But the paper cited by the blog post also covers the smoking prevalence data and finds it to not be statistically significant. Now you can look at their model and decide if you find that conclusion convincing, but I don't think just pulling a number from a different report (especially one from the Australian government, which has its own agenda in motivating its own policy) is really such a gotcha here.

To be clear, I don't want to play the game of paper vs. paper. I simply think there is more depth of detail here than can be nicely summarised in a comment.