You would expect the number of deaths to rise due to demographics, but just the distribution alone should not affect the rate. Now because a group is demographically together, that group of peers may have collectively had a significantly different integrated set of life experiences than previous groups - different food and food habits, different environmental exposures, work environments, etc. (all of which could add up to more obesity). So the dip could be the first sign that those integrated differences add up to a shift.

Though, re: obesity, aren't there a number of papers that seem to say that some extra weight in your latter years may be an indicator of better survivability for some medical events?

It's not very clear, but I think the article discusses that. It says the death rate above 65 is unaffected, so this is not coming from the aging boomers dying off. It's middle-aged people.

OH NOES!