The Kansas City Star ran an opinion piece by a former senior Wall Street Journal reporter, Stephen Moore. (Moore is now the chief economist at the conservative Heritage Foundation). The Star discovered a few weeks later that Moore had made substantial factual errors. The paper immediately printed a correction in print, put a warning at the top of the online version and put notes within the online version highlighting the mistakes. It also promised to run no more pieces by Moore. The full story is reported here by the Columbia Journalism Review. “We did not factcheck his … figures that we found out later were incorrect, and that was a failure on our part,” said Miriam Pepper, editorial page editor of the Kansas City Star, even though the corrections did not change the conclusions that Moore had reached. (In the interests of transparency, I should say that met Moore a few times when he was based in London and liked him).

Contrast this with the behaviour of the British Medical Journal. As we have reported before , the BMJ ran two articles in October 2013. One by Abramson and colleagues (BMJ 2013;347:f6123, doi:10.1136/bmj.f6123) made a series of basic errors. The Abramson paper said that an earlier study by Zhang et al had found that 18% of patients had side -effects from statins and that the same number had withdrawn, at least temporarily, from treatment, because of side effects. In fact, the Zhang study was not placebo controlled; its authors had highlighted the limitations and; the study had not said that 18% had discontinued therapy but said that it might be “up to 9%”. As if all this weren’t enough, Abramson et al made dismissive references the vast number of randomised, placebo-controlled studies that show no difference between the side effects from statins and those from placebos. The other paper also misstated the Zhang data

After months of complaints by Professor Sir Rory Collins and other epidemiology experts, the BMJ finally put a notice of correction on both papers online, acknowledging that “the article [cited in the BMJ papers] did not reflect necessary caveats and did not take sufficient account of the uncontrolled nature of the study.” It did not acknowledge the mathematical errors, made no apology in print and its editor said that the papers may have been justified because “randomised control trial data is notoriously poor at reporting adverse events” (An argument that neither article had made at the time — they just got the numbers wrong).

Now, a panel convened by the BMJ has said that it was right not to retract the papers in their entirety. As Heartwire reports (registration required), Professor Collins thinks this is no surprise: “Collins said he was ‘not surprised’ by the panel’s conclusions, pointing out that the entire panel was invited by the BMJ editor, all but one of the panel members have close links to the BMJ, and three of the seven members have written ‘papers about the hazards of statins that are not supported by the evidence.'”

As if that weren’t enough, Prof Collins says that the BMJ refuses to correct other errors in the papers, according to Heartwire: “Collins, for his part, believes there are additional errors in the paper that the BMJ has not addressed, despite taking six months to address the statin-side-effects figure, including a separate figure citing myopathy rates of 5%. The correct number, says Collins, is 0.05%. Letters to the BMJ, including one from the Cochrane group, have pointed out this error, yet it still stands, Collins alleges.”

So, the correct answer is that you should probably believe it if you read it in the Kansas City Star but be a great deal more sceptical if it’s in the British Medical Journal.

Hannen Swaffer, one of the early 20th century pioneers of British tabloid journalism, said that “freedom of the press in Britain is the freedom to print such of the proprietor’s prejudices as the advertisers don’t object to.” Surely it is time for the advertisers to start objecting?

bmj