“Evidence-based medicine requires evidence. It’s not optional.”—these golden words in science- and evidence-based medicine were re-emphasized by Dr. Angie Rasmussen, virologist at Columnbia University and prolific science communicator, on Twitter recently.

For proper appreciation of the magnitude of this quote, let me elaborate on the fascinating context in which Rasmussen wrote it.

In a world where global societies are suffering under the relentless onslaught of a deadly pandemic, and people are desperate for scientists to find a cure, the morning of June 16 brought tremendous news: “Cheap drug is first shown to improve COVID-19 survival.” wrote the Associated Press. And just like that, we were suddenly apprised of the ‘first life-saving coronavirus drug’—a title reserved in the recent past for hydroxychloroquine, the malaria/lupus/arthritis drug.

Glory Be! What empyrean elixir granting eternal health was this?

Turns out, this was the humble corticosteroid, dexamethasone, the same class of drug that is in current clinical use for various type of inflammatory, autoimmune, and allergic conditions in both humans and animals. Dexamethasone was suddenly thrust into spotlight, receiving glowing testimonials in the media (including social media) about its newfound wonder-drug status for the treatment of COVID-19, the severe respiratory disease caused by SARS-CoV-2, the novel coronavirus of 2019.

How did this revelation come about? How was the drug studied? How was it used and in what group of patients? Most crucially, how reliable was the data, and had others had the same clinical experience?

All excellent and necessary questions, especially given the recent debacle around hydroxychloroquine, which demonstrated some antiviral activity in the laboratory at very high concentrations—concentrations not safely achievable in the human body—and was immediately hyped (including by a certain citrous political leader who shall remain nameless) as the cure for COVID-19, but whose claims withered away under the glare of rigorously conducted clinical trials.

Strangely, we did not know the answers to most of these questions re dexamethasone when the hype started mid-June. Michelle Roberts, the health editor of BBC, breathlessly reported on the results freshly released to The Beeb by the UK clinical trial which had studied dexamethasone—the RECOVERY trial (site accessed June 17, 2020); they had, however, not yet published any of the actual data beyond some general information at their site, and their observations had not been validated or reproduced by any other peer group. In the world of science communication, this phenomenon is referred to as Science by Press Release, wherein scientific observations are presented as facts by the group which has made them, with no independent fact-check conducted at that point.

This is how the trial site described the data:

A total of 2104 patients were randomised to receive dexamethasone 6 mg once per day (either by mouth or by intravenous injection) for ten days and were compared with 4321 patients randomised to usual care alone. Among the patients who received usual care alone, 28-day mortality was highest in those who required ventilation (41%), intermediate in those patients who required oxygen only (25%), and lowest among those who did not require any respiratory intervention (13%).

Dexamethasone reduced deaths by one-third in ventilated patients (rate ratio 0.65 [95% confidence interval 0.48 to 0.88]; p=0.0003) and by one fifth in other patients receiving oxygen only (0.80 [0.67 to 0.96]; p=0.0021). There was no benefit among those patients who did not require respiratory support (1.22 [0.86 to 1.75]; p=0.14).

Overall dexamethasone reduced the 28-day mortality rate by 17% (0.83 [0.74 to 0.92]; P=0.0007) with a highly significant trend showing greatest benefit among those patients requiring ventilation (test for trend p<0.001). But it is important to recognise that we found no evidence of benefit for patients who did not require oxygen and we did not study patients outside the hospital setting. Follow-up is complete for over 94% of participants.

In plainspeak, this is what it all meant:

  • For all people getting the currently recommended treatments, those who required ventilation (i.e. those who are critically ill) died in the greatest numbers within 28 days of symptoms. According to their numbers, nearly half (41%) of COVID-19 patients needing a ventilator and one fourth (25%) needing supplemental oxygen died. (NOTE: This is something we knew already from the clinical experience gathered so far.)
  • Addition of low dose dexamethasone to the current treatments appeared to reduce the risk of death in 1 of every 3 patients on ventilators, and 1 in every 5 patients requiring oxygen (which means, they were still severely ill, but less critically).
  • In terms of raw numbers, this meant the risk of death was reduced from 41% to 27% in ventilated patients, and from 25% to 20% in patients getting oxygen. Taken together, it represented an overall reduction of approximately 17% in the risk of death. (This means, where 100 patients could have died, now only 83 may die.)
  • And finally, dexamethasone did nothing for those who did not require external respiratory support (which means, they were not as critically or severely ill).

However, this bare minimum description did not provide any additional information (which is natural, since it was a Press Release) on patient demographics, especially their conditions at entry, the staging of their COVID-19, their baseline and post-treatment values for various functional and inflammatory biomarkers (important since COVID-19 has an intensely inflammatory phase), and so forth. Without this information, it is rather difficult to properly evaluate the claims of efficacy of dexamethasone (or any other therapeutic intervention, for that matter). Even the numbers, from the perspective of the severely or critically ill patients, were likely not that impressive (even if statistically significant).

At the time of the press release, this data from the RECOVERY trial had not yet been independently evaluated, verified, or peer-reviewed—much like preliminary data presented in preprints—and it should have been approached with the appropriate caution. Sadly, that did not prevent the media outlets from expressing the minimal results in superlative terms. “Major breakthrough. . .” gushed the BBC. It multiplied those percentage numbers with then-current UK death toll from COVID-19 to repeat the claim that “Had the drug had been used to treat patients in the UK from the start of the pandemic, up to 5,000 lives could have been saved.”

Various prominent British personages led the cheers for this hyped-up claim, as did some prominent science journalists (who shall remain unnamed), and at least one doctor on Twitter, Scott Gottlieb, formerly of the FDA, who really should have known better.

The Preprint with Data Arriveth

As it happens, a week after the press release, the study was put up as a preprint, along with supplementary data, on MedRxiv. The primary numbers were not different from what was presented on the trial website and communicated on the press release. However, the data in the preprint provided greater clarity on some aspects. (If this is of interest, there is a detailed ‘tweetorial’ (tutorial on Tweeter) by Nicholas Lee, MD, a medical resident at University of Texas Southwestern.)

Some salient points which stood out to me from the data:

  • Their severely and critically ill patients maximally belonged to a younger set, less than 70 years. (Interestingly, their older patients were less critically ill and many did not require oxygen support.)
  • As expected, longer the time from the first appearance of symptoms, more ill the patients became; and as noted earlier, ill patients suffering for at least a week or more benefited better from the drug.
  • Not only was dexamethasone relatively ineffective in the less critically ill group, statistical analysis showed a marginal risk of harm in terms of death. The authors have not addressed this adequately in the discussion.
  • For every 10 patients, dexamethasone administration was associated with relatively early (fewer than 28 days) release from the hospital for at least 1 and a reduction in the risk of disease progression to critical illness or death in every 1–4 patients. As before, these effects were maximally seen in critically ill patients.

From this data, both patient age and the stage of illness in terms of oxygen requirement present an important dilemma for the clinician: when to start dexamethasone supplementation in the therapeutic regimen, and whether to wait until the patient progresses to a more severe disease (or turns 70!). A vital lacuna of the data presented is that the authors did not study/report the baseline and post-treatment status of any of the common inflammatory biomarkers, substances in the blood whose levels go up with increased inflammation (e.g. C-reactive protein, procalcitonin, ferritin, and so forth); these could have served as another set of clinically relevant, objectively evaluable secondary or tertiary endpoints and perhaps guided therapeutic decisions by staging the level of inflammation.

Since there is no peer review of this study yet, I would heartily recommend this Twitter-based Journal Club session, where a bunch of talented and dedicated infectious disease physicians, pharmacists and other professionals tore into the heart of the preprint, supplementing the data with their own clinical experience; the discussion was deep and scintillating.

Why Dexamethasone in COVID-19?

It is important to understand why dexamethasone might even be considered for COVID-19. Corticosteriods of the type glucocorticoid (such as dexamethasone), when administered, work to switch off multiple genes responsible for inflammatory mechanisms. There does not appear to be any great mystery to it; dexamethasone’s mechanism of action, pharmacological properties (how the drug behaves in the body, and how the body responds to it), and interactions with other drugs are pretty well characterized, and the drug’s beneficial effects have been observed in clinical trials with patients of acute respiratory distress syndrome. So, how does it apply to COVID-19?

As it happens, the COVID-19 disease has stages, the first of which is of course the invasion by the virus, followed by the response of the infected individual’s immune system (much like what happens in influenza as well). Many of the symptoms of COVID-19—which can be classed together as ‘inflammatory’—actually arise as side-effects of the body trying to fight off the virus.

Therein lies the rub: even after the viral attack by SARS-CoV-2 has died down, the effects of excessive inflammation still continue to rage inside the patient’s body, causing tissue damage at various levels and continuing to contribute to the symptoms. For example, inflammatory damage to the airways and the lungs can be so severe that the patient may be unable to breathe or get enough oxygen on their own, thereby requiring a mechanical ventilator and/or supply of external oxygen. Extensive inflammation, along with damage to small blood vessels, may even spread to other organs, causing them to fail.

THIS is the stage where steroids like dexamethasone, as well as monoclonal antibodies (e.g. Tocilizumab) which inhibit certain mediators of inflammation, known as cytokines, like IL6, are helpful, because they stifle down the exuberance of the inflammation. They are not really helpful in the early stages, because (a) they are not antivirals, they do nothing to the coronavirus as such, and (b) we would want to retain the body’s ability to mount an appropriate immune response, of which inflammation is a necessary part.

Talking to the New York Times, Dr. Martin J. Landray, the RECOVERY trial’s senior author and a professor of medicine and epidemiology at Oxford University, acknowledged that it makes sense for the drug to have disparate effects at different stages of COVID-19, given its biology.

The thing is, physicians—especially those dealing with infectious diseases and allergy/inflammations—know this very well. So, they were not that enthused when the hype about dexamethasone suddenly surfaced. And now the data shows that their skepticism may be warranted. Dexamethasone may not be quite the wonder-drug—in the context of COVID-19—that it was suddenly made out to be by uncritical media inflation of the preliminary results of one clinical trial.

That, however, is not to say dexamethasone may not be beneficial in COVID-19—but likely only for severely or critically ill patients, and within the narrow window between the onset of the post-viral inflammatory phase and the tissue damage becoming irretrievably severe; any administration must be carefully monitored, with the understanding that the clinical decision to start the treatment may not be straightforward or lend itself to the usual clinical algorithms.

An unfortunate effect of glucocorticoids is that they, via a complex mechanism, can increase blood sugar (a condition called hyperglycemia) and impair glucose metabolism, eventually leading to diabetes mellitus, the effects being even worse in people with preexisting (whether diagnosed or not) diabetes. The short course (10 days) and low dosage (6 mg per day) of dexamethasone used in the RECOVERY trial may not be enough to trigger these effects. However, the danger is that, in plenty of places across the world, the drug may be available without prescription, or doctors may prescribe it indiscriminately to people with mild symptoms, in larger doses and for longer durations—thereby increasing the risk for metabolic disorders, such as diabetes.

As of now, we also have no data on downstream effects of dexamethasone therapy. The RECOVERY trial says it has completed 94% of follow-ups, but it did not actually trace the patients once they left the hospital. We also don’t know how dexamethasone treatment may work out for patients with clinical improvement but persistently detectable viral RNA in low quantities, or how the drug will behave in case of a true reinfection (a possibility if a recovered patient tests positive for viral RNA again 6-8 weeks after recovery). Much about COVID-19 is still unknown & that uncertainty devolves on its therapeutic interventions as well.

As an infectious disease researcher, I have an additional concern based on available evidence. Clinical experience with COVID-19 patients in France, Germany, Netherlands, and the United States has shown that many patients who contracted a fungal infection called aspergillosis (which can be deadly in people with reduced immunity) in the aftermath of a bout of influenza or COVID-19—both of which result in excessively inflammatory conditions—were treated prior to or during their viral disease with corticosteroids. While these observational studies have not established any causal relationship between anti-inflammatories and aspergillosis, a contributory role of steroid-mediated immune suppression is not beyond the realm of possibilities.

“Evidence is not optional.”

This circles me back to Rasmussen’s golden words, the absolute requirement of evidence in medicine. Without data and critical analysis thereof, claims of clinical efficacy of medical interventions cannot be properly evaluated, and until such evaluation is performed, unsubstantiated hypes may result in premature adoption of such interventions into care practices, risking more harm than good. As science communicators and journalists, we must be cognizant of hypes and strenuously strive not only to eschew them in our own work, but also take part in speaking out against them before they take root in public consciousness.

Finally, whether dexamethasone works or not, absent a vaccine, wearing a mask in public and practising social distancing still remain the premier plan to protect public health by reducing the risk of transmission.