Reliability of saliva lithium level-a prospective study

140 synchronous samples of serum and saliva, collected from 28 patients undergoing lithium therapy, were studied. The mean saliva/serum lithium ratio calculated for the first 120 synchronous samples from 24 patients was found to be 2.68 (population mean ratio). Regression line equation calculated for the same population came out to be Y=0.325 + 0.22X. Predictive value of saliva lithium level was tested by applying this regression equation and the population mean ratio on 20 samples from the next 4 patients. Prediction was also tried in the 24 patients who had given more than 3 synchronous samples by using the individual mean saliva/serum lithium ratio. An individual's mean ratio was calculated from the initial 3 synchronous samples and the predictive value of saliva was tested on subsequent samples in the same patient by using his mean ratio. This method of prediction was found to be better than predicting on the basis of population figures. But no method was found to be consistently reliable and therefore saliva lithium level is not a reliable indicator of serum lithium concentration.