Adjunctive Antidepressant for Bipolar Depression Against a Null Trial Result
A single patient, six weeks into a breakthrough depressive episode on lithium. STEP-BD found no average benefit from adding an antidepressant to a mood stabilizer — but this patient has his own prior positive response to the same drug, and the two kinds of evidence point in different directions.
T.B. is a 52-year-old man, a recently retired machinist and longtime amateur woodworker, married for 28 years, who has been maintained on lithium for bipolar I disorder for the past decade with generally good control. Six weeks ago he sank into a depressive episode — low mood, poor concentration, and a loss of interest in the woodworking shop he built in his garage after retiring, which his wife says is the clearest sign to her that something is wrong. His lithium level is therapeutic and was raised further two weeks ago without meaningful improvement. His psychiatrist then tried quetiapine XR augmentation, an FDA-approved option for bipolar depression, but he stopped it after four days because the daytime sedation left him unsteady enough that he did not feel safe running a table saw, which for him is not a minor inconvenience but the activity that structures most of his retired life.
The question now is whether to add a conventional antidepressant on top of his lithium. STEP-BD, the largest randomized trial to address this directly, found no significant benefit of adjunctive antidepressant over mood stabilizer plus placebo for bipolar depression — and yet adding one remains extremely common in practice, including for T.B. himself: six years ago, during a different depressive episode on the same lithium regimen, bupropion added for about four months produced what both he and his wife describe as a clear, meaningful improvement, well tolerated, with no hint of a manic or mixed switch. A population-level null trial result and one patient's own prior positive response are not making the same claim, and the disagreement here is really about which kind of evidence should carry more weight for the decision in front of them.
Weighing a trial result against his own history
STEP-BD is the best evidence we have on this exact question, and it found adjunctive antidepressant no better than mood stabilizer alone for the average patient with bipolar depression. Reaching for a bupropion trial because it is familiar and low-effort, rather than pursuing another agent with actual positive trial evidence for bipolar depression, is exactly the pattern that keeps this practice common despite the data.
The reassuring part of that same trial is that it also found no significant increase in switch risk from adding an antidepressant to a mood stabilizer — so my objection is about expected benefit, not about safety.
STEP-BD describes an average patient, and he is not an average patient with respect to this specific question — he has his own n-of-1 trial from six years ago, same lithium regimen, same drug, clear and well-tolerated response with no switch. That is a stronger predictor of what will happen to him than a population trial average that includes many patients for whom nothing in his history applies.
I don't think citing STEP-BD settles this the way it would for a treatment-naive patient. Individual response history has always been allowed to override a population-level null result in psychiatric prescribing — that's the premise behind trying a drug again after it worked once.
Whatever gets decided about the antidepressant question, the quetiapine intolerance is a real, separate problem that needs its own answer — he lost a treatment option not because it failed to work, but because it interfered with something that matters to his daily functioning and his sense of purpose in retirement. That should weigh in the decision, not just get noted and set aside.
Bupropion XL 150mg daily added to his unchanged lithium regimen, based on his own documented prior response rather than the population-average STEP-BD result. Follow-up scheduled at two weeks to assess depressive symptoms and screen explicitly for any early signs of hypomanic switch, and again at six weeks to reassess whether the response is repeating what happened six years ago.
The disagreement about how much weight a null population trial should carry against one patient's own history was not resolved in principle — both positions were left standing, with the plan itself serving as the actual test: if bupropion works as it did before, the individualized read is supported; if it does not, the pharmacologist's caution about defaulting to a familiar drug over one with real trial support for bipolar depression specifically will be revisited.