Kiwimedic - I don't think the numbers or the calculations are shonky, rather it is a case of what is included (and more importantly, what is left out). If I recall correctly, most of the research coming out of Seattle only includes witnessed, Vf or VT arrests. As we all know, these are generally the most survivable, so if for example traumatic arrests (which are particularly unsurviable) are excluded, all of a sudden the 'save' rates start looking very good.
Theoretically, all systems should be the same, as they should all be using the Utstein style for reporting, which is a consensus based system to ensure like is being compared with like.
Utstein Style Original
Or more recently
Newer Utstein Style
The impression I get however, based on what happens locally, is that many systems work off raw data that includes all arrests, and the Utstein style is only used at the national or research level.
To quote my epidemiology lecturer;
"That b*tard is lying to me.
How do I know he is lying to me?
Because he's a b*tard."