Lewis · The New England journal of medicine 2021 · double-blind randomized controlled trial · n=478

Maintenance or Discontinuation of Antidepressants in Primary Care.

Cited 196 times in the scientific literature.

Level 2 - randomized trial

Individual double-blind randomized controlled trial.

PubMed 34587384 · doi:10.1056/NEJMoa2106356 · record verified 2026-08-30

What was done

A double-blind randomized controlled trial was conducted across 150 primary care practices in the United Kingdom. Adults with a history of at least two depressive episodes or who had taken antidepressants (citalopram, fluoxetine, sertraline, or mirtazapine) for 2 years or longer, and who felt well enough to stop, were randomized 1:1 to maintain their current antidepressant therapy (n = 238) or taper and discontinue to matching placebo (n = 240). The primary outcome was time to first depression relapse over 52 weeks. Secondary outcomes included depressive, anxiety, physical, and withdrawal symptoms, quality of life, and global mood.

What was found

Of 1466 screened patients, 478 were enrolled (mean age 54 years; 73% women). Trial assignment adherence was 70% in the maintenance group and 52% in the discontinuation group. By 52 weeks, relapse occurred in 39% (92/238) of the maintenance group versus 56% (135/240) of the discontinuation group (hazard ratio, 2.06; 95% confidence interval, 1.56 to 2.70; P < 0.001). Patients in the discontinuation group also experienced higher rates of depression, anxiety, and withdrawal symptoms.

Why it matters

This study provides clear randomized evidence that discontinuing long-term antidepressants in primary care patients who feel recovered significantly increases the risk of depressive relapse over a 1-year period compared to continuing maintenance therapy.

Limits

Substantial non-adherence occurred during the 52-week period, particularly in the discontinuation arm (48% non-adherence vs. 30% in maintenance). The trial only evaluated four specific antidepressants and excluded patients who did not feel ready to stop medication. Follow-up was limited to 1 year, leaving longer-term outcomes unmeasured. The study cohort was predominantly female (73%) and had a high baseline risk (recurrent episodes or at least 2 years of therapy), limiting generalizability to shorter-term or single-episode depression.

Cited by