Memorial Sloan Kettering Cancer Center researchers, led by first-author Sarah Borderud, claimed on September 22, 2014, that e-cigarettes did not help cancer patients quit smoking (media story here). They based that statement on a study they published online in Cancer, a journal of the American Cancer Society (abstract here).
The researchers had enrolled 1,074 cancer patients in a smoking cessation program. They subsequently found that “E-cigarette users were as likely to be smoking at the time of follow-up as nonusers (odds ratio, 1.0; 95% confidence interval, 0.5-1.7).”
Upon reading the study, I found a significant error: the main results table reported the exact opposite of the text. On October 16, I submitted a letter to the Cancer editor, co-signed by my colleagues Nantaporn Plurphanswat and Carl Phillips, requesting a correction.
Six weeks later, on November 25, a correction was published on the Cancer website. It said: “The authors discovered some errors…in Table 2.” The circumstances strongly suggest that the authors didn’t “discover” the errors, we did. The journal office had our letter on October 16, six weeks before the correction appeared.
Our letter, published online on March 4 (here), described other problems with the study; one is particularly important. Borderud et al. claimed: “Using an intention-to-treat analysis, E-cigarette users were twice as likely to be smoking at the time of follow-up as nonusers (odds ratio, 2.0; 95% confidence interval, 1.2-3.3).” In other words, e-cigarettes were harmful.
They reached this striking result by assuming that anyone lost during follow-up had continued to smoke. We pointed out that “Smokers who were E-cigarette users were twice as likely to be lost to follow-up as the other smokers (66% vs 32%)… The conclusion that E-cigarette users were twice as likely to be smoking is purely an artifact of the assumption. An equally plausible counter-assumption is that dropouts left the program because they had quit smoking...” They didn’t consider this possibility, but they should have.
Carl Phillips’ goes into more detail about the technical irregularities here.