depression treatment Follow

This is one of the oddest studies I've seen in a while. It was published in the American Journal of Psychiatry earlier this month and it's literally an example of trying to pull the wool over the public's eyes concerning the efficacy of atypical antipsychotics used to augment an anti-depressant in the treatment of major depression. The study authors are J. Craig Nelson of UCSF and George Papakostas of Harvard/Mass General. Their stated purpose was: "The authors sought to determine by meta-analysis the efficacy and tolerability of adjunctive atypical antipsychotic agents in major depressive disorder.trial, depression scale used, response and remission rates, and discontinuation rates for any reason or for adverse events." They then focused on published studies or studies presented as posters at various conferences for the use of Zyprexa, Risperdal, Seroquel and Abilify as augmentation treatments. Oddly, only Abilify is approved for such use by the FDA. Seroquel's maker, AstraZeneca, is still awaiting FDA approval of its drug as an augmentation--and has been waiting for almost six months since an FDA panel recommended--quite gingerly--that the agency approve the drug. That's an unusually long wait and tells you that something odd is afoot. Anyway, the researchers didn't report in their paper the efficacy of the four drugs. Instead, they pulled an odd trick and reported the odds ratio of all the studies of the four drugs as depression augmenters to placebo. In other words, they were telling readers the chance that this class of drugs beat placebo--not the efficacy of the drugs (except in a slippery manner I'll come to). That strikes me as a consumer and science dork as an utterly useless measure with little real world clinical significance, especially since several of the 16 studies were of only four weeks duration--some of the shortest depression treatment studies I've ever heard of. They reported the odds of beating placebo as: "Adjunctive atypical antipsychotics were significantly more effective than placebo (response: odds ratio=1.69, 95% CI=1.46–1.95, z=7.00, N=16, p