MGH FLARE - June 27 - The RECOVERY Trial
Les interrogations et réserves (mais pas le rejet) du « Massachusetts General Hospital » concernant l’étude #RECOVERY ;
Étude non encore acceptée par les pairs, ouverte ; la mortalité du groupe contrôle se situe à l’extrémité maximale de la fourchette habituellement observée (ce qui soulève l’hypothèse de « facteurs de confusion ») ; résultats donnés après un court suivi (ne tenant pas compte des effets secondaires potentiels à plus long terme du corticostéroïde utilisé).
As the data from RECOVERY are currently limited to a pre-print, there remain other important questions left yet unanswered. The mortality in the untreated patients on mechanical ventilation (40%) is at the higher end of the reported range of mortality in COVID-19 cohorts, with a lower end of the range in the high teens (Ziehr et al. 2020). This mortality rate does not invalidate the benefit of steroids, but raises the possibility of confounders if the control group differed from reference populations in important ways for which we don’t yet have data. For example, in this open label trial, concomitant medications may have been differentially given to the control ventilated patients whose providers knew they were not getting steroids. The authors state only that patients did not receive significant treatment with medications being tested in other arms of the trial. However, if other, potentially harmful medications were given more often in the control group this could lead to confounding. This potential confounder is made more plausible by the relatively high control group mortality.
The high mortality observed in the trial also means that the number needed to treat is likely higher at other centers. Again, this does not invalidate the results but, given the potential risk of long term sequelae in survivors treated with steroids , it is worth noting. The current preprint reports outcomes at 28 days; it therefore remains possible that late harm associated with steroids would manifest at 60 or 90 day follow-up. Lastly, the trial is open label, which means that other aspects of care could have been unconsciously influenced by knowledge of the trial intervention, including anticipation of complications of critical illness, increased attention to the patient receiving dexamethasone given predictable effect of steroids such as hyperglycemia, and decisions around duration of life support.
Overall, the data do seem generalizable. However, the large number of patients who died without being intubated, the relatively high mortality in the control group (as mentioned above), and the ability of attending physicians to opt their patients out of the trial if they felt they were likely to be harmed by steroids may all impact the ability to extrapolate the data to another healthcare setting or patient cohort.