Per usage alcohol has a higher death rate then either tobacco or pools.

Actually, the best available study suggests that drinking appears to increase life expectancy on average even in fairly heavy amounts. (Ungated paper available here -- see the striking graphs in figures 1 and 2.)

In practice, of course, this varies enormously between individuals, and it's somewhat correlated with ancestry. Some people with particularly bad predispositions are indeed better off as teetotalers, but the idea that total abstinence would make everyone (or even the majority of people in Western countries) healthier is just ludicrous.

As for those supposed total alcohol death statistics, these numbers are completely arbitrary. There is simply no reasonable unique way to define deaths as due to alcohol, and with convenient enough definitions you can make the numbers vary by orders of magnitude.