In 1990 we reported that the mean suicide rates in 27 Texas counties over a 10-year period were consistently lower in those with 'high' natural lithium content in the drinking water (70-160 mug/l) than in counties with 'medium' (12-60mug/1) or 'low' (0-10mug/l) water lithium levels.Ohgami et al have since argued, without proof, that these associations may have been spurious owing to what they considered an arbitrary division of the data. It is necessary, therefore, to emphasise that the data were partitioned in accord with accepted methods of statistical trend analysis and not in an arbitrary fashion, and that tests were conducted to assure that the partitioning of the data did not produce spurious associations.
展开▼