Anxiety & Mood

The framework: depression and anxiety are biopsychosocial

Major Depressive Disorder (MDD) and Generalized Anxiety Disorder (GAD) are not single diseases with single fixes. They are syndromes with overlapping biology, psychology, and social substrate. Twin studies put heritability around 30–40% for MDD and slightly higher for GAD, which means roughly 60% of the variance is environment, behavior, and circumstance. Any honest framework respects both halves.

Pharmacology is one lever. It is the most-studied lever and, in moderate-to-severe disease, it is the lever with the largest acute effect size — but it is not the only lever, and in mild disease the effect size shrinks toward the upper bound of what placebo and clinician contact can produce. Sleep architecture, aerobic capacity, social connection, daylight exposure, and structured psychotherapy each move the needle in their own randomized trials. The mistake is treating those as alternatives. They are layers.

The frame for this hub is simple: name what the literature shows for each lever, name the dropout rates and unblinding problems honestly, and let the reader and their clinician decide which layers to engage first. We do not write prescriptions on this site. We write the conversation you take to a clinician.

SSRIs and SNRIs: what STAR*D actually showed

Selective serotonin reuptake inhibitors (SSRIs) and serotonin– norepinephrine reuptake inhibitors (SNRIs) remain the first-line pharmacology for MDD and GAD in essentially every major guideline. The strongest real-world dataset on SSRI response is STAR*D — the Sequenced Treatment Alternatives to Relieve Depression trial — which followed 4,041 patients through up to four sequential treatment steps [Rush 2006]. At Step 1 (citalopram monotherapy), the remission rate was about 28% by HAM-D and 33% by QIDS-SR. By the end of four cumulative steps, cumulative remission climbed to roughly 67% — but each successive step recruited fewer responders and lost more patients to dropout.

Two grey areas matter here. First, STAR*D had high attrition; analyses treating dropouts as non-responders pull the headline numbers down. Second, more recent re-analyses have argued the original paper reported remission rates more favorably than the protocol's primary outcome would have permitted. Translation: SSRIs work for a real fraction of patients, but they do not work for everyone, and the original framing oversold the picture.

SSRIs are neither the miracle they were sold as in the 1990s nor the sugar pill they are sometimes painted as today. They are a moderate-effect-size intervention that helps roughly a third on the first try.

Discontinuation, sexual side effects, and the genetic-testing claim

Discontinuation syndrome is real. Roughly one in five patients stopping an SSRI develops symptoms — dizziness, electric-shock sensations, irritability, GI symptoms — and a subset of those experience prolonged or severe withdrawal. Tapering schedules of weeks-to-months, sometimes using hyperbolic taper strips, are now standard in the patient community for anyone who has been on treatment longer than six months.

Sexual side effects affect a substantial minority — depending on the agent and the survey method, anywhere from 25% to 70% report some reduction in libido, arousal, or orgasm latency. Post-SSRI sexual dysfunction (PSSD), the persistence of these effects after discontinuation, is recognized by the European Medicines Agency but the prevalence is contested.

Pharmacogenomic testing (CYP2D6, CYP2C19, and panel products marketed as "find your antidepressant") has been heavily marketed. The honest read of the evidence is that genotyping reliably predicts metabolic rate — slow and ultra-rapid metabolizers do exist — but the leap from metabolic rate to clinical response is much weaker than the marketing suggests. Useful for dose-adjustment in known outliers. Oversold as a crystal ball.

Ketamine and psychedelics: the new neuroplasticity class

The most important shift in mood-disorder pharmacology this century is the recognition that NMDA antagonism (ketamine) and serotonergic psychedelics (psilocybin, LSD, ayahuasca's DMT) can produce antidepressant effects with a fundamentally different time course than monoaminergic drugs. The original signal: a single IV ketamine infusion produced rapid antidepressant effects within hours and persisting for days, in patients with treatment-resistant depression [Berman 2000]. Two decades of replication and a 2017 consensus statement from the American Psychiatric Association have established ketamine as a reasonable option for TRD in appropriate settings [Sanacora 2017].

Sublingual troches, intranasal esketamine (the FDA-approved enantiomer), and IV protocols all show signal, with IV producing the most reliable response and esketamine being the only route currently FDA-approved specifically for TRD. The unresolved questions are durability and maintenance frequency; for most patients, a single dose does not produce a months-long remission.

On the psychedelic side, two trials anchor the modern revival. Davis et al. (JAMA Psychiatry 2021) ran an open-label trial of psilocybin- assisted therapy in MDD, reporting large effect sizes that persisted at one month [Davis 2021]. Carhart-Harris et al. (NEJM 2021) ran a six-week head-to-head of psilocybin versus escitalopram in MDD; the primary outcome (QIDS-SR-16 change) did not show a statistically significant superiority for psilocybin, but several secondary outcomes did [Carhart-Harris 2021].

The grey areas are honest and important. Psychedelic trials are small. Blinding is essentially impossible — patients know whether they received an active dose by the third minute — which means expectancy effects are not controllable. Long-term safety data in clinical populations remains thin. The mechanism (5-HT2A agonism, increased dendritic spine density, default-mode-network disintegration) is biologically plausible and increasingly well-characterized, but the translation from a controlled trial with therapist contact to "psilocybin works for depression" is a longer bridge than the headlines imply.

Lifestyle as first-line: exercise, meditation, sleep

The most under-prescribed antidepressant in the literature is structured aerobic exercise. A 2013 Cochrane review covering 39 trials found exercise produced a moderate clinical effect on depressive symptoms — comparable in effect size to psychotherapy and approaching antidepressant medication, though heterogeneity was high and many trials were short [Cooney 2013]. The dose that shows the cleanest signal is three sessions per week of moderate-to-vigorous aerobic work, 30–45 minutes, sustained for eight-plus weeks. Resistance training shows similar signal in a smaller body of literature.

Mindfulness and meditation hold up, with caveats. The Goyal et al. (JAMA Internal Medicine 2014) meta-analysis of 47 trials and 3,515 participants found moderate evidence for mindfulness meditation reducing anxiety, depression, and pain — but effect sizes were modest (around 0.3 standardized mean difference at 8 weeks) and not clearly superior to other active treatments [Goyal 2014]. Translation: meditation is a real lever, not a miracle. Eight weeks of structured practice produces measurable effects; expecting antidepressant-class response from a habit app is overcorrecting.

Sleep is the strangest of the three. Chronic insomnia is among the strongest predictors of incident MDD and GAD; treating insomnia (with cognitive behavioral therapy for insomnia, CBT-I) reduces depression risk and severity. And yet acute total sleep deprivation produces a rapid antidepressant effect in roughly 50–60% of depressed patients — a phenomenon known and replicated since the 1970s — that almost always reverses on recovery sleep. The contradiction (chronic sleep loss is depressogenic, acute sleep loss is antidepressant) is real and instructive: it tells you that mood regulation runs through circadian and homeostatic sleep machinery that is not yet fully understood. See our sleep hub for the protocol layer.

Supplements with real signal

Most of the supplement aisle for mood is noise. Three interventions have credible randomized signal and survive a tier framework.

Omega-3, EPA-heavy. Meta-analyses of omega-3 in MDD find a modest antidepressant effect, but the signal is driven almost entirely by formulations with high EPA (eicosapentaenoic acid) relative to DHA (docosahexaenoic acid). Mocking et al. (Translational Psychiatry 2016) re-analyzed 13 RCTs and concluded that EPA-dominant supplements (≥60% EPA, doses of 1–2 g EPA/day) produced clinically meaningful improvement, while DHA-dominant products did not [Mocking 2016]. Buying a "fish oil" off the shelf without checking the EPA:DHA ratio is the most common error.

Vitamin D in deficiency. The data here is cleaner for correction of deficiency than for supplementation above sufficient serum levels. If 25-OH vitamin D is below ~30 ng/mL, correction produces measurable mood improvement in a fraction of patients. Above ~40 ng/mL, the antidepressant signal disappears. The intervention is "test, then dose," not "everyone takes 5,000 IU."

Magnesium glycinate for anxiety. A 2017 systematic review of 18 trials found modest but consistent anxiolytic signal for magnesium supplementation, especially in mildly anxious and premenstrual cohorts [Boyle 2017]. Glycinate and threonate forms have better tolerability than oxide. The effect is not in SSRI-class territory; it is a sleep-onset and edge-of-the-day intervention, not a panic-disorder fix.

The gut–brain angle: hype vs signal

The gut–brain axis is real biology and oversold marketing in equal measure. The signal: the vagus nerve carries bidirectional signaling between enteric and central nervous systems; gut microbial metabolites (short-chain fatty acids, tryptophan-derived molecules) cross or signal across the blood-brain barrier; and roughly 90% of the body's serotonin is synthesized in the gut, though that pool does not directly cross into the brain.

The hype: the leap from "the gut influences mood" to "this proprietary psychobiotic blend treats depression" is several orders of magnitude longer than the marketing suggests. Psychobiotic trials — randomized studies of specific bacterial strains in depressed or anxious cohorts — exist, but are small, short, and produce small effect sizes. The field is interesting. The product category is ahead of the evidence.

One mechanism worth knowing: there is increasing evidence that part of the SSRI effect may be peripheral, via gut serotonin signaling and vagal afferents, rather than purely central. This is not the textbook explanation, but it is increasingly the working model. It does not change how an SSRI is prescribed today. It does suggest why GI side effects are so common, and why response heterogeneity is so large.

When pharma + therapy + lifestyle stack

The most consistent finding in the modern depression literature is that combination treatment beats monotherapy. Cuijpers et al. (World Psychiatry 2014) meta-analyzed 52 trials comparing combination cognitive behavioral therapy (CBT) plus pharmacotherapy against either alone, and found combination superior to medication alone (g = 0.49) and superior to psychotherapy alone (g = 0.46) [Cuijpers 2014]. The effect sizes are small-to-moderate, but they are consistent across study populations and across decades.

Add the lifestyle layer — aerobic exercise, sleep regularization, daylight exposure, social structure — and you have the actual first-line treatment package for moderate MDD and GAD as it would be practiced in a well-resourced clinic. The reason it is not delivered this way to most patients is access, not evidence: therapist availability is the rate-limiter, not the science.

The synergy data is unambiguous. CBT plus medication beats either alone. Add lifestyle and the package starts to look like what the guidelines have implied for twenty years but rarely delivered.

The limits of self-treatment

Most of this hub is readable as a framework for someone managing their own care. Some of it is not. Moderate-to-severe MDD, suicidal ideation, psychosis-spectrum symptoms, bipolar features, eating disorders, and active substance dependence are not self-treatment territory. They are clinician territory. The honest synthesizer position is: read everything, take it to a clinician, and use the literature to be a better-informed patient — not to replace the relationship.