As the author/presenter says, it's because it's not the same statistic. They're not "inverses" or something of each other. Survival rate can be really REALLY odd. Take something like a gunshot wound. You either survive it, or you don't, right? Survival rates there are HIGHLY correlated to the mortality rate, and quickly... for the most part. But there are cases of lead poisoning in cases where the slug is left in (that happens, and is accepted medical practice) where they die DECADES later. Of a gunshot wound, but it just took a LOT longer. But Survival rates are "5 years out" (or 10 sometimes) for Cancer. But detection affects that number, as well as treatment. Hence why mortality rates are more of an "Aggregate over the years" regardless of how long it takes a particular disease/condition to kill you. Compared year-on-year, you can tell if it's just detection is getting better, or if you're actually reducing the overall number of deaths due to a specific cause. Yes the numbers are smaller because it's based on the WHOLE population, but it's more accurate year-on-year that you're having an effect through policy. The Thumb Cancer example is great, in that the survival rate skyrocketed, but the mortality rate stayed exactly the same. That was the point of the example.
So because of the HUGE amount of money spent on detection in the USA, your 5-year survival rates for many cancers is better, and even very markedly so versus many "socialized medicine" countries. But your MORTALITY rates (% of population that will die from it each year) are barely better at all, and only in some cases. According to my original link, MORE of your population per year dies of all cancers than in Canada at the least, which was your original comparison.
you guys must be surfing at work or something. Shame on you!