for mild depression, does cbt work better than antidepressants, with fewer side effects?
“for the milder forms of depression, I think CBT works better than the drugs without some of the side effects that the the drugs have.”
a guest, historian of psychiatry (Andrew Scull) said on Lex Fridman Podcast, 2026-09-20, at 18:03. hear it in context →
first, the part that holds
antidepressants add little over placebo for milder depression, and they carry side effects that talking therapy does not. people also drop out of psychotherapy less often than out of drug treatment, and the benefit of psychotherapy holds up better at 6–12 months.
what we found when we went and looked
we read the head-to-head evidence and the drug side of the claim holds up. the part saying cbt "works better" does not. a 2020 network meta-analysis of 101 trials (11,910 patients) found no difference in response between psychotherapy and medication at the end of treatment (rr 0.99, 95% ci 0.92–1.08). an individual-patient meta-analysis of 16 cbt-versus-drug trials (1,700 patients) found a small edge for medication on one symptom scale, no difference in response or remission, and no sign that baseline severity changes the comparison. so the trials do not show cbt pulling ahead in milder cases. psychotherapy did come out ahead at 6–12 months and on dropout, and drug labels list side effects at roughly twice the placebo rate. for mild depression, the record supports "about as well, with fewer drug side effects and more lasting gains", not "better".
assertion by assertion
| assertion | verdict | what the record says |
|---|---|---|
| for milder depression, cbt works better than antidepressant drugs | overstated | at the end of acute treatment, the trials show the two are about equal. psychotherapy vs medication response rr 0.99 (95% ci 0.92–1.08) across 101 trials. in individual patient data from 16 cbt-vs-drug trials there was a small medication advantage on the hamilton scale (β −0.88, p=.03), no difference in response (or 1.24) or remission (or 1.18), and severity did not moderate the gap (β 0.00, p=.96). psychotherapy was ahead at 6–12 months (rr 0.85, 95% ci 0.74–0.98). limitation: most trials enrolled moderate-to-severe depression, and the network meta-analysis says there were too few mild-depression trials to analyse that group separately. [1] |
| cbt comes without the side effects that antidepressant drugs have | confirmed | a representative ssri label (sertraline) lists nausea at 26% vs 12% on placebo, diarrhea 20% vs 10%, insomnia 20% vs 13%, ejaculation failure 8% vs 1% and decreased libido 6% vs 2%, plus a boxed warning on suicidal thoughts and behaviors. cbt carries no pharmacological effects of this kind. psychotherapy also had better acceptability (all-cause dropout favoured psychotherapy, rr 1.17, 95% ci 1.02–1.32), and the authors suggest side effects as one reason. limitation: dropout is not a direct measure of side effects, and harms of psychotherapy are rarely recorded systematically in trials. [1][3] |
| antidepressants add little over placebo in mild-to-moderate depression | confirmed | in a patient-level meta-analysis of six placebo-controlled trials, the drug–placebo difference was below cohen's d 0.20 for patients with baseline hamilton scores under 23. the benefit grew with severity and crossed nice's threshold for clinical significance at a baseline score of 25. limitation: six trials, and only two drugs (imipramine and paroxetine). |
| cbt's advantage over placebo is larger than the drugs' in milder depression | false | in an individual-participant meta-analysis of pill-placebo-controlled trials, cbt's effect over pill placebo was small (smd −0.22). the authors put this alongside an antidepressant effect of about 0.31, and cbt's effect did not change meaningfully with initial severity. so the weak drug effect in mild depression is not matched by a stronger cbt effect there. both treatments show small effects over placebo. [2] |
both directions
for the speaker: the drug-versus-placebo gap is small for mild-to-moderate symptoms (d under 0.20 below a hamilton score of 23). psychotherapy is about as effective as medication at the end of treatment and does better at 6–12 months (rr 0.85). fewer people drop out of psychotherapy (rr 1.17). an ssri label lists nausea, diarrhea, insomnia and sexual side effects at about twice the placebo rate, plus a boxed warning. none of these drug effects come with cbt. if the claim were "at least as good, with fewer drug side effects", the record would back it.
against: no head-to-head analysis we read finds cbt beating medication at the end of acute treatment. the largest one (101 trials) shows a tie, rr 0.99. the patient-level cbt-versus-drug analysis found a small medication edge on one scale and no severity effect, so there is no evidence that cbt does relatively better in milder cases. cbt's own effect over pill placebo is also small (smd −0.22) and does not grow as severity falls. the network meta-analysis says there were too few mild-depression trials to test the speaker's exact subgroup.
what this cannot say
most comparison trials enrolled moderate-to-severe depression, so "mild" is inferred from severity-moderation analyses rather than tested directly. dropout is a rough stand-in for tolerability, and psychotherapy trials seldom record harms the way drug trials do. the placebo analysis covers six trials of older drugs. we could not open the nice ng222 guideline (it returned 403) or several pubmed pages, so national guideline wording is not cited here. our check covers adults with depressive disorders, not children, adolescents or subthreshold symptoms.
sources, each one fetched and read
- a network meta-analysis of the effects of psychotherapies, pharmacotherapies and their combination in the treatment of adult depression. World Psychiatry, 2020;19(1):92-107 What we read there: 101 studies, 11,910 patients, depression mostly moderate to severe; psychotherapy vs pharmacotherapy response RR 0.99 (95% CI 0.92-1.08); psychotherapy more acceptable (RR 1.17, 95% CI 1.02-1.32); psychotherapy better at 6-12 months (RR 0.85, 95% CI 0.74-0.98); 'there were too few studies in patients with mild depression.' Fetched 2026-10-21 (HTTP 200). https://pmc.ncbi.nlm.nih.gov/articles/PMC6953550/
- Initial severity of depression and efficacy of cognitive-behavioural therapy: individual-participant data meta-analysis of pill-placebo-controlled trials. British Journal of Psychiatry, 2017;210(3):190-196 What we read there: CBT standardized mean difference over pill placebo −0.22; 'Patients suffering from major depression can expect as much benefit from CBT across the wide range of baseline severity.' Fetched 2026-10-21 (HTTP 200). https://pmc.ncbi.nlm.nih.gov/articles/PMC5331187/
- US National Library of Medicine, DailyMed, sertraline tablet, film coated, prescribing information What we read there: Boxed warning: suicidal thoughts and behaviors. Most common adverse reactions (≥5% and twice placebo): nausea, diarrhea/loose stool, tremor, dyspepsia, decreased appetite, hyperhidrosis, ejaculation failure, decreased libido. Nausea 26% vs 12%, diarrhea 20% vs 10%, insomnia 20% vs 13%, ejaculation failure 8% vs 1%, decreased libido 6% vs 2%. Fetched 2026-10-21 (HTTP 200). https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=444ca054-e450-4c10-a31b-71ad43946cb8
speakers are described by role, not named. one sentence is quoted under fair use, with attribution and a link to the original at the timestamp; we do not re-host audio or video. nothing on this page is a reason to start, stop or change a medication. if you have the file behind anything marked unverifiable, the research board is where people can answer. adam