But a 19% increased risk for lung cancer is enough to shut down businesses across the country because they want to allow tobacco smoking.
For statistical significance an increased increased risk (a/k/a relative risk) needs to be over 200% and preferably over 300%.
As a professional statistician, I know your comment is nonsense. Statistical significance has to do with how sure you are that an effect exists, not how large the effect is. With a large sample, small effects can be identified and found significant.
The sample sizes in the studies described in the article are easily large enough for effects of the size described to be statistically significant; therefore there is high confidence the effect is real, assuming the researchers did their jobs competently and honestly.