I contributed an article reviewing the evidence for the GeneSight genetic test, which is being quite heavily marketed as a way to choose the right medications for patients. According to company's website:
"Multiple clinical studies have shown that when clinicians used GeneSight to help guide treatment decisions, patients were twice as likely to respond to the selected medication."
That's misleading, I found. The vast majority of the GeneSight data are based on studies with an unreliable methodology. These were so-called "open label" studies in which patients were non-randomly assigned to two groups: guided treatment vs. unguided treatment. All patients and clinicians knew which patients were assigned to which group, leading to the very real possibility of various biases--along with a heavy helping of the placebo effect.
One single randomized, blinded study has been published (not even properly blinded, since the clinicians knew which patients were in which group). It enrolled 49 patients (25 in the guided group, 24 in the unguided group). There was no significant difference in the depression improvement scores between the groups. There was a secondary analysis of 13 patients showing a potential benefit for those whose meds were categorized by the test as being particularly problematic.
13 patients? I don't think I would use a genetic test based on good results from an N of 13--and I would suggest that you think twice before you do so!
By the way, I'm at the APA meeting in Toronto and will be going to a lecture today sponsored by Assurex, the maker of GeneSight--if I learn anything new I'll let you know.
Subscribe to our RSS Feed!
