Lithium, as everyone knows, is a song by Nirvana. (It's also a different, and worse, song by Evanescence). It's also a chemical element. And it's a drug, used in psychiatry to treat bipolar disorder and, sometimes, severe depression.
One thing which lithium seems to do rather well - or at least better than other drugs - is make people less likely to commit suicide. At least, that's what most authorities say, so let's assume it's true.
Now a Japanese team report that the amount of lithium in tap water are negatively correlated with suicide rates - Lithium levels in drinking water and risk of suicide.
(Someone has helpfully put this up on Scribd for people without academic access).
They analyzed the water supplies in the 18 subdivisions of Oita prefecture in southern Japan, and compared the lithium levels in the water with the average suicide rate (normalized for age and gender). Here's what happened -
Now, on paper that's a pretty solid correlation. But how believable is it? The maximum lithium level in the water was found to be 59 micrograms per liter. By contrast, if someone were taking lithium for bipolar disorder, they would on average be taking about 1 gram of lithium carbonate (Li2CO3) per day, which works out at about 100 mg of lithium (and 900 mg of inert carbonate).
Assuming that people drink about two liters of tap water per day (almost certainly they drink less), that's about 120 micrograms per day. That's 0.12 mg. So about a thousand-fold less than you'd take if you were bipolar.
Could such a tiny amount of lithium do anything? Well, maybe. It's not impossible that it would have some small effect on suicide rates, but it seems very unlikely that it would account for the difference between a rate of 120 and a rate of 80 which is what the graph seems to imply. The effect is just too large to be credible, if you ask me.
Although that said, another paper just out found that tiny amounts of antidepressants (the amounts that you find in rivers because so many people take them and some of them are stay in urine) could affect the behaviour of baby fish. And presumably also baby humans. Unfortunately I can't access this study, so I don't know whether that's nonsense or not, but it's food for thought.
Ohgami, H., Terao, T., Shiotsuki, I., Ishii, N., & Iwata, N. (2009). Lithium levels in drinking water and risk of suicide The British Journal of Psychiatry, 194 (5), 464-465 DOI: 10.1192/bjp.bp.108.055798