations during all this.

Because all the staff teaching and learning on the programme are– they are clinicians in critical care. So we have taken the decision just to pause things just now. But what that’s given us the ability to do is to look at some of the education resources that we’ve put together and see whether or not they can be repurposed for NHS staff and others working at the front line. And so at the moment, I would say that the person leading that is Graham Nimmo, who’s one of the other programme directors.

And I would ask you, if that’s something that sounds like would be of interest, if you’re somebody who may get drawn into managing critically ill patients with COVID-19 or other conditions, and you haven’t been in an ICU for ages, if you just take a photograph of that code or scan that code, you’ll be taken to a questionnaire where you can contribute to helping us understand what we might be of use. And so I’ll take any questions now, please.

Well, David, thanks very much. I said at the very beginning of all this that there would be more questions than we would have time to answer. And I know that all the speakers are going to attempt to pick off some of the ones that we haven’t been able to address. But in fact, in years there, as you’ve been going through– and I’ve been looking at the questions, some quite technical– about ventilation have been coming in. You seem to me to have been picking a lot of them off. So thanks for that. I’m not going to actually repeat some of the themes that we’ve already heard about.

For example, about a pre-ITU proning and pre-ITU fluid balance– people have been asking about whether to run people dry, whether to–

particularly, given the cardiac problems that are happening. One area that we’ve not really touched on at all is the fact that many patients are on immunosuppressants. And a couple of people are asking you whether you’re seeing, or the literature is telling us anything about these people’s course being any different. Or whether you should continue or stop immunosuppressant treatment.

I think, undoubtedly, these patients on immunosuppressants are going to be more susceptible to severe disease. Whether or not use you use stop therapy very much depends on why you’re on the treatment. So I guess it’s a sort of individual patient thing. I think by and large, you’re probably going to be continuing them, unless they’re really not needed. But I think it comes down to an individual assessment. But I can’t really give you any hard facts or data to back any of that up. But that’s OK. Because the next question is going to challenge that part of you even more.

Lots of people are asking you to speculate on why there might be this quite striking difference in prevalence according to male and female.

I have no idea– no idea. That’s honest. I admire honesty. It’s important we say, we don’t know. I’ve got somebody here saying, would you wait until the patient is presumed to be virus negative before you perform a tracheostomy?

No, I don’t think so. I think the decision about whether or not you perform a tracheostomy really comes down to how long you think they’re going to be ventilated for. And that’s going to take five days or more anyway, generally speaking, to come to that sort of conclusion. So no, I think the things are independent. I suppose what the question might be getting at is risk to personnel of performing an aerosol-generating procedure. I can see Gregor nodding. But I suppose what I’m wondering is, if a patient needs a tracheostomy, they need a tracheostomy. And you just sort of treat them as you normally would. I’m looking to Gregor to see if he can add to that.

I’m not sure how easy it is to link back. And because we’re a wee bit over time due to our tech glitch, and my apologies to you– well, Gregor for the interruption, and David, for you being cut off completely. But I think I’ll draw it to a close there. As I’ve said, all the talks will be available on the website by this time tomorrow night, we would hope. And the speakers are going to attempt to pick off a considerable number of questions that we’ve not been able to address. So I really just would want to close again by emphasising what I said at the beginning. Thanks to everybody for joining with us tonight.

Thanks to the staff here at the college who’ve made this happen. Thanks to our speakers. Do use Slido to post ideas for future sessions. Some of you have done that tonight. And finally, and as I said in my weekly bulletin, do not hesitate to contact me directly at the college– the email address is there in my bulletin– to tell me about your own experiences of what’s going on wherever you work in the world. Individual clinicians’ stories do make a difference, is my experience, when we’re trying to exert influence on the distribution of resources in different places. So thank you again. Thanks to the speakers. And I look forward to you joining us again next week. Thank you now.