Dr. Valenstein and Colleagues Reply

To the Editor: Given the research evidence, we understand Dr. Terao’s surprise regarding the infrequent use of lithium and the more common use of a second antidepressant or an atypical antipsychotic as antidepressant augmenting agents in U.S. clinical settings. Only our familiarity with clinical treatments in the United States and the literature on providers’ treatment preferences tempered our own surprise at these findings (1) .

Dr. Terao suggests the observed low use of lithium may have resulted because providers “automatically” diagnose patients who are prescribed lithium as having bipolar disorder rather than depressive disorders, and the latter were excluded from study analyses. He notes that lithium has an FDA indication for bipolar disorder but not for antidepressant augmentation. He further suggests that the definition of antidepressant augmentation, which required >60 days of overlap between an augmenting agent and an antidepressant, might have excluded patients who initiated but discontinued lithium in <60 days because of rapid treatment responses.

Unfortunately, we do not believe that these methodological issues explain our findings.

First, U.S. physicians commonly use medications “off-label” for non-FDA approved indications. Few would feel compelled to re-diagnose depressed patients as having bipolar I disorder—a diagnosis that requires a manic episode and has different prognostic and treatment implications—to be concordant with FDA indications. However, physicians may make subtle diagnostic “shifts” when there are less distinct boundaries. Thus, we included patients in our cohort who received both depression and bipolar II diagnoses. Even then, rates of lithium use were low.

We required ≥60 days of overlap between potential augmenting agents and antidepressant medications to avoid misclassifying patients as receiving augmentation when they were actually switching or cross-tapering medications. We assumed that physicians generally continue augmentation when patients have a good response. When we reduced the required days of overlap, rates of lithium augmentation remained low. Only 1,672 (0.68%) of the 244,859 patients in our study had ≥30 days overlap of lithium and an antidepressant and only 2,035 (0.8%) received any lithium at all during the year.

The data, in our judgment, continue to indicate that lithium augmentation is seldom used in clinical practice. The reasons are uncertain. Concerns about safety, tolerability, convenience and stigma may play a role. Lithium may be less effective when coupled with selective serotonin reuptake inhibitors than with tricyclic agents (2) , and the lithium use is not actively promoted by large pharmaceutical companies. Rigorous randomized controlled trials that directly compare lithium and the more popular augmenting agents are urgently needed.