Depression is common, and it is treatable. It runs on a spectrum, from a low mood that lifts on its own to a major depression that lasts most of the day, most days, for weeks, and the more severe it is, the more the strongest treatments can do. The two first-line treatments are talking therapy and antidepressant medication, and across trials they work about equally well.
Alongside them, several things you can start yourself have trial evidence behind them, exercise most of all, with morning light, treating broken sleep, and staying in contact with other people close behind. The popular story that depression is simply a shortage of serotonin has not held up, and the medications still help many people, because a treatment can work without the illness being a shortage in the pathway it acts on. If you are thinking about suicide, that is a medical emergency, and the crisis routes are near the top of this page.
This page covers:
- what helps
- in what order
- the Chinese medicine reading
- the signs that need help the same day
Practice Ranking
Every practice we track for Depression, ranked by how well the evidence supports it for this condition. Strength describes the evidence, not our endorsement.
10 practices · 3 to start with
| # | Practice | Evidence | Type | Cost | Effort | Results In | Add to plan |
|---|---|---|---|---|---|---|---|
| 1 | Cognitive Behavioral Therapy First-line talk therapy; behavioral activation matches full CBT and lowers relapse. | Strong | Self-Directed | Free to $$ | Moderate to Hard | Weeks | |
| 2 | Walking Aerobic exercise, especially walking or jogging, was the strongest movement lever for depression, though only 1 of 218 trials was low-risk. | Moderate | Self-Directed | Free | Easy | Days to Longer | |
| 3 | Morning Light Bright morning light rivals antidepressants for seasonal depression and beat placebo in non-seasonal cases. | Moderate | Self-Directed | Free | Easy | Days to Weeks | |
| 4 | Sleep Restriction & Stimulus Control Treating co-occurring insomnia nearly doubled depression response, and insomnia roughly doubles the odds of later depression. | Moderate | Self-Directed | Free to $$ | Hard | Weeks | |
| 5 | Omega-3 & Fish Oil EPA-rich omega-3 eased existing depression by about 2.5 Hamilton points but did not prevent it. | Moderate | Supplement | $ to $$ | Easy | Weeks to Months | |
| 6 | Vitamin D Vitamin D did not prevent depression; the strength of this null does not make it a starting move. | Strong | Supplement | $ | Easy | Weeks to Months | |
| 7 | Saffron Saffron eased mild-to-moderate depression by about 0.89 SD over placebo. | Moderate | Supplement | $ to $$ | Easy | Weeks to Months | |
| 8 | Connection & Belonging Frequent loneliness more than doubled depression odds, so rebuilding connection is a genuine lever. | Emerging | Self-Directed | Free | Moderate | Weeks to Longer | |
| 9 | Time in Nature A 90-minute nature walk lowered brooding where an urban walk did not; early evidence. | Preliminary | Self-Directed | Free | Easy | Days to Weeks | |
| 10 | Acupuncture Small effect on depression, low-quality trials; education-level. | Preliminary | Self-Directed | Free to $$$ | Easy to Moderate | Weeks to Months | |
Default order puts the best-supported practices first, with self-directed care ahead of clinical options. Click any row to open the practice.
What It Is
If you are thinking about suicide right now, treat it as a medical emergency.
Contact your local emergency services, or the suicide and crisis line your own country's health service publishes, and if someone is with you, ask them to stay. Those routes are set out again in the "When to See Someone" section below. Asking someone directly whether they are thinking of ending their life does not put the idea there.
Depression is a medical condition, not a weakness and not a mood you should be able to think your way out of. Major depression means low mood or a loss of interest and pleasure that lasts most of the day, most days, for two weeks or more, usually alongside changes in sleep, appetite, concentration, energy, and a sense of your own worth. It sits on a spectrum. A low mood that tracks events and lifts within a normal week is ordinary and usually not a medical depression. At the other end, a severe depression can make it hard to work, eat, or get out of bed, and it can be dangerous.
For decades depression was explained as a chemical imbalance, a shortage of serotonin in the brain. Careful reviews of that idea have not found consistent support for it: the metabolite studies, the brain-imaging studies, and the two largest gene studies of the serotonin transporter all come up empty. This is often misread as "antidepressants do not work," and that is the wrong conclusion. A treatment can help without the illness being a shortage in the pathway it acts on. The drugs help many people, modestly on average and more clearly in severe depression, and the evidence for that sits in the research below.
A few things produce the same low mood and are worth checking once, because they change the treatment. These can all mimic depression:
- An underactive thyroid
- Low vitamin B12
- Sleep apnea
- Heavy alcohol use
- Some medications
Bipolar disorder can look like plain depression between its high phases, and it matters to catch, because an antidepressant on its own can tip someone with bipolar disorder into mania. Depression after a birth, postpartum depression, is common, treatable, and sometimes urgent. A doctor can rule the common causes out with a few questions and a blood test, and you can order the thyroid and B12 checks yourself through direct-to-consumer lab testing if that is easier.
What Helps
Depression is one of the most heavily studied conditions in medicine, and most people improve with treatment. Match the effort to the severity. For mild depression, the levers you can start yourself, with or without a course of therapy, are often enough. For moderate to severe depression, those levers still help, and therapy or medication does more of the work. If low mood or loss of interest has lasted most of the day, most days, for two weeks or more, that is the point to talk to a doctor. Nothing here asks you to have tried harder first.
The two first-line treatments are talking therapy and antidepressant medication. Neither is the automatic choice, so it comes down to a few things:
- what you can get access to
- whether you would rather work with a therapist or start a medication
- what you have tried before
They combine well, and for many people the two together beat either one alone. Among the therapies, cognitive behavioral therapy, behavioral activation, and interpersonal psychotherapy (IPT) all have solid trial evidence, and the type matters less than getting into one. Behavioral activation can be delivered by less specialized staff, which shortens the wait. Among the medications, the average benefit is small and grows with severity, so they earn their place most in moderate to severe depression, and they take a few weeks to work.
Several things you can start on your own have trial evidence behind them, and they fall in a clear order. Exercise is the strongest. Walking or jogging had the largest effect of any exercise mode, and the number to trust is the one measured against active comparators, a Hedges g of -0.62; the softer waiting-list comparisons that circulate are inflated. Two other lines of evidence point the same way, a dose-response across large cohorts and a genetic study built to separate cause from effect, so the direction is more trustworthy than the exact decimals.
Morning light comes next, with its clearest evidence in winter, seasonal depression. Treating broken sleep is one of the higher-return moves, because insomnia and depression each raise the risk of the other, and a short structured program for the insomnia lifts the depression along with the sleep. Staying in contact runs the same two ways: loneliness raises the risk, and contact lowers it.
Supplements are the most oversold part of this list. In the large VITAL trial, vitamin D and fish oil did not prevent depression, and those two results come from one randomization reported in two papers, not two independent trials; fish oil eases an existing depression a little, but as prevention it did nothing. Among herbs, three have trial evidence for milder depression:
- St John's wort matched standard antidepressants with fewer side effects, but its interaction risk is the deciding factor and sits in the Cautions section, and three of the review's authors had past funding from a manufacturer.
- Saffron did about as well as antidepressants in mild-to-moderate depression, on small trials from mostly one country.
- Chinese herbal formulas beat placebo and roughly matched antidepressants, but nearly all that literature comes from journals that almost never publish a negative result, so the true effect is likely smaller.
The Research & Studies
Everything here is based on the research we have collected and checked, sorted into groups and ordered with the strongest evidence first. Click any claim to open the studies behind it.
Mood & stress
All 21 antidepressants beat placebo, odds ratios 1.37 to 2.13
Every one of the 21 antidepressants tested worked better than a dummy pill, with the strongest roughly twice as likely to produce a response. The benefit is modest on average, and larger the more severe the depression.
All 21 antidepressants studied were more effective than placebo on response rate, with odds ratios from 2.13 (95% CrI 1.89 to 2.41) for amitriptyline down to 1.37 (1.16 to 1.63) for reboxetine. Only agomelatine and fluoxetine were better tolerated than placebo; clomipramine was worse. In head-to-head trials, agomelatine, amitriptyline, escitalopram, mirtazapine, paroxetine, venlafaxine and vortioxetine were more effective than the other drugs. Measured in: 116,477 adults with major depressive disorder across 522 double-blind randomized trials, published and unpublished, to January 2016. 46 of the 522 trials (9%) were at high risk of bias and 380 (73%) at moderate, and the certainty of evidence was moderate to very low. Trials run about 8 weeks, which is short next to how long these drugs are taken, and the analysis excluded treatment-resistant and psychotic depression, so it does not describe the people for whom first-line treatment has already failed.
Who this may not transfer to:62.3% women.
The study · 1
SSRIs beat placebo by 1.94 Hamilton points, just short of the 3-point clinical-significance bar
On the standard depression rating scale, SSRIs beat placebo by about 2 points on the 17-item Hamilton depression scale, which the reviewers judged just short of their pre-set bar for a change a patient clearly feels. The group average hides a wide split between strong responders and non-responders.
SSRIs reduced the 17-item Hamilton Depression Rating Scale by a mean of 1.94 points more than placebo (95% CI -2.50 to -1.37, 49 trials), which is below the 3-point threshold the review set in advance for clinical significance. They reduced the risk of no remission (RR 0.88) and increased serious adverse events (OR 1.37), corresponding to 31 per 1,000 on SSRIs against 22 per 1,000 on placebo. Measured in: 27,422 adults with major depressive disorder across 131 randomized placebo-controlled trials to January 2016. The review judged all 131 trials to be at high risk of bias, and its conclusion that harms outweigh benefits is a contested reading of the same data that other groups read as a modest benefit. A pre-set 3-point threshold is one convention among several, and a group average conceals the split between strong responders and non-responders.
Who this may not transfer to:The review does not report the pooled sex composition of the included trials.
The study · 1
Jakobsen et al., selective serotonin reuptake inhibitors versus placebo in patients with major depressive disorder, systematic review with meta-analysis and Trial Sequential Analysis · BMC Psychiatry 2017
All mainstream psychotherapies beat usual care, effect sizes -0.32 to -0.81
Every mainstream talking therapy beat doing nothing, and most beat a dummy pill, with effect sizes against usual care from about -0.32 to -0.81. The therapies did not clearly differ from each other, so the type matters less than getting into one.
All therapy types beat care-as-usual and waiting list, and all except non-directive supportive counseling and psychodynamic therapy beat pill placebo. Standardized mean differences against care-as-usual ranged from -0.81 for life-review therapy to -0.32 for non-directive supportive counseling. Individual therapies did not differ significantly from each other apart from non-directive supportive counseling, which was less efficacious than the rest. Results held when only low risk-of-bias studies were included. Measured in: 34,285 adults with depression across 331 randomized trials of cognitive behavioral, interpersonal, psychodynamic, problem-solving, behavioral activation, life-review and third-wave therapies. Nobody in a psychotherapy trial is blinded and the outcome is self-reported by the same unblinded participant, so expectation is inside every one of these numbers. The finding that the therapies do not differ from each other may also reflect trials being underpowered to detect differences between two active treatments.
Who this may not transfer to:The network meta-analysis does not report the pooled sex composition; adult depression trials typically enrol more women than men.
The study · 1
Cuijpers et al., psychotherapies for depression, network meta-analysis covering efficacy, acceptability and long-term outcomes · World Psychiatry 2021
Cognitive behavioral therapy cuts depression, effect size 0.75 (0.65 after publication bias)
Cognitive behavioral therapy produces a solid improvement in major depression, an effect size of about 0.75. It stays clear at 0.65 even after correcting for the studies that go unpublished when results are weak.
The pooled effect for major depression was g = 0.75, falling to g = 0.65 after adjustment for publication bias. Effects were large when the comparator was a waiting list and small to moderate when it was care-as-usual or a pill placebo. Measured in: 144 randomized trials (184 comparisons) in adults with major depression, generalized anxiety disorder, panic disorder or social anxiety disorder diagnosed by structured interview. Only 17.4% of the included trials were rated high quality, and the authors conclude the effects remain uncertain and should be treated with caution. The number quoted for CBT in general use is usually the waiting-list figure, which is the least demanding comparison available.
Who this may not transfer to:The review does not report the pooled sex composition of the 144 trials.
The study · 1
Cuijpers et al., how effective are cognitive behavior therapies for major depression and anxiety disorders, meta-analytic update · World Psychiatry 2016
Behavioral activation matched full CBT, both arms at PHQ-9 8.4 a year later
Behavioral activation, a simpler therapy that can be delivered by less specialized staff, worked as well as full cognitive behavioral therapy: both groups landed at the same PHQ-9 score of 8.4 a year later. That matters because it can reach the many people stuck on therapy waiting lists.
Behavioral activation delivered by junior mental health workers was non-inferior to CBT delivered by qualified psychological therapists at 12 months: PHQ-9 8.4 points in both arms, mean difference 0.1 points (95% CI -1.3 to 1.5, p = 0.89), against a pre-set non-inferiority margin of 1.9 PHQ-9 points. The per-protocol analysis agreed. Measured in: 440 adults meeting DSM-IV criteria for major depressive disorder recruited from primary care and psychological therapy services in Devon, Durham and Leeds, randomized 221 to behavioral activation and 219 to CBT. Treatment was open label, so participants and therapists knew which arm they were in, and 21% of the behavioral activation arm and 14% of the CBT arm had no primary outcome data at 12 months. People who were acutely suicidal or had attempted suicide in the previous two months were excluded, which is the group a reader in crisis belongs to. Non-inferiority is not superiority.
Who this may not transfer to:The trial enrolled adults of both sexes from UK primary care; the published abstract does not give the split.
The study · 1
Interpersonal psychotherapy beat control conditions, effect size 0.63
Interpersonal psychotherapy, which works on relationships and life changes, clearly beat no treatment (effect size 0.63) and matched other talking therapies. Kept up alongside medication, it also cut the chance of relapse.
Against control conditions, IPT gave d = 0.63 (95% CI 0.36 to 0.90) across 16 studies, a number needed to treat of 2.91. It did not differ from other psychological treatments (d = 0.04). Pharmacotherapy was slightly more effective than IPT after removal of one outlier (d = -0.19). Combined maintenance treatment with medication prevented relapse better than medication alone (OR 0.37). Measured in: 4,356 patients across 38 randomized trials including acute treatment, combination and maintenance studies. The control condition in most of the 16 comparisons was no treatment or usual care rather than an active alternative, which inflates the effect relative to a placebo comparison. Combination acute treatment was not better than IPT alone, but too few studies tested it to draw a conclusion.
Who this may not transfer to:The review does not report the pooled sex composition; the IPT literature includes several perinatal trials in women only.
The study · 1
Cuijpers et al., interpersonal psychotherapy for depression, meta-analysis · Am J Psychiatry 2011
Vitamin D did not prevent depression, hazard ratio 0.97
A daily vitamin D pill taken for over five years did not prevent depression in older adults, matching placebo almost exactly (hazard ratio 0.97). This tested prevention in people who were not deficient, so it says nothing about treating an existing depression or about people who are truly low in vitamin D.
Over a median 5.3 years, 2,000 IU a day of vitamin D3 did not change the risk of depression or clinically relevant depressive symptoms compared with placebo (609 events, 12.9 per 1,000 person-years, against 625 events, 13.3 per 1,000 person-years; hazard ratio 0.97, 95% CI 0.87 to 1.09). Mean change in PHQ-8 mood score did not differ from zero between groups. Measured in: 18,353 US adults aged 50 or older within the VITAL trial, mean age 67.5, 49.2% women; 16,657 with no depression history and 1,696 with a history but untreated for two years. This tests prevention in people without clinically relevant depressive symptoms at baseline, so it says nothing about treating an existing depression, and nothing about people who are actually vitamin D deficient, since participants were not selected for low levels. Everyone was 50 or older. This and the other supplement-prevention row on this page come from ONE randomization reported in two papers, not from two independent trials.
Who this may not transfer to:49.2% women, so the null applies about equally to both sexes in this age band.
The study · 1
Drug advantage is minimal below Hamilton 23 and clinically meaningful by 25
The more severe the depression, the bigger the drug's edge over placebo. It is small for milder depression, starting from a baseline Hamilton score under 23, and becomes clearly meaningful once the baseline score reaches 25.
In patient-level data, the medication minus placebo difference was below Cohen's d of 0.20 for patients with baseline Hamilton scores under 23, and crossed the National Institute for Clinical Excellence threshold for a clinically significant difference at a baseline score of 25. Measured in: 718 adult outpatients across 6 randomized placebo-controlled trials for which the authors supplied individual patient data. Six trials and 718 patients is a narrow base for a widely-quoted conclusion, and the drugs studied were paroxetine and imipramine rather than the modern range. Baseline severity is also confounded with regression to the mean, since the most severe scores have the most room to fall on any arm.
Who this may not transfer to:The paper does not report the pooled sex composition of the six trials.
The study · 1
Fournier et al., antidepressant drug effects and depression severity, patient-level meta-analysis · JAMA 2010
Walking or jogging cut depression most, Hedges g of -0.62, but only 1 of 218 trials was low-risk
Exercise treats depression. Walking or jogging had the biggest effect, a Hedges g of -0.62, with yoga, strength training, and tai chi close behind. The weakness is study quality: only one of 218 trials met the top bar for low risk of bias, so trust the direction more than the exact number.
Against active controls the strongest mode was walking or jogging (Hedges g of -0.62, 95% credible interval -0.80 to -0.45), then yoga (-0.55), strength training (-0.49), mixed aerobic exercise (-0.43), and tai chi or qigong (-0.42). Across all 218 trials only one met the Cochrane criteria for low risk of bias, so CINeMA confidence was low for walking or jogging and very low for the other modes. The analysis also reports a sex moderator: strength training and cycling looked better in women, while yoga, tai chi, and aerobic exercise combined with psychotherapy looked better in men. Measured in: 14,170 participants meeting clinical cut-offs for major depression across 218 randomized trials with 495 arms. The confidence ratings run from low to very low, so this evidence carries the direction more reliably than the exact effect sizes, and the page leans on the cohort dose-response and a Mendelian randomization result for that reason. The sex moderator comes from the same low-certainty pool, so it is a lead, not a prescription.
Who this may not transfer to:The analysis reports results moderated by sex, which is unusual in this literature and is why the moderator appears above rather than being folded into a general caveat.
The study · 1
Half the recommended activity tracks 18% less later depression, the full dose 25%
People who are more active are less likely to become depressed later, and most of the gain comes early: half the recommended amount of activity tracked 18% less depression, the full recommended amount 25%. This watches groups over time, so part of it is that early, undiagnosed depression itself reduces activity.
An inverse curvilinear dose-response, steepest at low volumes. Compared with adults reporting no activity, those accumulating half the recommended volume (4.4 marginal MET-hours per week) had 18% lower risk of depression (95% CI 13% to 23%), and those at the recommended 8.8 marginal MET-hours per week had 25% lower risk (18% to 32%). Benefits diminished and uncertainty grew beyond that. Measured in: 191,130 adults across 15 prospective cohort studies with 2,110,588 person-years of follow-up, each with at least 3,000 adults and 3 years of follow-up. What could explain it instead: Reverse causation is the main one: early, undiagnosed depression reduces activity months before it is recorded as an outcome, which makes inactivity look like a cause. Healthy-user bias runs the same way, since people who exercise also sleep, eat, drink and socialize differently.. Heterogeneity was large and significant (I2 = 74%), activity was mostly self-reported, and prospective association is not a treatment effect: nothing here says that adding activity to an already-depressed person produces the same 25%.
Who this may not transfer to:The pooled cohorts include both sexes; the paper does not report a single pooled proportion.
The study · 1
Pearce et al., association between physical activity and risk of depression, systematic review and dose-response meta-analysis · JAMA Psychiatry 2022
Genetics point from activity to less depression, odds ratio 0.74 per 1-SD
Using genetics to sort cause from effect, the arrow ran from more activity to less depression (odds ratio 0.74 per standard-deviation increase), not the other way. It held only for activity measured by a wearable, not for self-reported activity, and only in people of European ancestry.
Using genetic instruments, accelerometer-measured physical activity was protective against major depressive disorder (OR 0.74 per 1-SD increase in mean acceleration, 95% CI 0.59 to 0.92, p = 0.006). There was no significant effect in the reverse direction, and self-reported activity showed no significant relationship with depression in either direction. Measured in: 611,583 adults of European ancestry across non-overlapping genome-wide association studies: 91,084 with accelerometer data, 377,234 self-reported, and 143,265 in the major depressive disorder sample. What could explain it instead: Horizontal pleiotropy is the standing threat in any Mendelian randomization: the variants used for activity may affect depression through some other route, such as general health or body composition. The authors used weighted median, MR Egger and MR-PRESSO to test for it, which reduces the concern without removing it.. The result holds only for objectively measured activity and not for self-reported activity, which is a discrepancy the design cannot explain. All participants were of European ancestry, so it does not transfer automatically to other populations, and the odds ratio describes lifelong genetic tendency rather than the effect of taking up walking this month.
Who this may not transfer to:GWAS samples included both sexes; the paper does not report a pooled split, and all participants were of European ancestry.
The study · 1
Choi et al., assessment of bidirectional relationships between physical activity and depression among adults, 2-sample Mendelian randomization study · JAMA Psychiatry 2019
Bright light for seasonal depression, effect size 0.84, in the antidepressant range
For winter, seasonal depression, a morning light box gives a large improvement, an effect size of 0.84, which the reviewers put in the same range as antidepressant drugs. Blinding is the weak point, since a person sitting at a light box knows it.
Bright light treatment reduced depression symptom severity in seasonal affective disorder with an effect size of 0.84 (95% CI 0.60 to 1.08) across eight studies, and dawn simulation gave 0.73 (0.37 to 1.08) across five studies. The authors describe these as equivalent to the effect sizes seen in most antidepressant drug trials. Measured in: Randomized controlled trials of light therapy for mood disorders published between January 1975 and July 2003, of which only 13% of screened studies met the inclusion criteria. The screening rate is the finding underneath the finding: 87% of the light therapy literature was not built to answer the question. Blinding is the structural problem, since a person sitting in front of a light box knows it, and the dim-light and deactivated-device shams used are not obviously inert. The review is from 2005 and predates the LED devices most people now buy.
Who this may not transfer to:The review does not report pooled sex composition. Seasonal affective disorder is diagnosed more often in women, so the trial populations are likely to be weighted that way.
The study · 1
Golden et al., the efficacy of light therapy in the treatment of mood disorders, review and meta-analysis · Am J Psychiatry 2005
Omega-3 eased existing depression by about 2.5 Hamilton points, but did not prevent it
Fish-oil (omega-3) eased symptoms in people who already have depression by about 2.5 points on the 17-item Hamilton scale, a small effect on shaky evidence. As a preventive in people without depression it did not help and slightly raised risk, which reads as two different questions rather than a contradiction.
Against placebo, omega-3 supplementation gave a standardized mean difference of -0.40 (95% CI -0.64 to -0.16) on depressive symptoms in people with major depression, which the review translates to about 2.5 points on the 17-item Hamilton scale against a minimal clinically important change of 3 points. In prevention, a separate 18,353-person trial found 1 g a day slightly increased the risk of depression or clinically relevant depressive symptoms (hazard ratio 1.13, 95% CI 1.01 to 1.26). Measured in: 1,848 adults with major depressive disorder across 33 placebo-controlled trials for the treatment estimate; 18,353 adults aged 50 and over for the prevention trial. The Cochrane review rated its own evidence very low certainty and the confidence interval spans both a clinically important effect and a negligible one. The treatment and prevention results point in opposite directions because they are different questions in different populations, not a contradiction. This and the other supplement-prevention row on this page come from ONE randomization reported in two papers, not from two independent trials.
Who this may not transfer to:The Cochrane review does not report pooled sex composition; the prevention trial was 49.2% women.
The studies · 2
Appleton et al., omega-3 fatty acids for depression in adults, Cochrane systematic review · Cochrane Database Syst Rev 2021
St John's wort beat placebo and matched antidepressants, response ratio 1.28 to 1.87
St John's wort beat placebo (response ratio 1.28 in the larger trials, 1.87 in the smaller ones) and matched standard antidepressants for major depression, with fewer side effects. The serious catch is interactions, since it speeds up the breakdown of many medications, which is the basis for caution.
Against placebo, the combined response rate ratio was 1.28 (95% CI 1.10 to 1.49) in the nine larger trials and 1.87 (1.22 to 2.87) in the nine smaller ones. Against standard antidepressants the results were homogeneous and equivalent: RR 1.02 versus tricyclics and tetracyclics, RR 1.00 versus SSRIs. Fewer patients dropped out for adverse effects than on older antidepressants (OR 0.24) or SSRIs (OR 0.53). Measured in: 5,489 patients with major depression across 29 randomized double-blind trials, 18 placebo-controlled and 17 against synthetic antidepressants. Placebo-controlled results were markedly heterogeneous, smaller trials gave much larger effects than larger ones, and trials from German-speaking countries reported findings more favorable to hypericum than trials from elsewhere. Three of the review authors declared past funding, fees, or travel reimbursement from a hypericum manufacturer. The trials used specific standardized extracts, so a supermarket product is not the tested intervention, and none of this addresses the interaction risk that dominates the practical decision.
Who this may not transfer to:The review does not report pooled sex composition across the 29 trials.
The study · 1
Linde et al., St John's wort for major depression, Cochrane systematic review · Cochrane Database Syst Rev 2008
In children and teenagers, only fluoxetine clearly beat placebo, effect size -0.51
Among the antidepressants tested in children and teenagers, only fluoxetine was clearly more effective than placebo (effect size -0.51), which is why it is the first-choice drug in this age group. The evidence for the rest was rated very low quality.
Only fluoxetine was statistically more effective than placebo (standardized mean difference -0.51, 95% CrI -0.99 to -0.03). Fluoxetine was better tolerated than duloxetine and imipramine. Imipramine, venlafaxine and duloxetine caused more discontinuations for adverse events than placebo. Measured in: 5,260 children and adolescents with major depressive disorder across 34 double-blind randomized trials of 14 antidepressants, published and unpublished, to May 2015. The quality of evidence was rated very low for most comparisons, and the authors concluded that these drugs do not appear to offer a clear advantage in this age group overall. A network meta-analysis borrows strength across indirect comparisons, so a single positive result among 14 drugs deserves care.
Who this may not transfer to:The network meta-analysis does not report pooled sex composition; participants were children and adolescents.
The study · 1
Saffron eased mild-to-moderate depression, effect size 0.89 over placebo
Saffron, taken as a standardized extract, eased mild-to-moderate depression more than a dummy pill and about as well as standard antidepressants, at an effect size of 0.89. Most trials are small and from one country, so read it as promising rather than settled.
Across 11 randomized trials, a standardized saffron (Crocus sativus) extract reduced depression severity more than placebo (Hedges' g 0.891, 95% CI 0.369 to 1.412, p = 0.001) and was not inferior to standard antidepressant drugs (g -0.246, 95% CI -0.495 to 0.004, p = 0.053) in mild to moderate depression. Measured in: Adults with mild to moderate depression across 11 placebo-controlled or antidepressant-controlled randomized trials. Most of the trials are small and from a single country (Iran), which raises the same regional-positivity and publication-bias concerns seen in other herbal literatures, and the pooled effect is heterogeneous. The trials used standardized extracts at pharmacological doses, so a culinary pinch is not the tested intervention, and saffron is costly at trial doses.
Who this may not transfer to:The review does not report pooled sex composition across the 11 trials.
The study · 1
Tóth et al., the efficacy of saffron in the treatment of mild to moderate depression, meta-analysis · Planta Med 2019
Morning light beat placebo in non-seasonal depression, effect size 0.80
For year-round, non-seasonal depression, 30 minutes of morning light beat a dummy device in a careful trial (effect size 0.80), and light plus an antidepressant did best of all. The evidence here is younger and less settled than for seasonal depression, and other trials disagree on the add-on question.
In an 8-week randomized, double-blind, sham-controlled trial, light monotherapy at 10,000 lux for 30 minutes each morning beat placebo on MADRS change (d = 0.80, 95% CI 0.28 to 1.31), and light combined with fluoxetine beat placebo by more (d = 1.11). Fluoxetine monotherapy did not beat placebo (d = 0.24). Response rates were 50.0% for light, 75.9% for the combination, 29.0% for fluoxetine and 33.3% for placebo. Measured in: 122 adults aged 19 to 60 with non-seasonal major depressive disorder of at least moderate severity, recruited from outpatient psychiatry clinics in Canadian academic medical centers. Thirty people per arm is small, the fluoxetine arm failing to beat placebo suggests the trial was not powered as expected, and the sham was an inactive negative ion generator, which does not resemble a light box. An earlier meta-analysis found no effect at all from adding bright light to antidepressant medication across five trials, so this literature disagrees with itself on the adjunct question.
Who this may not transfer to:The trial enrolled adults of both sexes; the abstract does not give the split.
The studies · 2
Lam et al., efficacy of bright light treatment, fluoxetine and the combination in patients with nonseasonal major depressive disorder, randomized clinical trial · JAMA Psychiatry 2016
Tao et al., light therapy in non-seasonal depression, updated meta-analysis of 23 randomized trials · Psychiatry Res 2020
Anti-inflammatory drugs eased depression, effect size -0.55, on fragile trials
Anti-inflammatory drugs, mostly added on top of an antidepressant, improved depression scores in pooled trials (effect size -0.55). The trials are short and small and disagree with each other, and quality of life did not move, so this is an early lead rather than settled treatment.
Pooled across 26 trials, anti-inflammatory agents reduced depressive symptoms against placebo with a standardized mean difference of -0.55 (95% CI -0.75 to -0.35, I2 = 71%), with higher response (RR 1.52) and remission (RR 1.79) rates. Separate subgroup analyzes of NSAIDs, omega-3, statins and minocycline were each significant. In trials enrolling only women, no difference between groups was found. Quality of life did not change. Measured in: 1,610 adults with major depressive disorder across 30 randomized controlled trials to January 2019, mostly testing the agent as an add-on to an antidepressant. Heterogeneity was high (I2 = 71%) and most trials were short and small, so the pooled figure is fragile. The null in women-only trials is unexplained and sits awkwardly with the overall result. Quality of life did not move even where symptom scores did, which is the outcome a patient would notice.
Who this may not transfer to:Both sexes were included, and the subgroup of trials enrolling only women showed no difference between groups, so the pooled effect may not transfer to women.
The study · 1
Bai et al., efficacy and safety of anti-inflammatory agents for the treatment of major depressive disorder, systematic review and meta-analysis of randomised controlled trials · J Neurol Neurosurg Psychiatry 2020
Chinese herbal medicine beat placebo by 4.53 Hamilton points, on a positive-only literature
Chinese herbal formulas beat placebo by 4.53 points on the 17-item Hamilton scale and roughly matched antidepressants with fewer side effects. Nearly all the trials come from Chinese journals that almost never publish a negative result, a publication bias that means the true effect is likely smaller.
Chinese herbal medicine as monotherapy beat placebo on the 17-item Hamilton scale by 4.53 points (95% CI 3.37 to 5.69, GRADE moderate certainty) against a minimally important difference the authors set at 4 points. It did not differ from conventional antidepressants on total effective rate (RR 0.99) or Hamilton score (mean difference 0.44), and added to conventional medication it improved the total effective rate (RR 1.16). Adverse events were fewer than with conventional drugs. Measured in: 3,549 patients across 40 randomized controlled trials scoring at least 4 on the Cochrane risk-of-bias tool, searched to April 2018. Nearly all of this literature comes from Chinese journals, where a review of 1,100 controlled-trial abstracts found 99% reporting the test treatment as effective and none reporting it as ineffective, against 75% positive in trials published in England. The formulas differ from trial to trial, so the pooled figure describes a category rather than a treatment, and total effective rate is a soft dichotomized outcome that inflates apparent benefit.
Who this may not transfer to:The review does not report pooled sex composition across the 40 trials.
The studies · 2
Wang et al., efficacy and safety of Chinese herbal medicine for depression, systematic review and meta-analysis of randomized controlled trials · J Psychiatr Res 2019
Vickers et al., do certain countries produce only positive results, systematic review of controlled trials · Control Clin Trials 1998
Acupuncture beat sham by 1.69 Hamilton points, below a clearly felt difference
Against a convincing fake, acupuncture lowered depression scores by 1.69 points on the 17-item Hamilton scale, below what a person would clearly notice. Against no treatment the effect looked bigger, which is the weight of attention, travel, and a weekly appointment rather than the needles alone.
Against sham acupuncture, the reduction in depression severity was 1.69 points on the Hamilton scale (95% CI -3.33 to -0.05, 14 trials, 841 participants, low-quality evidence). Against no treatment, waiting list or usual care the effect was larger (SMD -0.66, five trials, 488 participants, low-quality evidence). Against medication the benefit was small (SMD -0.23, 31 trials, 3,127 participants, very low quality). Measured in: 7,104 adults with depression across 64 randomized trials, diagnosed by DSM-IV, ICD, Research Diagnostic Criteria or the Chinese Classification of Mental Disorders. Most studies were at high risk of performance bias and at high or unclear risk of detection bias. The sham-controlled effect of 1.69 Hamilton points sits below any usual threshold for clinical meaning, and the gap between the sham comparison and the no-treatment comparison is the size of what attention, travel and a weekly appointment contribute.
Who this may not transfer to:The review does not report pooled sex composition across the 64 trials.
The study · 1
Smith et al., acupuncture for depression, Cochrane systematic review · Cochrane Database Syst Rev 2018
A 90-minute nature walk lowered brooding, an urban walk did not
A 90-minute walk in nature lowered self-reported brooding and quieted a brain region tied to it, while the same walk in a city changed neither. This is a single walk in healthy volunteers, so it points to a plausible pathway rather than a treatment.
A 90-minute walk in a natural setting reduced self-reported rumination and activity in the subgenual prefrontal cortex, while a 90-minute walk in an urban setting changed neither. Measured in: Healthy urban-dwelling adults randomly assigned to a single 90-minute walk in a natural or an urban setting, with functional MRI before and after. A single walk, a small sample, healthy volunteers rather than people with depression, and no follow-up. Rumination is a risk factor for depression and not depression itself, so this describes a plausible pathway rather than a treatment effect, and the participants knew which walk they had taken.
Who this may not transfer to:The paper does not give a sex breakdown in its abstract; participants were healthy urban-dwelling adults.
The study · 1
Bratman et al., nature experience reduces rumination and subgenual prefrontal cortex activation · Proc Natl Acad Sci U S A 2015
How it works
The serotonin-deficiency theory of depression is not supported by the evidence
The idea that depression comes from a shortage of serotonin does not hold up: careful reviews of the metabolite, imaging, and gene studies find no consistent link. This does not mean antidepressants fail, since a treatment can help without the illness being a shortage in the pathway it acts on.
Across the main strands of serotonin research, no consistent association with depression was found. Meta-analyzes of the metabolite 5-HIAA and of plasma serotonin showed no relationship. Receptor and transporter imaging showed weak, inconsistent reductions in binding that were not separable from prior antidepressant use. Tryptophan depletion had no effect in most healthy volunteers. The two largest SERT gene studies, a genetic association study of 115,257 people and a collaborative meta-analysis of 43,165, showed no association with depression and no gene by stress interaction. Measured in: 17 systematic reviews, meta-analyzes and large data-set analyzes covering body-fluid serotonin, 5-HT1A binding, transporter imaging, tryptophan depletion and SERT genetics. An umbrella review inherits the limits of the reviews inside it, and the quality of those was variable. The paper drew a long series of published replies disputing its framing, its handling of the tryptophan depletion literature and its inference from absence of association. It says nothing about whether antidepressants work, which is a separate question with its own evidence.
Who this may not transfer to:Sex composition is not reported across the pooled reviews; the largest genetic samples were population-based and included both sexes.
The study · 1
Moncrieff et al., the serotonin theory of depression, systematic umbrella review of the evidence · Mol Psychiatry 2023
About 27% of people with depression have raised inflammation, roughly three quarters do not
About a quarter of people with depression, 27%, have raised inflammation on a blood test (CRP above 3 mg/L), more than healthy people, but roughly three quarters do not. So inflammation is part of the story for some, not the cause of depression in general.
27% of people with depression had C-reactive protein above 3 mg/L (95% CI 21% to 34%), and 58% had CRP above 1 mg/L. Compared with matched healthy controls the odds ratio for low-grade inflammation was 1.46 (1.22 to 1.75). Prevalence was not associated with sample source, antidepressant treatment, age, BMI or ethnicity. Measured in: 13,541 patients with depression and 155,728 controls across 37 studies, 30 of them contributing to the prevalence estimate. This is a prevalence and association finding rather than a causal one, and it cuts against the popular framing in both directions: raised inflammation is present in a substantial minority, and roughly three quarters of people with depression do not have it. CRP is a single non-specific marker that rises with infection, obesity and smoking.
Who this may not transfer to:Median 36% male across the pooled studies.
The study · 1
Osimo et al., prevalence of low-grade inflammation in depression, systematic review and meta-analysis of CRP levels · Psychol Med 2019
Sleep
Treating the insomnia lifted depression response from 17% to 32%
When insomnia comes with depression, treating the sleep with a short structured program (cognitive behavioral therapy for insomnia) raised the share of people whose depression responded from 17% to 32%, and the gain reached beyond the sleep questions on the scale.
Cognitive behavioral therapy for insomnia produced a depression response more often than control conditions (OR 2.28, 95% CI 1.67 to 3.12, GRADE moderate certainty), lifting the post-treatment response rate from 17% on control to 32% (95% CI 26% to 39%) at a median of 8 weeks. Insomnia remission also improved (OR 3.57). Depression improvement extended beyond the sleep items of the scale. Measured in: 4,808 adults with major depressive disorder and comorbid insomnia across 19 randomized trials, mean age 33.2, 73.2% women. Dropout was higher on CBT-I than on control (OR 1.69, low certainty), which is consistent with a treatment that asks people to spend less time in bed while they are already exhausted. Control conditions varied widely, and depression and insomnia scales share items, so part of the overlap is measurement rather than mechanism.
Who this may not transfer to:73.2% of participants were women, so the estimate is weighted towards women and towards younger adults.
The study · 1
Furukawa et al., cognitive behavioral therapy for insomnia to treat major depressive disorder with comorbid insomnia, systematic review and meta-analysis · J Affect Disord 2024
Insomnia roughly doubles the odds of later depression, odds ratio 2.60
People with insomnia but no depression were about twice as likely to become depressed later (odds ratio 2.60). Part of that is insomnia being an early sign of a depression not yet diagnosed, which is why treating the sleep, on the separate trial evidence, matters.
Non-depressed people with insomnia had roughly twice the odds of developing depression at follow-up: overall odds ratio 2.60 (95% CI 1.98 to 3.42) in the random-effects model, and 2.10 (1.86 to 2.38) in the fixed-effects model after adjusting for outliers. Measured in: 21 longitudinal epidemiological studies published between 1980 and 2010 that measured insomnia complaints and later depression in the same cohorts. What could explain it instead: The included studies did not consistently adjust for other intervening variables, and insomnia is itself an early symptom of a depression not yet diagnosed, so part of this association is one illness being detected twice.. Heterogeneity was present until outliers were removed, insomnia was defined by self-reported complaint rather than by diagnostic interview in most cohorts, and prediction does not establish that treating the insomnia prevents the depression. The CBT-I trial evidence is the separate line that speaks to that.
Who this may not transfer to:The review does not report the pooled sex composition of the 21 cohorts.
The study · 1
Baglioni et al., insomnia as a predictor of depression, meta-analytic evaluation of longitudinal epidemiological studies · J Affect Disord 2011
Social Connection
Frequent loneliness more than doubled the odds of depression, odds ratio 2.33
Adults who were often lonely were more than twice as likely to become depressed later (odds ratio 2.33). It runs both ways, since depression also pulls people into withdrawal, so contact is worth starting small and early.
Adults who were often lonely had a pooled adjusted odds ratio of 2.33 (95% CI 1.62 to 3.34) for new onset of depression compared with people who were not often lonely. Measured in: Eight independent general-population cohorts pooled from a systematic review of 32 longitudinal studies, most of them focused on depression. What could explain it instead: Reverse causation is the central one: depression that has begun but is not yet diagnosed causes withdrawal, so being lonely at baseline can be an early symptom rather than a cause. Poor health, unemployment and bereavement drive both, and adjustment for them varied between cohorts.. The authors flag heterogeneity and say the pooled figure should be read with caution. Loneliness was self-reported with different instruments across cohorts, and only eight of the 32 studies could be pooled.
Who this may not transfer to:The review does not report pooled sex composition, and notes that its cohorts were not diverse in age or population.
The study · 1
Mann et al., loneliness and the onset of new mental health problems in the general population, systematic review and meta-analysis · Soc Psychiatry Psychiatr Epidemiol 2022
Where the Evidence Runs Thin
No pill on the supplement shelf substitutes for treatment when depression is moderate or severe. The evidence does not support fixing depression with willpower, positive thinking, or a tidier morning routine. And waiting to feel motivated before you start gets it backward: in depression, the doing usually comes before the wanting to. Nothing on this page asks you to face it alone.
What To Do This Week
None of this needs a prescription to begin, and it works alongside whatever a clinician has you on. Start small, and let one thing build the next.
Say it out loud to one person you trust, and if low mood or loss of interest has run for two weeks or more, talk to a doctor this week. They can check the thyroid and the other explanations, and ask what else you take, since some herbs and supplements interact with antidepressants.
Walking or jogging had the largest effect of any exercise studied, and around 150 minutes a week tracks with the biggest drop in risk. If that feels far off, a ten-minute walk today counts and is the start of it. The doing comes first; the wanting follows.
Bright light within an hour of waking lifts mood and helps set the body clock that steadies your sleep. Outdoor light is free and strongest; in dark-winter months or a windowless routine, a 10,000-lux lamp for the morning stands in, and it has its largest evidence in seasonal depression.
Broken sleep both follows depression and worsens it. Hold a steady wake time, get light early and dim it at night, and if insomnia has taken hold, a short structured sleep program improved the depression itself in trials, and the gain reached past the sleep.
Loneliness more than doubles the odds of depression, and contact lowers it. It runs both ways, so start small: one message, one short visit, one regular thing with other people that you do not have to feel like doing first.
Therapy, medication, exercise, and light all take a few weeks to show. Pick two or three of these and hold them steadily, and measure yourself against where you were a month ago, not against yesterday.
Go Deeper
- Walking and staying active: the free, low-effort base with the strongest lifestyle evidence for depression, and how to build the daily amount.
- Strength training: another exercise mode that lowered depression in the same analysis, with some of the best acceptability.
- Morning light: bright light within an hour of waking, its largest effect in seasonal depression and a smaller signal beyond it.
- Insomnia: the sleep problem that both follows and worsens depression, and the short program that treats it.
- Tai chi and qi gong: a gentle movement practice with a small but measured effect and a low bar to entry for anyone with little reserve.
- Social connection: staying in contact as one of the few free levers, with the circumstances that make it hard named plainly.
The Chinese Medicine View
The Chinese Medicine View
Chinese medicine reads this as constraint, yu zheng (郁证), something pent up and unable to move, most often a stagnation of the Liver qi. It treats several different pictures rather than one condition, so the pattern is read before the formula. Hold this as an interpretive lens on constitution and circumstance, not a claim that it maps onto a diagnosis or a rating scale. Two cautions come from the tradition itself. Constraint that arises from a person's circumstances moves only as far as those circumstances allow, and the qi-moving formulas that drain a constrained qi can deplete someone who is already depleted, which is why the deficient patterns are strengthened rather than drained. Any herbs also belong with a practitioner and a traceable supply, both because the formulas differ from trial to trial and because St John's wort, the best-known herbal option, has the serious interaction profile set out in the Cautions section.
Tightness or fullness across the chest and ribs, frequent sighing, irritability that flares and passes, mood that shifts with circumstances, and in many women a premenstrual pattern. The most-named picture. Xiao Yao San is the classical direction.
Exhaustion that rest does not fix, poor appetite, loose stools, palpitations, poor memory, and broken sleep, with a pale tongue. Attributed to prolonged overthinking. Gui Pi Tang is the reference formula.
Constraint that has sat long enough to generate heat: a quicker temper, a bitter taste, red eyes, disturbed sleep, a red tongue with a yellow coating. Dan Zhi Xiao Yao San adds cooling herbs to the base.
The sensation of a lump in the throat that swallows away and returns, plum-pit qi, along with a heavy head, nausea, and a greasy tongue coating. Ban Xia Hou Po Tang is the classical formula.
Zang zao (脏躁): sadness and weeping without a clear cause, restlessness, and frequent yawning. Treated with Gan Mai Da Zao Tang, a formula of three foods.
Waking in the small hours, night sweats, heat in the palms and soles, a dry mouth, and a red tongue with little coating. Tian Wang Bu Xin Dan is the usual direction.
Acupuncture belongs with this lens. Against a convincing sham it lowered depression scores by a small amount, below what a person would clearly feel, and against no treatment the effect looked larger, which is the weight of attention and a weekly appointment. It is a reasonable option for someone drawn to it, sitting alongside the first-line treatments rather than in place of them.
Cautions For This Practice
Extra restraintEverything to be aware of is here, in one place. This practice suits most healthy people; a few situations call for real care.
Stopping an antidepressant brought symptoms in 31%, against 17% on placebo
At least one discontinuation symptom occurred in 31% (95% CI 27% to 35%) of people stopping an antidepressant, against 17% (14% to 21%) stopping placebo, a summary difference of 8 percentage points in the randomized trials. Severe symptoms occurred in 2.8% against 0.6% on placebo. Desvenlafaxine, venlafaxine, imipramine and escitalopram were associated with higher frequency, and imipramine, paroxetine and desvenlafaxine or venlafaxine with greater severity. Heterogeneity was substantial and the authors point to factors outside diagnosis, drug and trial design, including how investigators and patients report. Returning illness is difficult to separate from withdrawal in these data. An earlier review put the incidence above 50% using a less restrictive study base, an estimate that is disputed; the disagreement is about size, and both readings undercut the older guideline description of a brief, mild syndrome.Henssler et al., incidence of antidepressant discontinuation symptoms, systematic review and meta-analysisDavies and Read, systematic review into the incidence, severity and duration of antidepressant withdrawal effects
Antidepressants raised reported youth suicidal thoughts 0.7%, and helped 11% more respond
Across all trials and indications, the risk difference for spontaneously reported suicidal ideation or suicide attempt was 0.7% higher on drug than on placebo (95% CI 0.1% to 1.3%), a number needed to harm of 143. Within each individual indication the difference was not statistically significant. There were no completed suicides in any trial. The same analysis found an 11% response benefit in pediatric major depression, a number needed to treat of 10. Suicidal ideation was captured from spontaneous reports rather than by systematic questioning, which undercounts in both arms. Trials are short and exclude the highest-risk young people, and untreated depression carries its own suicide risk, so the comparison is against placebo rather than against no illness.Bridge et al., clinical response and risk for reported suicidal ideation and suicide attempts in pediatric antidepressant treatment, meta-analysis of randomized controlled trials
St John's wort interacts with many medications
The practical decision about St John's wort turns on its interactions, not its benefit. It speeds up the liver enzymes and the transporter that clear a long list of drugs, so it can lower the levels, and the effect, of the contraceptive pill, blood thinners, some heart, HIV, and transplant medicines, and other antidepressants. Combined with an SSRI or another serotonin-raising drug it can also push serotonin too high. It is effective for mild-to-moderate depression, so treat it like a drug: tell your prescriber and pharmacist before starting or stopping it, especially if you take anything else.
Serotonin syndrome from combining serotonin-raising agents
Starting or combining serotonin-raising medicines, including two antidepressants, an antidepressant with tramadol, triptans, linezolid, or St John's wort, can rarely push serotonin activity too high. Agitation, confusion, a fast heart, sweating, tremor, and twitching muscles within hours of a change are the signs, and it is a same-day emergency. It is uncommon and avoidable, and it is why every new combination goes through a prescriber.
Do not stop an antidepressant abruptly
Stopping suddenly, or missing several doses, commonly brings dizziness, electric-shock sensations, nausea, flu-like feelings, and vivid dreams. These were understated in guidelines for years. They are not a sign of addiction, and they are largely avoidable: a prescriber can taper the dose down slowly, more slowly for the shorter-acting drugs, so if a medication is no longer wanted, that is a plan to make together rather than a decision to act on alone.
Tell your prescriber everything you take
Readers usually arrive already taking something. Herbal products, supplements, and over-the-counter medicines can interact with antidepressants, and St John's wort is only the most-cited example. A single conversation listing everything, prescription and not, is the cheapest way to avoid the interactions on this page.
Start slow, be smart, read the research, and consult a professional if you have any concerns. This is here to inform your choice, not make it for you.
When to See Someone
Depression is treatable and most people improve, and a few of the signs below need help the same day. If you are thinking about suicide, or about harming yourself, treat it as a medical emergency: call your local emergency number, go to an emergency department, or call the suicide and crisis line your own country's health service publishes. Crisis numbers differ from country to country, so use the one your health service lists rather than one copied from a website. If someone is with you, ask them to stay. Asking someone directly whether they are thinking of ending their life does not put the idea there.
- A suicide attempt, an overdose of any size, or a new plan or stated intention to end your life(seek urgent care)
- Hearing or seeing things others do not, or losing contact with reality(seek urgent care)
- A new mother with thoughts of harming herself or her baby, which is postpartum depression and is both treatable and urgent(seek urgent care)
- Agitation, confusion, a fast heart, sweating, tremor, and twitching muscles within hours of starting or combining medicines, which can be serotonin syndrome(seek urgent care)
- A sudden swing into unusual energy, racing thoughts, little need for sleep, or reckless behavior, especially soon after starting an antidepressant, which can point to bipolar disorder and needs a different treatment
- The first weeks of an antidepressant in someone under 25, when a young person should have someone checking in, since suicidal thoughts can rise briefly early on
- Low mood or loss of interest most of the day, most days, for two weeks or more: a reason to talk to a doctor this week, who will check the thyroid and other explanations and ask what else you take, including supplements and herbal products, since some interact with antidepressants
- Stopping an antidepressant abruptly, which can bring dizziness, electric-shock sensations, nausea, and vivid dreams; a prescriber can taper it slowly instead
None of this is here to frighten you. Depression is common and it responds to treatment, and reaching for help early is the ordinary thing to do, not the last resort. When in doubt, tell a doctor.
Common Questions
Do antidepressants actually work, or is it all placebo?
They work, more clearly the more severe the depression is. The largest comparison, covering 116,477 adults, found all 21 antidepressants studied more effective than placebo. The average difference on the rating scale is small, which one review read as below the bar for clinical significance, but that average hides a severity gradient: the benefit is minimal in milder depression and clearly meaningful in severe depression. The drugs also take a few weeks to work and are not the only first-line option, since talking therapy performs about as well.
Isn't depression just a chemical imbalance, a shortage of serotonin?
That story has not held up. Careful reviews of the serotonin-deficiency idea found no consistent link between serotonin and depression, including in the two largest gene studies. This is not the same as saying antidepressants fail. A treatment can help without the illness being a shortage in the pathway it acts on. The drugs help many people. The chemical-imbalance explanation of why is the part that did not hold up.
What is the strongest thing I can do myself?
Move your body, and stay in contact with people. Walking or jogging had the largest effect of any exercise studied for depression, a Hedges g of -0.62, and about 150 minutes a week tracks with the biggest drop in risk. If your energy is low, a ten-minute walk today is enough to count, because in depression the action usually comes before the motivation for it. None of this replaces therapy or medication for moderate to severe depression, but it is treatment in its own right and it costs nothing.
Does therapy work as well as medication?
About as well, and it lasts. A network meta-analysis of 331 trials found every mainstream therapy beat doing nothing, and the different therapies did not clearly differ from each other, so the type matters less than getting into one. Behavioral activation, a simpler version, matched full cognitive behavioral therapy in a head-to-head trial and can be delivered by less specialized staff, which helps with the waiting lists. Therapy and medication also combine well, and for many people the two together beat either one alone.
Do St John's wort or other supplements help?
St John's wort has trial evidence for mild-to-moderate depression, matching standard antidepressants in trials, but it speeds up the clearance of many medications, so tell a pharmacist before you start it. Saffron extract also beat placebo in mild-to-moderate depression, on small trials. Fish oil helps existing depression only a little, vitamin D did not prevent depression at all, and an anti-inflammatory diet is not a treatment on its own. None of these replaces therapy or medication when depression is moderate or severe.
I was told to just exercise and think positive. Is that fair?
Only partly. Exercise is one of the strongest self-directed levers, but "think positive" is not a treatment, and neither is waiting for motivation, which in depression tends to follow action, not precede it. Mild depression may lift with the self-directed levers and therapy. Moderate to severe depression needs therapy or medication doing more of the work, with exercise, light, sleep, and connection supporting it. Nothing about depression means you have to fix it alone.
Explore Related
Other pages this one connects to, by the evidence they share, the outcomes they touch, and the ground they cover.
All 32 sources on this page independently checked and cross-referenced.
Thomas Dehli, Founder & Editor, Sacred Lotus
Sacred Lotus has published Chinese medicine reference material since 2001. Integrative pages are held to the same standard as the herb and formula library: cite the source, grade the claim at its real strength, and say where the research has not looked. This page is educational and it is not medical advice. Last reviewed and updated August 9, 2026.
Evidence strength
How confidently the research supports a claim. Strength describes the evidence, not our endorsement.