Anxiety cases rose 50% in three decades. The age-standardised rate did not move.

The Global Burden of Disease Study's answer to "is anxiety increasing?" is that both answers are true and they measure different things. The absolute number of people with an anxiety disorder rose by 50% between 1990 and 2019, while the overall age-standardised burden rate remained stable across those same three decades [s1]. More people, at about the same per-person risk.

Meanwhile the rate at which anxiety gets diagnosed in a clinical encounter has moved a great deal, and that is a third quantity again [s2].

The GBD estimates, stated in full

For 2019, the study estimated 45.82 million incident cases of anxiety disorders globally (95% uncertainty interval 37.14 to 55.62), 301.39 million prevalent cases (95% UI 252.63 to 356.00), and 28.68 million disability-adjusted life years (95% UI 19.86 to 39.32) [s1]. The analysis covered 204 countries and regions from 1990 to 2019, and used the estimated annual percentage change to quantify temporal trends by sex, region and age [s1].

The headline temporal finding is the one that gets lost: the overall age-standardised burden rate of anxiety disorders remained stable over the past three decades, while the latest absolute number increased by 50% from 1990 [s1]. Age-standardisation strips out the effect of a population that is both larger and older. What is left, on these estimates, did not rise.

That is not a small distinction. A 50% increase in cases is a real increase in need — in clinicians required, treatments delivered and years lived with disability. It is simply not evidence that something has made individual people more anxious than they used to be.

The study also found large disparities in both the age-standardised rate and its trend across sex, country and age, so the global stability conceals movement underneath it [s1]. And it reports one attributable-risk finding directly: in 2019, 7.07% of global DALYs due to anxiety disorders were attributable to bullying victimisation, mainly in the population aged 5 to 39, with the proportion increasing in almost all countries and territories compared with 1990 [s1].

Diagnosis is measured separately, and it moved

A study of the National Ambulatory Medical Care Survey — a nationally representative annual survey of US office-based visits — tracked anxiety diagnosis and treatment in children, adolescents and young adults across three periods between 2006 and 2018 [s2].

The proportion of office visits carrying an anxiety disorder diagnosis rose from 1.4% (95% CI 1.2–1.7; n = 9,246,921 visits) in 2006–2009 to 4.2% (95% CI 3.4–5.2; n = 23,120,958 visits) in 2014–2018 [s2]. That is a change from 1.4% to 4.2% in the share of clinical encounters in which the diagnosis appears [s2].

What happened inside those visits changed too, and not in the direction the diagnosis trend might suggest. The proportion of visits including any therapy fell from 48.8% (95% CI 40.1–57.6) to 32.6% (95% CI 24.5–41.8), while there was no significant change in the overall use of medications [s2]. The likelihood of receiving medication alone during a visit was significantly higher in the last period than the first, with a relative risk ratio of 2.42 (95% CI 1.24 to 4.72) [s2].

So diagnosis went up, therapy went down as a share of what those visits contained, and medication-only care became relatively more common [s2].

Three quantities, three different questions

Prevalence is how many people meet criteria. Age-standardised rate is how likely a person of a given age is to meet them. Diagnosis rate is how often a clinician records the condition. Public argument about whether anxiety is "really" rising routinely swaps one for another, and each swap produces a different conclusion from the same underlying reality.

There is also a live hypothesis that the third quantity may feed back into the first two. Researchers writing in New Ideas in Psychology have proposed what they call the prevalence inflation hypothesis — asking whether mental health awareness efforts are contributing to the rise in reported mental health problems — and framed their paper explicitly as a call to test it [s3]. It is a hypothesis and a research agenda, not a finding, and it should be read as exactly that.

What none of this establishes

The GBD figures are modelled estimates built from a systematic review, and their uncertainty intervals are wide — the prevalence estimate spans 252.63 million to 356.00 million [s1]. Modelled global estimates depend heavily on where primary data exist, and they exist unevenly.

The diagnosis study covers office-based visits in one country and does not follow individuals, so a rise in the proportion of visits with an anxiety diagnosis could reflect more anxious young people, more recognition of the same anxious young people, changes in who visits an office at all, or coding practice [s2]. It ends in 2018, before the pandemic period that dominates most current discussion.

And the prevalence inflation hypothesis has, on its own authors' framing, not yet been tested [s3].

The defensible summary

More people have anxiety disorders than in 1990, by about half again [s1]. On the best available global estimates, the per-person risk did not rise over that period [s1]. In US office practice among people under 26, the diagnosis appeared in 4.2% of visits in 2014–2018 against 1.4% in 2006–2009, alongside a fall in the share of those visits that included therapy [s2].

Whether the underlying experience of anxiety has become more common is a question these data are not built to answer, and saying so is more useful than picking whichever of the three numbers supports a prior.

Sources

• Global, regional and national burden of anxiety disorders from 1990 to 2019: results from the Global Burden of Disease Study 2019 — Epidemiology and Psychiatric Sciences, 2021-05-06
• Trends in Office-Based Anxiety Treatment Among US Children, Youth, and Young Adults: 2006-2018 — Pediatrics, 2023-06-07
• Are mental health awareness efforts contributing to the rise in reported mental health problems? A call to test the prevalence inflation hypothesis — New Ideas in Psychology, 2023-02-10