We noted with interest that the article "For Nonremitting Depression, Add Rather Than Switch" in the April 21 issue describes the results of the NIMH Sequenced Treatment Alternative to Relieve Depression trial (STAR*D). Unfortunately, the headline gives the wrong impression of the study results.

After treatment with citalopram failed to bring patients to remission, patients and clinicians had some choice in the groups of treatments that were available for next-step randomization. They could either choose to add another medication (bupropion-SR or buspirone), they could switch to another medication (bupropion-SR, sertraline, or venlafaxine-XR), or they could accept options that included either adding cognitive-behavioral therapy (CBT) or switching to CBT (Rush et al., 2004). Another option was for patients to choose to be randomized to any one of all available treatments. This strategy is called "equipoise randomization" (Lavori et al., 2001). This design allows patient preference to be a factor in treatment allocation. Consequently, patients were not forced to a randomization to either a switching or adding strategy.

This patient preference may introduce a confounding variable, which can preclude a direct comparison of the two strategies. Indeed, in our study most patients and clinicians chose to either add or switch, but not both; consequently, the results from adding medication cannot be compared with the results from switching medications. Therefore, results do not indicate that adding is preferred over switching. Instead, we believe the results suggest that either adding or switching appears reasonable.