- Signs & symptoms
- Diagnostic approach for admitted patients
Treatment: General protocols
- ED patients getting admitted to the hospital
- Inpatient management of hypoxemic, non-intubated patients
- Intubated patients
Treatment: Specific anti-viral & immunosuppressive therapies
- Immunomodulation: Stages of illness & timing of therapies
- Anti-viral therapies & why they may not work in critical illness
Treatment: Additional issues by organ system
- Renal failure
- ID – Anti-bacterial therapy
- Hematology: COVID-associated coagulopathy
- Endocrine – Glycemic control & diabetes
- COVID-19 is a non-segmented, positive sense RNA virus.
- COVID-19 is part of the family of coronaviruses. This contains:
- (i) Four coronaviruses which are widely distributed and usually cause the common cold (but can cause viral pneumonia in patients with comorbidities).
- (ii) SARS and MERS – these caused epidemics with high mortality which are somewhat similar to COVID-19. COVID-19 is most closely related to SARS.
- COVID-19 binds via the angiotensin-converting enzyme 2 (ACE2) receptor located on type II alveolar cells, intestinal epithelia, and the vascular endothelium (Hamming 2004).
- This is the same receptor as used by SARS (hence the technical name for the COVID-19, “SARS-CoV-2”).
- When considering possible therapies, SARS (a.k.a. “SARS-CoV-1”) is the most closely related virus to COVID-19.
- COVID-19 is mutating, which may complicate matters even further. Virulence and transmission will shift over times, in ways which we cannot predict. Ongoing phylogenetic mapping of new strains can be found here.
nomenclature used in this chapter
- Technically, the virus is supposed to be called “SARS-CoV-2” and the clinical illness is called “COVID-19.” This gets confusing, so for this chapter the term COVID-19 will be used to refer to both entities.
- The term “SARS” will be used to refer to the original SARS virus from 2003 (which has currently been renamed SARS-CoV-1).
- (1) Hypoxemic respiratory failure
- The primary organ failure is hypoxemic respiratory failure.
- COVID-19 can reduce surfactant levels, potentially leading to atelectasis and de-recruitment.
- Pneumocytes with viral cytopathic effect are seen, implying direct virus damage (rather than a purely hyper-inflammatory injury; Xu et al 2/17).
- Autopsy studies show lymphocytic pneumonitis, acute fibrinous organizing pneumonia (AFOP), and diffuse alveolar damage. More discussion of lung pathology here.
- (2) Systemic inflammation
- This has some similarities to hemophagocytic lymphohistiocytosis and CAR-T cell cytokine release syndrome, but it appears to be a distinct entity.
- Clinical markers of this may include elevations of C-reactive protein and ferritin, which appear to track with disease severity and mortality (Ruan 3/3/20).
- This is discussed further below in the section on immunomodulation.
- (3) Disseminated intravascular coagulation
- A major driver of pathogenesis appears to be a hypercoagulable state, likely due to both systemic inflammation and to direct endothelial injury by the virus.
- This is discussed further below in the section on hematology.
stages of illness – this is discussed in the section on immunomodulatory therapy.
modes of transmission
- COVID has various routes of transmission (e.g., contact and airborne). There is increasing evidence that airborne is the dominant mode of transmission. However, airborne transmission seems to be relatively inefficient (more akin to tuberculosis than measles).
- Full contact precautions should be maintained for hospitalized patients (e.g. full-length gowns and gloves).
- People commonly transmit the disease while asymptomatic. For example, speaking or singing can aerosolize the virus, which is then transmitted to others, especially within an enclosed indoor space (potentially triggering super-spreader events).
- Lower humidity might serve to promote transmission, as aerosols can dehydrate and remain aloft for longer duration.
strategies to avoid aerosol virus transmission
- Masking is a primary strategy to reduce transmission. Universal masking in the community (preferably with surgical masks) serves to reduce dispersal and inhalation of aerosols. N95 masks (“FFP2” in Europe) are more effective and are utilized in medical contexts with patients who are known or suspected to have COVID-19.
- Physical distancing and adequate ventilation are extremely important. In a hospital context, negative pressure ventilation or HEPA filters may reduce the aerosol virus burden. Modulation of temperature and humidity might provide additional avenues to limit spread of the virus.(Zhao et al. 2020)
- There is much concern regarding “aerosol-generating procedures” (AGPs). The number of aerosol-generating procedures has gradually been increased to encompass nearly any procedure involving the airway or mouth. However, this is probably a failed paradigm. Patients with COVID-19 are constantly generating aerosols by talking, coughing, or even simply breathing – so the concept of “aerosol-generating procedures” is a false dichotomy. A better approach is to assume that patients with COVID-19 are continually generating aerosols and to act accordingly. More on this here.
- COVID is spread via aerosols: First10EM by Justin Morgenstern
- Want to stop COVID? Be quiet (PulmCrit)
personal protective equipment (PPE)
- (1) Contact precautions (waterproof gown and gloves)
- (2) N95 mask or a powered, air-purifying respirator (“PAPR”).
- (3) Goggles or eye shield (goggles which seal onto the face may be ideal, especially for intubation).
- (4) Hair cover, especially for aerosol generating procedures (listed in the section above).
- (5) Hood may be used, especially during intubations.
- Shoe covers aren't recommended, as removing them may increase exposure (ANZICS guidelines).
- Shoes that are easily cleaned and don't need to be touched might be preferable (e.g. Danskos).
applying and removing PPE (donning & doffing)
- Understanding how to put on (don) and remove (doff) personal protective equipment is extremely important (especially if contact transmission is a dominant mode of transmission).
- Removing soiled PPE is the most critical and difficult aspect.
- Applying and removing PPE should ideally be practiced before patients arrive (e.g. using simulation).
- This video describes how to use PPE (you may skip the first 5 minutes).
some pearls about personal protective equipment
- Pay attention to the junction between gloves and gowns. The gown should be tucked into the gloves (leaving no gap in-between). Using gloves with extended cuffs facilitates this (similar to sterile surgical gloves). Gloves with long cuffs may facilitate removal of the gown and gloves as a single unit (see 12:30 in the above video if this doesn't make sense).
- When removing PPE, always start by first applying alcohol-based hand sanitizer to your gloves.
- After fully removing PPE, sanitize hands and wrists with alcohol-based hand sanitizer again.
- Create a step-wise protocol for PPE removal. Two examples are shown below, but this may very depending on your exact gear. Follow the steps slowly.
- Consider doffing with someone watching you (to ensure good technique). If this isn't possible, doffing in a mirror may be helpful.
signs and symptoms
👁 Table of symptoms described by various studies.
signs & symptoms
- COVID-19 may cause constitutional symptoms, upper respiratory symptoms, lower respiratory symptoms, and, less commonly, gastrointestinal symptoms. Most patients will present with constitutional symptoms and lower respiratory symptoms (e.g. fever and cough).
- Gastrointestinal presentations: 10-20% of patients can present initially with gastrointestinal symptoms (e.g. diarrhea, nausea), which precede the development of fever and dyspnea (Wang et al. 2/7, Goyal et al).
- “Silent hypoxemia” – some patients may develop hypoxemia and respiratory failure without dyspnea (especially elderly)(Xie et al. 2020).
- This can lead to some unusual presentations (e.g. knee pain… as a result of syncope, which in turn resulted from profound hypoxemia).
- Physical examination is generally nonspecific.
- ~2% of patients may have pharyngitis or tonsil enlargement (Guan et al 2/28).
typical disease course
- Incubation is a median of ~4 days (interquartile range of 2-7 days), with a range up to 14 days (Carlos del Rio 2/28). Rare patients may have a longer incubation, however (graphed out nicely by Lauer et al).
- Typical evolution of severe disease (based on analysis of multiple studies by Arnold Forest).
- Dyspnea ~ 6 days post exposure.
- Admission after ~8 days post exposure.
- ICU admission/intubation after ~10 days post exposure. However, this timing may be variable (some patients are stable for several days after admission, but subsequently deteriorate rapidly).
👁 Table of general laboratory findings described in several studies.
complete blood count
- WBC count tends to be normal.
- Lymphopenia is common, seen in ~80% of patients (Guan et al 2/28, Yang et al 2/21). The optimal cutoff for defining lymphopenia is unclear; a cutoff of <1,500 b/L may increase sensitivity for diagnosis of COVID-19 (Goyal et al).
- Mild thrombocytopenia is common (but platelets are rarely <100). Lower platelet count is a poor prognostic sign (Ruan et al 3/3).
- Abnormal coagulation labs are frequently seen. Generally, the most notable finding is markedly elevated D-dimer levels.
- For more see the hematology section below.
- C-reactive protein (CRP)
- COVID-19 increases CRP. This seems to track with disease severity and prognosis. In a patient with severe respiratory failure and a normal CRP, consider non-COVID etiologies (such as heart failure).
- Young et al. 3/3 found low CRP levels in patients not requiring oxygen (mean 11 mg/L, interquartile range 1-20 mg/L) compared to patients who became hypoxemic (mean 66 mg/L, interquartile range 48-98 mg/L).
- Ruan et al 3/3 found CRP levels to track with mortality risk (surviving patients had a median CRP of ~40 mg/L with an interquartile range of ~10-60 mg/L, whereas patients who died had a median of 125 mg/L with an interquartile range of ~60-160 mg/L).
- 👁 Image of prognostic labs including CRP.
- Severe COVID-19 can moderately increase procalcitonin levels (e.g., within a range of roughly 1-10 ng/ml). For example, among patients with severe disease, 14% had a procalcitonin level >0.5 ng/ml (Guan et al 2/28).
- Among patients with known COVID-19, an elevated procalcitonin is a poor prognostic sign (likely reflecting of systemic inflammation)(Lippi et al. 2020).
- A markedly elevated procalcitonin (>>10 ng/ml) might suggest the presence of a bacterial infection, rather than COVID-19.
evaluation for competing diagnoses
- PCR for influenza and other respiratory viruses (e.g. RSV) may be helpful. Detection of other respiratory viruses doesn't prove that the patient isn't co-infected with COVID-19
- ~5% of patients may be co-infected with both COVID-19 and another virus)(Wang et al.). However, the rate of viral co-infection is dynamic, depending on the prevalence of other viruses in the community. Recent studies from New York suggested a co-infection rate of only 1-2% (Goyal et al, Richardson et al).
- Conventional viral panels available in some hospitals will test for “coronavirus.”
- This test does not work for COVID-19!
- This PCR test for “coronavirus” is designed to evaluate for four coronaviruses which usually cause mild illness.
- Ironically, a positive conventional test for “coronavirus” actually makes it less likely that the patient has COVID-19.
- Blood cultures should be performed as per usual indications.
specific testing for COVID-19
- (1) Nasopharyngeal swab should be sent.
- (2) If intubated, tracheal aspirate should be performed.
- (3) Bronchoalveolar lavage or induced sputum are other options for a patient who isn't intubated. However, obtaining these specimens may pose substantial risk of transmission.
- It's dubious whether these tests are beneficial if done for the sole purpose of evaluating for coronavirus (see the section below on bronchoscopy).
limitations in determining the performance of RT-PCR
- (1) RT-PCR performed on nasal swabs depends on obtaining a sufficiently deep specimen. Poor technique will cause the PCR assay to under-perform.
- (2) COVID-19 isn't a binary disease, but rather there is a spectrum of illness. Sicker patients with higher viral burden may be more likely to have a positive assay. Likewise, sampling early in the disease course may reveal a lower sensitivity than sampling later on.
- (3) Most current studies lack a “gold standard” for COVID-19 diagnosis. For example, in patients with positive CT scan and negative RT-PCR, it's murky whether these patients truly have COVID-19 (is this a false-positive CT scan, or a false-negative RT-PCR?).
- Convalescent serologies may help resolve this problem (although they may to have their own limitations as well).
- Specificity seems to be high (although contamination can cause false-positive results).
sensitivity may not be terrific
- Sensitivity compared to CT scans
- In a case series diagnosed on the basis of clinical criteria and CT scans, the sensitivity of RT-PCR was only ~70% (Kanne 2/28).
- Sensitivity varies depending on assumptions made about patients with conflicting data (e.g. between 66-80%)(Ai et al.).
- 👁 Image of analysis of Ai et al. to determine sensitivity & specificity of PCR here.
- Among patients with suspected COVID-19 and a negative initial PCR, repeat PCR was positive in 15/64 patients (23%). This suggests a PCR sensitivity of <80%. Conversion from negative to positive PCR seemed to take a period of days, with CT scan often showing evidence of disease well before PCR positivity (Ai et al.).
- Bottom line?
- PCR seems to have a sensitivity somewhere on the order of ~75%.
- A single negative RT-PCR doesn't exclude COVID-19 (especially if obtained from a nasopharyngeal source or if taken relatively early in the disease course).
- If the RT-PCR is negative but suspicion for COVID-19 remains, then ongoing isolation and re-sampling several days later should be considered.
CXR & CT scan
general description of imaging findings on chest x-ray and CT scan
- The typical finding is patchy ground glass opacities, which tend to be predominantly peripheral and basal (Shi et al 2/24). The number of involved lung segments increases with more severe disease. Over time, patchy ground glass opacities may coalesce into more dense consolidation.
- Infiltrates may be subtle on chest X-ray.
- Findings which aren't commonly seen, and might argue for an alternative or superimposed diagnosis:
- Pleural effusion is uncommon (seen in only ~5%).
- COVID-19 doesn't appear to cause masses, cavitation, or lymphadenopathy.
sensitivity and time delay
- Limitations in the data
- Data from different studies conflict to a certain extent. This probably reflects varying levels of exposure intensity and illness severity (cohorts with higher exposure intensity and disease severity will be more likely to have radiologic changes).
- Sensitivity of CT scanning?
- Sensitivity among patients with positive RT-PCR is high. Exact numbers vary, likely reflecting variability in how scans are interpreted (there currently doesn't seem to be any precise definition of what constitutes a “positive” CT scan).
- Among patients with constitutional symptoms only (but not respiratory symptoms), CT scan may be less sensitive (e.g., perhaps ~50%)(Kanne 2/27).
- CT scan abnormalities might emerge before symptoms?
- Shi et al. performed CT scanning in 15 healthcare workers who were exposed to COVID-19 before they became symptomatic. Ground glass opacification on CT scan was seen in 14/15 patients! 9/15 patients had peripheral lung involvement (some bilateral, some unilateral).
- Emergence of CT abnormality before symptoms explains the existence of an asymptomatic carrier state (discussed above).
- Chest X-ray
- An illustrated guide to the chest CT in COVID-19 (PulmCCM, by Jon-Emile Kenny)
- Series of COVID-19 chest X-rays, courtesy of @ChestImaging
- In order to achieve sensitivity, a thorough lung examination is needed (taking a “lawnmower” approach, attempting to visualize as much lung tissue as possible).
- A linear probe may be preferable for obtaining high-resolution images of the pleural line (to make the distinction between a smooth, normal pleural line versus a thickened and irregular pleural line).
- COVID-19 typically creates patchy abnormalities on CT scan. These will be missed unless ultrasonography is performed overlying the abnormal lung tissue.
- More on this from 5 minute sono here.
- The findings on lung ultrasonography appear to correlate very well with the findings on chest CT scan.
- With increasing disease severity, the following evolution may be seen (Peng 2020)
- (A) Least severe: Mild ground-glass opacity on CT scan correlates to scattered B-lines.
- (B) More confluent ground-glass opacity on CT scan correlates to coalescent B-lines (“waterfall sign”).
- (C) With more severe disease, small peripheral consolidations are seen on CT scan and ultrasound.
- (D) In the most severe form, the volume of consolidated lung increases.
- 👁 Image of these patterns here.
- Other features:
- Peripheral lung abnormalities can cause disruption and thickening of the pleural line.
- Areas of normal lung (with an A-line pattern) can be seen early in disease, or during recovery.
- Tiny pleural effusions may be seen, but substantial pleural effusions are uncommon (Peng 2020).
- As with CT scans, abnormalities are most common in the posterior & inferior lungs.
- For excellent examples of the correlation between CT scan and lung ultrasonography see Huang et al.
- Sensitivity of lung ultrasonography isn't clearly defined.
- Sensitivity will depend on several factors (most notably disease severity, presence of obesity, and thoroughness of scanning).
- My guess is that a thorough ultrasound exam might have a sensitivity somewhere between CT scanning and chest X-ray (e.g., perhaps sensitivity ~75%?)(Huang et al.). There isn't solid data yet, but it's probably reasonable to extrapolate from our experiences regarding other types of pneumonia.
- Specificity is extremely low. A patchy B-line or consolidation pattern can be seen in any pneumonia or interstitial lung disease. Thus, clinical correlation is necessary (e.g., evaluation of prior chest imaging studies to see if chronic abnormalities are present).
- Note that supine, hospitalized patients may have B-lines and consolidation in a posterior and inferior distribution due to atelectasis. Thus, the lung ultrasonography may have greatest sensitivity and specificity among ambulatory patients.
general approach to imaging
all imaging modalities are nonspecific
- All of the above techniques (CXR, CT, sonography) are nonspecific. Patchy ground-glass opacities may be caused by a broad range of disease processes (e.g. viral and bacterial pneumonias). For example, right now in the United States, someone with patchy ground-glass opacities on CT scan would be much more likely to have a garden variety viral pneumonia (e.g. influenza or RSV) rather than COVID-19.
- Imaging cannot differentiate between COVID-19 and other forms of pneumonia.
- Imaging could help differentiate between COVID-19 and non-pulmonary disorders (e.g. sinusitis, non-pulmonary viral illness).
- Ultimately, the imaging is only one bit of information which must be integrated into clinical context.
possible approach to imaging in COVID-19
- Below is one possible strategy to use for patients presenting with respiratory symptoms and possible COVID-19.
- The temptation to get a CT scan in all of these patients should be resisted. In most cases, a CT scan will probably add little to chest X-ray and lung ultrasonography (in terms of actionable data which affects patient management).
- From a critical care perspective, CT scanning will likely add little to the management of these patients (all of whom will have diffuse infiltrates).
- 👁 Schema for imaging patients with respiratory symptoms and suspected COVID-19.
- RSNA focus page on coronavirus (contains fantastic slide show that provides an appreciation of possible imaging findings in a few minutes)
- Risks of bronchoscopy:
- May cause some deterioration in clinical condition (due to instillation of saline and sedation).
- Enormous risk of transmission to providers.
- Considerable resource allocation (requires N95 respirators, physicians, respiratory therapists) – all resources which will be in slim supply during an epidemic.
- Benefits of bronchoscopy:
- Benefit of diagnosing COVID-19 is dubious at this point (given that treatment is primarily supportive).
- Bottom line on bronchoscopy?
- Bronchoscopy might be considered in situations where it would otherwise be performed (e.g. patient with immunosuppression with concerns for Pneumocystis pneumonia or fungal pneumonia).
- Bronchoscopy should usually not be done for the purpose of ruling COVID-19 in or out (Bouadma et al.).
diagnostic approach for admitted patients
👁 Checklist of tests to consider when evaluating a patient with respiratory failure and suspected COVID-19
👁 One possible diagnostic flow chart for an ill patient admitted to hospital with suspected COVID-19.
- This approach is based on the availability of a PCR assay for COVID with a reasonably short turn-around time. This currently isn't a reality in most locations in the United States. Hopefully it will be soon.
- Requiring a negative influenza PCR before testing for COVID isn't desirable, because ~5% of patients may be co-infected (Wang et al.). Thus, a positive influenza PCR cannot exclude COVID. The rate of double-positivity may decrease over time, as the rate of influenza in the community decreases.
- The largest challenge may be determining who needs to be ruled out for COVID (i.e., who needs to be entered into this algorithm in the first place). Currently there is no simple answer for this – clinical judgement is required.
- Ruling out too many patients will result in excessive consumption of masks in patients who don't have COVID. Additionally, placing patients under COVID precautions may impair their care (e.g., isolation may serve as a barrier to obtaining scans or to family visitation).
- Ruling out too few patients may result in nosocomial transmission of COVID.
approach to ED patients getting admitted to the hospital
- Chest X-ray (useful to prognosticate patients and avoid missing non-COVID pathology – even in the era of POCUS).
- CBC with differential.
- Electrolytes, coagulation studies.
- COVID prognostication labs: C-Reactive Protein (CRP), Lactate dehydrogenase (LDH), D-dimer.
- Blood cultures x2.
- Swab for COVID & respiratory viruses.
- Additional studies as clinically warranted (EKG, POCUS, etc.).
- CT chest
- Generally not needed solely for purpose of diagnosing COVID (especially if there are characteristic abnormalities on CXR and POCUS).
- However: if the patient is going to the scanner for another reason (e.g. trauma, abdominal pain, etc) and you are concerned about COVID – then strongly consider adding a chest CT while the patient is in the scanner.
cardiovascular & Bp support
- A fluid-conservative strategy is often advisable (for example, avoid reflexive use of 30 cc/kg fluid boluses).
- For patients with a history of diarrhea and clinical evidence of hypovolemia, titrated fluid administration may be beneficial.
- For patients with significant dyspnea or hypoxemia, try to stabilize with one of the following techniques:
- When in doubt, err on the side of avoiding intubation.
- For patients with hypoxemia, start dexamethasone 6 mg daily (or an equivalent dose of another steroid, for example 40 mg prednisone or 32 mg IV methylprednisolone).
- Treatment pathways are evolving rapidly – have a low threshold to consult with MICU to assess the patient and collaborate on plan (i.e. ward vs. ICU).
- Beware of “silent hypoxemia” – patients may be very hypoxemic but look good (without much dyspnea). The first sign of deterioration is often escalating oxygen requirement, rather than dyspnea.
- There is an enormous tendency for patients to develop acute kidney injury.
- Aggressively avoid all nephrotoxins (especially NSAIDs and vancomycin).
- (Note: Contrast dye probably isn't nephrotoic. If you need to get a scan with contrast, then just get it with contrast.)
- For patients with infiltrates and possible bacterial pneumonia: usual treatment is azithromycin plus ceftriaxone.
- Avoid vancomycin (if high index of suspicion for MRSA pneumonia consider linezolid or ceftaroline).
approach to inpatient management of non-intubated patient
daily examination: focus on
- Do not use a stethescope (this is a fomite that poses risk of disease transmission).
- Cardiac and lung ultrasonography may be performed as indicated for changes in clinical status.
- Lung ultrasonography (not ascultation) is the preferred modality for evaluating pulmonary status.
- Daily labs
- Electrolytes, Creatinine, Magnesium, Phosphate
- CBC with differential
- C-reactive protein
- Admission labs: all of the above plus:
- Urine pregnancy test in reproductive-age women
- Blood culture x2
- Liver function tests
- Coagulation tests including INR, PTT, fibrinogen
- Ferritin, LDH
- Usually target a roughly even fluid balance for patients on the hospital ward (unless there are ongoing fluid losses such as diarrhea, or objective evidence of hypovolemia).
- Consider discontinuation of home antihypertensive agents (especially ACE-inhibitors or ARBs). The use of ACE inhibitors or ARBs among outpatients is controversial, but among ill inpatients these agents are potentially nephrotoxic and should be avoided.
- Oxygen supplementation
- (1) Start with low-flow nasal cannula (e.g. 1-6 liters/minute).
- (2) For dyspnea or worsening desaturation, consider early implementation of either HFNC (ideally with awake proning) or CPAP/BiPAP.
- The approach to respiratory support is discussed further below.
- Consult ICU early for deteriorating patients, as this can escalate rapidly.
- Provide dexamethasone 6 mg daily for patients requiring oxygen (or an equivalent dose of steroid – 40 mg prednisone daily or 32 mg IV methylprednisolone).
- Avoid nebulized bronchodilators
- Only use bronchodilators if truly indicated.
- Instead of nebulizers, use a metered dose inhaler (4-8 puffs may be roughly equivalent to one nebulizer treatment).
- ⚠️ Avoid nephrotoxins (especially NSAIDs).
- Initially most patients will be on empiric antibiotics for bacterial pneumonia (e.g. azithromycin plus ceftriaxone).
- Follow microbiologic studies.
- DVT prophylaxis (continue unless platelets <30, as COVID-19 may cause a pro-coagulable form of DIC despite low platelet count).
- For patients with marked D-dimer elevation, consider higher doses of enoxaparin such as 0.5 mg/kg BID (more on this below).
- Conservative transfusion strategy (generally avoid transfusion unless HgB <7 mg/dL, or <8 mg/dL with active myocardial ischemia).
- May use acetaminophen 1 gram enterally q6hr for antipyretic and analgesic effects.
- Melatonin 5 mg QHS for sleep. (Zhang et al 2020, Zhou et al. 2020)
- ⚠️ Avoid NSAIDs (may cause nephrotoxicity and possibly up-regulate the ACE2 receptor, thereby worsening infection)
approach to intubated ICU patient
daily examination: focus on
- Ventilator settings & synchrony with ventilator.
- Confirm ETT depth at the upper teeth (ensure no migration of the tube).
- Tighten connections between ETT, connecting tubing, and ventilator (to prevent accidental disconnection).
- Neurologic status.
- Cardiac and lung ultrasonography if clinical question.
- Do not use a stethescope (this is a fomite that poses risk of disease transmission).
- Daily labs
- Electrolytes, Creatinine, Magnesium, Phosphate
- CBC with differential
- C-reactive protein
- Intermittent labs
- Triglycerides every 72 hours for patients on propofol (surveillance for propofol infusion syndrome).
- Liver function tests every other day.
- Admission labs: all of the above plus:
- Urine pregnancy test in reproductive-age women
- Blood culture x2
- Tracheal aspirate for gram stain & culture
- Urine legionella & pneumococcal antigens
- Complete set of coagulation labs (INR, PTT, fibrinogen, thromboelastography if available)
- Initially, patients may be hypovolemic (e.g., due to diarrhea and poor oral intake). Titrated fluid resuscitation may be helpful initially, based on physical examination and history. It may be reasonable to allow patients to have a net positive fluid status over the first couple days in ICU.
- After the first couple days in ICU, avoid fluid boluses (more on this here and here) & avoid maintenance fluid infusions. Follow fluid balance and generally target an even fluid status.
- Consider using low-dose vasopressor as necessary to support MAP (rather than fluid).
- Consider discontinuation of home antihypertensive agents (especially ACE-inhibitors or ARBs). Sedation and positive-pressure ventilation will tend to reduce the blood pressure, so antihypertensive agents may be unnecessary.
- Lung-protective ventilation
- APRV might be the ideal ventilator mode if providers are trained in this (a primary pathophysiological problem seems to be atelectasis, which APRV manages well).
- Conventional low-tidal volume ventilation is also effective.
- Start dexamethasone 6 mg daily or equivalent steroid dose.
- ⚠️ Avoid ABG/VBG if possible.
- Consider trending etCO2 and minute ventilation instead of obtaining serial ABG/VBG measurements.
- Oxygen saturation generally does appear to track with pO2 in these patients and can be used to titrate oxygen administration.
- Enteral nutrition.
- Stress ulcer prophylaxis.
- ⚠️ Avoid nephrotoxins (especially NSAIDs).
- Initially most patients will be on empiric antibiotics for bacterial pneumonia (typically azithromycin plus ceftriaxone).
- Discontinue ceftriaxone within <48 hours if no evidence of bacterial infection.
- ⚠️ Avoid vancomycin. These patients don't tend to have MRSA, but they do often develop kidney injury. If MRSA coverage is truly necessary, consider linezolid or ceftaroline.
- Follow microbiologic studies.
- DVT prophylaxis (continue unless platelets <30, as COVID-19 may cause a pro-coagulable form of DIC despite low platelet count). For patients marked elevation of D-dimer, consider higher doses of enoxaparin (e.g. 0.5-1 mg/kg BID). This is discussed further below.
- Conservative transfusion strategy (generally avoid transfusion unless HgB <7 mg/dL, or <8 mg/dL with active myocardial ischemia).
- Follow glucose levels periodically.
- Insulin as needed to avoid severe hyperglycemia (consider allowing some permissive hyperglycemia, to reduce the need for frequent glucose checks).
- Acetaminophen 1 gram enterally q6hr scheduled (for antipyretic and analgesic effects).
- Opioid bolus PRN pain (e.g. fentanyl 50 mcg IV q30 min PRN breakthrough pain).
- Low-dose propofol as a titratable sedative (e.g. ideally around 0-40 mcg/kg/min).
- COVID-19 patients appear prone to developing hypertriglyceridemia (possibly due to systemic inflammation).
- Ideally keep propofol doses low, to avoid hypertriglyceridemia (which may necessitate stopping propofol entirely).
- Adjunctive atypical antipsychotic (e.g. 10-20 mg olanzapine per tube QHS, or quetiapine).
- For ongoing pain, consider adding a pain-dose ketamine infusion (0.1-0.3 mg/kg/hr)(more on this here).
- Melatonin 10 mg QHS for sleep.(Zhang et al 2020, Zhou et al. 2020).
- ⚠️ Avoid NSAIDs (may cause nephrotoxicity and possibly up-regulate the ACE2 receptor, thereby worsening infection).
- More on ICU analgesia and ICU sedation.
lines & tubes
- (1) Orogastric tube or small-bore post-pyloric feeding tube.
- (2) Central line
- Low threshold to place a quad-lumen central line with meticulous sterility.
- Best site may be left internal jugular vein (save the right internal jugular for dialysis or ECMO).
- Consider early transition to a PICC catheter for patients with an ongoing ICU stay.
- (3) Arterial line
- Potentially useful for a patient in shock on multiple vasopressors.
- For non-shocked patients, utility of an arterial line is dubious. This may serve only to encourage frequent ABG/VBG draws (which are unlikely to materially improve care and will cause anemia).
stages of illness & timing of therapies
The above staging system was proposed by Siddiqi et al. Patient courses may vary, making discrete staging challenging. However, this provides a useful conceptualization of the disease process.
stage I (early infection)
- Clinically: Incubation followed by non-specific symptoms (e.g. malaise, fever, dry cough). This phase may last for several days, with fairly mild symptoms. Patients often don't require hospital admission.
- Biologically: Viral replication occurs. An innate immune response follows, but this fails to contain the virus. Symptoms reflect a combination of direct viral cytopathic effect and innate immune responses (e.g. Type-I interferon release).
- Anti-viral therapies could be beneficial, especially in patients predicted to be at higher risk for poor outcome. Anti-viral therapies probably have maximal efficacy when given early, during this phase.
- Interferon I-beta could theoretically be useful to augment the innate immune system response to the virus. This involves rendering cells resistant to viral infection, an intervention which would probably be most effective if deployed as early as possible (however this is a theoretical consideration, which currently is not recommended).
- Immunosuppression could theoretically be dangerous at this point, as it could delay the development of an adequate adaptive immune response. For example, early initiation of steroid has been shown to prolong virus shedding in SARS (Lee et al 2004).
stage II (pulmonary phase)
- Clinically: Despite being stable for several days during Stage I, as patients enter Stage II they may abruptly deteriorate (often with worsening hypoxemic respiratory failure). Patients will often present to the hospital at this point. They may progress rapidly to ARDS, requiring intubation. Markers of systemic inflammation are often moderately elevated (e.g. C-reactive protein, ferritin).
- Biologically: An adaptive immune response occurs, which causes a reduction in viral titers. However, this also leads to increased levels of inflammation and tissue damage.
- Antiviral-therapy could be beneficial (although the later on that antiviral treatment is initiated, the less effective it is likely to be).
- Some immunosuppression could be beneficial for patients with more severe manifestations (e.g., moderate dose steroid).
stage III (hyperinflammation phase)
- Clinically: Patients deteriorate with progressive disseminated intravascular coagulation and multi-organ failure (e.g. vasodilatory shock, myocarditis). Laboratory abnormalities include marked elevation of D-dimer, C-reactive protein, and ferritin. Patients may initially respond well to intubation and ventilation during stage II, but subsequently develop increasing levels of inflammation, which leads to clinical deterioration.
- Biologically: The adaptive immune response spirals into an immunopathological dysregulated inflammation.(Mehta et al.)
- All the treatments from Stage II may be continued (e.g. moderate-dose steroid and antiviral therapy). Depending on the level of inflammation, a higher dose of steroid could be considered.
steroid isn't indicated in early disease
- Early administration of steroid may increase viral shedding (e.g. administration during the replicative phase)(Lee et al 2004)
- Most patients recover well without severe sequelae – so obviously steroid cannot benefit such patients.
- Steroid should not be used in patients with normal oxygenation.
steroid is indicated for patients with acute hypoxemic respiratory failure
- The RECOVERY trial demonstrated a mortality benefit using dexamethasone 6 mg daily for up to 10 days among hospitalized patients requiring supplemental oxygen or mechanical ventilation.
- Indications for steroid include the following:
- (1) Acute hypoxemic respiratory failure (true increase in oxygen requirement compared to baseline)
- (2) Requirement for mechanical ventilation
- (3) Another accepted indication for steroid (e.g. COVID plus asthma or COPD exacerbation)
dose and duration of steroid
- Dexamethasone 6 mg daily for up to 10 days was studied in the RECOVERY trial, so this is the most evidence-based dose.
- If dexamethasone isn't available, other equivalent doses of steroid may be utilized:
- Oral betamethasone 6 mg (overall most similar to dexamethasone)
- IV or oral methylprednisolone 32 mg
- Oral prednisone 40 mg or prednisolone 40 mg
- Higher doses of steroid (e.g. dexamethasone 10-20 mg daily or equivalent doses of methylprednisolone) could be considered in patients with ARDS (based on the DEXA-ARDS trial and the CoDEX trial on COVID), especially in the face of elevated or rising inflammatory markers. Laboratory markers which could support a need for higher doses of steroid might include CRP >125 mg/dL, Ferritin >1,000 ng/mL, LDH > 300 U/L, and D-dimer > 1,000 ng/mL. If higher doses of steroid are used, the dose may be reduced to ~6 mg/day dexamethasone or equivalent as soon as improvement occurs (e.g. falling CRP; note that ferritin may lag behind CRP).
- The RECOVERY trial protocol involved dexamethasone 6 mg/day for up to 10 days. However, the median duration of steroid utilization in that study was only 7 days. Therefore, if patients are making solid clinical improvement then it may be safe to discontinue dexamethasone prior to 10 days.
- Dexamethasone has a long biological half-life, so it will auto-taper and thereby prevent rebound inflammation. If using a shorter-acting steroid (e.g. prednisone or methylprednisolone) it could be reasonable to taper off over ~3 days to mimic the pharmacokinetics of dexamethasone.
background on antiviral therapy 🛑
- For maximal benefit, antiviral therapy probably needs to be started very early after the initial develop of symptoms (i.e. during the viral response phase). Unfortunately, most patients present to the hospital with severe illness after about a week of clinical illness.
- To date, evidence with numerous anti-viral therapies has proven to be disappointing. Overall, the use of antiviral therapy for critically ill patients with COVID-19 may be limited.
- Remdesivir is a nucleoside analogue developed in response to the 2015 Ebola outbreak. It didn't really work for Ebola, so further approval or testing wasn't pursued at that time.
- Remdesivir is an investigational drug which is not currently FDA approved for any indication, including COVID-19. However, remdesivir has received an emergency use authorization (EUA) for COVID-19 (a lower bar than FDA approval).
safety & side effects
- To date, published experience with remdesivir involves well under a thousand patients. As such, this is a very new drug which we don't fully understand. Little is known regarding side-effects. Over time, it's likely that additional side-effects will emerge.
- Known side-effects at this point:
- Infusion-related reactions (may include hypotension, nausea/vomiting, diaphoresis).
- Elevated liver enzymes (AST, ALT, hyperbilirubinemia).
- Volunteers given remdesivir have reported phlebitis, constipation, headache, ecchymosis, nausea, and extremity pain.(Jorgensen CJ et al)
- Renal failure
- Remdesivir may be contraindicated in renal insufficiency. To date, studies involving remdesivir in COVID-19 have excluded these patients due to concern that the intravenous vehicle sulfobutylether beta-cyclodextrin could accumulate – so the safety of remdesivir in this context is unknown.
- Given that remdesivir is a nucleoside analogue it might be teratogenic. In the ACTT-1 trial, women of child-bearing age were required to use contraception for a month after exposure to remdesivir.
contraindications to remdesivir 🛑
- Renal failure (The only positive study on Remdesivir, ACTT-1, excluded patients with GFR<50 ml/min).
- AST or ALT above five times normal.
- Many studies have aggressively excluded women who might become pregnant.
- Teratogeinicity remains unclear. As a nucleoside analogue, this is possible.
- Remdesivir does not affect mortality (based on a meta-analysis of three RCTs involving >7,000 patients).
- Remdesivir does not appear to reduce the risk of intubation (based on the SOLIDARITY trial results).
- Remdesivir does not have a robust or reproducible effect on hospital length of stay.
- Remdesivir reduced the time to recovery by ~1-4 days in the ACTT-1 trial (e.g., duration of oxygen utilization). This generated the popular misconception that ACTT-1 demonstrated decreased hospital length of stay. However, it's actually unclear whether remdesivir caused a statistically significant reduction in length of stay within the ACTT-1 trial (extraneous factors often affect the length of stay, making it a difficult outcome to reduce).
- Remdesivir increased the hospital length of stay in the SOLIDARITY trial.(WHO et al., 2020) This is likely because SOLIDARITY was an open-label trial (which may more closely mirror clinical reality in this situation). Open-label use of remdesivir likely prolonged length of stay because patients were kept in the hospital longer in order to receive remdesivir. The SOLIDARITY trial found that both remdesivir and hydroxychloroquine extended length of stay to the same extent, suggesting that both drugs are equally ineffective for COVID-19. Further discussion of this trial here.
dosing & monitoring
- 200 mg IV once, followed by 100 mg IV for four days for a five-day total course.
- Follow liver function tests.
bottom line on remdesivir?
- The World Health Organization currently does not recommend the use of remdesivir.(Rochwerg et al. 2020) The above trials don't exclude a minimal benefit from remdesivir – but if a benefit exists it is likely to be small. Remdesivir has some known harms and probably additional unknown side effects (it has never been broadly utilized, so rare side effects may remain unknown).
- Practitioners will continue to be pressured to administer remdesivir, especially in the United States. A five-day course might be reasonable in patients with no contraindications. When possible, shared decision-making may be beneficial.
- Bamlanivimab is a recombinant antibody which binds and neutralizes the spike antigen on COVID-19.
- When administered to rhesus macaques prior to infection with COVID, bamlanivimab reduced the viral load.
- A trial on the use of bamlanivimab in hospitalized COVID patients was stopped early due to futility. Publication of this trial is pending (currently only a press release is available).
- The BLAZE-1 trial evaluated the use of bamlanivimab among outpatients. Bamlanivimab was administered within <3 days after collecting the COVID sample.(Chen et al. 2020)
- The primary endpoint was viral load after 11 days. Bamlanivimab had no consistent effect on viral load (the intermediate dosage of bamlanivimab did seem to reduce viral load, but when considering the multiplicity of statistical tests involved, this may have been due to chance).
- A secondary endpoint was hospitalization, which was redefined in a post-hoc fashion as either admission to the hospital or the emergency department. Bamlanivimab did seem to reduce “hospitalization” – but given that this is a secondary endpoint which was redefined, its meaning is unclear.
- Bamlanivimab can cause anaphylaxis. The precise incidence of this adverse event remains unclear, but it seems to be rare.
bottom line on bamlanivimab
- There is very little evidence that bamlanivimab is effective.
- The NIH and CDC guidelines currently recommend against the use of bamlanivimab.
- Studies are ongoing, including the use of bamlanivimab combined with another monoclonal antibody into a cocktail.
- Bamlanivimab has received an Emergency Use Authorization (EUA) in the United States, with some centers administering it. This is likely an unwise use of healthcare resources.
More on bamlanivimab here.
thoughtful fluid resuscitation
- Gentle fluid administration could be considered for patients with evidence of hypoperfusion and a history suggestive of total body hypovolemia (e.g. prolonged nausea/vomiting and diarrhea).
- Aggressive fluid resuscitation (e.g. blind administration of 30 cc/kg fluid) should be avoided.
- Patients rarely are shocked on admission (even among critically ill patients, admission blood pressure is generally normal and lactate elevations are mild-moderate)(Yang et al 2/21).
- Overall, the rate of reported “sepsis” is generally low (<5%). The virus doesn't seem to generally cause a septic shock picture (but of course, patients may always suffer from superimposed bacterial septic shock).
- More discussion on fluid therapy for COVID-19 is here.
- Troponin elevation is common (especially high-sensitivity troponin).
- This is a strong predictor of mortality. Among non-survivors, troponin tends to increase steadily from day 4 of illness through day 22 (Zhou et al. 2020).
- Potential causes of troponin elevation in COVID-19 patients may include:
- Myocardial injury (troponin elevation without symptoms/ EKG/echo findings of myocardial ischemia)
- Type-I MI (plaque rupture) – this is probably among the least common causes.
- Type-II MI (stress MI)
- Stress cardiomyopathy (a.k.a. Takotsubo cardiomyopathy)
- Viral cardiomyopathy
- Investigation should focus on integration of EKG and echocardiographic findings as well as clinical context.
- In most cases, specific therapies for acute coronary syndrome will not be indicated.
- Fulminant cardiomyopathy can occur. This may be a late feature, which can occur even after patients are recovering from respiratory failure.
- Cardiogenic shock appears to be an important cause of death, contributing to ~7-33% of deaths.(Ruan 3/3/20) However, this wasn't a prominent feature in the series of patients at Cornell in New York City (Goyal et al).
- It's unclear whether this represents a viral cardiomyopathy (virus can be recovered from myocardial tissue), stress/Takotsubo cardiomyopathy, or cardiac dysfunction due to systemic inflammation (i.e., a feature of virus-induced hemophagocytic lymphohistiocytosis).
- Evaluation: EKG, echocardiography, and troponin levels to evaluate for acute coronary occlusion.
- Palpitations were reported in 7% of patients in one cohort (Liu 2020).
- A large series reported arrhythmia in 17% of patients, but didn't specify further (Wang 2/7/30).
- These studies lack control groups, so it's unclear to what extent COVID may be causing arrhythmias (or whether arrhythmias simply occur in sick patients).
- Rarely present upon admission, but can be a late finding among critically ill patients in ICU.
- Potential causes:
- Cardiogenic shock (i.e. myocarditis)
- Secondary bacterial infection with septic shock
- Systemic inflammation / hemophagocytic lymphohistiocytosis
- Pulmonary embolism
- Pulmonary hypertension due to excessive mean airway pressures (e.g. PEEP or APRV)
- Anaphylactic reaction to medication
- Complete septic workup (e.g. blood cultures, sputum culture, chest X-ray, examination of line sites)
- Bedside echocardiogram and physical examination
- Review of serial labs (hemophagocytic lymphohistiocytosis labs should be measured routinely).
- Vasopressor support as guided by echocardiography and physical examination.
- Empiric antibiotic therapy if concern for septic shock.
- Corticosteroid therapy may be considered (although most patients will be on this allready).
- Inhaled pulmonary vasodilator could be considered for intubated patients with acute cor pulmonale.
high flow nasal cannula
safety of HFNC
- There is widespread concern that using HFNC could increase the risk of viral transmission. This doesn't appear to be evidence-based.
- Guidelines say HFNC is safe.
- ANZICS guidelines on COVID-19 state the following:
- “High flow nasal oxygen (HFNO) therapy (in ICU): HFNO is a recommended therapy for hypoxia associated with COVID-19 disease, as long as staff are wearing optimal airborne PPE.”
- “The risk of airborne transmission to staff is low with well fitted newer HFNO systems when optimal PPE and other infection control precautions are being used. Negative pressure rooms are preferable for patients receiving HFNO therapy.”
- Surviving Sepsis Guidelines state: “For acute hypoxemic respiratory failure despite conventional oxygen therapy, we suggest using HFNC over conventional oxygen therapy (weak recommendation, low quality of evidence”
- WHO guidelines on COVID-19 state that “Recent publications suggest that newer HFNC and NIV systems with good interface fitting do not create widespread dispersion of exhaled air and therefore should be associated with low risk of airborne transmission.”
- ANZICS guidelines on COVID-19 state the following:
- Reasons that HFNC might not increase viral transmission are:
- HFNC supplies gas at a rate of ~40-60 liters/minute, whereas a normal cough achieves flow rates of ~400 liters/minute (Mellies 2014).
- Therefore, it's doubtful that a patient on HFNC is more contagious than a patient on standard nasal cannula who is coughing.
- HFNC typically requires less maintenance than invasive mechanical ventilation. For example, a patient who is on HFNC watching television may be less likely to spread the virus compared to an intubated patient whose ventilator is alarming every 15 minutes, requiring active suctioning and multiple providers in the room.
- The intubation procedure places healthcare workers at enormous risk of acquiring the virus, so intubation with a goal of reducing transmission is probably counterproductive (see figure above from Tran 2012).
- 👁 Image of risk factors for nosocomial SARS transmission from Tran et al. here.
- Existing evidence does not support the concept that HFNC increases pathogen dispersal substantially (although the evidence is extremely sparse). This includes a small study of patients with bacterial pneumonia (Leung 2018) and an abstract regarding particulate dispersal by volunteers (Roberts 2015).
evidentiary basis for HFNC
- HFNC is generally a rational front-line approach to noninvasive support in patients with ARDS (based partially on the FLORALI trial).
- One case series from China suggested that HFNC was associated with higher rates of survival than either noninvasive or invasive ventilation (of course, this could reflect its use in less sick patients)(Yang et al, see table 2).
- A management strategy for COVID-19 by a French group used HFNC preferentially, instead of BiPAP (Bouadma et al.).
- Westafer et al. found no increase in COVID infections due to HFNC use. (more discussion of this study here)
noninvasive ventilation (BiPAP & CPAP)
(1) continuous positive airway pressure (CPAP) might be the best modality of noninvasive support
- Atelectasis leading to hypoxemia seems to be a major problem among these patients.
- 👁 Image of progressive alveolar collapse.
- CPAP could have major advantages here:
- CPAP can provide the greatest amount of mean airway pressure, and thus most effective recruitment.
- 👁 Image comparing mean airway pressure due to CPAP vs. BiPAP.
- CPAP doesn't augment tidal volumes, so this could facilitate more lung-protective ventilation.
- CPAP can provide the greatest amount of mean airway pressure, and thus most effective recruitment.
- Possible approach to CPAP therapy in COVID-19:
- Increase the CPAP pressure to 15-18 cm if tolerated.
- Titrate FiO2 against oxygen saturation. Falling FiO2 requirements indicate effective recruitment, whereas rising FiO2 requirements suggest CPAP failure.
- Monitor tidal volumes and minute ventilation
- 👁 Image illustrating how a noninvasive ventilator can be used as a monitor.
- Further discussion of CPAP in COVID-19.
(2) role of BiPAP?
- BiPAP could provide benefit beyond CPAP by providing some mechanical support for the work of breathing. However, this does carry a theoretical risk of possibly encouraging the patient to take excessively large breaths (thereby inducing lung damage).
- The ideal BiPAP settings probably involve using a high level of end-expiratory pressure with a low driving pressure (e.g., 16 cm inspiratory pressure with 12 cm expiratory pressure). This will closely resemble CPAP, with a little added support.
- BiPAP could be particularly useful in patients with combined syndromes (e.g. COPD plus COVID-19).
(3) helmet interface
- A helmet interface may have several advantages:
- It could reduce environmental contamination (Cabrini 2020; Hui 2015).
- There is a decreased risk of aspiration if emesis occurs.
- In one RCT investigating ARDS, the helmet reduced intubation rates and possibly mortality (Patel 2016).
- The helmet can be set up without requiring a ventilator, thereby potentially sparing ventilators for other patients.
- Helmets have previously not been used in the United States, but they have become available currently.
- The low compliance of the helmet interface may make it difficult to synchronize with the patient when performing BiPAP. Thus, these devices might work a bit better with CPAP.
safety when using CPAP, BiPAP, or helmets
- Viral filters are essential to create a closed system and limit transmission.
- A discussion of how to configure this is located here.
- This is possible with either a two-limb system involving a full featured mechanical ventilator, or a one-limb system involving a dedicated BiPAP machine (e.g. Respironics V60).
- Improved mask seal may improve safety.
- Helmet masks might theoretically have an advantage here.
awake prone positioning
- This involves a non-intubated patient who prones themselves by lying on their belly. For patients with difficult lying in a prone position, alternating between lying on different sides might also be beneficial.
- Can be combined with simultaneous use of any other noninvasive support device (e.g. low-flow nasal cannula, high-flow nasal cannula, BiPAP, CPAP, or even helmet noninvasive ventilation).
- Requires cooperative patient with intact mentation.
- Could be useful especially in situations where access to invasive ventilation is limited.
physiology: why it works?
- Same physiology as proning a patient who is intubated (proning is proning). For example:
- May improve secretion clearance.
- May recruit atelectatic lung tissue in the dependent lung basis (this seems to be a major issue in COVID-19 patients).
- Proning intubated patients with COVID-19 is widely reported to be successful in improving oxygenation. It stands to reason that similar success could be obtained by proning a patient who isn't intubated.
- Awake proning was recommended by Sun et al. as one technique which could be used to avoid intubation for patients with COVID-19 (Sun et al.).
nuts and bolts
- Help patient lie on their belly in a prone position. For patients with difficulty maintaining this position, other positions may be used (e.g. rotating between lying on alternate sides and sitting bolt upright).
- The most important aspect might be to avoid spending considerable time in a supine position (which promotes atelectasis).
- Make sure support devices are well secured to the patient (e.g. it could be helpful to use tegaderm to anchor a nasal cannula).
- Encourage proning as much as is tolerated (ideally ~12-18 hours/day, but this may be difficult for some patients).
- Follow oxygenation and FiO2 requirement. Ideally an improvement in oxygenation should be seen within a few hours. If no change in oxygenation is observed, ongoing pronation may have less merit.
- Proning the non-intubated patient (PulmCrit blog).
overall schema for noninvasive support
- Many “rules” are circulating regarding COVID-19 (e.g. you must never use HFNC or BiPAP). These don't appear to be evidence-based or guideline-supported.
- Patients vary widely, so use common sense.
indications for intubation?
- An “early intubation” strategy involving intubation of every patient who requires >6 liters nasal cannula will lead to unnecessary intubations and likely iatrogenic harm. Thus, efforts should be made to avoid intubation if possible (using HFNC, noninvasive support, and/or awake proning).
- Patients who can avoid intubation have a substantially better prognosis than patients who are intubated. Although this obviously doesn't prove causality, it does suggest that intubation should be avoided whenever possible.
- Ultimately, the decision to intubate is based on the clinical judgement of the bedside practitioner. Key factors to consider in this decision may include:
- (1) Oxygenation
- There is no well-defined oxygen saturation trigger for intubation.
- Inability to maintain saturation >80% may be considered an indication for intubation, but not necessarily an absolute one.
- (2) Respiratory distress & work of breathing
- It's important to differentiate between tachypnea versus respiratory distress. Tachypnea without increased work of breathing is less concerning.
- Increased work of breathing (e.g., accessory muscle use, sensation of air hunger, diaphoresis) is more worrisome.
- (3) Clinical trajectory
- A relatively stable or improving trajectory may favor ongoing observation.
- Ongoing decline over time may favor intubation.
- (1) Oxygenation
- Additional discussion of intubation timing
See the COVID-19 Airway Page by Scott Weingart for this.
invasive mechanical ventilation
pathophysiology: COVID & ARDS
- (1) Patients with COVID who are intubated will generally meet the Berlin definition of ARDS ([#1] above). However, since this definition is extremely broad, meeting it doesn't have any specific clinical implication.
- (2) A predominant problem seems to be atelectasis. COVID patients may respond favorably to positive airway pressure (e.g., higher levels of PEEP or APRV) with recruitment of lung tissue and improved oxygenation.
- (3) Many COVID patients appear to have PseudoARDS ([#4] above)
- PseudoARDS refers to patients with severe hypoxemia who have improvements in the P/F ratio to higher than 150 following ~12 hours of optimization on the ventilator.
- PseudoARDS is clinically relevant because these patients are less likely to benefit from proning. Prone ventilation does appear to work well for patients with COVID, but it may increase requirements for sedation and paralytics (thereby potentially extending time on the ventilator). Thus, if basic ventilator optimization is capable of obtaining a P/F ratio >150, then proning may not be beneficial.
- (4) Gattinoni's H/L model has been disproven and should not be used to guide ventilator management.
- Further discussion:
airway pressure release ventilation (APRV)
- Early APRV could be very useful for these patients (i.e. used as the initial ventilator mode, rather than a salvage mode).
- Benefits of APRV include:
- (1) A primary physiologic problem in COVID appears to be de-recruitment, which is well managed by APRV. A drop in FiO2 requirement to ~50% is often seen within 6-12 hours on APRV (full recruitment takes time).
- (2) APRV often allows for improvement in hypoxemia without paralysis and/or proning. This may avoid iatrogenic complications from these interventions (e.g. delirium, myopathy).
- (3) APRV is a more comfortable mode than conventional volume-cycled ventilation. This may allow us to render patients comfortable and awake on the ventilator more easily, while using fewer medications (an especially important challenge as we run out of many sedatives).
- A practical guide to using APRV in COVID can be found here.
- APRV initiation can cause hemodynamic shifts, so pay careful attention to blood pressure during initiation.
- True failure to respond to APRV within 12-24 hours (e.g. with PaO2/FiO2 <100-150) would be a strong argument to move towards prone ventilation (discussed here). However, when started early APRV may be more likely to succeed – thereby avoiding the need for proning.
- The main limitation to APRV is that many centers aren't familiar with it or don't have ventilators which can provide APRV.
conventional low tidal-volume ventilation
- This will likely be the most commonly used mode of ventilation, given a strong evidentiary basis as well as widespread experience.
- Tidal volumes should be targeted to a lung-protective range (6 cc/kg ideal body weight, with some liberalization to 8 cc/kg if necessary).
- MDCalc can be used to calculate appropriate endotracheal tube depth & tidal volumes.
- There is no consensus regarding exactly how to titrate PEEP. ARDSnet PEEP tables may represent a reasonable starting point. Titration to clinical effect may be useful if there is sufficient time and experience to do this.
- 👁 Image of ARDSnet low-PEEP & high-PEEP tables here.
- The concept of using unusually low levels of PEEP does not appear to be evidence-based and is not recommended for COVID. Low levels of PEEP may cause partial atelectasis of the lungs, leading to atelectotrauma (repeated opening and closing of the alveoli with each respiratory cycle). Substantial atelectotrauma may be even more dangerous than barotrauma.
permissive hypercapnia & optimization of metabolic acid/base status
- Regardless of the ventilator mode, permissive hypercapnia may be useful. The safe extent of permissive hypercapnia is unknown, but as long as hemodynamics are adequate, a pH above roughly ~7.15 may be tolerable (hypercapnia is preferred over lung-injurious ventilation).
- A common error is to focus solely on respiratory parameters in order to improve the pH, while ignoring metabolic acid/base status. For example:
- ICU patients often have non-anion-gap metabolic acidosis (NAGMA). Treatment of NAGMA with bicarbonate may be the safest way to address a low pH (rather than increasing the intensity of mechanical ventilation and thereby threatening the lung).
- Even if the metabolic acid/base status is normal, IV bicarbonate may still be considered to improve pH, while simultaneously continuing lung-protective ventilation (discussed here). Targeting a mildly elevated serum bicarbonate can facilitate safe ventilation with low tidal volumes (more on different forms of IV bicarbonate here).
- Prior to consideration of proning, optimization on the ventilator for 12-24 is generally preferable (discussed here).
- For failure to respond to initial ventilator optimization (e.g. with persistent PaO2/FiO2 below 150 mm), prone ventilation should be considered.
- Proning is effective at increasing oxygenation, but it has the drawback of requiring deeper levels of sedation. Paralysis may be needed, but many patients can tolerate proning without paralysis (simply with deep sedation). Increasing levels of sedation (with or without paralysis) may increase the risk of delirium and myopathy, potentially prolonging the length of ventilation.
inhaled pulmonary vasodilators
- Inhaled pulmonary vasodilators offer potential efficacy with few drawbacks:
- i) Improved ventilation/perfusion matching may improve oxygenation.
- ii) Pulmonary vasodilation may off-load the right ventricle, avoiding cor pulmonale.
- iii) Inhaled vasodilators generally have little effect on systemic hemodynamics (thereby avoiding systemic hypotension).
- Potential indications:
- (1) Refractory hypoxemia
- (2) Hemodynamic instability with evidence of cor pulmonale (e.g. right ventricular dilation on echocardiography)
- Mechanical ventilation and coronavirus pneumonia (Giuseppe Natalini, ventilab blog, Google translation from Italian)
disaster ventilation strategies
 splitting ventilators
- In a dire emergency, one ventilator can be used to support several patients.
- Pressure-cycled ventilation should be used, with a driving pressure <13-15 cm (Aoyama et al. 2018).
- Blog exploring the general strategy to setting the ventilators here.
- Columbia Presbyterian protocol for splitting ventilators here.
- Some additional ideas about how to hook everything up here.
 outpatient-design BiPAP machines for intubated patients?
- It could be concievable to connect outpatient BiPAP machines to endotracheal tubes.
- FiO2 might be limited (bleeding in wall oxygen might only achieve a limited FiO2, maybe around 50-60%) ???
- This would require the patient to spontaneously trigger breaths, so light sedation would be needed.
- Compared to the split ventilator technique (above), BiPAP devices could be used with less ill patients:
- Splitting the ventilator requires deep sedation and provides full ventilator support – this is better for the sickest patients.
- BiPAP machines require light sedation and provides partial support – this could be used for less ill patients.
- Home-design BiPAP masks are often unable to generate high flow rates, so they won't be able to support patients who are dyspneic with high flow demands.
- On a related note, Trilogy devices could probably easily be repurposed to be used as ventilators.
 oxylator resuscitator
- Small automated device which can provide pressure-cycled ventilation.
- Relatively inexpensive and there still seems to be a reasonable supply available.
- Allows for delivering titratable levels of PEEP.
- More information:
 votran automatic resuscitator
- These are small plastic devices which can provide pressure cycled ventilation (more here). In some ways, they may be conceptualized as a simplified and primitive version of an oxylator resuscitator. They are designed for field use in a disaster.
- Unlike the oxylator resuscitator, this device doesn't allow for the addition of higher levels of PEEP (PEEP is fixed at a relatively low level, around ~5-8 cm).
- i) This renders it unsuitable for use in patients with ARDS.
- ii) In a dire emergency, the votran automatic resuscitator might be used in less ill patients (e.g. a patient with trauma or drug intoxication), thereby freeing up other ventilators to be used on sicker COVID patients.
overall strategy for ventilator shortage
- There is no one-size-fits all solution.
- Different strategies may work for different types of patients.
- Any way that we can free up ventilators is beneficial.
- For example:
- Splitting ventilators: Could be used for extremely ill patients (intubated, on deep sedation).
- BiPAP machines attached to endotracheal tubes: Could be used for patients who are close to weaning off ventilation.
- Votran automatic resuscitator: Might be used for patients intubated for non-pulmonary reasons (patients with normal lungs).
- Patients with COVID-19 often respond well to intubation and positive pressure ventilation (probably reflecting lung recruitment). Unfortunately, they may continue to have a tendency to de-recruit their lungs. Consequently, there may be an increased risk of deterioration after extubation.
- Overall it seems that patients with COVID-19 can be weaned off ventilation similarly to other patients, with the exception that post-extubation BiPAP might be a stronger consideration.
- ANZICS guidelines state that HFNC and/or noninvasive ventilation (with a well fitted facemask and separate inspiratory and expiratory limbs) can be considered as bridging therapy post-extubation, but must be provided with strict airborne PPE.
- CPAP therapy or BiPAP (with high end-expiratory pressure) might be useful to prevent de-recruitment in these patients. (More on COVID-19 & CPAP here).
- By the time of extubation, patients will often have been ill for well over a week. It's likely that their viral load will be decreasing at that point, so the risk of virus transmission may be lower (compared to the initial intubation). More on transmission above.
- COVID can cause mild elevation of transaminases (e.g. in 200's). However, fulminant hepatitis or liver failure hasn't been reported (B&W guidelines).
- Potential mechanisms of liver injury include (B&W guidelines)
- Direct viral infection
- Drug hepatotoxicity
- Shock liver
- Systemic inflammation / hemophagocytic lymphohistiocytosis (this might be associated more closely with bilirubin elevation)
- Many medications used in these patients may also elevate transaminases, so liver function test abnormality mandates medication review.
epidemiology & timing
- Renal failure requiring dialysis is reported in a subset of patients admitted to ICU (probably ~5%).
- It tends to be a late finding, occurring 1-2 weeks after admission.
pathology & pathogenesis
- Acute tubular necrosis due to generalized multi-organ failure is probably the predominant mechanism.
- Complement deposition in the tubules was observed in six patients within an autopsy study, raising the question of whether this may be a contributory mechanism. (Diao B et al.)
- Virus can bind to proximal tubular epithelial cells (which express the ACE2 receptor), so direct viral infection is possible.
treatment is supportive
- Avoid nephrotoxins.
- Re-dose renally cleared medications.
- Hemodialysis indications seem to be the same as for other patients.
- The prognosis of patients requiring dialysis appears poor.
- COVID-19: one study found a mortality of 10/10 patients in a recent study on COVID-19 (Zhou et al).
- SARS: Renal failure correlated with poor prognosis (92% mortality with renal failure versus 9% without). In multivariable analysis, renal failure was the strongest predictor of mortality (more-so even than ARDS)(Chu et al. 2005).
- Goals of care should be explored prior to proceeding to hemodialysis.
- Avoid giving excess fluid, as this may necessitate dialysis to remove fluid.
- COVID-19 patients may be hyper-coagulable, so heparin or citrate anticoagulation anticoagulation may be important to maintain a CRRT circuit.
- The prognosis of patients requiring dialysis appears poor.
initial empiric antibiotics
- Initially, there may be concerns regarding the possibility of a superimposed bacterial pneumonia. When in doubt, it may be sensible to obtain bacterial cultures, prior to initiation of empiric antibiotic therapy. Based on culture results, antibiotics might be discontinued in <48 hours if there isn't evidence of a bacterial infection (this is exactly the same as management of influenza pneumonia).
- MRSA coverage?
- COVID doesn't seem to increase the risk of MRSA (unlike influenza). This based on anecdotal reports, with a very low level of evidence.
- MRSA therapy could be instituted based on typical indications for a patient with community-acquired pneumonia (further discussion here and here).
- Excessive use of vancomycin should be discouraged, as patients are at substantial risk of developing renal failure.
- 👁 Algorithm for who needs MRSA coverage in context of community-acquired pneumonia.
delayed bacterial superinfection
- Bacterial pneumonia can emerge during the hospital course (especially ventilator-associated pneumonia in patients who are intubated).
- Among patients who died from COVID-19, one series found that 11/68 (16%) had secondary infections.(Ruan 3/3/20)
- This may be investigated and treated similarly to other ventilator-associated pneumonias, or hospital-acquired pneumonias.
COVID associated coagulopathy
- COVID produces a hypercoagulable state, for several reasons:
- (1) Inflammation (e.g., IL-6) stimulates up-regulation of fibrinogen synthesis by the liver.(Carty 2010)
- (2) Virus infects endothelial cells.
- COVID-associated coagulopathy (CAC) has some similarities to disseminated intravascular coagulation (DIC). However, CAC appears to be a distinct entity from DIC. The pathophysiological hallmark of DIC is systemic coagulation activation, which generally seems to be lacking in CAC (e.g., the INR and PTT aren't substantially elevated).
- The primary consequence of CAC appears to be thrombosis. Microthrombi may contribute to organ dysfunction (especially ARDS). Macrovascular thrombosis often leads to deep vein thrombosis and pulmonary emboli.
- 🔑 If patients with COVID develop frank DIC (e.g., elevation of INR and low fibrinogen), then consider the possibility of a superimposed bacterial infection.
hematologic abnormalities seen in COVID-19
- Dramatic elevations in D-dimer are the hallmark laboratory abnormality of CAC (see above).
- Patients with D-dimer >1,000 at admission are twenty times more likely to die than patients with lower D-dimer values. (Zhou et al.)
- In clinical practice, fibrinogen is generally elevated or normal.
- Some reports describe severely ill patients with hypofibrinogenemia. However, it's possible that some of these patients may have been suffering from superimposed bacterial pneumonia leading to DIC. (Han et al. 2020)
- Thrombocytopenia can occur, but this is generally mild.
- PT/INR are often slightly elevated.
- aPTT (activated partial thromboplastin time) may be reduced slightly. Markedly elevated factor VIII levels may tend to pull the aPTT downwards.
- Thromboelastography (TEG) (Panigada et al.)
- Reduced R-time indicating enzymatic hypercoagulability in 50% of patients (but it may rarely be increased in some patients).
- Increased maximal amplitude (MA), indicating excess platelet/fibrinogen function in 83% of patients.
- Antithrombin levels may be slightly diminished.
- Factor VIII and von Willebrand factor are considerably increased.
evaluation for clinical thrombosis
- Bedside ultrasonography to evaluate for deep vein thrombosis may be considered, especially if there are other clinical features of DVT/PE.
- CT pulmonary angiography
- May be useful in select patients.
- Logistically, large-volume CT scanning of patients with COVID is often impossible (e.g., risk of disease transmission and inability to transport unstable patients to the scanner).
heparin resistance in COVID-associated coagulopathy
- Heparin resistance is common in CAC, with a severity increasing in parallel with the illness severity. In ICU patients, this is a very common problem.
- The mechanism of heparin resistance appears to be that COVID stimulates the production of acute-phase reactive proteins which bind to heparin and thereby reduce the effective “free” heparin which is available.
- For patients on a heparin infusion, heparin resistance will be clinically obvious (due to inability to achieve a therapeutic PTT level or anti-Xa level).
- For patients who are being treated with low-molecular-weight heparin (LMWH), heparin resistance may go unnoticed. This may cause LMWH therapy to be ineffective.
- Further discussion of heparin resistance in general is here, and more evidence on heparin resistance in COVID is here.
- DVT prophylaxis should generally be maintained – unless platelets are below 25,000 .(ISTH guidelines 3/25).
- 🛑 Standard dosing of DVT prophylaxis fails in ICU patients due to heparin resistance, as discussed above. For example:
- Klok et al found that despite DVT prophylaxis, about 27% of patients had venous thromboembolic events and 4% had arterial thromboembolic events (which is likely an underestimate, due to lack of systematic screening for these events and truncated observation periods for some patients). Consequently, these authors suggested doubling the typical dose of prophylactic heparin (e.g., enoxaparin 40 mg twice daily, rather than once daily).
- Spiezia L et al. reported that despite DVT prophylaxis, 5/22 of patients (23%) were noted to develop deep vein thrombosis (a figure which likely underestimates the true burden of venous thromboembolic disease).
- Dutt et al. found that 40 mg enoxaparin daily failed to achieve therapeutic anti-Xa levels among COVID patients in the ICU. This provides biological evidence of heparin resistance, explaining the clinical failure of standard DVT prophylaxis observed across many trials.
- Moll et al. from Brigham & Women's hospital recently reported that standard dose DVT prophylaxis was ineffective among ICU patients with COVID.
- How should DVT prophylaxis in the ICU be provided? This is currently unclear.
- (1) Standard doses of low-molecular-weight heparin are inadequate based on numerous studies.
- (2) One reasonable approach could be augmented or “intermediate” doses of low-molecular-weight heparin (e.g., 0.5 mg/kg enoxaparin BID). This dosing scheme is already supported by a considerable body of evidence in other types of critically ill patients (discussed further here).
- (3) Adjustment of enoxaparin based on anti-Xa levels could be a rational and evidence-based strategy, if this is logistically feasible.(Trunfio 2020) This strategy is compatible with using a higher dose of enoxaparin (e.g., 0.5 mg/kg enoxaparin BID could be started, and subsequently adjusted based on anti-Xa levels).
management of known PE or DVT (or highly suspected PE or DVT in a patient too unstable to get a CT scan)
- Therapeutic anticoagulation is indicated for known or highly suspected venous thromboembolic disease.
- Based on heparin resistance, standard doses of low-molecular-weight heparin may be inadequate among critically ill patients (e.g., 1 mg/kg BID enoxaparin).
- It could be preferable to use a heparin infusion if this is feasible, as it will help ensure that therapeutic heparin efficacy is achieved.
- There are some reports of purpura fulminans occurring in COVID-19. Diagnosis is clinical based on the characteristic appearance of the extremities, as well as laboratory derangements (e.g., marked elevation of D-dimer).
- Treatment of purpura fulminans is challenging. For a discussion of this, see the chapter on purpura fulminans.
glycemic control & diabetes
- ACE2 receptor is present in the islets of langerhans within pancreas, raising the possibility that virus could directly affect the endocrine pancreas (Yang et al. 2010).
- SARS has been shown to induce a transient state of insulin resistance.
- Currently there isn't any evidence available regarding COVID-19.
- Possible predictions regarding COVID-19?? (Currently these are guesses).
- (1) Patients with Type-I diabetes and COVID-19 might present with diabetic ketoacidosis (rather than with typical pulmonary symptoms).
- (2) Patients without diabetes may develop hyperglycemia in the ICU which requires more aggressive management than the average patient.
analgosedation for the intubated patient
why optimal analgosedation is essential
- Achieving a patient who is mentating normally and is comfortable on the ventilator is enormously beneficial, for numerous reasons:
- (1) This may reduce the respiratory rate and thereby promote lung-protective ventilation.
- (2) An awake and cooperative patient is vastly easier to extubate.
- (3) Avoidance of delirium may improve long-term neurocognitive function?
- (4) It's just a nice thing to do for the patient.
unique challenges faced in COVID patients
- (1) Patients often remain on the ventilator for relatively long periods of time (e.g. >7-14 days). Prolonged use of some medications may cause dependence and even withdrawal (e.g. opioids or dexmedetomidine).
- (2) Patients seem to develop hyperlipidemia rapidly if exposed to higher doses of propofol (possibly related to a component of hemophagocytic lymphohistiocytosis).
- (3) Medication shortages are beginning to emerge (especially intravenous medications). This may necessitate a transition to oral agents (which seem to be in better supply).
construction of a multi-modal analgosedative regimen
- The key concept here is using relatively low doses of multiple different medications to function in a synergistic fashion.
- Using relatively low doses of each individual medication optimizes the efficacy/toxicity ratio of that medication.
- One example of an analgosedative ladder for COVID is shown below;
- For ongoing pain, analgesics are added on sequentially.
- For ongoing anxiety, sedatives are added on sequentially.
- A useful combination may be: melatonin, olanzapine, propofol, acetaminophen, ketamine, and PRN opioid. A surprising number of patients can be rendered awake and comfortable on the ventilator with this combination (especially when using a comfortable ventilator mode, such as APRV).
analgesic #1: scheduled acetaminophen
- Acetaminophen should be scheduled at a dose of 1 gram q6hr. In cirrhosis or severe alcoholism, the dose may be cut in half (500 mg q6hr).
- Acetaminophen provides mild analgesia as well as antipyresis (both effects with a goal of improving patient comfort).
analgesic #2: pain-dose ketamine infusion
- A low-dose ketamine infusion has numerous potential benefits:
- (1) Provides mild analgesia (reducing the amount of opioid required).
- (2) Ketamine attenuates the development of opioid tolerance and opioid-induced hyperalgesia, thereby blunting opioid side-effects. (Barr 2013, Angst 2003).
- (3) Ketamine exerts anti-depressant effects which may improve patient mood, even at low doses (Rasmussen 2013, Zarate 2006).
- (4) Ketamine might weakly inhibit IL-6 (deep dive on this by Adam Thomas et al. here). This isn't a real reason to use ketamine, but perhaps a fringe benefit.
- The usual dose is 0.1-0.3 mg/kg/hr. At the higher end of this range, mild psychotomimetic effects may be seen. These effects are often beneficial (e.g. mild sedative effect), but occasional patients will have disturbing hallucinations.
- Empirical dose-titration can generally find a sweet spot where there is analgesia, but no problematic psychotomimetic side-effects.
- When in doubt, it's safer to stay closer to the 0.1 – 0.15 mg/kg dose range.
analgesic #3-4: opioids
- PRN boluses of opioid are generally preferable:
- Since bolus doses given only when necessary, this limits the total dose of opioid. Thus, it's probably preferable to use large PRN boluses, compared to a continuous infusion.
- Opioid infusions are prone to numerous problems:
- They tend to run longer and at higher doses than necessary, thereby increasing toxicity.
- Fentanyl tends to accumulate in fat tissue over time, which can be extremely problematic.
- Fentanyl infusions expose patients to massive cumulative doses of opioid (e.g. 50 mcg/hr fentanyl for a day is equivalent to ~240 mg oxycodone).
- If opioid infusions are used, a daily interruption or dose-reduction should be performed (to verify that the dose of opioid being used is indeed necessary).
- For patients on an opioid infusion, always ensure that a substantial amount of opioid (e.g. ~25-50%) is being given in the form of PRN boluses. If a patient is on a continuous infusion and receiving no PRN boluses, that implies that the infusion rate is unnecessarily high.
sedative #1: melatonin
- (1) This may help maintain day-night circadian rhythm, thereby providing very weak sedative activity at night (Mistraletti et al. 2015).
- (2) Some evidence suggests that melatonin may prevent delirium, although this is controversial.
- (3) Melatonin could theoretically have some anti-viral effects (Zhang et al 2020, Zhou et al. 2020).
- The usual dose is 5 mg QHS.
sedative #2: atypical antipsychotics with sedative-predominant properties (olanzapine, quetiapine)
- (1) These may promote a hemodynamically stable sedative regimen (by reducing the dose of propofol required).
- (2) Antipsychotics provide sedation without promoting delirium (“non-deleriogenic sedatives”).
- (3) Timed administration may promote sleep.
- The major advantage of olanzapine is that it doesn't prolong QT or cause Torsades de Pointes.
- The most logical dosing schedule might be 5-20 mg QHS.
- The drawback of olanzapine is that it has a relatively low maximum dose (20 mg), which may limit its potency.
- The major advantage of quetiapine may be a higher maximum dose (800 mg/day). At these doses, it may be a bit more powerful than olanzapine.
- The drawback of quetiapine is that it does increase the QT interval.
- Quetiapine has a shorter half-life than olanzapine, so it should be dosed twice daily (for use as a maintenance sedative). A higher dose may be given at night to promote sleep (e.g. 50 mg in the morning and 100 mg before sleep).
sedative #3: propofol or dexmedetomidine infusion
- Generally not ideal in COVID patients, due to the long duration of sedation necessary (patients will become dependent on dexmedetomidine and subsequently withdraw from it).
- Propofol infusion is generally preferable here.
- COVID patients appear prone to developing hypertriglyceridemia due to propofol (possibly because of underlying hemophagocytosis). This is problematic, because severe hypertriglyceridemia may necessitate completely stopping propofol.
- Using low doses of propofol (e.g. 5-30 mcg/kg/min) may avoid the development of hypertriglyceridemia.
sedative #4: PRN IV haloperidol
- IV haloperidol could be useful for patients with agitated delirium.
- Haloperidol may increase the QT interval, so caution is required.
sedative #5: phenobarbital
- Phenobarbital is particularly effective in patients with a history of alcoholism. However, as we encounter shortages of other sedatives, low-dose phenobarbital as an adjunctive, general-purpose sedative may become increasingly useful (Gagnon et al. 2017).
- Due to balanced effects on the glutamate and GABA systems, phenobarbital may be less deleriogenic than benzodiazepines.
- Typical dosing regimen:
- Loading dose: 5-10 mg/kg once.
- Maintenance dose: 1-2 mg/kg daily (either IV or PO).
- For patients on ongoing maintenance therapy, check levels occasionally (targeting a level of ~15-25 ug/mL or ~64-107 uM/L).
- Phenobarbital may also be given enterally with 100% bioavailability and fairly rapid absorption.
sedative #6: benzodiazepines
- Benzodiazepines are generally an agent of last resort in the ICU, due to their tendency to cause delirium.
- Sometimes PRN benzodiazepines are necessary. In this situation, the dose of benzodiazepine should be minimized. Simultaneous efforts should be made to augment other sedatives and analgesics, with a goal of minimizing benzodiazepine exposure.
other neurologic problems
- Patients with COVID-19 are at risk for a variety of neurological problems, especially if critically ill. It will be difficult sorting out common complications of critical illness versus unique features of COVID-19. (Aaroe et al. 4/16)
- Common complications of critical illness
- Critical illness myopathy and neuropathy (especially among patients receiving extended paralysis)
- Cerebrovascular disease
- More unique complications related specifically to COVID-19
- Guillain-Barre Syndrome
- Acute Disseminated Encephalomyelitis (ADEM)
- Acute necrotizing encephalopathy
- Direct viral encephalitis
- Common complications of critical illness
Guillain-Barre Syndrome (GBS)
- Data is currently limited to a five-patient case series. (Toscano et al. 4/17)
- This seems to begin ~5-10 days after the initiation of clinical illness (coincident with development of adaptive immunity).
- Weakness is the predominant clinical finding (most often ascending paralysis). Dysautonomia doesn't seem to be a prominent issue.
- Guillain-Barre Syndrome may tend to blend in with critical illness neuropathy & myopathy, which may be more frequent (especially among intubated patients).
- The diagnosis may be supported by neuroimaging (excluding other lesions) and bedside electrophysiologic studies.
- Intravenous immune globulin (IVIG) is generally the front-line therapy for Guillain-Barre Syndrome (with equal efficacy compared to plasmapheresis and superior tolerability).
- More on GBS in the IBCC chapter on this here.
Acute Disseminated Encephalomyelitis (ADEM)
- This has been reported only in a single patient with COVID-19, so its incidence remains unclear (Zhang et al.).
- Acute Disseminated Encephalomyelitis is often seen after viral illnesses. It causes multifocal demyelinating lesions scattered throughout the white matter within the brain, spinal cord, and optic nerves.
- A variety of symptoms may occur, depending on the location of the lesions (e.g., confusion, coma, seizure, weakness, sensory abnormality).
- The diagnosis is based largely on neuroimaging (with multiple lesions present, resembling those of multiple sclerosis).
- This may be treated with steroid.
Acute necrotizing encephalopathy
- This is a rare disorder caused by various viral infections. It has been reported in only a single patient with COVID-19, so its incidence in this situation remains unclear (Poyiadji et al.).
- The pathogenesis seems to involve systemic inflammation, which damages the blood-brain barrier (Wu et al. 2015).
- Clinical features may include confusion, seizure, or focal neurologic deficits.
- Radiographically this causes multi-focal, symmetric lesions on CT scan and MRI involving the thalami, brainstem, cerebral white matter, and cerebellum (with involvement of the bilateral thalami being the most consistent finding). At various stages, there may be edema, petechial hemorrhages, or eventually necrosis.
- There is no established therapy. Steroids have been utilized with mixed results.
- Overview: Ischemic stroke can occur to any cohort of patients under physiologic stress. However, it seems that COVID-19 causes an unusually large number of ischemic strokes (including in patients with otherwise mild disease manifestations). This is likely a manifestation of hypercoagulability due to COVID-19.
- A report of 221 patients with COVID-19 detected acute ischemic stroke in 11/221 patients (5%), cerebral venous sinus thrombosis in one patient (0.5%), and intracranial hemorrhage in one patient (0.5%)(Li et al 3/13). Notably, 12/13 patients with cerebrovascular complications from COVID-19 had extremely high levels of D-dimer (with an average level of 6,900 ug/L).
- Oxley et al. reported a series of five patients <50 YO with COVID-19 who presented to medical care due to symptoms of large-vessel occlusive stroke. Aside from stroke, these patients had either none or mild symptoms from COVID-19. Three of the patients had D-dimer levels over 1,500 ng/mL, suggesting that COVID-19 is capable of inducing a hypercoagulable state even in the absence of severe hypoxemic respiratory failure.
- COVID-19 can invade the brain and directly cause a viral encephalitis. Fortunately this seems to be rare. To date, only one case of encephalitis is reported (Moriguchi et al).
- Patients with COVID-19 can be relatively young and suffering from single-organ failure due to a reversible etiology, so many would be excellent candidates for ECMO.
- VV ECMO could be used for respiratory failure (although it's unclear how common true refractory hypoxemia is).
- VA ECMO could be useful in patients with fulminant cardiomyopathy and cardiogenic shock
- Exact indications and timing are unclear.
- In an epidemic, ECMO capabilities would probably rapidly become saturated. Very thorny ethical issues could arise (e.g. how long of an ECMO run is one patient allowed to have before the withdrawal of life-sustaining therapy, in order to allow the circuit to be used for another patient).
- Infographics on ECMOed by M Velia Antonini
prognostication of individual patients
overview: three general domains
- (1) Epidemiological risk factors
- Age above ~55-60 years old
- Morbid obesity
- Chronic kidney disease
- Coronary artery disease, heart failure
- Chronic pulmonary disease
- Transplant or other form of immunosuppression
- (2) Vital signs
- Respiratory rate >24 breaths/min
- Heart rate > 125 b/m
- Oxygen saturation <90% on room air
- (3) Labs (rough cutoffs only; greater elevations are worse prognostically and individual studies vary regarding cutoff values)
- C-reactive protein >100-125 mg/L.
- Absolute lymphocyte count <0.8.
- Neutrophil/Lymphocyte ratio >3-5.
- LDH >245-300 IU/L.
- (D-dimer >1,000 ng/ml. However, D-dimer values vary between labs, clouding generalization of this data).
- (Ferritin >300 ug/L. However, ferritin may lab behind current clinical changes and is often harder to obtain rapidly, making it less useful for acute risk stratification).
The 4C mortality score currently appears to be the best prognostic score
- This score was derived and validated from a consortium of 260 hospitals in England, Scotland, and Wales. It has been validated subsequently in Italy and among a variety of different ethnic groups.(Knight et al. 2020)
- An online calculator that includes risk stratification can be found here. Low-risk patients may be appropriate for home disposition, intermediate-risk patients for ward disposition, and higher-risk patients for step-down unit or ICU disposition.
- ⚠️ Important limitations:
- The 4C mortality score predicts the risk of death (not requirement for advanced medical care).
- Absolute mortality rates shift over time, so the score is more useful as an indicator of relative severity than absolute mortality risk.
- The area under the ROC curve is ~0.79, so the score isn't perfect. The score has never been compared to best clinical judgement. While the 4C score can be used as an electronic second opinion, it should not supplant clinical judgement.
- More on COVID prognostic models (including the 4C score) here.
more on laboratory prognostication
- Blood cell count abnormalities
- Lymphopenia and its trends over time (prolonged or worsening lymphopenia portends poor outcome).(Chu et al. 2004)
- Neutrophil/lymphocyte ratio (NLR) appears to be a superior prognosticator when compared to either lymphopenia or C-reactive protein.(Liu et al. pre-print) As shown in the second figure below, neutrophil/lymphocyte ratios >3 could suggest a worse prognosis.
- Other predictors of poor outcome include markers of inflammation (C-reactive protein and ferritin), lactate dehydrogenase, and D-dimer. D-dimer elevation over 1,000 ng/ml was the strongest independent predictor of mortality. (Zhou et al. 3/9/20)
- Troponin is a prognostic factor, but it may be challenging to compare values obtained across different laboratories.
- 👁 Image of prognostic labs
- Preliminary indicators of mortality based on data from China and South Korea (MDCalc, Shahriar Zehtabchi and Joe Habboushe).
- (1) It remains unclear what fraction of patients are hospitalized.
- There may be lots of patients with mild illness who don't present to medical attention and aren't counted.
- The vast majority of infected patients (e.g. >80%) don't get significantly ill and don't require hospitalization.
- (2) Among hospitalized patients (Guan et al 2/28)
- ~10-20% of patients are admitted to ICU (note – as the pandemic progresses and fewer patients present to hospital, this percentage is growing).
- ~3-10% require intubation.
- ~2-5% die.
- (3) Longer term outcomes: Prolonged ventilator dependency ?
- Patients who survive the initial phases of the illness may still require prolonged ventilator support (possibly developing some radiographic elements of fibrosis)(Zhang 2020).
- As the epidemic progresses, an issue which may arise is a large volume of patients unable to wean from mechanical ventilation.
- Overall mortality
- (Caveat: There are numerous sets of numbers published and they vary a lot. However, from the clinician's standpoint the precise numbers don't really matter.)
Update #8, 3/10:
Update #7, 12/4:
Update #6, 5/17:
Update #5, 5/6:
Update #4, 4/21:
Update #3, 4/13:
Update #2, 3/30:
Update #1, 3/22:
First COVID cast, 3/11:
questions & discussion
To keep this page small and fast, questions & discussion about this post can be found on another page here.
- Journal & Society homepages on COVID-19
- Treatment guidelines
- FOAMed on COVID-19
- WHO guidelines on fluid administration for COVID-19 are dangerous (PulmCrit)
- EMCrit RACC on airway management in COVID-19 (Weingart & Brian Wright)
- COVID-19 on RebelEM (Salim Rezaie)
- COVID-19 on St. Emlyns (Ashley Liebig)
- COVID-19 on Radiopaedia (Daniel Bell)