Letters Section Editor: Stephen J. Lurie,
MD, PhD, Senior Editor.
In Reply: Drs Bowden and Fawcett point out
the many potential sources of bias or confounding in our comparison of lithium
and divalproex. We did in fact measure and adjust for comorbid substance abuse
and anxiety disorder, but we lacked data on many other potential risk factors.
This is an inevitable shortcoming of large epidemiologic studies based on
administrative claims data. Our analyses do indicate that 1 year of treatment
was a far stronger predictor of drug choice than any clinical characteristics
of individual patients, arguing against significant confounding by indication.
Goodwin FK, Simon G, Revicki D, Hunkeler E, Fireman B, Lee J. Pharmacotherapy and Risk of Suicidal Behaviors Among Patients With Bipolar Disorder—Reply. JAMA. 2004;291(8):939-940. doi:10.1001/jama.291.8.940-a