While I want to believe the result, it seems like the method is far too susceptible to confounding cohort effects. I'd be curious if a similar trend is seen in other countries that ended rationing at different times, and if any such change is seen in countries that didn't ration. The cancer rates may not be directly comparable due to other differences, but if the rate change is consistently centered around whenever sugar consumption changed, that would be much more compelling evidence; whereas if post-1953 children have higher cancer rates across the board it's probably something else.