20.7 — Depression

Depression is one of the most treatable conditions in medicine.

Around 60 to 70 percent of people respond to the first adequate treatment, and with persistence through second and third options, the great majority get substantially better.

And it is one of the most under-treated, because the illness itself argues against seeking help. That is not a personality flaw in the person; it is a symptom of the condition — hopelessness makes treatment feel pointless, and low energy makes the phone call feel impossible.

Knowing that is useful, both for yourself and for someone you care about.

What it actually is

Not sadness. Sadness is a response to something, it moves in response to events, and it lifts.

Depression is a persistent change in how the brain regulates mood, energy, sleep, appetite, concentration and the ability to feel pleasure.

The core features, present most days for at least two weeks:

Low mood.

Anhedonia — loss of pleasure and interest. This is the more specific one. Sadness is common; the flattening where things that used to be enjoyable simply are not, is what points to depression.

Then:

Fatigue, disproportionate to activity.

Sleep disturbance — classically waking in the early hours and being unable to return to sleep; sometimes sleeping far too much.

Appetite and weight change, in either direction.

Poor concentration, which people frequently mistake for a memory problem.

Feelings of worthlessness or excessive guilt, and the guilt is characteristically out of proportion.

Psychomotor change — either slowed movement and speech, or agitation.

Thoughts of death or suicide.

And a point worth making plainly: depression frequently shows up as physical symptoms — pain, fatigue, digestive trouble — particularly in cultures where emotional distress is not readily discussed. A person presenting with unexplained physical symptoms may be describing depression accurately in the vocabulary available to them.

Why it happens

And the honest answer is more interesting than the version people are usually given.

The "chemical imbalance" explanation is too simple. Serotonin depletion as the cause of depression is not supported by the evidence, and it never explained why antidepressants take weeks to work when they change serotonin levels within hours.

The better current picture:

Antidepressants appear to work by promoting neuroplasticity — the brain's ability to form and reorganise connections — which takes weeks, and which matches the delay.

Chronic stress reduces the size and connectivity of the hippocampus and the prefrontal cortex (Chapter 11.6), and effective treatment appears to reverse this.

Inflammation is involved in a subset — depression is commoner in inflammatory conditions, and inflammatory markers are raised in some people with depression.

Genetics contribute around 35 to 40 percent of the risk, spread across many genes with small effects.

And the biggest single environmental factor is adversity, particularly in childhood — which is why the biological and the psychological explanations are not competing accounts.

The useful framing: something in life, something in biology, and something in how a person has learned to interpret events, all interacting. Which is also why treatment works from several different directions.

The kinds

Major depressive disorder — episodes as described above.

Persistent depressive disorder — a lower-grade depression lasting years. Frequently untreated because the person considers it their personality, and frequently responsive when finally treated.

Bipolar depression — depression in someone who also has episodes of elevated mood (Chapter 20.9). And this distinction is critical, because antidepressants alone can trigger mania. Anyone with depression should be asked about periods of unusually elevated mood, reduced need for sleep, or uncharacteristic behaviour.

Seasonal pattern — winter onset, with increased sleep and appetite. Light therapy has genuine evidence here.

Postnatal depression — affecting around 10 to 15 percent of new mothers, and fathers too (Chapter 15.8). Distinct from the "baby blues" of the first week, which is common and self-limiting. Postpartum psychosis is rare, different, and an emergency.

Premenstrual dysphoric disorder — severe mood symptoms in the luteal phase, and a real condition with effective treatment.

And depression secondary to something else: hypothyroidism, B12 deficiency, sleep apnoea, chronic pain, anaemia, and some medications. Which is why basic blood tests are done — not to look for a cause of most depression, but to catch the treatable exceptions.

Treatment

And the choice depends on severity, on what has worked before, and substantially on preference.

Psychological therapy.

Cognitive behavioural therapy — identifying and testing the thought patterns that maintain low mood. As effective as medication for mild to moderate depression, and with lower relapse rates after stopping.

Behavioural activation — the deceptively simple observation that depression reduces activity, which reduces reward, which deepens depression. The intervention is scheduling activity before the motivation returns rather than waiting for it. It is as effective as full CBT and simpler to deliver.

Interpersonal therapy, focusing on relationships and role changes.

And digital CBT programmes have real evidence, which matters because access to therapists is the binding constraint almost everywhere.

Medication.

SSRIs first line — sertraline, escitalopram, fluoxetine, citalopram, paroxetine.

What to expect, because these expectations determine whether treatment succeeds:

Side effects come first, benefit comes later. Nausea and headache in the first week or two, usually settling. Mood improvement takes 2 to 4 weeks and continues improving for 6 to 8.

Stopping at two weeks because "it isn't working" is the commonest reason treatment fails, and it is entirely preventable by saying this in advance.

Sexual side effects affect a substantial minority and are consistently under-discussed, which leads to people stopping without saying why.

A warning worth knowing: in the first weeks, particularly in under-25s, there can be increased agitation and, rarely, increased suicidal thinking. Which is why early follow-up is standard, and why the energy returning before the mood lifts is a period that needs watching.

Continue for at least 6 to 12 months after recovery, because stopping early doubles the relapse rate. Longer after repeated episodes.

Stopping is done by tapering slowly. Discontinuation symptoms — dizziness, electric-shock sensations, irritability, flu-like feelings — are real, sometimes prolonged, and were understated for years. Paroxetine and venlafaxine are the worst; fluoxetine the easiest, because it clears slowly on its own.

Other classes: SNRIs — venlafaxine, duloxetine, useful where pain coexists; mirtazapine, which helps sleep and appetite and causes weight gain; bupropion, with fewer sexual side effects; tricyclics, effective and more toxic in overdose; and MAO inhibitors, effective and now rarely used because of dietary restrictions.

And an honest note about how well they work: the average benefit over placebo across all severities is modest, and the benefit is larger in more severe depression. For mild depression, therapy and exercise are reasonable first choices; for severe depression, medication clearly works. Both of those statements are true, and quoting only one of them has produced a great deal of confused public debate.

When the first treatment does not work

And this is the section that matters most, because it is where people give up.

First: check the basics. Was the dose adequate? Was it taken for long enough? Is there an untreated cause — thyroid, alcohol, sleep apnoea, ongoing adversity that no drug will fix?

Then: switch to a different drug, or add therapy, or add a second agent.

Around a third recover with the first drug; the large trials found that with sequential switching, around two thirds eventually did. Which is the number to hold on to.

For treatment-resistant depression:

Ketamine and esketamine. Given as an infusion or nasal spray, they can improve mood within hours rather than weeks, which was genuinely unexpected and has reshaped thinking about mechanism. The effect wears off over days to weeks, so repeated treatment is needed, and long-term safety is still being established. It is used in specialist settings.

Electroconvulsive therapy. And it deserves a fair hearing, because its reputation comes from how it was done in the 1950s.

Given under general anaesthetic with a muscle relaxant, so there is no convulsion to observe. It is the most effective treatment for severe depression that exists, with response rates around 70 to 80 percent, and it works faster than drugs, which is why it is used when someone is not eating or drinking or is acutely suicidal.

The real cost is memory — gaps around the treatment period, and sometimes for events in the preceding months. Usually improving, sometimes not entirely. Modern unilateral techniques reduce this.

Transcranial magnetic stimulation — non-invasive, no anaesthetic, effective for moderate treatment-resistant depression, requiring daily sessions for weeks.

And psilocybin-assisted therapy is in trials with promising early results. The honest position is that it is not established treatment and the trials are ongoing.

What helps alongside

Exercise. The evidence is genuinely good — comparable to medication for mild to moderate depression in several meta-analyses. Aerobic or resistance, and the dose that shows benefit is around 30 minutes most days.

And the catch is stated honestly: depression removes the motivation to exercise, which is why "just exercise" is unhelpful advice on its own and why it works far better as part of a structured plan with support.

Sleep. Insomnia both precedes and worsens depression, and treating it improves mood independently (Chapter 20.11).

Social connection. Isolation is both a cause and a consequence, and interrupting the loop matters.

Alcohol. A depressant, a disruptor of sleep, and a very common form of self-treatment that makes the condition worse. Reducing it is one of the higher-yield changes.

Sunlight and daylight exposure.

Diet — the evidence for a Mediterranean-style dietary pattern is reasonable, and the trial evidence is smaller than the headlines suggest.

Suicide — the part that must be said plainly

Around 700,000 people die by suicide each year.

Asking someone directly whether they are thinking of ending their life does not plant the idea. This has been studied, and it does not. What it does is give permission to say something the person is frequently desperate to say and cannot start.

So ask directly, and plainly: "Are you thinking about killing yourself?"

What increases immediate risk: a specific plan, access to means, previous attempt, recent loss, alcohol or drug use, isolation, and a sudden calm after a period of despair, which can mean a decision has been made.

What to do:

Stay with them. Listen without rushing to solve. Do not leave someone at immediate risk alone.

Reduce access to means. This is the single most effective practical action — removing medications, firearms, and means of access. Suicidal crises are frequently short-lived, and survival of an attempt is followed by long-term survival in the large majority of cases. Putting distance between a person and a method saves lives.

Get help now — emergency services, a crisis line, or an emergency department. Any country's emergency number works, and most countries have a dedicated crisis line.

And do not promise secrecy about a life-threatening risk. It is a promise that cannot be honoured safely.

For yourself: if you are having these thoughts, they are a symptom of a treatable illness distorting your judgement about your own future. Depression lies specifically about permanence. Tell one person today, and go to an emergency department if the thoughts are immediate.

What is genuinely hopeful

Treatment works, and it works for most people who stick with it.

Sequential treatment matters — the first thing not working says little about the second.

The options have expanded substantially — rapid-acting treatments, brain stimulation, better-delivered therapy, and digital access.

And recovery is the normal outcome. Most episodes end. That is not encouragement; it is what the data shows.

What the next page fixes

Chapter 20.8 covers anxiety disorders — the commonest mental health conditions of all, what separates them from ordinary worry, and why the treatment that works best is the one that feels most counter-intuitive.