Steroids in COVID19: The Pendulum Swings Back

(This article appeared on CriticalCareNow on 9/11/2020)

The benefit of corticosteroids in the critically ill is a perennial subject of debate, and over the last two decades the pendulum has swung several times. It should come as no surprise that this debate has been reinvigorated by COVID19.

In the early days of the pandemic it was uncertain if steroids would be beneficial of harmful. On one hand, emerging evidence suggests that corticosteroids may be beneficial in ARDS. One recent high quality RCT, the DEXA-ARDS study, demonstrated more ventilator free days and decreased mortality in patients with ARDS who were treated with dexamethasone vs. usual care. One the other hand, studies of non-COVID coronavirused (SARS and MERS) have failed to show benefit. One study of corticosteroids in MERS found prolonged viral shedding, increased likelihood of mechanical ventilation, and higher mortality. Given this uncertainty, there was clear equipoise for RCTs.

Then, in June, the UK RECOVERY trial was published. This large, open label, multi-arm RCT of hospitalized patients with COVID19 found a significant decrease in 28-day mortality from 25.7% in the control group to 22.9% among patients treated with a relatively low dose of 6 mg/day dexamethasone. Dexamethasone also reduced the likelihood of requiring invasive mechanical ventilation (IMV) by 2.1% and shortened hospital LOS by 1 day. The benefit of dexamethasone was particularly striking among patients who required IMV, where it reduced 28-day mortality from 41.4% to 29.3%, signifying a NNT of 8. There are very few other interventions of comparable efficacy in the ICU. Interestingly, while this study found benefit for dexamethasone among patients with hypoxemia, it found an apparent increase in mortality among patients not on supplemental O2. 

This large well-designed study was immediately practice changing, but it left some questions unanswered: would a higher doses of steroid be more beneficial? Are these effects generalizable to other steroids?

Fortunately several other studies of steroids in COVID19 were already in progress. Last week, THREE (CAPE-COVID, CODEX, RE-MAP-CAP) were published in JAMA one day, which, combined with a few smaller studies, brought the total up to 7. An accompanying meta-analysis provides a fuller picture of the impact of steroids in COVID.  These studies differed in choice of steroid (hydrocortisone, dexamethasone), dose, population enrolled (ICU vs. wards, intubated vs. not), and primary endpoint (mortality, duration of IMV, hospital LOS, etc.) All of these studies were methodologically sound, though several were stopped early due to limited enrollment after RECOVERY was published. Only the RECOVERY study showed a statistically significant mortality reduction with steroids, though REMAP found a reduced likelihood of organ failure, and CODEX found a reduction in time on IMV by 2.2 days. When we put these studies together, however, we see a clear and consistent effect.

Study	Country	Population	Steroid/daily dose	Planned N	Actual N
RECOVERY	UK	Hospitalized patients	Dexamethasone 6 mg	1071	1007
DEXA-COVID	Spain	ICU patients on IMV	Dexamethasone 20 mg	200	19
STEROID-SARI	China	ICU patients on HFNC, NIPPV, or IMV	Methylprednisone 80 mg	80	47
COVID-STEROID	Denmark	ICU patients on HFNC, CPAP, or IMV	Hydrocortisone 200 mg	1000	29
CAPE-COVID	France	ICU patients on supplemental O2 who were not septic	Hydrocortisone 200 mg	300	149
CODEX	Brazil	ICU patients on IMV	Dexamethasone 20 mg	350	299
RE-MAP CAP	Multiple	ICU patients on NIPPV/IMV or on vasopressors	Hydrocortisone 200 mg	7100	197

I did a quick meta-analysis of these studies using a random effects model. The heterogeneity in effect sizes, measured by I^2, is low at at only 15%. Looking at the Forest plot 5/7 studies show a trend towards benefit, while the two that do not are quite small, with less than 30 patients each. We can see that the benefit of steroids appears robust to country, hospital, and concomitant medications. Furthermore, we see benefit regardless of the specific corticosteroid or the dose used.

View fullsize

Forest plot of the seven RCTs examining corticosteroids for hospitalized patients with COVID19.

In the excellent meta-analysis published this week in JAMA, they explore some additional subgroups. They also found that most patient groups seemed to benefit: corticosteroids were effective in men and women, in older and younger patients, and in those on IMV and supplemental O2.  In contrast to earlier studies, they found that there was benefit to corticosteroids both when patients were treated within a week of symptom onset and when they were treated later.

Summarizing what we now know about corticosteroids in patients with COVID19:

• Corticosteroid treatment is associated with numerous benefits: less time in hospital, lower likelihood of requiring mechanical ventilation, and decreased mortality.

• All hospitalized COVID19 patients who are hypoxemic enough to require supplemental O2 seem to benefit from steroids. As there is a trend towards increased mortality in patients who are not on oxygen, hospitalized COVID patients not on supplemental oxygen should not receive corticosteroids.

• The choice of steroid does not seem to matter, and there is no obvious advantage to higher doses. It is therefore reasonable to use the lowest effective dose:

  • The NIH COVID19 Treatment Guidelines and the WHO Guidelines both recommend Dexamethasone 6mg IV or PO daily for 10 days or until discharge (whichever comes first).

  • If Dexamethasone is not available you can substitute other corticosteroids: Prednisone 40 mg/day, Methylprednisone 32 mg/day, or Hydrocortisone 160 mg/day.

• There does not appear to be a right or wrong time to give steroids; patients appear to benefit if corticosteroids are started promptly or >1 week after symptom onset.

It’s not clear if we can extrapolate these lessons about COVID to all patients with ARDS or septic shock. The steroid pendulum will likely continue to swing for years to come. Many questions persist about COVID: should these patients be anticoagulated? Is there benefit to targeted immunomodulation? But nine months into the pandemic, at least a few questions have been put to rest: Hydroxychloroquine is not beneficial, and steroids are.

I also delivered this content as a lecture. You can watch a recording here or download the slides below: