A Call for Rational Intensive Care in the Era of COVID-19.

To the Editor:

As intensive care physicians, we have been trained to treat viral pneumonia and its attendant complications of acute respiratory distress syndrome (ARDS) and multiorgan failure. The coronavirus disease (COVID-19) pandemic has challenged our profession to revisit its paradigms. Specifically, do mechanical ventilation strategies optimized in ARDS trials still apply to this disease? Is our policy of waiting for proof of benefit before instituting novel therapeutics still sensible? In this commentary, we make the case that the ICU is already optimized for the care of patients with COVID-19 and that departures from our standard of care require evidence, not vice versa.

We have learned from decades of critical care research and experience that protocol-driven, physiologically based management strategies result in improved patient outcomes, particularly for ARDS (1). The Berlin Definition established criteria for ARDS based on its acute clinical presentation in the presence of hypoxemia and radiographic pulmonary edema not arising entirely from hydrostatic mechanisms (2). We, along with other intensivists, have observed that some patients with COVID-19–induced ARDS exhibit higher than expected lung compliance that seems out of proportion to the degree of shunt physiology. Importantly, although experience has shown that stiff lungs are a common finding in patients with ARDS in general, measures of static respiratory system compliance are not included in the Berlin Definition. ARDS is a syndrome, not a disease, and is heterogeneous by its nature. Regardless, findings in COVID-19 have led some to believe that COVID-19–related respiratory failure is an ARDS variant (3). A worrisome corollary of this belief is that the accumulated database of proven ARDS management strategies (e.g., intubation and low-tidal-volume ventilation, prone positioning, and surveillance for nosocomial infections) can be disregarded. In fact, the patients enrolled in the ARMA (Ventilation with Lower Tidal Volumes as Compared with Traditional Tidal Volumes for Acute Lung Injury and the Acute Respiratory Distress) trial of low-tidal-volume ventilation and the PROSEVA (Proning Severe ARDS Patients) trial of prone positioning exhibited myriad etiologies, compliances, and shunt fractions but nevertheless benefited from the targeted interventions (4, 5). We should not deny the benefits proven by rigorous randomized controlled trials to our patients with COVID-19.

Biological plausibility is insufficient justification to administer a medication to a critically ill patient outside of a clinical trial. Indeed, our specialty’s history is littered with examples of agents that carried a strong mechanistic rationale and even positive in vitro signals yet failed or were shown to be harmful in clinical trials, such as surfactants, N-acetylcysteine, statins, and β-agonists, to name a few in ARDS alone (6). Currently, numerous agents are being administered to patients with COVID-19 outside of controlled trials, including hydroxychloroquine, azithromycin, doxycycline, remdesivir, lopinavir-ritonavir, heparin, low-molecular-weight heparin, tissue plasminogen activator, glucocorticoids, tocilizumab, eculizumab, IFN-β, IFN-γ, IL-1 inhibitors, mesenchymal stem cells, convalescent plasma, nitric oxide, vitamin C, and others. We do not suggest that physicians never use unproven medications off-label or off-trial; in the ICU, we frequently must give therapies based on strong signals in disease processes that are similar to the one in front of us. In contrast, the routine use of the agents listed above for COVID- 19—outside of controlled trials—strains credulity. Many of these compounds have failed in trials of viral infection and ARDS. Continued use of lopinavir-ritonavir is even more shocking in light of a negative randomized controlled trial in COVID-19 that was published early in the pandemic (7).

Why are physicians abandoning standards of critical care in the era of COVID-19? Emotion, stress, fatigue, and political proclamations amplify our innate desire to help our patients and try something—anything—that might provide benefit and give hope to providers and patients alike. This data-free approach will ultimately harm more patients than it helps, as one-off administration of medications ruins clinical equipoise about their use. When a medication is administered to a patient who then improves, the natural human bias is to believe that the drug caused the improvement. Nevertheless, if the patient succumbs to the disease, our biases do not confirm the counterfactual logic. Instead, we believe that the disease was too severe for the drug to overcome, while we minimize the possibility that the drug was ineffective or toxic. The only known strategy to overcome these biases lies in the scientific method and the application of controlled trials to determine whether an agent is effective and the degree to which it is harmful. The possibility of persistent COVID-19 is real, and the emergence of new viral pandemics in the future is certain. For our patients’ sake, we need to know what works and what does not. The straw man argument—that patients with COVID-19 improve with protocol-driven supportive care—needs to serve as a null hypothesis to be rejected or accepted in controlled trials. To act as if we know otherwise is irrational, hubristic, and reckless. Pending data from ongoing clinical trials, we must resist the innate human desire to act on emotion and instead rely on our creed: first, do no harm.