Half of People, One in Twenty Today: The Statistic Trap You Keep Falling Into

Here is a number that changes how you read a room: by age 75, roughly half of all people will have experienced at least one diagnosable mental health disorder. Not half at some unlucky point in history — half of ordinary people, across 29 countries, living ordinary lives.

That number only makes sense once you understand the model it comes from, and the model matters far more than the headline.

The situation

Most health statistics give you a snapshot. The Global Burden of Disease study estimates that in 2023 about 5% of people in the United States had depression. That tells you how many are depressed right now. It tells you almost nothing about your odds across a whole life — because mental illness tends to come and go in episodes, sometimes once, sometimes returning decades later.

Asking "how many people have depression today?" is like asking "how many people have a cold today?" and concluding colds are rare. The useful question is: over a full life, what are the chances it ever happens to you?

The model: lifetime risk and the cumulative curve

The tool that answers this is lifetime risk — the probability that something happens to a person at any point before a given age — reconstructed from a cumulative incidence curve. It is a standard instrument in epidemiology and actuarial science, the same math insurers use to price whether you will ever make a claim, not whether you are claiming this year.

You cannot measure lifetime risk directly without following one person for 75 years. So researchers interview people of every age at once — 20-year-olds, 50-year-olds, 70-year-olds — ask each about their full history, and stitch the answers into one curve. A 25-year-old shows the early part of the story; a 70-year-old shows the whole arc. Laid together, they give the share of people who have experienced a condition by each age.

The shape carries the insight. The curve climbs steeply when people are young and flattens later, because for most mental health conditions the first episode arrives early: roughly half of everyone who will ever experience depression has their first episode before 30.

The figures here come from a 2023 study by John McGrath and colleagues, drawing on the World Mental Health Surveys — more than 150,000 face-to-face structured diagnostic interviews conducted between 2001 and 2022.

What the curves say

By age 75:

• Mood disorders (depression, bipolar): more than one in three women, one in four men.
• Anxiety disorders (generalised anxiety, phobias, panic, social anxiety, PTSD): almost one in three women, nearly one in five men.
• Substance use disorders (alcohol or drugs): more common in men.

Two sex differences point in opposite directions: depression and anxiety are more common in women; alcohol and drug problems more common in men. Bipolar disorder hits both equally, so the mood-disorder gap is driven almost entirely by depression.

You cannot simply add these up — the same person often carries several conditions, and counting each would count that person many times. After correcting for overlap, the authors conclude: about 46% of men and 53% of women will develop at least one of these disorders by age 75. Around half of everyone.

The two honest objections

Is it too high? The instinct is to suspect inflation — loosened definitions turning normal sadness into "depression." This is a real debate. Definitions do matter: the broad category "alcohol abuse" captures far more people than strict "alcohol dependence," and where you draw the line moves the number.

Or too low? The data pushes this way. These figures are probably conservative, because of recall bias: the surveys depend on people remembering their own past episodes, and memory fades — older respondents under-report episodes from decades earlier. Studies correcting for this tend to find higher lifetime risk.

So: the measured lifetime prevalence is roughly 50% (fact). Whether that reflects real suffering or diagnostic drift is contested (interpretation). The exact true figure is unknown, because no one perfectly recalls a lifetime.

Where else this shows up

The transferable lesson is point-in-time versus lifetime thinking, and it reaches well beyond mental health.

Cancer looks rare on any given day, yet lifetime risk is close to one in two. Unemployment sits in single digits at any moment, yet most working adults spend some stretch out of work. Annual divorce rates look small; the lifetime probability does not. In each case a condition that is uncommon at any instant becomes near-universal across a span — because it moves in and out of lives, and a long life gives it many chances to arrive.

Carry one habit away: whenever you hear a prevalence figure, ask whether it is a snapshot or a lifetime. The two can differ tenfold and imply opposite attitudes. A 5% snapshot says this probably won't be me. A 50% lifetime risk says plan as though it might — buy the insurance, build the savings buffer, learn the warning signs.

What to watch

The thing that would settle the inflation debate: longitudinal studies that re-interview the same people over time — removing recall bias — and still find lifetime risk near or above half. Several high-income-country studies already point that way.

My expectation: roughly 70% that future recall-corrected longitudinal estimates land above today's 50% figure rather than below it, because memory loss works in one direction and almost nobody invents a mental illness they never had. If the next wave of such studies comes in well under 50%, treat this model's application here as weaker than claimed.

The practical point is quieter than the statistic. If half of people will face this, a mental health episode is not an aberration from a normal life. It is part of what a normal life, lived long enough, tends to contain — for you, or for someone sitting close to you right now.

Distilled from Our World in Data