Treatment for depression does not always end when a person starts feeling better. Even after symptoms have noticeably eased, there is still a risk that they will return. That is why one of the main goals of treatment is not only to achieve improvement, but also to sustain it over time. But are all antidepressants equally effective at doing this?
An international group of researchers compared the effectiveness of 21 antidepressants, focusing on sustained treatment response–the ability of medications to maintain the initial improvement in symptoms during the subsequent course of treatment.
The results showed that escitalopram, duloxetine, and agomelatine had an advantage over fluoxetine in terms of sustained response. This means that patients taking these medications were more likely to maintain their improvement throughout the treatment period examined.
A systematic review and network meta-analysis published on November 10, 2026, in The Lancet Psychiatry offers a new way to assess the long-term effectiveness of antidepressants. However, the findings do not mean that these three medications are a better fit for every patient than the others. Treatment choice depends on symptom severity, coexisting conditions, tolerability, and an individual’s response to treatment. The study also does not provide grounds for changing or stopping an antidepressant that is already working without consulting a doctor.
What the researchers compared
The review included 115 double-blind randomized studies involving 30,428 participants. The studies compared antidepressants with one another or with a placebo, and treatment lasted more than 12 weeks. However, the main comparison of the effectiveness of individual medications drew on a narrower subset of the data: 35 studies involving around 10,000 participants.
The authors assessed sustained treatment response. As explained in the Kyoto University materials, this measure combines two events: symptoms first improve, and then that improvement is maintained without relapse during the continuation or maintenance phase of treatment.
This approach addresses a practical question about the overall course of treatment. A medication should help a person feel better and sustain that progress. However, a treatment response does not necessarily mean full remission: some symptoms may remain even when there is noticeable improvement.
That is why describing the findings as a “relapse prevention ranking” oversimplifies them. The authors examined whether people achieved and sustained improvement, not only whether depression returned in people who had already recovered.
How escitalopram ranked
Escitalopram, duloxetine, and agomelatine showed a statistically significant advantage over fluoxetine in terms of sustained response. The odds ratio was 1.62 for escitalopram, 1.57 for duloxetine, and 1.40 for agomelatine.
These figures should be interpreted carefully. An odds ratio of 1.62 does not mean a 62% reduction in relapse risk or a 62% increase in the chance of improvement. Odds and probability are calculated differently. Without knowing the baseline probability of the outcome, the relative estimate cannot be translated into an intuitive number of additional people out of 100 who would benefit from one medication rather than another.
Overall, the findings identify these three medications as promising options for achieving and sustaining a response. On their own, they do not show that escitalopram is definitively better than duloxetine or agomelatine, or that any of the three is superior to all other medications.
Escitalopram is a selective serotonin reuptake inhibitor. It blocks the transporter protein that returns serotonin to the nerve cell, thereby altering serotonin signaling. But its mechanism of action does not automatically explain its advantage in this comparison. The meta-analysis assessed clinical outcomes, not the biological reasons for differences between medications.
How reliable are the comparison results?
A network meta-analysis makes it possible to compare medications even when they have rarely been compared directly. For example, the results of trials comparing two medications with a shared comparator can help estimate differences between them indirectly. This expands the scope of analysis, but makes the conclusions dependent on how comparable the participants and conditions were across different studies.
An expert review by the Science Media Centre notes that 19 of the 35 studies in the main comparison had a high risk of bias. In other words, features of how these trials were conducted may have skewed the effectiveness estimates. The large overall size of the review does not eliminate the limitations of individual studies.
A second important limitation is the timeframe. The typical follow-up period was around six months. This is useful for assessing whether initial improvement is sustained, but the comparative results cannot be confidently extrapolated to taking antidepressants for years.
Finally, treatment acceptability should be distinguished from safety. If a person stops taking a medication for any reason, this may be due to side effects, insufficient benefit, or other circumstances. Rates of treatment discontinuation cannot establish that a particular medication has the fewest adverse reactions.
What the findings mean for patients
The review gives doctors additional evidence to discuss when choosing a medication, especially when treatment is just beginning. But the decision also depends on previous responses to treatment, coexisting conditions, interactions with other medications, and tolerability. In a statement from Radboud University Medical Center, co-author Eric Ruhe emphasizes that the findings do not mean escitalopram should be prescribed to everyone.
Like other antidepressants, escitalopram can cause adverse effects, including nausea and changes in sleep and sexual function. So higher average effectiveness on a single measure does not determine the balance of benefits and risks for a particular person.
The diversity of depression is one reason treatment can be difficult to tailor. Attempts to link it to differences in brain function, including research into neuroprofiles of depression, have not yet provided a universal way to choose the right medication in advance.
If a prescribed medication is helping and is reasonably well tolerated, this review alone is not a reason to switch to escitalopram. The study did not examine whether switching would benefit patients who are stable. Abruptly stopping an antidepressant can cause withdrawal symptoms; any change to the treatment plan should be discussed with the prescribing doctor.
For that conversation, it is more useful to describe specific changes: whether interest in usual activities has returned, whether sleep has improved, which symptoms remain, and which side effects are getting in the way. Being able to keep working is not a substitute for this assessment–a point that is especially clear in discussions of “high-functioning depression”. The practical purpose of long-term treatment is to sustain clinically meaningful improvement, not simply to keep taking medication out of habit.
Can this review help me choose an escitalopram dose?
No. The comparative findings presented here do not determine the right dose for any particular person. The dose is chosen based on the clinical situation, response to treatment, tolerability, and other medications being taken.
Do the findings apply to the treatment of anxiety disorders?
They cannot be directly extrapolated to anxiety disorders. The review focused on depression; a medication’s effectiveness for another condition needs to be assessed separately, even if the medication is used for both disorders.
Does the study show that antidepressants are better than psychotherapy?
No, this analysis does not answer that question. It included comparisons of antidepressants with one another or with a placebo. These comparisons cannot establish whether medication is more beneficial than psychotherapy or whether combining the two is helpful.