I'd like to summarize articles I found interesting during this past year's survey of journals--I'll start with The American Journal of Psychiatry.
1. January 2013: "Adding moderate-dose lithium does not help patients with bipolar disorder". This is an editorial comment by Dunlop et al. about the "Litmus" study published by Nierenberg et al. Basically, this randomized study showed that adding small doses of lithium to the other treatments (including medication) that bipolar patients are receiving, does not lead to any symptomatic benefit over 6 months of follow-up.
I would conclude from this study that either 1) larger doses of lithium are necessary to be consistently useful or 2) subgroups may exist among the bipolar population which respond much better to such treatments, but such effects would not be noticed in a large cohort of patients sharing only the bipolar label as it is currently defined.
**an interesting alternate opinion is given in this article: http://www.ncbi.nlm.nih.gov/pubmed/21525518 Here, the authors show that trace lithium levels in drinking water (!) are inversely associated with suicide rates. The amounts of lithium in drinking water could provide the equivalent of about 1-10 mg of oral intake (much less than the hundreds of milligrams per day which are typical in bipolar management). Of note, small amounts of lithium could be absorbed not only orally but transdermally (e.g. from showering). As I look at the authors' data graphs, I have to wonder how strong this result really is. Possibly a small number of outlier points have substantially boosted the apparent effect size. Furthermore, despite the authors' attempts to control for obvious confounding factors (such as SES variation etc.) I have to wonder if this could be a non-causal association. We would have to see more data about supplementing with tiny doses of lithium to be more sure of a protective effect. At the very least, we need to see studies of this type replicated elsewhere. The authors mention a study showing a similar result in Japan (http://www.ncbi.nlm.nih.gov/pubmed/19407280), but in this case there was no attempt to control for confounding factors. A study of this type in England showed no association: http://www.ncbi.nlm.nih.gov/pubmed/21525523. Other studies referred to were done in the early 70's or earlier, and are of questionable quality.
Another positive line of evidence is typified by an article like this: http://www.ncbi.nlm.nih.gov/pubmed/22500970. Here, the case is presented that lithium has neuroprotective effects and may even have a role in reducing the risk of dementia. The evidence is based mostly on in vitro studies, but there are some randomized human studies coming out, such as http://www.ncbi.nlm.nih.gov/pubmed/21525519, in which one year of lithium titrated to a low serum level of 0.25 - 0.5 mMol, led to improvements in cognition and biomarkers in Alzheimer Disease patients. A small annoyance I have about this publication, as with many others, is that the precise information is not mentioned about exact doses of lithium each subject received, nor about the exact serum levels of each subject (0.25 - 0.5 is within a low range, but the top end of this range is 100% higher than the low end!).
A fair conclusion from existing evidence could be to consider a trial of low-dose lithium augmentation for patients with bipolar disorder, unipolar depression, or suicidal ideation from other causes. The evidence would tell us that the most likely result of such a trial would be no symptomatic benefit. But the possibility exists, with some reasonable support, that some patients may improve, perhaps because they are part of a subgroup who are more responsive to lithium effects.
This reminds me of another very interesting, albeit very controversial idea to research, which would be to look at population effects of various treatments. I think something like this would be most appropriate and ethical if it was a vitamin or nutritional supplement, or obvious positive community resource such as a recreational facility available to all, etc. But the idea would be to give every individual in the population the same treatment, and to see if this could lead to reduced incidences of various symptoms over long periods of time. Immunizations against infectious disease work on a similar principle: there is some protection for each individual receiving an immunization, but there can be a massive reduction in disease prevalence if every person in the population is immunized. Of course, psychological symptoms are obviously not the same as infectious diseases, but there is evidence that social factors strongly contribute to mental health events.
A benign study of this sort could, for example, be conducted in a prison environment, in which all inmates might receive access and formal time assigned to attend a new fitness facility. Or all inmates could receive a vitamin supplement, etc. with their consent.
Similar studies could be possible on a university campus: I think the most apt study could be to design better fitness facilities with much more convenient access, less crowding, and very low fees, as a giant public health experiment to reduce rates of depression and increase general health.