Thread by @PloederlM

1. Our paper on lithium in drinking water and suicides in Switzerland was just accepted for publication. It is to our knowledge the first pre-registered study on this topic. We found a near perfect null-result.

2. Background: lithium is considered to be suicide protective in therapeutic doses and some are convinced it is THE drug in suicide prevention. However, with new evidence, appropriate statistical methods, and results for suic attempts, uncertainty remains

https://x.com/PloederlM/status/1625778975917199362

3. There are now ca 20 studies on lithium in drinking water and suicide rates. Overall, results support the suicide preventive powers of lithium bc higher lithium levels were associated with lower suicide rates. Some even suggested to add lithium to tap water.

4. However, as soon as the number of studies allowed the analysis of publication bias, such a bias was confirmed and accounting for missing studies lead to a much smaller and non-significant correlation of lithium and suicides.

https://x.com/PloederlM/status/1319254086554877954

5. Not surprising, because none of the studies were pre-registered. Some had absurdly high correlations (> 0.60) and the choice of confounders varied substantially. So we thought we should do a pre-registered study on the topic. Protocol here: osf.io/fn3u9

6. Thanks to a colleague who lives in Switzerland now, we got access to measures of lithium (> 4000 measues) from over 1000 municipalities and the suicides in these municipalities from 1981 to 2021. Thus, our study was well powered to detects small correlations.

7. We found null-findings for crude associations, after adjusting for confounders, using crude suicide rates and age-adjusted suicide rate ratios, and in models accounting for spatial dependency.
R-code and detailled analysis are on the OSF osf.io/gb2rk/

8. As happens ofen with inconvenient or null-findings, our manuscript was desk-rejected by several journals until one editor decided for review. One reviewer said “The anti-suicidal effect of lithium in bipolar, unipolar and some other psychiatric disorders is absolutely sure.”

9. He/she requested analysis separately for men and women and we did that, with similar null-findings among men and women.

10. I will provide the link to the publication later, when the paper is going online.

@threadreaderapp unroll

@ESYudkowsky Wouldn't this be better investigated with a random sample and checking lithium in the serum?

@SocDoneLeft Let's call it an informative graph ;-)

@XueFisher And on second thought, why should it vary over time? Lithium is basically washed out from rocks/soil, and this is a rather constant process. But appreciate your point!

@XueFisher A challenge was that suicides at the municipality level varied greatly, so choosing a restricted time-frame would have come with much noise.

@jt_kerwin But honestly, I am 52, and I am an existential psychotherapist and not on an academic track. I am simply too old to cheat in this, wasting precious lifetime just to come up with some results. No way
;-)