Section 3: The Antidepressant Controversy

This section addresses the most debated prescribing question in bipolar pharmacotherapy: whether antidepressants have any role in the treatment of bipolar depression. The evidence is genuinely contradictory, the clinical stakes are high, and the answer depends on the specific patient sitting in front of you.

3.1 The STEP-BD Findings: The Trial That Changed the Field

For years, the standard approach to bipolar depression was to establish a mood stabilizer as an "antimanic ceiling" and then add a traditional antidepressant. This paradigm was fundamentally challenged by the STEP-BD findings (Sachs et al., 2007, New England Journal of Medicine).

The Systematic Treatment Enhancement Program for Bipolar Disorder was a large, NIMH-funded initiative that included a 26-week, randomized, double-blind, placebo-controlled trial of adjunctive antidepressants (bupropion or paroxetine) added to mood stabilizer treatment in 366 patients with acute bipolar depression. The primary outcome was "durable recovery," defined as resolution of mood symptoms sustained for at least eight consecutive weeks (Sachs et al., 2007, New England Journal of Medicine).

The results were striking: durable recovery was achieved by 23.5% of patients in the antidepressant group and 27.3% in the placebo group. The difference was not statistically significant. Adding an antidepressant to a mood stabilizer provided no therapeutic advantage over mood stabilizer plus placebo for acute bipolar depression (Sachs et al., 2007, New England Journal of Medicine).

Beyond the efficacy failure, the broader STEP-BD dataset revealed potential iatrogenic risks (harms caused by the treatment itself, rather than by the illness). Researchers identified a pattern they termed antidepressant-associated chronic irritable dysphoria (ACID), a persistent state of agitated, irritable depression that may be associated with antidepressant exposure. In an analysis of the first 1,500 STEP-BD enrollees, patients who received antidepressants for bipolar depression had an unadjusted hazard ratio of 9.95 (95% CI 1.10--89.72; p = 0.04) for developing ACID compared to those who remained antidepressant-free. However, when the analysis was adjusted for significant covariates (notably prior antidepressant-related manic switch and sex), the hazard ratio dropped to 1.05 (p = 0.99), leading the study authors to conclude that ACID may not represent an independent phenomenon but rather part of a broader spectrum of treatment-emergent affective switches (El-Mallakh et al., 2008, Journal of Affective Disorders). The clinical relevance of ACID as a distinct entity therefore remains uncertain, though the broader risk of antidepressant-associated mood destabilization is well-established through other lines of evidence (see Section 3.2).

Subsequent analysis showed that the presence of even subsyndromal manic symptoms (mixed features) at the time of antidepressant initiation was a severe negative prognostic indicator. In patients with concurrent manic symptoms at intake, adjunctive antidepressant use was associated with significantly higher mania severity at follow-up (p = 0.006) (Goldberg et al., 2007, American Journal of Psychiatry).

3.2 The Risk of Destabilization: Short-Term vs. Long-Term

The risk picture from antidepressants in bipolar disorder is more nuanced than either blanket prohibition or blanket acceptance would suggest. A critical distinction is the difference between short-term acute treatment risk and long-term maintenance risk.

Short-term switch risk may be lower than historically feared. A 2024 Lancet Psychiatry systematic review and network meta-analysis investigated switch rates across 13 RCTs involving 1,362 patients during acute treatment (up to 10 weeks). This analysis found that no single antidepressant was associated with a significantly higher risk of switch to mania compared to placebo during the acute window. The authors concluded that the risk of switching may be more strongly influenced by patient-related intrinsic factors (younger age, early illness onset, rapid cycling history, prior switch episodes) than by the specific drug class (Oliva et al., 2025, EClinicalMedicine).

This finding aligns with the McGirr et al. (2016) meta-analysis, which found that acute, short-term treatment (up to 16 weeks) was not associated with a statistically significant increase in the risk of treatment-emergent affective switch (TEAS, the term recommended by the ISBD to describe antidepressant-associated mood elevation), with an RR of 0.926 (95% CI 0.576--1.491, meaning the true risk likely falls somewhere in this range; p = 0.753) (McGirr et al., 2016, The Lancet Psychiatry).

Long-term maintenance risk is real and quantifiable. The same McGirr meta-analysis told a very different story when examining 52-week extension data: prolonged, long-term antidepressant use was associated with a significant and substantial increase in the risk of TEAS, yielding an RR of 1.774 (95% CI 1.018--3.091; p = 0.043) (McGirr et al., 2016). This distinction between short-term safety and long-term risk is clinically important: an antidepressant that appears safe during an 8-week acute trial may carry meaningful destabilization risk if continued for a year.

Cycle acceleration: when the NNH is worse than the NNT. Beyond acute switch events, antidepressants carry the insidious potential to induce cycle acceleration, sometimes called "roughening" of the illness course, producing more frequent mood shifts and fewer stable days over time. This longitudinal harm was quantified by Ghaemi and colleagues in a meta-analysis evaluating long-term antidepressant maintenance in bipolar I disorder (Ghaemi et al., 2008, Acta Psychiatrica Scandinavica). Adding an antidepressant to a mood stabilizer yielded only modest protection against depressive recurrences (RR = 0.73, a 27% risk reduction; NNT = 11). But that same maintenance antidepressant therapy was associated with a 72% increase in the risk of a new manic or hypomanic episode (RR = 1.72; 95% CI 1.23--2.41; NNH = 7) (Ghaemi et al., 2008, Acta Psychiatrica Scandinavica).

Because the NNH (7) is lower than the NNT (11), a clinician using long-term maintenance antidepressants is statistically more likely to induce a manic episode than to prevent a depressive recurrence. This is one of the most clinically actionable statistics in bipolar pharmacotherapy: it directly informs the decision to taper antidepressants after acute remission rather than continuing them indefinitely.

This layered risk picture, with short-term acute treatment appearing relatively safe and long-term maintenance carrying real and quantifiable harm, is what makes the antidepressant question genuinely difficult rather than simply answerable with a prohibition.

PsychHQ

The Antidepressant Controversy: STEP-BD & Destabilization Risk