I'm going to take these two together.
So, I'll start by saying I'm primarily an ICU doc these days so my denominator is only the people who get a pulse back because otherwise they don't come to me. And that's one way to adjust the numbers and show improvement in research; exclude a group. If you don't transport someone who doesn't have ROSC after 30 min of on scene CPR, and then look at the percent of people who make it to the hospital who ultimately have good neuro outcomes, then yes your numbers improve. Admittedly I haven't done an exhaustive lit-review, but that's been my problem with pre-hospital EMS studies. They would compare new data of hospital outcomes with old data which included all of the pre-hospital dead.
I've currently got 2 people on the service with >10 minutes of CPR time, and both are neuro wrecks. In the past month that number rises to 11, all of whom are neuro wrecks. I have to go back about 3 months to get one with a decent neuro outcome with greater than 10 minutes of CPR. That's purely anecdotal however, and the plural of anecdotes isn't data.
And good quality CPR is vitally important. Which is why it's such a problem pre-hospital. In a big well trained EMS system that (sadly) sees a lot of fresh full arrests, you can have well trained medics. But most EMS systems country wide aren't that way. Heck, we can't even get a lot of ED techs and hospital techs to give good quality CPR. What works in Seattle/King Co, etc, isn't really feasible in smaller agencies.
I don't know what the right answer is. It's probably not grab and run (lots of dead bodies will get transported that way). I'm not sure staying on scene for a half hour is the best plan either. But I'll leave it up to the EMS directors to figure out what makes the most sense for their system, and I'll just take care of the fallout.
When EMS systems look at cardiac arrest survival, they calculate it as (survivors/total resuscitations attempted). There has been an attempt to standardize definitions and data collection via the Utstein Template (
http://circ.ahajournals.org/content/110/21/3385.long). Many systems will report "survival" according to the "Utstein definition", which is generally accepted as the % survival of patient who presented to EMS with a bystander-witnessed arrest (EMS-witnessed arrests are excluded from this) for whom the initial rhythm was shockable. There is NO EXCLUSION from these calculations if the patient is not transported, only if a resuscitation isn't attempted (e.g. obvious death or DNR).
To be sure, places like Seattle/King Co., Boston, and Wake Co benefit from a population that is healthier at baseline than places like Baltimore or Philadelphia. However, Baltimore and Philadelphia are not known for quality EMS. While some cities may never be able to achieve the same survival rates as others (if all EMS factors were the same), there is no question in my mind that most cities could improve survival if they modeled themselves on "high performing" systems.
As far as poor CPR, you get what you expect/demand. If you do not expect good CPR and thus do not demand good CPR, you will not get it. But, to get techs, EMTs, and paramedics to consistently give high-quality CPR, you need to train, measure, and actively work to improve. Most physicians don't want to make that effort. Too many think of cardiac arrest resuscitation as futile and and an act of "going through the motions" whether prehospital or in-hospital.
As far as early or late transport - there's never been an RCT as far as I know of rapid or late transport. Almost all (if not all) of the systems with the best outcomes do not routinely transport with CPR in progress. I know of many services that had large improvements in survival when they stopped rapidly transporting cardiac arrests. I think the future of cardiac arrest resuscitation will be identifying which small subset of patients would benefit from rapid transport for an intervention such as ECMO (though if you're in Paris, a doc will come to the scene and implement ECMO, even if you're in the middle of a gallery in the Louve). As of right now, the standard of care should be to work it on scene unless it's trauma. But, we should realize that even in the best systems, the majority will die - death will be very common, but survival is possible