Dear Editor,

The RECOVERY study recently reported significant mortality improvement from dexamethasone in hospitalized patients affected by Coronavirus Disease 2019 (COVID-19) [1]. The impressive 35% reduction in the relative risk of mortality achieved in the ventilated subset of patients was surprising given that such a clear signal has not been seen in previous extensive research either into viral pneumonia or acute respiratory distress syndrome (ARDS). This prompted us to examine past randomised controlled trials (RCTs) in ARDS more closely in terms of impact of the type of glucocorticoid, dosing and commencement of treatment on mortality.

Previous studies of glucocorticoid use in viral influenza have lacked rigorous RCT interrogation. An association was however suggested between steroid use and an increased risk of mortality [2]. This may relate to impaired production of type I interferons, the first line of defence against severe viral respiratory infections [3]. The use of steroids in previous coronavirus outbreaks (SARS and MERS) was overall associated with more complications and delayed viral clearance [4, 5]. Expert opinion and clinical trials support the use of corticosteroids in severe community-acquired pneumonia [6, 7]. However, no clear outcome benefit has been shown for steroid use in ARDS despite decades of research (Table 1) (Search methodology available in Online Resource). Accordingly, the World Health Organisation initially advised against routine use of corticosteroids in COVID-19 disease outside a clinical trial [8] and the Surviving Sepsis Campaign gave a weak recommendation with some of the experts deferring judgment until higher quality direct evidence was available [9].

Table 1 Randomised controlled trials of steroids in ARDS
Full size table

At first glance, it is difficult to reconcile how steroids, with no clear evidence in ARDS and potential harm in viral pneumonitis, can be associated with such a mortality benefit in COVID-19 associated severe respiratory failure in RECOVERY. Methodological queries will doubtless be raised as this was an open-label study with withdrawals and crossovers and an unspecified treatment duration. While an impressive outcome was achieved in the sicker cohorts, it is difficult to explain why benefit was only seen in male patients under 70 years old; a signal-to-harm was seen in a sizeable proportion of patients who were not receiving oxygen at the time of enrolment; and a marked outcome difference was seen using a 7-day cut-off from symptom onset, especially as few would have been hospitalised within the first 5 days.

The RECOVERY study may have shed some light on the different results achieved in the various ARDS RCTs where low-dose dexamethasone appears to provide the greatest benefit. Of all glucocorticoids, only dexamethasone lacks any mineralocorticoid activity. The role of mineralocorticoid activity in the progression of pulmonary hypertension is suggested by pre-clinical data [10]. The lower dose regimen may strike the right balance between anti-inflammatory and immunosuppressive effects. Indeed, higher doses of steroids (methylprednisolone) were associated with harm, and this was most evident in early sepsis studies [11]. The benefit from later commencement in COVID-19 disease may relate to a beneficial effect on the resolution of lung injury with less clinical impact from any delay in viral clearance after 7 days. The timing of viral clearance with COVID-19 is shorter than SARS or MERS [12].

While the results of the RECOVERY cannot be extrapolated beyond COVID-19, and does warrant further scrutiny, the use of early low-dose dexamethasone in ARDS merits further attention.