Not yet published but the results of a new randomized placebo controlled trial at Oxford University showed that 2,000 hospital patients on ventilators given the glucocorticoid dexamethasone had death rates of 28% compared to the 40% death rates of people receiving standard of care. This is the first drug shown to reduce mortality in a proper prospective clinical trial and it is a quite significant effect.
It is cheap, available everywhere, and it works through well known anti-inflammatory effects to reduce cytokine release in the most severely ill people. It had no obvious beneficial effect at earlier stages (see further below).
What could be confusing to some people is that there had been publicized warnings that people who were taking this class of drugs (glucocorticoids) on a more chronic basis (e.g. for asthma, COPD, rheumatoid arthritis, inflammatory bowel disease, Cushing's disease, some kidney diseases, etc. etc.) might be at higher risk because these drugs are immunosuppresive. At the time those warnings were NOT accompanied by any actual data and it could be hard to tease out the effect of the drug from the co-morbidities that are the reason people are taking the drug. In fact, one systematic review that has looked at this fairly closely in patients with respiratory diseases found that in terms of the Coivd-19 pandemic "there is no evidence to support the withdrawal of inhaled corticosteroids in patients [with asthma or COPD] , and to do so is likely to be harmful. Patients with asthma and COPD who are stable while using ICS should continue on their treatment". In other words, there are lots of things that in theory seem like they ought to be true (the idea that chloroquine should be useful in Covid-19 was an example, warnings about corticosteroids and ACE inhibitors were another); but at the end of the day you actually have to do carefully controlled clinical trials, or at the very least have some data from experiments in animals to go on.
In any case the situation in later stages of Covid-19 is quite different than at the beginning because what kills a lot of people is the overwhelming excessive response to the infection (which leads to multi-organ failure) and not the direct effect of the virus per se. It is widely believed that this disease passes through about three different stages (although people with mild disease really only go through the first).
There were earlier hints at this result in the literature, including a systematic review from Spain that concluded that methylprednisolone (another drug in the same class) could lower mortality rate in more severe forms of the condition.
For people with more technical backgrounds or any physicians here, here is an earlier discussion of the complexities involved (before this controlled trial). It argues that what you should do for these patients depends on the stage of the disease, defines the three stages, and pretty much predicts the results reported by the Oxford trial.
https://emcrit.org/pulmcrit/steroid-covid/