[nostrademons]

That's not necessarily a disproof - if opioids are killing people who are disproportionately young, then they could have an outsize effect on life expectancy. Say life expectancy is 79 years. If a new cause of death occurs that affects 10% of the population and kills, on average, at age 78, then life expectancy goes down to 78.9. If a new cause of death occurs that affects 10% of the population and kills at age 29, then life expectancy goes down to 74. Same reason that infant mortality had a disproportionate affect on life expectancy: the numbers of deaths might be small, but the reduction in lifespan is large.

Personally I would bet on obesity being a bigger contributor to declining lifespans than opioids, but I'm just showing that raw numbers of deaths cannot automatically disprove the hypothesis that a decline in lifespan is due to opioids.

[OscarCunningham]

For a cause of death affecting 40000/2700000 it could only reduce life expectancy from 79 to 77.8, even if all the victims were age 0.

EDIT: I suppose it could be lowered a bit more if all those who died were those who would otherwise have lived longest.

[bpicolo]

> I suppose it could be lowered a bit more if all those who died were those who would otherwise have lived longest.

That would average out in a population size of 40k

[OscarCunningham]

No, I mean there could be some actual effect that selected for those people.

For example if rich people were culturally more likely to use heroin than poor people, then overdoses would lower the life expectancy more than you would expect, because rich people (when not overdosing on heroin) have better access to medical care and hence longer life expectancy.

[amsilprotag]

There's a great breakdown here [0]. The first chart shows age 15-44 deaths going up ~10% from 2010 to 2016. The third chart shows drug overdose deaths going up 50%(15-24), 80% (25-34), and 60% (35-44) from 2012 to 2016.

[0] http://www.calculatedriskblog.com/2018/05/lawler-us-deaths-j...