You know, one thing I’ve really enjoyed about the last year has been its comfortingly retro qualities. We’ve had a 1970s fashion renaissance, a massive roaring twenties recession, and thrown in there was some classic World War II era espousal of the heath benefits of cigarettes.

Do you remember that? It was a weird subplot, around last summer, where newspapers started to report that smoking might protect you from Covid. It came from a July 2020 paper in the European Respiratory Journal, reported that smoking was not associated with adverse events in patients admitted to hospital with Covid, and that smokers were at a significantly lower risk of acquiring the virus.

Well, anyway, it was a load of rubbish. It’s been retracted because the authors failed to disclose conflicts of interest – several of them worked in the tobacco industry. In an unsurprising turn of events, doing pervasive and persistent damage to your respiratory system doesn’t actually protect you from a virus that predominantly affects the respiratory system. Who would’ve thought?

It doesn’t actually appear that the authors did anything untoward with the European Respiratory Journal data. No, it seems this is a tale as old as time: scientific studies that are funded by or affiliated with a particular industry deliver results that are favourable to that industry, in a way that government-funded studies do not. They call it funding bias, and it’s sufficiently common that there’s a term to describe it and research examining it. The tobacco industry is a prime example of this phenomenon, but it’s also common across the pharmaceutical industry.

Well, obviously, you might think. The pharmaceutical and tobacco industries aren’t exactly known as shining examples of morality. It’s not surprising that they’re fixing the data in their favour. You wouldn’t be entirely wrong, of course: there are notable examples of scientists being either deliberately deceptive or grossly negligent and making spurious claims as a result. Look at Andrew Wakefield, who accepted tens of thousands of pounds in return for demonstrating a (make no mistake, non-existent) link between autism and the MMR, or Malcom Pearce, who falsely claimed to have performed a ground-breaking operation involving the transplant of an ectopic pregnancy. Or, to give them their full medical titles, Andrew Wakefield and Malcom Pearce.

It isn’t known exactly how common scientific misconduct is, but the general consensus is probably not very, at least when compared to the metric tonne of scientific research produced on an annual basis, and there’s no evidence that misconduct is more common in industry-funded studies. There’s opposite, in fact, is true: the quality of industry-funded studies is at least as high, if not higher, than those funded through other sources.

There are other ways to guide the results of a study in a particular direction than out-and-out fraud, however.  As David Michaels, epidemiologist and professor of public health, wrote in the Washington Post in 2008, it’s often not so much the answers of industry studies that are biased, but the questions. There are a plethora of industry tricks that encourage the outcomes of studies one way or another: testing a new drug against an established drug that either does not work or does not work very well, using a dose of the established drug that is too high or too low, publishing the results of a trial several times to make it appear that multiple studies reached the same conclusion, publishing only studies which had favourable conclusions for your drug and burying the rest. In reviews or meta-analyses, where the author selects a group of papers and synthesises an overall outcome, the papers that are selected and how heavily they are weighted can have an enormous effect on the conclusions drawn.

There are all kinds of morals to this story – about not needing to lie to fail to tell the truth and not believing everything you hear and such. But once again, it appears that the principal takeaway is that you should stop smoking. Tale as old as time, indeed.

Image hosted by Wikimedia Commons.