Different conclusions for pancreatic cancer and cardiovascular diseases
In some cases, our assessment differs from that of the authors of the original study. One example concerns pancreatic cancer. In a study from 2007, Luo et al. investigated a cohort of Swedish construction workers who had been followed up for a period of up to 27 years (3). With close to 280,000 participating men, this is the largest available Swedish cohort with registered snus habits. Luo et al. found that the use of snus doubled the risk of pancreatic cancer. Nine years later, the same construction workers were included in a pooled analysis with men from eight other Swedish cohorts. Araghi et al., who conducted the pooled analysis, found a hazard ratio (HR) of 1.07 (95 % CI 0.77–1.50) for pancreatic cancer among men who used snus only, compared with men who did not use snus and did not smoke (6). The confidence interval suggested a potential risk reduction of up to 23 per cent as well as a potential risk increase of up to 50 per cent.
Users of snus who quit after myocardial infarction halve the risk of dying in the aftermath
We consider this result of the pooled analysis to be inaccurate. In their article, Araghi et al. put particular emphasis on an analysis conducted amongst men who used snus, irrespective of their smoking habits, compared to men who did not use snus, also irrespective of smoking habits (6). The authors adjusted (in the pooled analysis) for smoking, but had no data for volume or duration, and found a hazard ratio of 0.96 (95 % CI 0.83–1.11). They interpreted this to indicate that snus does not cause pancreatic cancer in men. As already discussed above, this methodology may lead to overadjustment for smoking and underestimation of the effect of using snus. Furthermore, the long follow-up period may carry an increased risk of misclassification, because many may have quit their snus habit in the course of the follow-up period. Consequently, several factors may render it impossible to reveal an association.
We therefore have most confidence in the study conducted by Luo et al. (3), which shows a doubling of the risk, i.e. an effect that is sufficiently strong to qualify for an upgrade according to the GRADE method (7). We also know that the tobacco-specific nitrosamine NNK (nicotine-derived nitrosamine ketone) causes pancreatic cancer in animal studies and that exposure to other tobacco products such as cigarette smoking leads to a similar increase in the risk of pancreatic cancer. It has been proven that the risk of this form of cancer decreases to the background level after smoking cessation. It is reasonable to assume that the same is the case on snus cessation.
The conclusions drawn in the studies on cardiovascular disease and cancer have incorrectly been cited in support of the claim that using snus does not carry a health risk (4, 6). However, the absence of a statistically significant difference is not synonymous with an absence of effect in the exposure group.