2. How do practices in key clinical domains vary across institutions? (12 Key Items Addressed)

Background / what is known: The decisions to admit and discharge a patient are among the most consequential decisions that health care providers make. In general, for all comers with pneumonia, the IDSA recommends use of a risk stratification tool, such as the pneumonia severity index, to guide the selection of outpatient versus inpatient treatment.⁴ However, no specific guidelines exist for COVID-19, and at many institutions, admission and discharge decisions are informed as much, if not more, by cultural norms and available resources as by formal guidelines. In addition, COVID-19 poses unique challenges, because the trajectory of illness is highly variable and difficult to predict (at least with current medical science) and patients can suddenly worsen in the second and third weeks of illness.^5,6 In an ideal world, one might expect health care providers to be more conservative in their admitting and discharging practices. At the same time, however, surging case rates and strains on capacity have created opposing pressures to admit fewer patients and discharge them earlier.

The HOMERuN COVID-19 Discharge Criteria Working Group: A much more in-depth analysis of discharge practices can be found here.

The gatekeeping functions: In an overburdened hospital system, the decision to admit or discharge a patient becomes an important mechanism to regulate access to a limited resource. We observed that hospitals were likely to adjust criteria for discharge but not for admission. Even during surge conditions, only 13% of hospitals raised thresholds for admitting patients with COVID-19 whereas nearly half (46%) lowered thresholds for discharging patients with COVID-19 from their pre-pandemic standard of care for patients with lower respiratory tract infections.

SpO2 / O2 requirement cutoffs: The degree of hypoxia warranting admission and acceptable for discharge varied widely across institutions. Admission thresholds ranged from SpO2 <94% or any O2 requirement to SpO2 <88% or O2 requirement >2L, with nearly 1 in 3 (29%) respondents generally requiring SpO2 <90% or O2 requirement > Discharge thresholds ranged from SpO2 >94% or no O2 requirement to stable O2 requirement up to 4L, with 43% of hospitals discharging patients on supplemental O2.

Home monitoring: One of the innovations that may reduce barriers to discharge is programs to monitor patients for worsening shortness of breath and/or hypoxia at home. Such programs were more likely to exist at institutions that relaxed their discharge criteria (79% versus 63% overall).

ICU admission criteria: Although hospitals have not, for the most part, modified general admission criteria, it appears that they have been more willing to raise the bar for ICU admission, especially during surge conditions. Of the 27 sites who provided this type of granular data, 8 (30%) permit the use of maximal or near maximal high flow nasal cannula (50-70L, 0.8-1 FiO2) and/or non-rebreather mask outside of the ICU.

Observation status: Only 27% of institutions have explicitly developed a pathway or program to identify patients with COVID-19 suitable for a short stay, although 52% of institutions provide care to COVID-19 patients in observation units. This represents an opportunity for institutions to provide guidance and improve alignment between ED and Hospital Medicine providers.

Takeaways and areas for future study: In the absence of clear clinical guidelines and in the face of varying degrees of resource strain, many hospitals have adjusted their admission and especially their discharge thresholds for patients with COVID-19. While one might expect more risk-averse decision-making in the setting of a novel disease entity, a significant minority of institutions have actually raised the bar for admission and simultaneously lowered barriers to discharge. Clearly, more research is needed to determine whether length of stay can be safely reduced in this manner, to identify patient factors that predict post-discharge outcomes such as readmission and to study the impact of innovations like home monitoring programs. Practice change in this area, borne out of necessity during the COVID-19 pandemic, may have implications for patients with other lower respiratory tract infections.

Background / what is known

• Screening: A large proportion of COVID-19 cases result from asymptomatic or pre-symptomatic transmission.^7,8 In order to prevent such transmission within the health care setting, the CDC recommends that institutions “consider targeted SARS-CoV-2 testing of patients without signs or symptoms of COVID-19.”⁹ The IDSA suggests a more nuanced approach, recommending testing of asymptomatic individuals who are being hospitalized when there is a high prevalence of COVID-19 in the community (defined as >10%), but not when there is a low prevalence (defined as <2%).¹⁰ Both groups acknowledge that the best practice in any given scenario will depend on a number of real-world factors, including the availability of testing, test turnaround time, and other local infection control practices, e.g., isolation precautions for asymptomatic individuals.
• Serology: The IDSA recommends against routine use of antibody testing to diagnose SARS-CoV-2 infection during the first 2 weeks following symptom onset. It suggests that immunoglobulin (IgG) testing can be considered as an adjunct to nucleic acid amplification tests (NAAT) when there is high clinical suspicion for COVID-19 and NAATs are repeatedly negative, though sensitivity and specificity will be reduced when patients are less than 3-4 weeks from symptom onset.¹¹

Testing of asymptomatic individuals: The vast majority (90%) of institutions have implemented universal testing of all patients admitted to the hospital. In addition, 17% test all patients who present to the ED. To further reduce the risk of in-hospital transmission, 27% of institutions regularly re-test patients requiring aerosol-generating procedures and 6% re-test all patients at regular intervals.

Serologic testing: The majority (79%) of sites have no routine indications for antibody testing. Few (10%) sites use antibody testing for diagnosis in patients with symptoms of COVID-19 and negative NAATs. Only 2 of 52 sites (4%) routinely use antibody testing for diagnosis in all patients with symptoms of COVID-19, against IDSA recommendations.

Recommendations / takeaways: The vast majority of institutions have implemented screening of all hospitalized patients at the time of admission, going above and beyond existing CDC and IDSA recommendations in order to protect their workforces and patients. This is understandable given the real impact of even rare nosocomial transmission events on health care worker and patient morale and confidence.

Limitations of this study / areas for future study: We did not obtain any information on the types of testing platforms used or turnaround times. As rates of immunization go up and as rates of COVID-19 in the community fall and/or become more seasonal, it will be interesting to see how institutions adjust their screening and testing practices.

Background / what is known: The COVID-19 pandemic has strained health care capacity in many areas and forced clinicians and health care systems to consider rationing care by deciding which patients to prioritize when allocating scarce resources such as hospital beds and limited supplies of some medications. At the same time, even in the absence of severe resource constraints, one of the major challenges in COVID-19 management is the unpredictability of individual patient trajectories, with the potential for sudden or late decompensation. Both of these challenges would be mitigated by accurate methods to risk stratify patients.

This urgent need has driven rapid innovation. Indeed, since the beginning of the pandemic, over 100 COVID-19-specific clinical prediction models for prognosis have been published.¹² Nearly all, however, suffer from bias, a lack of external validation, and poor methodological reporting; furthermore, it is not clear how they should be employed to guide clinical decision making. The 4C Mortality Score, developed by the International Severe Acute and emerging Infections Consortium (ISARIC), is one tool for the inpatient setting that has been peer-reviewed and externally validated in large datasets.^13,14 This 8-variable score predicts in-hospital mortality for patients with COVID-19, and its authors have suggested that the score could be used to classify patients into low-, intermediate-, and high-risk classes, with low-risk patients being managed in the community and intermediate- and high-risk patients being admitted to the hospital. In addition, they suggest that high-risk patients could be targeted with early initiation of steroids and escalation to ICU level care. However, there is no evidence to support any of these applications.

In addition to COVID-19-specific models, there are existing tools, such as the CURB-65 score and Pneumonia Severity Index, that have been validated as decision aids to guide the initial site of treatment for patients with community-acquired pneumonia. However, it is unclear how well these scores apply to patients with COVID-19 or how they should be applied.¹⁵

Clinical practice: 37% (19) of sites have implemented standardized methods of risk assessment to identify COVID-19 patients at high risk for ICU admission, mechanical ventilation, and/or mortality. Interestingly, 68% of these sites utilize a homegrown risk score or checklist, and 47% use a published risk score, such as the 4C Mortality Score. These risk assessments are most commonly applied to determine eligibility for certain treatments (79%) and to determine whether patients are admitted or discharged (53%). Other applications include determination of appropriate level of care, whether in the hospital (observation vs. floor vs. ICU) or outpatient setting (intensity of monitoring for discharged patients). At 2 sites, although risk assessment is a standardized aspect of care, it is not formally incorporated into clinical decision making.

Recommendations / takeaways: The need to efficiently allocate limited health care resources has driven rapid and widespread experimentation, at least among academic medical centers, in the area of risk stratification. Although there are no well-validated tools for this purpose, over one-third of hospitals have already implemented some form of standardized risk assessment, including homegrown tools in over two-thirds of cases. These tools are being used to guide a number of clinical decisions, including ED disposition, triage to different levels of care, and treatment eligibility. Prognostication in COVID-19 is challenging for many reasons, and the accuracy of any method may change rapidly in the face of evolving standards of care, rates of vaccination, viral variants, and other factors. Given the potential for patient harm from misclassification, the efficacy of risk stratification policies should be monitored on an ongoing basis, and the rationale for and processes used to develop such policies should be transparent to frontline clinicians.

Limitations of this study / areas for future study: These findings raise many important and interesting questions. For example, regarding homegrown risk scores, what was the process for development and what types of patient variables are included? In addition, for all assessment methods, what was the rationale and intent (e.g., to prevent unnecessary admissions vs. inappropriate discharges), how were they validated in the local setting and for specific clinical indications, and how are efficacy and potential adverse consequences monitored?

Background / what is known: Remdesivir is a nucleoside analogue that blocks viral replication by inhibiting the SARS-CoV-2 RNA polymerase. In May 2020, the FDA issued an Emergency Use Authorization for the use of remdesivir to treat hospitalized patients with severe COVID-19, and in October 2020, the FDA approved remdesivir for hospitalized patients with COVID-19.

• The evidence: The data on remdesivir's efficacy are nuanced and potentially conflicting. The ACTT-1 trial sponsored by the NIH found that remdesivir shortened the time to recovery for all-comers hospitalized due to COVID-19, and there was a trend toward improved mortality that did not reach statistical significance.¹⁶ Subgroup analysis, however, clarified that this benefit was concentrated among patients on low-flow oxygen — the only disease severity subgroup to experience significant reductions in time to recovery or mortality. In contrast, patients not on oxygen and patients on high flow nasal cannula/noninvasive positive-pressure ventilation (HFNC/NIPPV) experienced small benefits that were not statistically significant, and there was no suggestion of benefit in mechanically vented patients. In additional subgroup analyses, the benefit of remdesivir was also found to be greater when given earlier in the course of illness.In contrast to the ACTT-1 trial, the SOLIDARITY trial sponsored by the World Health Organization (WHO) found no benefit for remdesivir on mortality, progression to mechanical ventilation, or time to discharge.¹⁷ There was, however, a trend toward improved mortality in the subgroup of patients on supplemental oxygen that did not reach statistical significance. Importantly, in light of ACTT-1 findings, patients on low-flow and high-flow oxygen were not separated in the analysis.Altogether, a parsimonious interpretation of the data suggests a modest benefit of remdesivir on time to recovery and mortality in patients on low-flow oxygen decreasing to no benefit for patients on mechanical ventilation. In addition, remdesivir appears to be more beneficial when administered earlier in the disease course. This is compatible with our current understanding of the natural history and pathophysiology of COVID-19, in which an early, mildly symptomatic phase of illness characterized by high viral replication is followed in some patients by a later, more severe phase characterized by excessive inflammation.^18,19
• The guidelines: Just as there are conflicting data on remdesivir's efficacy, so too major guidelines from the IDSA and NIH differ somewhat in their recommendations regarding remdesivir use. The IDSA suggests remdesivir for all patients with severe illness (SpO2 <94%), including those not requiring oxygen, and all patients with critical illness requiring mechanical ventilation or extracorporeal membrane oxygenation (ECMO).² The NIH recommends remdesivir more narrowly: remdesivir is recommended for patients requiring low-flow supplemental oxygen, optionally recommended for patients requiring HFNC or NIPPV, and not recommended for patients requiring mechanical ventilation or ECMO.¹

Summary of practice: Evidence, guidelines, and clinical practices related to remdesivir are summarized in the table below. There is strong consensus in areas where major guidelines from the IDSA and NIH agree, e.g., in patients on low-flow and high-flow supplemental oxygen, despite weak underlying evidence in some scenarios. The greatest practice variation occurs in areas where the IDSA and NIH guidelines are not aligned, e.g., in mechanically ventilated patients and patients with moderate illness at high risk of progression. Finally, a sizeable minority of sites are providing remdesivir to patients with moderate illness — a practice that runs counter to both existing evidence and guidelines.

Evidence gaps and innovation: It is interesting to note that both the IDSA and NIH recommend broader use of remdesivir than is currently supported by empiric evidence, possibly due to expected benefit for patient subgroups that were not explicitly studied. In this evidence gap, a significant minority of institutions are prioritizing patients by using formal risk stratification scores (29%, see section on risk stratification) and/or restricting remdesivir use in certain subgroups to patients in the "early or viral phase of illness" (35%). This latter approach is most commonly used to determine which patients with critical illness or moderate illness at high risk for progression are eligible for remdesivir. "Early or viral phase of illness" is variably defined by time from symptom onset (range: 3-14 days, median: 10) or by a negative antibody test.

Recommendations / takeaways: In spite of a potential mortality benefit in patients on low-flow supplemental oxygen, 1 in 6 respondents are not using remdesivir all or most of the time in this patient population. At the same time, there is high utilization of remdesivir among patients requiring HFNC/NIPPV and, to a lesser extent, mechanical ventilation, despite no or weak evidence of benefit in these patient populations. These trends may reflect multiple competing influences, including limited access to remdesivir and systemic tendencies toward overtreatment. The practice of providing remdesivir to patients with moderate illness not at high risk of progression should be discontinued as it is not evidence-based nor guideline-driven.

Limitations of this survey / areas for future study: Neither the IDSA nor NIH guidelines address timing in relation to symptom onset or phase of illness, although there is some clinical and scientific evidence that remdesivir's benefit is greater earlier in the course of illness. Given that roughly 1 in 3 sites are incorporating phase of illness into treatment decisions for at least some patient subgroups, this is an area that deserves additional study. Patients on low-flow supplemental oxygen are the most likely to benefit from remdesivir; however, due to limitations in hospital capacity, there may be pressure to care for some of these patients in the outpatient setting. Our survey did not ask about programs to provide remdesivir to outpatients, including recently discharged patients. Likewise, we did not ask whether patients who are otherwise stable for discharge are being held in-house in order to complete their courses of remdesivir, although anecdotally this is a common source of inter-provider and inter-site practice variation.

Background / what is known: During the early months of the COVID-19 pandemic, the NIH and IDSA recommended against the routine use of corticosteroids due at least in part to their association with worse outcomes during the SARS epidemic.^20,21 In July 2020, the RECOVERY trial demonstrated that dexamethasone decreased mortality in hospitalized patients with COVID-19.²² Subgroup analysis showed greater benefit in more severely ill patients with a roughly one-third reduction in mortality in mechanically ventilated patients, a roughly one-fifth reduction in mortality in patients on supplemental oxygen (importantly, low-flow and high-flow were not distinguished), and a trend toward increased mortality that did not reach statistical significance in patients not requiring oxygen.

Both the IDSA and NIH recommend dexamethasone use in patients who require mechanical ventilation or ECMO and avoidance of dexamethasone in patients who do not require oxygen.^1,2 For patients on supplemental oxygen, the IDSA recommends dexamethasone use and does not distinguish between those on low-flow and high-flow oxygen; whereas, the NIH recommends dexamethasone use in those who require "increasing amounts of supplemental oxygen," HFNC, or NIPPV.

Summary of practice: Evidence, guidelines, and clinical practices related to dexamethasone are summarized in the table below. There is universal or near universal consensus on use for patients requiring 4-6L NC, HFNC/NIPPV, and mechanical ventilation — areas where both major guidelines and the underlying evidence are in alignment. There is slightly less consensus for patients with severe illness (SpO2 <94%) not requiring oxygen or those requiring only 1-2L NC — areas where the IDSA and NIH guidelines disagree and where the underlying evidence is weak, as these subgroups were not explicitly examined in RECOVERY and were instead grouped into separate categories where dexamethasone had conflicting effects. Finally, a sizeable minority of sites are providing dexamethasone to patients with moderate illness — a practice that runs counter to both existing evidence and guidelines.

• Takeaways: Given evidence of a mortality benefit, it is encouraging how rapidly clinical practice has converged around dexamethasone use in patients with critical illness and severe illness requiring greater than minimal supplemental oxygen. The practice of providing dexamethasone to patients with moderate illness should be discontinued, as it is not evidence-based nor guideline-driven.
• Limitations of this survey / areas for future study: The degree of practice variation between patients with severe illness not requiring oxygen and those requiring minimal supplemental oxygen is striking, especially given the clinical and physiological similarity between these groups. It would be helpful to have additional outcomes data to guide dexamethasone use in these patients. Our survey focused on the care of hospitalized patients, though as discussed above (see admission and discharge criteria), many sites are caring for patients requiring supplemental oxygen in the outpatient setting. It would be interesting to know whether dexamethasone is being prescribed to outpatients and how criteria for use and monitoring differ from the inpatient setting.

Background / what is known: Convalescent plasma is plasma collected from a donor who was previously infected with SARS-CoV-2. It contains a polyclonal mixture of antibodies including variable titers of neutralizing antibodies that are hypothesized to provide passive immunity by blocking the binding of viral particles to their cellular receptors. In August 2020, the FDA issued an Emergency Use Authorization (EUA) for convalescent plasma to be used in the treatment of hospitalized patients with COVID-19. Although the biological rationale for convalescent plasma is clear, randomized clinical trials, including the RECOVERY, CONCOR-1, and REMAP-CAP trials, have failed to demonstrate any benefit in hospitalized patients.^23-25 In a separate study population, high-titer convalescent plasma was shown to significantly reduce the progression to severe disease in older patients who were treated within 72 hours of the onset of mild symptoms, suggesting that plasma quality and early administration may be essential.²⁶ At this time, the IDSA and NIH recommend against the use of convalescent plasma in hospitalized patients, although the NIH does allow that there is currently insufficient to recommend for or against its use in patients with impaired immunity.^1,2

Convalescent plasma: The majority (69%) of institutions have no routine indications for convalescent plasma. A minority consider it during the early phase of illness (20%), for immunocompromised patients who may be unable to mount an effective immune response (14%), and for patients with undetectable or low antibody titers (10%), which may capture both patients in the early phase of illness and immunocompromised patients. In this setting, "early phase of illness" was variably defined as within 72 hours up to 10 days of symptom onset. Infectious Diseases consultation was only routinely obtained at 33% of sites prior to administration of convalescent plasma.

Recommendations / takeaways: The majority of institutions have no routine indications for convalescent plasma in keeping with both evidence and clinical practice guidelines. Even where convalescent plasma is still in use, it seems to be reserved for a limited and targeted patient subset. Obtaining ID consultation prior to use of convalescent plasma may be one mechanism to improve standardization of care.

A note about monoclonal antibodies: Monoclonal antibodies directed against the receptor binding domain of the SARS-CoV-2 spike protein are thought to work through a similar mechanism of action to convalescent plasma, albeit with the advantage of having uniformly high titers of neutralizing antibodies. Given this similarity, it is not unexpected that randomized clinical trials have failed to demonstrate a benefit of anti-SARS-CoV-2 monoclonal antibodies in patients who are hospitalized due to COVID-19.^27,28 The FDA has issued EUAs for the use of monoclonal neutralizing antibodies, including bamlanivimab, bamlanivimab/etesevimab, and casirivimab/imdevimab, for ambulatory patients with mild to moderate COVID-19 at high risk for progression to severe disease and/or hospitalization. The FDA has not authorized the use of these products in patients who are hospitalized due to COVID-19 and, in fact, warns of potentially worse clinical outcomes in patients on high-flow oxygen or mechanical ventilation. The IDSA and NIH both recommend against the use of anti-SARS-CoV-2 monoclonal antibodies among hospitalized patients with severe COVID-19.^2,27 We did not survey hospitals about their use of these therapies, because they have not been authorized for patients who are hospitalized due to COVID-19. However, we have heard anecdotal reports that some sites are providing anti-SARS-CoV-2 monoclonal antibodies to patients with mild to moderate COVID-19 who are hospitalized for other reasons and thus meet EUA use criteria.

Background / what is known: Baricitinib is an inhibitor of Janus kinase, a kinase that mediates signaling downstream of multiple inflammatory cytokines, including IL-6. In November 2020, the FDA issued an Emergency Use Authorization (EUA) for baricitinib in combination with remdesivir for the treatment of hospitalized patients with severe COVID-19 requiring supplemental oxygen, mechanical ventilation, or ECMO. The ACTT-2 trial demonstrated that patients who received this combination were more likely to experience clinical recovery and less likely to progress to mechanical ventilation compared to patients who received remdesivir alone.²⁹ There was also a trend toward improved mortality that did not reach statistical significance in the group that received baricitinib and remdesivir. This trial was conducted prior to the widespread use of dexamethasone, which has been shown to improve mortality. The IDSA and NIH currently recommend the combination of baricitinib and remdesivir only for the rare patient with severe COVID-19 requiring supplemental oxygen in whom corticosteroids are contraindicated.^1,2 They recommend against the use of baricitinib in patients who can receive corticosteroids and against the use of baricitinib in combination with corticosteroids pending ongoing clinical trials.

Clinical practice: The majority (70%) of institutions have no routine indications for baricitinib use. A minority (25%) consider it for patients with severe disease in whom steroids are contraindicated, as recommended by the IDSA and NIH.

Recommendations / takeaways: Although patients with contraindications to short-term corticosteroid therapy are relatively rare, it appears that baricitinib is likely underutilized in this patient population. Contraindications to steroids include previous hypersensitivity and uncontrolled fungal infection. Delirium, psychosis, and hyperglycemia may also be relative contraindications.

A note about other immunomodulators: Many other immunomodulators, including IL-6 antagonists, IL-1 antagonists, and TNF antagonists, have been studied in patients with COVID-19. The NIH and IDSA have both recommended use of tocilizumab, an antibody directed against the IL-6 receptor, in addition to dexamethasone for patients with severe or critical illness who are exhibiting “rapid respiratory decompensation” or “progressive” illness with elevated inflammatory markers, respectively.^1,2 However, the FDA has not issued an EUA for tocilizumab use in patients with COVID-19. Our survey did not include tocilizumab or any other immunomodulators other than corticosteroids and baricitinib.

Background / what is known: COVID-19 is associated with an increased rate of venous thromboembolism (VTE), with estimated prevalence of 22.7% in ICU patients and 7.9% in non-ICU patients in one meta-analysis of 66 studies including 28,173 patients.³⁰ Some studies have also reported increased rates of arterial thrombosis, in particular stroke.³¹ Experts have been divided on the optimal dose of preventive anticoagulation — prophylactic vs. therapeutic vs. intermediate — as well as on which, if any, patients should be selected for intensified dosing.³²

In October 2020, ASH released guidelines "suggesting" prophylactic-intensity anticoagulation over intermediate- or therapeutic-intensity anticoagulation for both ICU and non-ICU patients with COVID-19 in the absence of known clot.³ The Society acknowledged very low certainty of evidence and cautioned clinicians to make "an individualized assessment of the patient's risk of thrombosis and bleeding" when deciding on anticoagulation intensity. The NIH also recommends standard prophylactic dose anticoagulation for hospitalized patients with COVID-19 and states that there are currently insufficient data to recommend for or against higher doses.¹

At the end of January 2021, the results of a planned interim analysis of ATTACC, ACTIV-4a, and REMAP-CAP, a multiplatform randomized controlled trial, were described in a press release.³³ This large trial compared the impact of prophylactic and therapeutic dose anticoagulation on organ support-free days in ICU and non-ICU patients with COVID-19, with non-ICU patients further stratified by D-dimer levels (2x the upper limit of normal). In ICU patients, therapeutic-dose anticoagulation was inferior to prophylactic dosing and met the pre-defined stopping point for futility. In contrast, therapeutic-dose anticoagulation was superior to prophylactic dosing in non-ICU patients, regardless of D-dimer, and appeared to reduce the risk of death or need for organ support by approximately 30%. These results have not yet been published, and ASH has announced that it will update its guidelines after the full data are available for review.

The HOMERuN COVID-19 Prevention and Treatment of Venous Thromboembolism Working Group: A much more in-depth analysis of practices related to VTE diagnosis, prevention, and treatment can be found here.

Therapeutic-dose anticoagulation: The majority (71%) of institutions are restricting therapeutic-dose anticoagulation to patients with confirmed clot, consistent with current ASH and NIH guidelines. The remaining 29% are also treating a subset of patients with very high D-dimer (21%) or critical illness (8%). The most commonly used threshold for D-dimer elevation is >3,000ng/mL.

Limitations of this survey / areas for future study: 87% of our survey responses were obtained prior to the announcement of the ATTACC, ACTIV-4a, and REMAP-CAP results. It will be interesting to see how practice evolves in light of this emerging data. As ICU admission criteria vary from site to site and even within sites during surge conditions, it will be important to better define markers of illness severity that better define which patients benefit from vs. are harmed by therapeutic anticoagulation. Our survey only inquired about indications for therapeutic dose anticoagulation for hospitalized patients with COVID-19 and did not address screening or testing for VTE, post-discharge VTE prophylaxis, or antithrombotic therapy.

Background / what is known: Proning is thought to improve recruitment of alveoli in dependent areas of the lung and to reduce ventilation-perfusion mismatch. A number of small observational studies and case series have suggested that awake proning, also called self-proning, improves oxygenation in non-intubated patients with COVID-19.^34-36 For the most part, these studies have measured short-term changes in oxygenation rather any impact on the need for escalation of care or mechanical ventilation. Larger clinical trials are in progress, but no results have yet become available. Proning is not tolerated by all patients, and the risks include airway obstruction, particularly in patients with altered mental status; dislodgement of oxygen delivery devices; and pressure ulceration and neuropathy.

Awake proning: A majority (65%) of hospitals routinely implement awake proning for at least some non-intubated patients with COVID-19, most commonly in patients on HFNC or NIPPV (57%) followed by patients on 1-6L NC (49%) and then those not on oxygen but at high risk of progression (41%).

Recommendations / takeaways: In stable patients with the physical and cognitive status to supinate themselves, awake or self-proning is a relatively simple, low-risk, and potentially beneficial intervention that may currently be underutilized. In patients with more severe illness, the potential benefits and safety risks are both elevated; therefore, proning should be implemented with a standardized protocol that includes inclusion/exclusion criteria, safety monitoring, and nursing care instructions.

Limitations of the survey / areas for future study: We did not ascertain how proning was operationalized, e.g., as a physician-placed order set, a nursing-driven protocol, patient education, etc. It could be valuable to collect institutional protocols in order to extract best practices for dissemination, as this appears to be a practice that may be underutilized and could have benefit for patients with COVID-19 as well as other causes of hypoxic respiratory failure.

Background / what is known: From the beginning of the pandemic, certain lab findings have been described as diagnostic clues to potential SARS-CoV-2 infection )such as lymphopenia, mild transaminitis, elevated ferritin to procalcitonin ratio) or predictors of disease severity (such as elevated inflammatory markers [e.g., c-reactive protein, erythrocyte sedimentation rate, ferritin, lactate dehydrogenase, IL-6], elevated D-dimer, elevated neutrophil to lymphocyte ratio, elevated troponin, and coagulopathy).^37-40 The CDC, IDSA, and NIH have not issued guidelines related to laboratory testing or monitoring of hospitalized patients with COVID-19.

Admission labs: At a majority of institutions, it is routine to send a relatively large battery of tests at the time of admission including:

• Complete blood count (CBC) with differential (98%)
• Lung function tests (LFTs) (96%)
• D-dimer (84%)
• Coagulation tests, including prothrombin time (PT), partial thromboplastin time (PTT), and fibrinogen (84%)
• C-reactive protein (CRP) (78%)
• Lactate dehydrogenase (LDH) (73%)
• Ferritin (65%)
• Troponin (57%)

In addition, procalcitonin was routinely checked in just shy of half (49%) of institutions. IL-6, which received much attention early in the pandemic as a predictor of severe disease, is sent routinely at 31% of institutions.

Post-admission monitoring: In contrast, only a small subset of labs are routinely trended every 24-48 hours post-admission at a majority of institutions:

• CBC with differential (78%)
• LFTs (59%)
• D-dimer (59%)
• CRP (59%)

Other labs are left to the provider’s discretion

Recommendations / takeaways: Based on these findings, it appears that lymphopenia, D-dimer, and CRP — 3 readily available and quick-resulting tests — may be the most commonly used laboratory markers of disease progression, trended every 24-48 hours at a majority of institutions. Routine trending of LFTs is also common and may reflect monitoring for liver injury in the setting of widespread remdesivir use. In the absence of formal guidelines, it seems reasonable to derive best practices from consensus in this domain.

Background / what is known

Infectious Disease (ID) consultation: ID consultation for management of COVID-19 has many potential benefits, including improved standardization of care and adherence to evidence-based practice, rapid synthesis and application of emerging information, stewardship of supply-limited medications, and accelerated learning from local experience.⁴¹ These benefits were most pronounced early in the pandemic and have diminished as the management of COVID-19 has become more standardized and as hospitalists and other primary team members have gained familiarity with the natural history and care of patients with COVID-19. In addition, the benefits of routine ID consultation must be balanced against the feasibility of scaling this service and costs in terms of clinician burnout and decreased attention to other priorities.

Palliative Care consultation: Much has been written about the importance of Palliative Care during the COVID-19 pandemic. Patients with COVID-19 have many symptoms, including dyspnea, anxiety, and loneliness. Those with severe illness face difficult decisions given the frequently prolonged course of illness with uncertain or poor prognosis. These needs are frequently compounded by visitor restrictions and overburdened clinicians. In this setting, Palliative Care teams can be instrumental in improving both patient and provider satisfaction and aligning care with goals and prognosis. However, available Palliative Care resources vary widely from institution to institution, and this time-intensive service is potentially more difficult to scale and convert to virtual models of care.^42-44

ID consultation: A majority (59%) of institutions have created COVID-19-focused ID consult teams. ID consults for the management of COVID-19 are available as both in-person consults (79% during non-surge, 69% during surge) and virtual e- or tele-consults (66% during non-surge, 75% during surge) at a majority of institutions. Indications for ID consultation varied widely across institutions, with 22% of sites involving ID for all patients with COVID-19 and 27% having no routine indications for ID consultation. Interestingly, utilization of ID consultations did not appear to correlate with the magnitude of the COVID-19 surge. ID consultation for therapeutic management varied by agent with ID being involved for initiation of dexamethasone at 25% of sites, for remdesivir at 40% of sites, and for immunomodulatory agents other than corticosteroids at 55% of sites. The relatively low rate of ID consultation for dexamethasone may reflect greater hospitalist comfort with this medication and/or clearer evidence and indications for use.

Palliative Care consultation: A sizeable minority (31%) of institutions have created COVID-19-focused Palliative Care consult teams. Palliative Care is routinely consulted for all patients with critical illness due to COVID-19 at 29% of sites and for all patients with risk factors for severe disease at 23% of sites. A single site recommended Palliative Care consultation for all patients with COVID-19, bandwidth-permitting. The majority of sites (69%) have no routine indications for Palliative Care involvement.

Takeaways / Areas for future study:

• ID consultation: Most sites developed COVID-19-focused ID consult teams that were operationalized as virtual or tele-consults. Although this was motivated by unique pressures created by the pandemic, this delivery model may be generalizable to other common inpatient scenarios. There was surprising variation in the indications for ID consultation with similar proportions of sites involving ID in all COVID-19 admissions and having no routine indications for ID consultation. It would be valuable to study whether ID consultation improved standardization of care, concordance with evidence/guidelines, and patient outcomes, as has been shown for other infections.
• Palliative Care consultation: Palliative Care was routinely involved in the care of sicker patients, whether already critically ill (1 in 3 hospitals) or at high risk factors for severe disease (1 in 4). On the one hand, this is a relatively low proportion of hospitals, given the general consensus on the importance of early goals of care discussions in the setting of COVID-19. On the other hand, this is also an impressively high proportion of hospitals, given the large numbers of patients involved and the difficulty scaling Palliative Care compared to other services. It would be interesting to study the impact of Palliative Care consultation on ICU admission, length of stay, mortality, and patient and provider satisfaction.

Background / what is known: The need for isolation precautions can have a substantial impact on a patient’s clinical care (e.g., delaying tests or procedures), wellbeing (e.g., more strict visitor restrictions), and discharge disposition as well as on the real and perceived safety of other patients and healthcare workers. The CDC has issued guidance for symptom-based discontinuation of precautions that applies to the majority of patients with COVID-19 who are not severely immunocompromised.⁴⁵ For patients with mild to moderate COVID-19 (SpO2 >94% on room air), the CDC recommends discontinuing precautions 10 days after symptom onset as long as symptoms have improved and the patient has been fever-free for 24 hours without the use of antipyretics. For patients with severe to critical COVID-19 (SpO2 <94%, RR >30, or lung infiltrates >50%), the CDC recommends discontinuing precautions when at least 10 days and up to 20 days have passed from symptom onset as long as the same conditions of clinical improvement have been met. Patients with “severe immunocompromise” pose a distinct challenge, as they may continue to shed replication-competent virus well beyond 20 days from symptom onset. In these cases, the CDC recommends consultation with ID specialists and consideration of a test-based strategy for discontinuing precautions.

Moderate illness: The majority (69%) of sites are following CDC guidelines and de-escalating isolation precautions 10 days after symptom onset. The remaining are more conservative than CDC guidelines and continue precautions for 14-20 days from symptom onset (15%), until discharge (12%), or until NAATs are negative (4%, note: this is against CDC guidance).

Severe illness: The majority of sites are following either liberal CDC guidelines and discontinuing precautions 10 days after symptom onset (14%) or conservative CDC guidelines and discontinuing precautions 20 days after symptom onset (67%). The remaining are more conservative than CDC guidelines and continue precautions until discharge (8%) or until NAATs are negative (10%, note: this is against CDC guidelines).

Recommendations / takeaways: A sizeable minority of institutions continue isolation precautions longer than currently recommended by the CDC, resulting in unnecessary PPE use and potentially negative impacts on patient care and wellbeing. Such practices should be reconsidered in light of scientific data on the natural history of SARS-CoV-2 transmission and rising rates of health care worker vaccination.

Limitations of this survey / areas for future study: We did not specifically address practices related to patients who are severely immunocompromised or how the degree of immunosuppression is being defined operationally. Given case studies that suggest such individuals may be an important source of viral variants, including those capable of immune escape, this may be an area where additional guidance is valuable.