From Wiki Journal Club
Jump to: navigation, search
Lewis G, et al. "Maintenance or discontinuation of antidepressants in primary care". The New England Journal of Medicine. 2021. 385(14):1257-1267.
PubMedFull textPDF

Clinical Question

Among adults on stable doses of antidepressant medications who are not currently depressed with ≥2 prior depressive episodes, what was the risk of depression relapse among those randomized to maintenance of their current therapy vs. discontinuation of their therapy?

Bottom Line

Among adults on stable doses of an antidepressant, discontinuation of antidepressant therapy was associated with a 2-fold higher risk of depression relapse when compared to maintenance of therapy.

Major Points

Depression is a common condition and antidepressants are commonly prescribed among adults, with approximately 19% of US adults aged ≥60 taking an antidepressant medication.[1] The evidence supporting the use of contemporary antidepressant medications comes from short-term trials.[2] The role for long-term continuation of antidepressant medications is unclear.

Published in 2021, the Antidepressants to Prevent Relapse in Depression (ANTLER) trial enrolled 478 adult primary care patients in the UK with a history of depression (~95% with ≥3 prior depressive episodes) on stable doses of certain antidepressant medications at specific doses. Individuals with a current depressive episode were excluded. Participants were randomized to maintenance of current therapy vs. discontinuation using a tapering of their current medication to full placebo. At 52 weeks, depression relapse occurred in 39% of those on maintenance vs. 56% on discontinuation (HR 2.06; 95% CI 1.56 to 2.70). The number needed to harm with discontinuation was 6.

ANTLER was limited by a large drop-off between screening to enrollment, as only 2% of those invited to participate were ultimately enrolled, lack of a diverse population, and only a small assortment of medications studied. This trial provides initial evidence for continuation of antidepressant medications among those on stable therapy.


As of September 2021, no guidelines have been published that reflect the results of this trial.


  • Multi-clinic, double-blinded, placebo-controlled RCT
  • N=478
    • Maintenance (n=238)
    • Discontinuation (n=239)
  • Setting: 150 general practices in England
  • Enrollment: The years in which enrollment occurred were not defined. Presumably it occurred in the late 2010s.
  • Follow-up: 52 weeks
  • Analysis: Presumably intention-to-treat
  • Primary outcome: Depression relapse


In-depth eligibility criteria are described in the supplementary appendix.[3]

Inclusion Criteria

  • Primary care patients aged 18-74 years
  • ≥2 prior depressive episodes or taking antidepressants for >2 years
  • Taking 1 of the following for ≥9 months at precisely this daily dose:
    • Citalopram 20 mg
    • Sertraline 100 mg
    • Fluoxetine 20 mg
    • Mirtazapine 30 mg
  • Recovered from most recent depressive episode and feeling well enough to stop antidepressants

Exclusion Criteria

  • Current depression at the time of trial entry, as assessed by a CIS-R computerized, self-administered tool
  • Use of escitalopram (not widely used in the UK), paroxetine (concern for withdrawal symptoms with discontinuation and declining use in the UK), and venlafaxine (concern for withdrawal symptoms with discontinuation and its status as a "second line" antidepressant agent)
  • Bipolar disorder, psychotic illness, dementia
  • Terminal condition
  • Unable to comprehend English questionnaires
  • Contraindication to medication or placebo
  • MAOI use
  • Women who were pregnant, planning pregnancy, or breast feeding

Baseline Characteristics

From the maintenance group

  • Demographics: Age 54 years, 71% female, 93% white race, 61% married, 59% currently employed
  • Depression details:
    • Medication: Sertraline 17%, citalopram 47%, fluoxetine 32%, mirtazapine 4%
    • Age of onset: 33 years
    • ≥3 prior depression events: 94%
    • On antidepressants for ≥3 years: 71%
    • PHQ9 score: 3.9
    • GAD7 score: 3.2
    • Short-Form Health Survey:
      • Physical component: 48
      • Mental component: 47
    • Modified Toronto Side Effect Scale: 4.2
    • Modified DESS score:
      • Number of new or worsening symptoms: 1
      • ≥1 new or worsening symptom: 50%
    • Mood worse than 2 weeks prior: 5%


  • Randomized to an arm:
    • Continuation - Participants continued their current antidepressant medication
    • Discontinuation - Participants tapered and discontinued their current antidepressant medication; the active drug was eventually replaced with placebo.
      • Details about the discontinuation protocol are documented in "Trial Treatments and Procedures" in the Methods section on page 1259.
  • Of note, the citalopram arm was closed early because the medication supply was exhausted prior to the end of recruitment.


Presented as maintenance vs. discontinuation.

Primary Outcome

Depression relapse at 52 weeks
39% vs. 56% (HR 2.06; 95% CI 1.56 to 2.70) NNH=6

Secondary Outcomes

Scores on symptom questionnaires at 52 weeks
12, 26, and 39 week scores are also shown in Table 2. EFfect size or difference shown in the parentheses.
PHQ9: 3.7 vs. 4.0 (0.4; 95% CI -0.3 to 1.1)
GAD7: 3.0 vs. 3.1 (0.3; 95% CI -0.4 to 0.9)
Modified Toronto Side Effect scale: 3.7 vs. 3.5 (0.0; 95% CI -0.4 to 0.5)
Modified DESS, # new or worsening symptoms: 0.8 vs. 1.1 (0.3; 95% CI 0 to 0.6)
Short-Form Health Survey Physical Component: 49 vs 49 (-0.6; 95% CI -2.1 to 0.9)
Short-Form Health Survey Mental Component: 47 vs. 46 (-1.6; 95% CI -3.4 to 0.2)

Additional Outcomes

Stopped taking study medication
30% vs. 48% (HR 2.28; 95% CI 1.68 to 3.08)
Adherence to protocol
70% vs. 52%

Subgroup Analysis

There was no difference in the primary outcome by medication subgroup, CIS-R depression score, CIS-R anxiety score, age of onset, and possibly 2 vs. >2 prior episodes of depression (only ~4% of the population had 2 prior depression episodes).

Adverse Events

Any serious adverse event
4% vs. 3%
Deaths: 0 vs. 0
Hospitalization: 3% vs. 3%


  • Only studied certain antidepressants at specific doses.
  • Little racial or ethnic diversity
  • There was a coding error in the capture of the baseline CIS-R's panic score so the maximum score for that section was 2 instead of the usual 4. This impacted the generation of the baseline anxiety variable.[3]
  • There was insufficient medication supply to provide full 12 months supply to 21 participants of those taking sertraline, so they received 8 to 11 months instead.[3]
  • For the final population of 478, 30,543 were pre-screened and 23,429 were invited to participate, of which 1,342 expressed interest in participating. Given the high melt-off from prescreening and invitation to enrollment, the generalizability of this study might be limited.


National Institute for Health Research in the UK

Further Reading

  1. Pratt LA et al. Antidepressant Use Among Persons Aged 12 and Over:United States,2011-2014. NCHS Data Brief 2017. :1-8.
  2. Rossi A et al. Fluoxetine: a review on evidence based medicine. Ann Gen Hosp Psychiatry 2004. 3:2.
  3. 3.0 3.1 3.2 Supplementary Appendix