Tuesday, February 15, 2011

Misrepresentation of Snus Use in Karolinska Institute Studies

As I documented in a previous blog entry, Karolinska Institute (KI) studies on the health effects of snus use have profoundly influenced regulatory action worldwide, but those studies are compromised by important, troubling and unresolved discrepancies. Following is evidence that KI investigators misrepresented snus use among the revolving-door cohort in multiple journal articles.

From 1971 to 1974, 135,000 Swedish construction workers were enrolled in a health program that served as the basis for epidemiologic follow-up studies investigating associations between many risk factors and diseases. The first report, claiming that snus use had risks for cardiovascular diseases, was published by Gunilla Bolinder (abstract here); subsequent studies variously included and excluded this group, prompting me to label it the “Bolinder Revolving-Door” cohort (here).

There are many other problems with the KI studies that are technical in nature, but one deserves special mention because it is so basic and egregious. It involves the definition of snus use.

When conducting epidemiologic research, it is critically important to classify subjects according to current or former exposure to a risk factor; there are often large differences in risk in these groups. For example, current smokers have higher risks for heart attacks than former smokers. Sometimes, researchers don’t collect information on current and former use, in which case they designate subjects as “ever-users” and “never-users.” The ever-never classification is not nearly as informative as current-former-never classification.

In her 1994 publication, Dr. Bolinder clearly stated that she investigated cardiovascular diseases among “present” (i.e., current) snuff users. Keep in mind that Bolinder’s cohort probably had “former” users of snus, but she didn’t mention them.

Now, fast-forward to 2007, when Maria-Pia Hergens published a study of snus use and heart attack among construction workers (abstract here). As I mentioned previously, Hergens excluded the Bolinder cohort, and this was her reason: “During the period 1971–1974 exposure information on snuff use was limited to ever or never use.” (emphasis added)

Thus, Bolinder claimed that she studied “present” (i.e., current) snus users in her cohort, but in 2007 Hergens said the only information available in the Bolinder cohort was ever or never use. To make matters more interesting, Gunilla Bolinder was a co-author on Hergens’ publication.

Did the Bolinder Revolving-Door Cohort have current snus users or ever snus users? The Bolinder and Hergens-Bolinder descriptions of snus use are not only conflicting but irreconcilable; one of them is a gross misrepresentation.

Last year, I described a 30-year misinformation campaign in the United States, almost entirely based on Dr. Deborah Winn’s misrepresentation of dry powdered snuff in her influential 1981 New England Journal of Medicine report (my entry here). Winn exaggerated and conflated the risks of a rarely used smokeless product (dry powdered snuff), and attached those risks to popular chewing tobacco and moist snuff products, which have virtually no risks (evidence here).

The credibility of the KI studies is contingent on the resolution of their many significant discrepancies. It would be a public health travesty if the KI misrepresentation persisted as the basis for a Winn-style misinformation campaign in Sweden and throughout the world.

No comments: