A thing with this study is that they measure survival mostly in an age bracket with relatively low chance of death, but also in a bracket where chances of death

A thing with this study is that they measure survival mostly in an age bracket with relatively low chance of death, but also in a bracket where chances of death by accidents due to lifestyle are relatively high. A base death rate of 2.2 per 10k years is one per 4500 years. Since you live about 80 years, the sample is skewed towards the not-so-likely to die. Perfectly acceptable btw, but it makes inferences to what happens to the life expectancy Curve harder. My prior would be: the cumulative survival curve for ADHD is markedly lower in the ages 5-30s (one dies more often when at risk due to accidents), and then converging to the curve UNLESS the medicines are harmful as well. My point in other words: aMRR is a fine measure, but you really want to identify causes to shifts in (parts of) the survival function. For now it's a finding that just confirms prior beliefs. [edit] Confirmation by the same author [1]. All that would be cool now is estimates of survival functions. They must have some people in their dataset dying of old age. "Children with ADHD had an increased risk of injuries compared with other children. Treatment with ADHD drugs reduced the risk of injuries by up to 43% and emergency ward visits by up to 45% in children with ADHD. Taken together with previous findings of accidents being the most common cause of death in individuals with ADHD, these results are of major public health importance." [1] http://pure.au.dk/portal/en/persons/soeren-dalsgaard(9da04734-659f-4d4e-bae8-a6ca57c22f05)/publications/effect-of-drugs-on-the-risk-of-injuries-in-children-with-attention-deficit-hyperactivity-disorder-a-prospective-cohort-study(80d398f9-9a15-48c3-bc55-83eef1e497fd).html http://pure.au.dk/portal/en/persons/soeren-dalsgaard(9da0473...