Translate

Friday, February 21, 2014

A Saliva Test for Depression?

A new paper, available on-line and soon to be published in the Proceedings of the National Academy of Sciences has attracted a lot of attention as evidence of a new saliva test for depression. The study is based on two cohorts of secondary school children in the UK. They were followed prospectively and completed depression rating scale scores as well as 0800H saliva cortisol repeatedly during the follow-up (which was 12 months for one of the cohorts and 36 months for the other). As both elevated depressive symptom levels and elevated AM cortisol are well known risk factors, the result is perhaps not surprising: those with persistently elevated symptom levels and persistently elevated AM cortisol had a higher risk for a major depressive episode. The effect was only significant in boys. The idea of a saliva test for depression is intriguing, which may partially explain the media interest. However, statistical prediction does not translate into a clinically meaningful test and certainly not a useful screening test. It has not been shown that this kind of a test is a more effective way to predict depression risk than prediction algorithms using more traditional risk information, or for that matter asking a few questions about important risk factors. The fascination with this kind of result seems to arise from a strong desire for convergence between mental health and physical health disciplines. Physicians base diagnoses on laboratory data quite often, whereas this does not occur to nearly the same extent in mental health. The difference between a measure of depressive symptoms and a depressive episode is that in the latter instance the symptoms are more severe and more persistent. The fact that more severe and more persistent symptoms need to be added to salivary cortisol to create a significant prediction is as much a sign of how difficult it remains to use biological data as it is an indication of an effective diagnostic or predictive tool.

No comments: