Comment You can't deduce causality from longitudinal study (Score 1) 237
After reading the full study report, the headline for this post is more than a bit sensational (disappointing for /.). This was a longitudinal study comparing a set of data from a certain time period and can't be used to determine causality. Also, the control were only matched on a few data points, making this difficult to apply to a general population or even interpret if there was an actually difference in mortality given that table 3 showing differences in co-morbid diagnoses shows large disparity in key areas like cardiovascular risks, hypertension, heart failure.
This point underlies why this study is not helpful. Many serious medical illnesses have insomnia has a symptom, and often treated with a sedative hypnotic. This DOES NOT mean that the sedative hypnotics are "killing" people. I hate studies like this.