The president being prescribed dexamethasone is a significant indication of the severity of his case.
— Ben Walsh (@BenDWalsh) October 4, 2020
Studies have shown that it isn't useful and is often harmful when used to treat mild Covid-19 cases https://t.co/J2zhiyB9rG pic.twitter.com/Q8V0uQ8XS8
Click for the thread.
There are three claims here and one piece of evidence. The argument is 1) use of dexamethasone is an indicator of the severity of the case, 2) that using dexamethasone in mild cases is not helpful, and 3) that using dexamethasone in mild cases can be harmful.
The evidence is a New York Times article from three months ago reporting results of a study indicating:
Scientists in Britain announced a major breakthrough in the battle against the coronavirus last week, reporting they had found the first drug to reduce deaths among critically ill Covid-19 patients.
The results were first made public in a sparsely detailed news release. Now the full study, neither peer reviewed nor published yet, has been posted online, and it holds a surprise.
The drug — a cheap, widely available steroid called dexamethasone — does seem to help patients in dire straits, the data suggest. But it also may be risky for patients with milder illness, and the timing of the treatment is critical.
The drug “may harm some patients, and we’re not entirely sure which patients those are,” said Dr. Samuel Brown, an associate professor of pulmonary and critical care medicine at University of Utah School of Medicine in Salt Lake City, who was not involved in the research.
Following the announcement last week, officials at some American hospitals said that they would begin to treat coronavirus patients with dexamethasone, and the World Health Organization called for accelerating production to ensure an adequate supply. U.K. health officials moved to limit exports of the steroid.
You always want to know the number of participants, the trial methodology and the effect size. Emphasis added.
The drug was tested in a clinical trial that included some 6,425 patients in Britain. One-third were randomly assigned to receive the drug, while the others received the usual care. Patients in the first group received a very low dose of the drug, given daily for up to 10 days.
Dexamethasone was beneficial for those who had been sick for more than a week, reducing deaths by one-third among patients on mechanical ventilators and by one-fifth among patients receiving supplemental oxygen by other means.
Patients given the steroid who were not receiving respiratory support, however, actually died at a slightly higher rate than similar patients who were not given the drug, although the difference was not statistically significant.
What stands out is that the headlined finding, that dexamethasone might be dangerous for mild cases, is clearly very weak, and quite possibly a statistical fluke.
The NYT got distracted in its reporting, headlining the weakest finding and omitting the strongest finding from the headline. Not wrong, just misleading.
And critically, this was from four months ago. What is the body of knowledge preceding this study and what research has been conducted since then? It seems that Walsh might not know.
It appears that the entirety of Walsh's conclusion rests on a single NYT article early in the research cycle and has no contextual knowledge or specialization.
Certainly that is what seems to be indicated by the long trail of responses to the tweet from practicing pharmacists, medical doctors, and recovered patients. They all more or less testify that Walsh's conclusions are wrong and that use of the drug provides no indication of case severity, that use of the drug can occur prophylactically and that there is little to no danger associated with the drug under proper use. Additionally, the pharmacists and doctors are indicating that there is a long and deep research history independent of Covid supporting their position.
My tentative take away is that dexamethasone is widely used with few robustly documented downsides and several robustly documented benefits. That its use provides no indication of patient condition.
What to make of the argument? I see that Walsh is a finance reporter at Barron's with prior positions at Reuters and HuffPost. Looking at his tweet history, there is no obvious strong partisan history. Was this a partisan attack on the President seeking to overdramatize the president's health condition?
It seems like Walsh remembered a three month old report which was badly reported in the first place and derived a set of conclusions that were otherwise unsupported. As far as I can tell, his primary sin is that he believed mainstream media, did not read the original reporting closely, and did not independently check the basis for his claim among actual researchers and medical practitioners.
He accidentally produced cognitive pollution, substantially driven by the original cognitive pollution produced by the NYTs.
But in an environment where people are learning to see the media as purveyors of hysterical emotion, often heavily ideologically or partisan tinged, rather than as seekers of truth or purveyors of accurate reporting, these kind of erroneous slips get blown out of proportion.
It seems an indicator that the news industry as a whole desperately needs to recommit to truth and factual reporting. I suspect that they cannot do that and make a profit, but that is a separate issue.
No comments:
Post a Comment