How-To Guides

Bupropion augmentation had the best overall balance of benefit and harm in the base-case model for nearly all patient subgroups. Compared with switching to bupropion, combination therapy with bupropion was preferred in every subgroup, and switching to bupropion was not the preferred strategy for any subgroup. The main exception was adults aged 851389 years with nonelevated weight, in whom aripiprazole augmentation was preferred because the modeled fall risk with bupropion augmentation weighed more heavily.

No. Aripiprazole augmentation was favored over switching to bupropion only in patients with nonelevated weight, while switching to bupropion was preferred in patients who were overweight at baseline. In adults under 65 years, aripiprazole provided 27.3 additional gross depression-free day equivalents versus switching to bupropion, but that benefit was partly offset by tardive dyskinesia in nonelevated-weight patients and more than offset by side effects in overweight patients.

Baseline BMI materially changed the treatment ranking. Among patients with elevated BMI at baseline, switching to bupropion was preferred over aripiprazole augmentation because the modeled harms of weight gain plus tardive dyskinesia outweighed aripiprazole's added antidepressant benefit. In overweight patients, weight-related harms from aripiprazole were estimated to be 95% to 131% as large as its depression benefit relative to switching to bupropion.

In adults younger than 65 years, aripiprazole augmentation provided 27.3 additional gross depression-free day equivalents compared with switching to bupropion. The authors also estimated that aripiprazole's efficacy advantage over bupropion augmentation was small, roughly equivalent to about 1 week earlier remission of depressive symptoms.

The modeled harms from tardive dyskinesia offset 56% to 81% of aripiprazole's depression benefit over switching to bupropion. The paper notes that the small risk of permanent tardive dyskinesia translated into an expected quality-of-life loss equivalent to about 2 to 3 weeks of additional depressive symptoms.

Weight gain was a major factor in the model, especially for patients who were already overweight. After 1 year of aripiprazole treatment, clinically significant weight gain was common, and in overweight adults the expected lifetime harms from weight gain were about as large as the depression benefit relative to switching to bupropion. The authors estimated that, for adults in whom further weight gain posed medical risk, the expected harm from weight gain was equivalent to about 1 month of active depression.

Fall risk reduced the net benefit of bupropion augmentation in older adults, but in the base case it did not outweigh its efficacy advantage over switching to bupropion at any age. The expected harm from falls was equivalent to about half a week of active depression in adults aged 65284 years and about 1.5 weeks in adults aged 851389 years. The impact of falls was most pronounced in the oldest subgroup, where aripiprazole became the preferred strategy for those with nonelevated weight.

Adults aged 851389 years with nonelevated weight were the only subgroup in which aripiprazole augmentation was preferred over bupropion augmentation. In that subgroup, aripiprazole offered 3.0 more depression-free day equivalents than bupropion augmentation and 10.9 more than switching to bupropion.

The oldest subgroup had the greatest uncertainty. Among adults aged 851389 years with nonelevated weight, probabilistic sensitivity analysis favored aripiprazole in 58.8% of runs, bupropion augmentation in 36.0%, and switching to bupropion in 5.2%. In adults aged 851389 years with elevated BMI, bupropion augmentation was preferred in 70.0% of runs, switching to bupropion in 15.8%, and aripiprazole in 14.2%.

Most rankings were robust to multiple sensitivity analyses. Changing the discount rate to 0% or 3%, shortening the time horizon from lifetime to 20 years, extending fall-related disability to 2 years, modeling weight gain as fully reversible after 3 years, or ignoring bupropion-related weight loss did not change the preferred treatment for any subgroup. One exception was that allowing some weight-gain harms to accrue even in nonelevated-weight patients gave bupropion augmentation a slight edge over aripiprazole in adults aged 851389 years with nonelevated weight.

The model focused on side effects that differed between treatments and could cause long-term harm. It explicitly modeled serious falls, clinically significant weight gain of at least 7% of baseline weight, and tardive dyskinesia. The authors did not include more readily reversible adverse effects such as anxiety, insomnia, akathisia, somnolence, and other acute extrapyramidal symptoms.

The main limitations were uncertainty in model parameters and model structure, incomplete capture of side effects, and limited generalizability of the underlying trials. The authors noted a relative lack of high-quality follow-up data on long-term consequences of short-term changes in depression status and one-time weight increases, and the model did not include all known or possible adverse effects. They also cautioned that VAST-D included some patients with only 1 prior failed antidepressant trial, OPTIMUM focused on older adults, and VAST-D enrolled veterans who were generally male and had high rates of comorbid PTSD, which may limit generalizability.