Airway Choices in the Era of Many Choices


In the beginning, there was awake blind nasal tracheal intubation, a brutal and often unsuccessful technique that thankfully disappeared when non-anesthesiologists gained access to paralytics. The primary tool for emergency airway management then became the traditional laryngoscope, a device little improved since the 1940s, until the advent of video laryngoscopy in the early 2000s which changed everything.

The age of VL was ushered in by the 2001 Glidescope, which simultaneously introduced two technologies: video (putting a camera at the end of the blade and projecting the image onto a screen), and hyperangulated geometry (blades with a much steeper curve that are designed not to move the tongue out of the way, but to advance around the tongue). It took us a decade to figure out that video was a transformational advance, but not hyperangulated geometry, which is a trade-off compared to conventional standard geometry (Macintosh or Miller) blades. But because they were introduced together, in the same revolutionary device, another decade has passed and there is still much confusion about how these technologies fit together.

Direct laryngoscopy uses a blade to push the tongue out of the way, establishing a direct line of site between the eye and the glottis; intubation is carried out by looking in the mouth. Video laryngoscopy uses a blade with a camera at the end; once the blade is positioned, the operator intubates looking at the screen. VL can employ either SG or HA blades. HAVL, which uses a blade that goes around the tongue, requires that the operator look at the screen. SGVL uses blades that move the tongue out of the way, allowing the operator to look in the mouth and perform DL (usually more difficult), or look at the screen and perform VL. Traditional laryngoscopy is using an old-school 1943 laryngoscope, which of course can only do DL.

The hopefully over VL vs. DL debate

For many years, there was a raging debate about VL vs. DL, but this has become less relevant with the rise of VL systems that incorporate hot-swappable SG and HA blades, allowing providers to switch between HAVL and SGVL in seconds. Because you can perform DL using an SGVL blade in the same way you would perform TL, SGVL is direct and video, SGVL contains TL, and now that we can instantly flip between SGVL and HAVL, it has become obvious to most (I hope) that there is no longer a role for TL. I still see pockets of enthusiasm for the 1943 laryngoscope, but this seems to me usually a demonstration of machismo similar to how an aged cardiologist recently condescended to me for relying on ultrasound to diagnose tamponade. All airway providers must be competent in DL, but only because video is not always available.

The real debate: SGVL vs. HAVL

The question now is which video blade to reach for. Practitioners who are highly skilled at either SGVL or HAVL will successfully intubate nearly everyone; both techniques are excellent.

HA blades produce an excellent view of the glottis almost every time with very little practice, and can get good views when SG blades can’t. This is a very important advantage, especially for folks who intubate infrequently, because seeing the cords calms you down, and you do a better job when you’re calm. HAVL requires less force, head/neck movement, and tissue displacement, which is an advantage in neck immobility, limited mouth opening, or an oral lesion that would bleed if disturbed. However, the fantastic view easily obtained with HAVL comes at the price of more difficult tube delivery, because the view from the camera points up, while the trachea points down. This can be overcome with maneuvers designed to flatten the angle between camera-view and trachea, but these maneuvers involve degrading your view, and are therefore against your instincts, and also don’t apply to SGVL or DL, where you want to optimize your view.

SGVL is quicker than HAVL, because tube delivery is (literally) straightforward, and this advantage is powerfully magnified by a bougie, which can be used with HAVL but is simpler with SG. Similarly, suction is easier to apply with an SG blade, and this can make a big difference in heavily soiled airways. The view-optimizing goals of SGVL are intuitive and work with your instincts, and the SGVL skillset applies directly to DL/TL, which is not true of HAVL.

The Era of Single-Use Flexible Endoscopy

Flexible endoscopes,  often called fiberoptic bronchoscopes, are immensely powerful airway tools that until recently have been available in a small minority of emergency settings, because the conventional devices are expensive, fragile, and require a sterilization process that often doesn’t fit into EM workflows. The development of single-use, disposable endoscopes has brought this technology downstairs, and these devices are now available from several vendors.

Most emergency providers have little experience with flexible endoscopy, but they are not hard to use and a reasonable degree of proficiency can be acquired in a single mannequin session. The gadgets are often referred to as fiberoptic, though most of them no longer use fiberoptic technology, and often referred to as bronchoscopes, though they are rarely used for bronchoscopy in the ED.

Their classic use is as a primary intubation modality using meticulous local anesthesia, in a patient who is fully or nearly fully awake, and breathing normally. Topicalized Awake Intubation (TAI) is a core skill for anesthesiology but not always easily applied to emergency practice because it requires time and some degree of patient cooperation, as well as materials and topicalization skills not always present. It is still a skill that should be taught and practiced in EM training, because it is the safest way to intubate patients with a high degree of known or predicted difficulty, and absolutely can be performed successfully in many emergency scenarios.  An alternative breathing intubation technique that provides some (not all) of the safety of TAI but is much easier to deploy in the ED is the use of dissociative dose ketamine without a paralytic (KOBI). More discussion around the merits and demerits of these two techniques can be found here and here.

Most ED patients, however, are (appropriately) intubated using a paralytic, and flexible endoscopy, now more available in the ED, is perhaps even more important as part of an RSI-based airway strategy.  Using an endoscope by itself (often through the nose) is the traditional way to execute TAI, and this can work well in other techniques where the patient continues to breathe (such as KOBI), but if used after a paralytic, flexible endoscopic intubation is best accomplished in combination with either a video laryngoscope (where you can go fast) or laryngeal mask (where you can continue to ventilate during the procedure).

I got my hands on a disposable flexible endoscope years ago and described an ED Double Setup that incorporates 2-provider video laryngoscopy + flexible endoscopy; a single-provider variation on this procedure using a channeled VL was published by Sowers & Kovacs.  Operating room-based case series and correspondence report on this “dynamic stylet” or “smart bougie” technique, and there is even an RCT.

SGVL or HAVL can be used, though the VL-flex endo technique uniquely leverages the strength of HAVL (obtaining the best view of the glottis) while addressing its key weakness (difficult tube delivery). If VL intubation is unsuccessful, a second operator slides in from the right wielding a flexible endoscope loaded with an endotracheal tube. The VL operator holds the best achievable view (left hand) and utilizes suction (right hand) as the FE operator advances the endoscope to the glottis and into the trachea viewing the mouth directly, then the VL screen, and finally the FE screen as needed.

Airway Strategy with Many Choices

Many modern emergency departments and ICUs have immediate access to VL with hot-swappable SG and HA blades in a variety of sizes, flexible endoscopy, and second generation laryngeal masks that are nearly foolproof to place and designed to easily transmit an ETT.  These new tools can work together in extremely powerful ways, and we haven’t properly updated our airway strategy to take advantage of this. Here is an example of a strategy that does.

The first step is to decide on paralysis vs. breathing; you’ll usually choose RSI, which is usually the right choice, but make sure you know when it isn’t the right choice and the alternatives.

When you’ve decided to move ahead with RSI, the next step is to slow down, think, and prepare for dangerous physiology, optimal oxygenation, and plans A, B, and C. Use a checklist.

Plan A is going to be HAVL or SGVL, and I recommend, unless you’re already excellent with HAVL, that your go-to approach should be SGVL with bougie. When combined with view-optimization maneuvers, you will rarely encounter an airway that cannot be intubated with SGVL+bougie.

When you do encounter such an airway, the best move is not bag-mask ventilation but laryngeal mask ventilation. We’ve known for many years that LMV is both easier to do and more likely to be successful than BMV, and this is even more true now with 2G supraglottics.

Once you have placed an intubating SGA and restored oxygenation, you can go for any Plan A2 you like, but you are now well positioned to intubate through the SGA using flexible endoscopy, which can be done as ventilation continues if you have a swivel adaptor. Alternatively, removing the SGA and performing HAVL will often be successful, especially if you can perform VL-Flex endo, as described above.

Recent years have brought incredible advances in emergency airway management that have made intubation easier to learn and less dangerous. Take a critical look at the devices you stock in your department, what you should stock in your department, and how you can best take advantage of many choices to improve airway outcomes.

Bougie Molding During Laryngoscopy

There is lotsofevidence / that / using / a / bougie will increase your intubation success rate, compared to a styletted endotracheal tube. This is because the bougie is much smaller and easier to maneuver than an ETT, and does not block your view of the target at the last moment, as is often the case with a styletted tube. The Coudé tip allows operators to successfully intubate  grade 3, epiglottis-only views, which means if you are practiced with the bougie and can get a view of the epiglottis, you will be able to intubate.

A less discussed but powerful feature of the bougie is its capacity to be molded; this can be done prior to laryngoscopy (e.g. to conform to the shape of a curved laryngoscope) or during laryngoscopy to allow the bougie to act as a poor man’s flexible endoscope by allowing real-time adjustments to the tip’s trajectory.

This 4-minute video discusses the conventional floppy vs. newer malleable bougies and presents a case where molding a malleable bougie during laryngoscopy turned what could have been a very difficult airway with limited mouth opening into straightforward procedure, even for a junior operator.

There are many manufacturers of floppy and malleable bougies, I have no relationships with any of them.

Simulation Package for ED Management of Opioid Use Disorder

Package includes course overview, materials for 3 cases, pre-session worksheet, post-session worksheet, and anonymous course feedback form as well as supplementary OUD reference materials.


OUD Simulation Package  [print-ready .pdf]

OUD Simulation Package  [editable MS word .docx]


Each case includes an overview, simulated patient notes, physician briefing, and reference materials.

Case 1: Prevention of OUD

Case 2: Management of opioid withdrawal

Case 3: Harm reduction in OUD

Credits: Amish Aghera, Reuben J. Strayer, Sergey Motov, Nubaha Elahi, Michael Lamberta

A Mass Casualty In Slow Motion: Emergency Medicine During the COVID-19 Surge in New York City


The first time I really took notice of coronavirus was when a few providers started wearing masks during their shifts. This was in mid-February when the virus was a Wuhan problem, and we were screening people with travel questions. If you weren’t recently in Hubei Province, weren’t in contact with someone recently in Hubei Province, and weren’t in contact with someone known to have coronavirus, you were deemed to not have coronavirus. If you screened positive on any of those criteria, we got on the phone with our city department of health, and if they agreed, they coordinated testing with Atlanta; at that time only the CDC had a diagnostic test for SARS-CoV-2. I asked my colleague Matt why he was wearing a mask. His response: You don’t think it’s coming here?

I’m pretty sure I hadn’t contemplated that question, until that moment. Meanwhile, our hospital administration was discouraging us from wearing masks, because they are needed for surgeries, and because it presents badly. Meanwhile, boxes of masks started disappearing. 

It wasn’t long after that, early March, when the well patients started coming in, mostly with fever but also chest pain, gastrointestinal symptoms, fatigue, and of course shortness of breath. Most of them were doing fine, ambulatory. They wanted to be tested for coronavirus. During this period we had some tests, then we had no tests, then we had tests, then no tests. We set up a “COVID clinic” across the street from the ED to keep these patients away from the department and process them quickly. Almost all of them were discharged home, with or without a test.

A small number of these early patients were sicker, and needed oxygen, so they had to be admitted. We started to notice alarmingly low saturations, much lower than expected for their clinical condition. We were doing a lot of chest xrays and labs, and some chest CTs. Nearly all these chest CTs showed bilateral ground glass opacities. COVID pneumonia. There was no treatment for COVID pneumonia, but there was a suggestion that hydroxychloroquine was effective, so we treated everyone with hydroxychloroquine. These patients were put in isolation rooms, which quickly filled up throughout the hospital, so they were boarding in the ED, waiting for an iso room.

It was becoming clear that circumstances in Italy were dire, and Italy isn’t China, Italy is across a much smaller ocean. NYC hospital administrators began to grasp the magnitude of the problem and were furiously trying to increase capacity. Elective surgeries were canceled and all the PACUs became ICUs, as did the ORs, then the cafeterias. Tents started going up outside the hospital. The first of three refrigerated trucks arrived. 

We separated the ED into cold and hot zones. In the hot zone, we were wearing N95s. Providers who had contact with patients who were found to be COVID+ were quarantined for 14 days. This seems so quaint in retrospect.

The arriving patients started getting sicker. The paradigm at the time was that if the patient was not stabilized with low-flow oxygen (6 liters/min nasal cannula),  they should be intubated. This was based on the notion that these patients would all end up requiring intubation anyway, and noninvasive high-flow oxygenation strategies such as high flow nasal cannula (HFNC) and NIV (CPAP/BiPAP) would aerosolize virus, infecting providers and other patients. We had never seen this disease before, so we accepted the intubate early paradigm, and sent dozens of intubated patients to the ICU in the first days of the surge, at least five times as many ICU admissions as usual.

By mid-March, we were practicing in disaster mode, our threshold to admit went way up. We stopped doing tests on patients who obviously had COVID but were ok–if you didn’t need oxygen, you went home, come back if you can’t breathe. At this point everyone in the ED is wearing cap, goggles, N95, surgical mask, gown and gloves their entire shift, regardless of where they were stationed in the ED. There were still nominally hot and cold zones but we all knew that was pointless because everyone had COVID, which we were diagnosing primarily by CT. Every patient with fever and respiratory symptoms had peripheral ground glass opacities, but, as time went on, every patient regardless of their symptoms had peripheral ground glass opacities. And we were doing CT chest on a lot of patients, because the upstairs teams for a while thought they could separate the COVID from the not, but then some non-COVID patients who came in with, for example, appendicitis, turned out to have COVID. And so we were asked to start doing a lot more CT chests, and they all showed COVID. I saw a middle aged man who was on mile 40 of a 60 mile bicycle ride, came in in full spandex after getting hit by a car, fractured femur. No chest complaints, fever, nothing. Was on a 60 mile bike ride. CT chest: peripheral ground glass opacities. That was when I realized how prevalent COVID was in New York City. On March 20, Governor Cuomo announced that starting two days later, March 22, New York State would go on lockdown.

Health care providers started getting sick with COVID. One of our docs, a healthy, active woman in her 30s, who intubated 6 patients in one shift, ended up in the ICU (she went home, is doing ok now). The PPE shortage captured the attention of the entire country. We had enough PPE for everyone to use one set of PPE per shift; this was of course PPE designed to be discarded after a single patient encounter. It is very unpleasant to wear PPE for an entire shift. It’s hard to breathe in a mask, it’s a pain to eat or drink or pee so you are less likely to do any of those things. Nobody can hear anyone so everyone’s shouting and misunderstanding, and of course the pressure sore on the bridge of your nose, a painful badge of honor.

On March 27 I retweeted my colleague Cameron’s video showing a pulse ox saturation of 44% with otherwise normal vitals. I suggested, “Consider withholding intubation in hypoxic patients who are otherwise OK.” Some folks who had noticed the same thing chimed in in agreement, but most of the replies were skeptical or disparaging. Over the next days, however, scores of providers broadcast similar findings, including another one of my colleagues, Eric, who photographed a woman texting comfortably with a saturation of 54%. Enter the term happy hypoxemic, and within a week, no one doubted that this was happening, that this was some strange feature of COVID. This was when we, as a community, started to question the early intubation strategy, especially given the outcomes of intubated patients emerging from China and Italy. We began to see intubation as something to be avoided until unequivocally required, and started using HFNC and NIV to manage critically ill COVID patients, despite the aerosolization concerns. 

That said, everyone was freaking out about aerosolization concerns. At least 100 aerosol-responsible intubation checklists were unleashed onto the world. The intubation box was born, and will hopefully soon die. Some hospitals allowed HFNC but not NIV, some hospitals allowed NIV but not HFNC. Some hospitals allowed neither, so everyone in those shops got intubated.

Everyone was freaking out about cardiac arrest. Chest compressions and bag mask ventilation were thought to be high risk for aerosolization, and since we were learning from Italy and China that so many who got intubated for COVID died, some in the EM community suggested that all patients who arrived to the ED in cardiac arrest should be immediately pronounced without resuscitation attempts. Fortunately this solution did not carry the day, and we developed techniques for balancing our responsibility to patients and our own safety. Our protocol is to halt chest compressions and rescue ventilation as soon as EMS arrives, until two fully-PPE-donned providers place an LMA attached to a viral filter; cardiac arrest care can then proceed as usual.

Everyone was freaking out about ventilators. Every day the media came up with some way to make a story out of physicians deciding who lives and dies because we don’t have enough ventilators, even though no hospital actually ran out of ventilators. I was asked to write the institution’s ventilator allocation policy; we didn’t need it and I don’t think anyone ever read it. We all had to learn to use a variety of newly acquired ventilators, and ventilators are tricky.

Everyone was lionizing healthcare workers, who, while the rest of the country was learning how to bake bread on lockdown, were going to work in a lake of coronavirus, putting themselves in harm’s way to care for the severely afflicted. Every media outlet was desperate for any frontline doc to tell them how terrible the situation is, and how terrible their lives are. Nevermind that putting themselves in harm’s way is what firefighters and police do every day (for much less money and esteem), nevermind that healthcare workers have a job and continued income while so many others were suddenly in dire financial straits, nevermind that being trapped at home feeling useless is for lots of folks harder than going in to work, even when your work is more challenging than usual. I noticed that shifts that were vacated (generally for providers getting sick) were immediately snatched up, much faster than usual. Some of this was camaraderie in a disaster, but some of it was: if I’m not at work, I’m trapped at home with my family, I’d rather be at work. Meanwhile, at 7p every day in NYC, everyone claps and celebrates healthcare heroes. I think the focus of attention shifted a little too much to healthcare heroes.

I did appreciate getting messages (and cookies, and brisket) from 100 people I hadn’t heard from in ages. 

Well patients stopped coming to the ED, and sick patients too. Strokes, MIs, opioid overdoses, mostly disappeared. It’s distressing to think about what happened to them.

In the last week of March, the emergency department became a place familiar only to those who do battlefield medicine. The census was lower than usual, but every patient who presented was extremely sick. There were no ICU beds left in the city, so critically ill patients started accumulating in the ED. And were they ever critically ill–these patients had oxygen requirements we had never seen before. They came in saturating in the 50s, we put them on HFNC and their sats went up, but as soon as the cannula fell off, the saturation would drop down to nothing, in seconds. We then understood why everyone was getting intubated in China and Italy.

The emergency department was now a huge, open ICU, filled with patients who were either intubated or on HFNC/NIV, and the patients on HFNC/NIV required more intensive care than the intubated patients, because they’re awake. They turn to one side and the cannula falls off, causing a sudden emergency. They need to use the restroom. They want to talk to their families, who are not allowed in the hospital. The ICU teams were overwhelmed upstairs and in fact all the doctors in the hospital were now repurposed to intensive care, with orthopedic surgeons adjusting vents. So we in the ED were on our own managing a sea of critically ill patients, and we had to pretend to be intensivists too. 

We learned that we could improve oxygenation by having these patients flip on their bellies. Awake proning was born, and a few days and some thousands of tweets later, everyone was awake proning their patients. 

The patients who seemed stable enough to go to a non-ICU setting upstairs often turned out to not be. Code blues were called overhead seemingly every 30 minutes. We learned that the HFNC/NIV patients simply could not be managed on wards behind closed doors, and the ICUs could only accommodate intubated patients, so we stopped admitting the HFNC/NIV patients. We formed our own “ED Admitting Team” and admitted the patients to ourselves, they stayed downstairs and we just watched them as closely as we could, never knowing when to intubate them. Some of these patients after many days on HFNC/NIV could be downgraded to low-flow oxygen and admitted safely, but many got intubated. One young man was prone on HFNC on an ED stretcher for 12 days, and he never turned the corner. We pulled the trigger and intubated him, he went upstairs and was immediately put on ECMO. He was with us for 12 days, we all got to know him and were collectively heartbroken to send him to the ICU on a vent.

There was a period of a couple weeks where the demand for intensive care resources so exceeded supply that the shifts felt like walking through a fog, moving from one dying patient to the next. The absence of any scientific guidance made us feel and behave like helpless homeopaths. We had no effective therapies so we threw all sorts of likely ineffective therapies (azithromycin, hydroxychloroquine, zinc, vitamin C, steroids, nitric oxide, prostacyclin, convalescent plasma, remdesivir, tocilizumab) at them. No one in the world knew when these patients needed to be intubated or how to manage the vent once they were intubated. Saturations in the 70s felt normal, pronouncing patients dead proceeded like a drumbeat, and there were moments when it perversely felt like a relief. Perhaps because we had never felt like we could do so little for so many who needed so much, there were moments when it seemed better to be dead than be a dying patient in a pandemic-overwhelmed hospital with your loved ones not allowed to visit, your care coordinated by doctors powerless to stop the disease laying waste to your lungs.

On April 11 I was much more tired than usual; the next day I was on shift and developed a pounding headache, I never get headaches. When I got home late in the evening I felt awful and that was when I realized. The next day was like a truck hit me. Sweats, chills, nausea, no appetite, no energy, aching all over. I told my boss I had coronavirus. My boss said you need to come in and get tested. I said why, I know I have coronavirus, I feel like shit, I don’t want to come in and get tested. He said you need to come in and get tested. I went in and got tested. Days #4 and #5 were the worst. I barely got out of bed. Constant high fevers, debilitating headaches, dizzy whenever I tried to stand up. I ate nothing. But the worst part was the uncertainty about how things would go over the next week; lots of healthy young healthcare workers were getting very sick and I’ve seen more than a few people my age killed by covid. My test came back positive. Things were a little bit dark. I was anxious and miserable. On day #6 I started to feel a bit better, and I very slowly improved over another week, until I could do my usual activities again. It was the sickest I’ve ever been, but no complaints, I feel fortunate. 

The surge abated in mid-April, but the arrivals numbers remained very low, everyone in the community (rightfully) scared to come to the hospital. Many of the admitted patients were discharged, and many died. Most of the ED patients went upstairs. 

Now the department is slow. There is still coronavirus, lots of people dying of coronavirus upstairs, and still people dying of coronavirus downstairs, and we expect this to continue for many months, but now we can handle the volume. And we have a lot more experience managing COVID, though still very little science to guide us.

The countrywide lockdown is a mitigation strategy designed to flatten the curve so that hospitals aren’t overwhelmed and so we can transition to a containment strategy which requires the development of a robust test/trace/isolate infrastructure. Americans were frightened by what they saw in NYC, so they isolated much better than anyone predicted, and we flattened the curve better than anyone predicted. But a sophisticated and committed federal government response is needed to quickly build a test/trace/isolate infrastructure, and that hasn’t happened and looks like it’s not going to happen.  So most states, including New York, are nowhere near ready to come off of lockdown, from a public health perspective. But the public won’t tolerate this much longer, so we’re going to open up, and new cases/deaths will start to rise. As long as most of the dead are old, and we don’t see conditions like the mass casualty that was visited upon New York City for three weeks in March and April, I think most americans will trade a steady stream of deaths for an open economy. And that’s what I think will happen, until we get a vaccine, a cure, or herd immunity. I hope this is a once in a lifetime event. For untold hundreds of thousands, it definitely will be.


Thanks to Lois Isaksen, Michael Turchiano, Nick Schwartz, Josh Schiller, Matt Friedman, Eric Lee and Ram Parekh for their review and suggestions. 

Illustration by Lian Chang





3 Weeks of Coronavirus in New York City

NYC at this moment (April 6, 2020) seems to have crested with a wave of very sick patients, many many of them requiring intubation or other aggressive forms of oxygenation. The patients are arriving more slowly in recent days, but arrivals are often very ill. Most of us have watched many patients–not all of them old–deteriorate and die, sometimes very quickly, and this is psychologically traumatic. I find it particularly demoralizing to manage so many patients who are succumbing to a disease that is so poorly understood, and where all our treatment strategies are currently based on anecdote and theory, which is just slightly better than being based on nothing. We desperately need experimental data to determine what works. It is very difficult to deploy a serious research effort when the clinicians are getting crushed and the research teams are on lockdown, but we need to do better. We don’t have time for the usual publication-focused machinery of research to churn out papers in 6 months, we need trials now, every covid patient should be enrolled in a trial so we can learn something about how to treat this awful disease.

So all of what follows is based on opinion and consensus, some observational data, because experimental data does not yet exist. It would therefore not surprise me if much of it is ultimately demonstrated to be entirely wrong. 


Assessment of Well Patients

Early on, there was an emphasis on COVID testing, ancillary testing (e.g. CXR, CT, labs), and discharge guidelines often based on oxygen saturation. As the epidemic has evolved, COVID testing has disappeared and the rest has become less useful. Currently, I think most of us have adopted a strategy where patients who present with COVID symptoms, but are able to walk around without oxygen, are assumed to have COVID and discharged with isolation precautions and indications for return. It is expected that some of them will return requiring admission. My main concern is not that we’re inappropriately discharging COVID patients who bounce back–it’s essential to keep people out of the hospital right now until they absolutely need admission–rather that we’re attributing every symptom to COVID and not adequately working up other causes of the patient’s symptoms. That said, it is astounding how many patients have classic COVID symptoms, it really does seem that everyone in the city has this disease. The positive test numbers you hear reported in the media are useless, we’re not testing anyone. Focus on admissions and deaths to get some sense of disease prevalence in a region.

If we had access to reliable and rapid coronavirus testing, we could have done much more, from a public health perspective, than discharge these patients with instructions to isolate, which is impossible for many folks to do in their homes. I hope such testing will become available in regions that aren’t already saturated with coronavirus, before they become saturated with coronavirus.


The ‘intubate early’ paradigm, where patients who “fail” low flow oxygen (e.g. 6 L/min nasal cannula) are intubated and mechanically ventilated, has been abandoned by most centers, because intubated patients with COVID lung disease are doing very poorly, and while this may be the disease and not the mechanical ventilation, most of us believe that intubation is to be avoided until unequivocally required. Also, a low threshold to intubate further strains already-strained hospital resources.

What should be the trigger for intubation is a very difficult question right now, but saturation should not be used in isolation. Reasonable markers are respiratory distress (not simply tachypnea), altered mentation, and rising CO2. In general, when not sure, wait, augment noninvasive therapies, and reassess. This is the converse of the usual EM paradigm, which is when you’re not sure whether or not to intubate, intubate.

Here is the guidance we have come up with based on Gattinoni’s two-phase theory of disease. Many are reporting much better results using APRV instead of conventional modes of ventilation.


Noninvasive oxygenation therapies 

  1. Proning. This should be done for all patients with COVID lung disease sick enough to be admitted to the hospital, regardless of oxygenation therapy. A “proning” team (e.g. providers otherwise idled by societal lockdown) can round on all the patients to prone them, or turn them on one side or another, which seems to improve oxygenation for many patients. 
  2. Encouragement of deep breathing. Many COVID patients take very rapid but very shallow breaths. We have also noticed that many COVID patients have painful breathing, and some present with a complaint of pain instead of dyspnea. Incentive spirometry, with or without treatment with opioids, may be of benefit.
  3. High flow nasal cannula. Consensus seems to be to start with high FiO2 and low flow rate, increase flow rate as needed. Cover with surgical mask to reduce aerosolization.
  4. Noninvasive ventilation. Again it is the high concentration of oxygen that seems to be of value here rather than the pressure, so some have advocated for CPAP with a low pressure, FiO2 100% as an initial approach. Use viral filter at the level of the mask to reduce aerosolization. 

To the extent that you can isolate or cohort patients on HFNC/NIV, do so. As the epidemic worsens, your ability to do this diminishes. 

Pulmonary hypertension seems to play an important role in very ill COVID patients and there is some enthusiasm for using pulmonary vasodilators like inhaled nitric oxide or prostacyclin.



Prothrombosis is one of the many not-yet-understood but repeatedly observed aspects of this disease. Many institutions are moving to aggressive anticoagulation practices in COVID patients, based on trending d-dimers. At a minimum, everyone admitted should probably be prophylaxed. I’ve heard two reports of COVID patients doing very poorly, not stable for CT, treated with lysis, to immediate improvement. Was it PE? Was it diffuse microthrombosis? Was it something else? Was it coincidence? We have no idea.


Cardiac Arrest

All sorts of chitchat on who should be resuscitated and how. If a patient is already intubated and receiving maximal therapies for COVID, and deteriorates and arrests, it does not seem appropriate to pursue further resuscitation, assuming no immediately reversible cause (e.g. ventilator disconnection/obstruction) is found. Undifferentiated patients arriving to the ED in cardiac arrest should, in my opinion, be managed like a patient arriving in cardiac arrest in normal times, with one exception: the patient should be assumed to have COVID, and appropriate measures should be taken to protect staff and other patients from aerosols generated during intubation and chest compressions. See this impossibly well-produced video.


Ventilator Allocation

Despite the media clickbait frenzy on this topic, to my knowledge there has been no need for any hospital to go on an allocation protocol, but that time may come. You do not want that time to come without a protocol that you have established and vetted. Here is one protocol.


Emergency Department and Hospital Flow

At first, departments try to separate into hot and cold zones, but as the prevalence of COVID increases in the community, most of us have noticed that everyone presenting for any reason has evidence of COVID (e.g. the man who was in the midst of a 40 mile bicycle ride gets hit by a car, breaks his leg, has no other symptoms, but CT shows lungs full of COVID, all of us have seen this repeatedly). So, at least in NYC, the entire ED becomes a COVID zone.

I have nothing to add to the national referendum on PPE other than to say that wearing PPE for the duration of an ED shift is difficult, and I think every ED worker in the city is now wearing a single full PPE getup for their entire shift, regardless of where they are in the department. Think about where providers are going to don/doff PPE, where providers are going to store PPE if it has to be reused. Where in the department/hospital/offices PPE will be allowed, and not allowed. For example, is PPE allowed in the breakroom? What will you do with the food that is donated by the community, that your providers will really want to eat? Seems like a trivial problem but it isn’t, frontline gotta eat. Useful to designate a (nonclinical, probably) person to manage food and PPE donations.

Non-COVID ED visits have dropped off precipitously, no one has any idea where all the strokes, heart attacks, intoxicated and withdrawing patients are. However the fraction of patients requiring admission has skyrocketed; the majority of people who arrive to the ED now require significant oxygen support and admission. 

Many if not most of these patients are on advanced oxygen therapies (HFNC, NIV, or MV) but most hospitals do not have even close to the personnel or structural resources required to optimally care for this volume of ICU/Stepdown patients. This is perhaps the greatest struggle in the latest stage of the epidemic in NYC: providing intensive care to 10x the number of patients the hospital is set up for. Many ED’s at the moment are functioning as huge ICUs, caring for enormous numbers of critically ill patients awaiting inpatient beds. It’s even worse upstairs, because standard medical wards are now also ICUs/Stepdowns filled with patients who have very high oxygen requirements, but unlike the ED, these patients are behind closed doors and wards are not resourced to provide the level of monitoring required, and if someone knocks off their NIV mask or HFNC, they can run into trouble very quickly. The number of “Anesthesia, STAT” calls to ward beds is a jarring reflection of the grim conditions. Plan to augment ward staff (physicians, nurses, technicians, anyone) to keep more eyes on these patients. 

Because the arrivals volume is so low, very few ED staff are needed to do emergency medicine, and many of us are repurposed to provide ICU/Stepdown inpatient care to the admitted patients. It would be wise to plan for this transition, because it’s very disorienting, especially when you’re trying to do it in a Tyvek suit for 10 hours. Many units are attempting to leverage other services that are inactive during societal lockdown, and there is a role for just about everyone: The patients, if nothing else (and there is a lot else) need food, water, hygiene, their home meds, an update on their condition (if they’re conscious), someone to talk to their families who are not allowed inside. Any healthcare provider can do this. In one NYC hospital, the surgical service has taken over an entire wing of the ED filled with ICU/Stepdown patients and is just running it, caring for these patients as though they’re in the SICU, because they might as well be. It is an awesome demonstration of our shared purpose and the petty illogic of our usual balkanized culture. People are ready to step up. Utilize them.

COVID-19 Data Sharing Project

Use the links below to access anonymized patient-level data for patients seen in our COVID-19 ICU. This data is updated in near real time.

Because New York City has been seriously affected by COVID-19 before most other cities, we have gained experience in managing many of these patients before severe cases have accumulated in other regions. This database aims to inform clinicians who haven’t yet cared for severely ill COVID-19 patients, but will, soon.

First sheet in each patient notebook is a summary; following sheets correspond to subsequent hospital days. Patients who are discharged or deceased are marked as such, unmarked patients are currently admitted. More patients are added daily.

Credits: Daniel Dove, Ashley R. Davis, Aneri Sakhpara, Cameron Kyle-Sidell.


Patient 1 (discharged)

Patient 2 (discharged)

Patient 3 (discharged)

Patient 4 (discharged)

Patient 5 (deceased)

Patient 6 (deceased)

Patient 7 (deceased)

Patient 8 (discharged)

Patient 9 (deceased)

Patient 10 (discharged)

Patient 11 (discharged)

Patient 12 (discharged)

Patient 13 (deceased)

Patient 14 (deceased)

Patient 15 (deceased)

Patient 16 (deceased)

Patient 17 (deceased)

Patient 18 (discharged)

Patient 19 (discharged)

Patient 20 (discharged)

Patient 21 (discharged)

Patient 22 (deceased)

Patient 23 (discharged)

Patient 24 (deceased)

Patient 25 (deceased)

Patient 26 (deceased)


[4/14/2020] Given the profusion of descriptive data now becoming available, we have ceased enrollment of this cohort at 26 patients. We will continue to update all these patents until disposition.

[7/23/2020] I presented on our ED Admitting Team (not the patients reported on on this page, but a similar cohort that was cared for downstairs, in the ED) at our hospital weekly COVID rounds. 20 minute presentation, followed by 20 minutes of questions/comments.