The answer is that, as one moves into more behaviors and more detail, the chance to meaningfully correlate outcomes with lithium exposure becomes more difficult. Also, because no one knows why those correlations exist (why lithium has the effect it does), it's not possible to turn a correlation into a rigorous, scientific cause-effect relationship.

The gold standard in science is not a correlation, but a cause-effect relationship accompanied by an explanation. With respect to lithium, we're a long way from that goal.