Checklists are probably good but looking at the actual paper[0] I don't think we can conclude "a third" with much confidence. The mortality rate for surgeries was already declining--the argument is that it declined faster during the implementation period (Fig 2; 2008-2010). Also it appears the types of admissions changed a lot during the same period based on the stark difference between Fig 1 (unadjusted) and Fig 2 (adjusted).
I'm surprised that the researchers didn't use better data or variation (e.g., hospital-level mortality or the timing of the rollout) to strengthen the case. So much can be lurking beneath annual country averages...
I'm surprised that the researchers didn't use better data or variation (e.g., hospital-level mortality or the timing of the rollout) to strengthen the case. So much can be lurking beneath annual country averages...
[0]: https://onlinelibrary.wiley.com/doi/full/10.1002/bjs.11151