Starting in March 2020, studies began to show that smokers were under-represented among COVID-19 patients, suggesting that something in tobacco may offer protection against SARS-COV-2 infection. The evidence remains inconclusive, but it seems that some public health experts and journalists don't want to get to the bottom of this mystery.
Smoking is a dangerous habit that kills many thousands of people every year; the world would be a healthier place if fewer people smoked. Available evidence overwhelmingly confirms those claims, but we have to avoid the temptation to go beyond those evidence-based statements as we investigate open questions surrounding the health effects of tobacco use.
In the wake of the COVID-19 pandemic, some public health experts and reporters have been unable to resist that temptation as they research the link between smoking and coronavirus infection. They have dismissed research suggesting that cigarettes or something in them may—and I emphasize “may” in this sentence—protect smokers from COVID-19, or at least minimize the severity of their symptoms if they do get sick.
Writing in the June 2nd British Medical Journal, investigative reporters Stéphane Horel and Ties Keyzer discounted this possibility, alleging that the authors of some of the relevant research were in the pocket of Big Tobacco and their studies have been discredited:
In the early days of the pandemic, media outlets around the world reported that smokers seemed to be under-represented among patients seriously ill with covid-19 in China and France. The headlines asked, does nicotine protect against covid-19? … It has since been roundly disproved that smoking protects against covid-19 … the BMJ can today also report on undisclosed financial links between certain scientific authors and the tobacco and e-cigarette industry in a number of covid research papers.
This, I think, is an inadequate assessment of the issue for two reasons. First, Horel and Keyzer too hastily dismissed interesting research that deserves follow-up. More importantly, they did so based on a standard they themselves don't meet.
What does the evidence show?
Although Horel and Keyzer asserted that the protective effect of smoking against COVID-19 has been refuted, they spent only two paragraphs dealing with the evidence; one was a quote from the World Health Organization (WHO) and the other, just 50 words long, referenced several studies that supposedly confirmed their thesis. The roughly 1,800 remaining words of the story were dedicated to commentary about conflicts of interest (COI).
We'll return to COIs shortly, but let's deal with the data first. Compare the BMJ analysis to a November 2020 review of all the research (up to that point) on COVID and smoking. Many studies have indeed found that smoking elevates the risk of infection, the authors explained, but the evidence on this association is “contradictory and inconclusive”:
While most of the studies to date have indicated an association between smoking and a worsening of COVID-19 symptoms, there are reports that have suggested an inverse relationship between smoking and COVID-19. In particular, smoking prevalence among patients hospitalized with COVID-19 has been reported to be lower than the smoking prevalence in the general population.
This article was curiously missing from Horel's and Keyzer's footnotes. Perhaps that was just an oversight, but more studies have subsequently raised awkward questions about smoking and COVID-19. For example, an April 2021 retrospective cohort study (also missing from the BMJ footnotes) found “that significantly more formers [sic] smokers were hospitalized and died from COVID-19 than current or never smokers.”
The researchers rightly noted that they weren't able to demonstrate causation and their results may have been confounded by age and comorbidities. They nonetheless pointed out that their results fit with those of several other studies from around the world that "have reported a low prevalence of current smokers among people that tested positive for SARS-COV2 and/or were hospitalized due to COVID-19.” 
A so-called “living review” of the evidence, originally published as a pre-print this March but subsequently updated, similarly found that “Compared with never smokers, current smokers appear to be at reduced risk of SARS-CoV-2 infection while former smokers appear to be at increased risk of hospitalization, greater disease severity and mortality from COVID-19.”
Assuming this protective effect is real (we don't know that it is), what might cause it? Several possible mechanisms have been suggested. It may be that nicotine inhibits “Interleukin-6, a key player in the cytokine release syndrome induced by SARS-COV2 infection.” Another possibility is that nicotine reduces levels of ACE2 receptors by which SARS-COV-2 enters host cells. Research has also linked tobacco use to elevated nitric oxide production in the lungs, and that could impair viral entry into cells and viral replication.
I hasten to add that this is all speculative. No one should take up smoking for fear of COVID-19, and nothing in tobacco has been shown to prevent or treat SARS-COV-2 infection. The point is that we don't have enough evidence to determine if this protective effect is real or, if so, what causes it. Anyone who says otherwise, whether they're trying to confirm or debunk it, is overstepping the research. We need (and should want) better data to find out what's going on.
Conflicts of interest
Conflicts of interest can influence the results that scientists publish, which is why most journals require prospective authors to report potential COIs. The problem, in this case, is that Horel and Keyzer seemed content to dismiss any link between reduced COVID infection risk and smoking because the tobacco industry has funded some (though not all) of the relevant research.
This is little more than an example of the genetic fallacy, making it useless as an argument. If we consistently applied this standard, we could disqualify anybody we disagreed with on any issue. For example, Keyzer is affiliated with the Investigative Desk, a Dutch nonprofit funded in part by the University of Bath, which has taken $20 million from the Bloomberg Foundation to finance a “global tobacco industry watchdog” group. In 2019, the foundation also launched “a $160 million push to ban flavored e-cigarettes.” Horel and Keyzer assured BMJ readers that
“The [Investigative] Desk is fully independent in the selection, investigation, and publication of our topics. Donors have no role or substantive say in this, and we do not align the timing of our publications with them.”
I believe them. But why should they be allowed to claim independence from their donors while denying the same opportunity to scientists who have reached a different conclusion than they have? Christopher Snowdon, director of lifestyle economics at the UK-based Institute of Economic Affairs, appropriately summed up the issue in response to Horel's and Keyzer's article:
Far from being 'roundly disproved,' the evidence that smokers are at reduced risk of SARS-CoV-2 infection is much stronger today than it was when the hypothesis first emerged last March. This evidence cannot be dismissed on the basis of tenuous financial links of a handful of researchers to the tobacco and vaping industries. Why do we keep seeing this strong inverse association between smoking and SARS-CoV-2 infection? Is it the nicotine? Is it the smoke? Is it something else? We do not know and we are not going to find out by burying our heads in the sand. 
 The five studies were references 10,12,18, 19, and 20 in the paper.
 Horel and Keyzer named Snowdon in the article, prompting his reply.