Early in the COVID epidemic, it became clear that the virus often causes a cytokine storm, leading to immunopathological organ damage. The first major study to suggest benefit from steroid in COVID was a retrospective study by Wang et al. released in early March, which found reduced mortality among COVID patients with ARDS who were treated with steroid.1 Several studies followed which generally supported the use of steroid in COVID, including a multi-center pre/post study at the Henry Ford Health System (previously reviewed here and here and here). Combined with prior evidence supporting the use of steroid in pneumonia and ARDS, this constituted a reasonable basis for the use of steroid in COVID. Consequently, many hospitals have been using steroid routinely for hypoxemic COVID patients for several months already.
This is a massive 176-center, randomized, open-label, pragmatic trial comparing several treatments to usual care for patients hospitalized with COVID-19. Recently, preliminary results were reported for the dexamethasone group (treated with 6 mg dexamethasone daily for up to 10 days).
Patients were randomized to dexamethasone or placebo in a 2:1 ratio. Baseline characteristics were well matched between the two groups:
The results were strongly positive, including the primary endpoint (improved 28-day mortality) and secondary endpoints (e.g., reduced need for intubation and reduced hospital length of stay):
The drug’s benefit was restricted to patients with acute hypoxemic respiratory failure (intubated patients, or patients requiring oxygen):
Although this is a subgroup analysis, this effect was already predictable ahead of time, based on the biology of the virus. Early in the course of infection, immune activation may be helpful to control the virus. Later on, as hypoxemia is developing, excess inflammation becomes harmful. Thus, the ideal timing for steroid initiation is probably after the patient begins requiring oxygen:
In a very crude attempt to kick the statistical tires of this study, I’ve re-calculated the statistics using an online calculator. They seem quite robust (table below). There is almost a statistically significant signal for harm among patients not receiving oxygen, which supports that this group shouldn’t be treated with dexamethasone:
The study does have some weaknesses, which are largely those of any massive, pragmatic study performed during a pandemic.
- Lack of granular data about patients (e.g., laboratory values, viral load). This lack of detail is a necessary sacrifice required to recruit over 6,000 patients.
- Open-label design (since designing matched placebos during a pandemic wasn’t feasible). This could introduce some systemic bias (e.g., imagine if patients not receiving dexamethasone were intentionally treated with tocilizumab). However, data on co-interventions suggest that this wasn’t the case:
but… the results are too good to be true ?!?
In critical care medicine, we’re not used to seeing such positive studies. For example, no drug has ever been proven to improve mortality in critical care (discussed here and here). So the RECOVERY trial results may seem too good to be true.
The explanation here is fairly simple. When is the last time a critical care RCT was performed on 6,000 patients with the same exact pathogen? Never. We’ve never even come close. One major limitation of critical care trials is that our patient populations are usually horrifically heterogeneous (e.g., patients with “sepsis” or “ARDS” – whatever those syndromes are supposed to be).
This pandemic provided a unique opportunity to perform a massive trial investigating thousands of patients infected with a single pathogen. It’s likely that dexamethasone could work just as well against some other pathogens as well (e.g., imagine 6,000 patients infected with Streptococcus pneumoniae). In the past, we simply haven’t been able to perform this study.
why dexamethasone? Is anything special about dexamethasone?
As a connoisseur of fine corticosteroids, dexamethasone has a special place in my heart. It has some uniquely useful properties:
- Long biological half-life (~36-48 hours). This a very convenient property because it allows dexamethasone to auto-taper itself gradually (thereby potentially avoiding rebound inflammation).
- Dexamethasone has little mineralocorticoid activity, which is potentially beneficial for a few reasons. Mineralocorticoid stimulation may promote fluid retention and hypernatremia (which are especially undesirable in patients with ARDS). Additionally, there are some weak hints in the literature that mineralocorticoid stimulation might conceivably be harmful in ARDS.2,3 As a pure glucocorticoid agonist, dexamethasone is one of the “cleaner” steroids mechanistically.
- Dexamethasone has superior penetration of the central nervous system compared to some other steroids. This is a desirable property among patients with hemophagocytic lymphohistiocytosis (HLH), but it’s unclear whether it is coming into play here.
what happens when dexamethasone runs out?
The paradox of pandemic medicine is that if a medicine is actually found to be beneficial, its supply may be immediately exhausted. Reasonable substitutions would include equivalent doses of methylprednisolone (32 mg), prednisone (40 mg), or prednisolone (40 mg). If using one of these agents, a very short taper at the end of the steroid course might be considered to mimic the prolonged half-life of dexamethasone.
Oral betamethasone sodium phosphate could be another option. Betamethasone is an attractive option because it has nearly identical properties as dexamethasone (long biological half-life and lack of mineralocorticoid activity). Betamethasone and dexamethasone have the same exact chemical formula, differing only by the chirality of a single methyl group (figure above). Most parenteral formulations of betamethasone contain a 50-50 mixture of betamethasone sodium phosphate (immediate-acting) and betamethasone acetate (which takes longer to absorb) – making this less desirable than the oral form.
the American medical establishment has failed
In a prior blog, I lamented our failure to perform an RCT on steroid in COVID-19:
Fifty years ago we put a man on the moon, but today it seems like an insurmountable challenge to organize an open-label randomized trial of prednisone.
It’s fortunate that the United Kingdom was able to achieve this much-needed research. However, the success of the United Kingdom highlights the failure of the United States. The United Kingdom has recorded far fewer numbers of COVID-19 patients than the United States (which has now seen over two million infections!). Thus, a well-organized national trial of dexamethasone in the United States could have completed its recruitment in a few weeks, finished months ago, and saved thousands of lives already.
Some useful trials are being performed in the United States, but overall it seems that our research infrastructure and IRB requirements are preventing coordinated efforts similar to the RECOVERY trial. This is particularly regrettable, as we are continuing to see ~25,000 new cases daily in the United States (representing an enormous opportunity for MC-RCTs). To date, trial output from the United States has focused largely on overhyped anti-viral drugs, rather than more practical interventions (e.g., steroid or heparin).
early versus late adoption in a pandemic
The tension between early-adoption versus late-adoption is a perpetual debate that will never be resolved (because it has more to do with the art of medicine than scientific facts). For more on this fundamental rift in medicine, see a prior blog about the battle between the ivory tower academics versus the rogue cowboys.
Recently, there has been a spate of articles bemoaning how clinicians have adopted treatments for COVID-19 which hadn’t been subjected to large RCTs. It has almost become a sign of academic virtue-signaling to disparage any therapy which isn’t supported by irrefutable evidence.
Exercising caution in a pandemic is certainly wise. However, being overly cautious could lead to delayed adoption of beneficial therapies. While many academics were eagerly publishing articles about eschewing anything not supported by large RCTs, other clinicians were quietly using steroid to save lives. This should remind us that there is no easy answer to the early-vs-late adoption debate. Erring too far in either extreme is potentially detrimental.
- Dexamethasone is now the standard of care for COVID-19 patients with acute hypoxemic respiratory failure (i.e., requiring oxygen or mechanical ventilation).
- Available evidence does not support the use of dexamethasone in COVID-19 patients who are not requiring oxygen. Using dexamethasone too early is potentially harmful.
- Optimal management of patients with COVID-19 might involve close monitoring of oxygenation (e.g., with home pulse oximetry). Dexamethasone should be initiated if patients become hypoxemic and require supplemental oxygen.
- When dexamethasone supplies are exhausted, equivalent doses of other steroid may be used (e.g., methylprednisolone).
- RECOVERY trial
- Study pre-print here (they even posted the supplemental appendix!).
- PulmCrit stuff on steroid in COVID
- 1.Wu C, Chen X, Cai Y, et al. Risk Factors Associated With Acute Respiratory Distress Syndrome and Death in Patients With Coronavirus Disease 2019 Pneumonia in Wuhan, China. JAMA Intern Med. Published online March 13, 2020. doi:10.1001/jamainternmed.2020.0994
- 2.Ruthman CA, Festic E. Emerging therapies for the prevention of acute respiratory distress syndrome. Therapeutic Advances in Respiratory. Published online May 22, 2015:173-187. doi:10.1177/1753465815585716
- 3.Atalay C, Dogan N, Aykan S, Gundogdu C, Keles M. The efficacy of spironolactone in the treatment of acute respiratory distress syndrome-induced rats. Singapore Med J. 2010;51(6):501-505. https://www.ncbi.nlm.nih.gov/pubmed/20658111
- PulmCrit – I’m so confused about bamlanivimab - November 30, 2020
- PulmCrit – Oscar awards for the best COVID prognostic models - November 28, 2020
- PulmCrit – Heparin resistance in COVID & implications for DVT prophylaxis - November 24, 2020