Mental Health and Loneliness
A generational rise in anxiety, a loneliness epidemic across age groups, and a public-health story that is reshaping how the next generation grows up. The data, the disagreements, and what specific things help.
(CDC; down from 42% in 2021, above 30% in 2015)
(about 24% in low-income countries, twice the high-income rate)
(up from 3% in a 1990 Gallup poll)
A note on framing. Mental health and loneliness are real measurable phenomena that have shifted in measurable ways over the last fifteen to twenty years. The page below tries to walk through the picture - what the data shows, where the disagreement is, and what specific things help - without either the panic that loud commentary sometimes produces or the dismissal that other commentary applies. The crisis is real and is also more multi-causal than any single explanation captures.
The shape of the crisis
Several dimensions of mental health have visibly worsened in most developed countries since around 2010-2012. The data is consistent enough across multiple measurement systems that the underlying shift is hard to dismiss as a measurement artifact.
Adolescent depression and anxiety rose sharply. US data from the National Survey on Drug Use and Health shows the share of adolescents reporting a major depressive episode roughly doubled between 2010 and 2020, with the increase concentrated in girls. Similar patterns appear in UK, Australian, Canadian, and parts of European data. The increase was not subtle, and since 2021 it has partly reversed in the US: the survey's depression figure fell from about 21% in 2021 to 15% in 2025. In the CDC's 2025 Youth Risk Behavior Survey, 33% of high school students reported persistent sadness or hopelessness (43% of girls), down from 42% in 2021 but above 30% in 2015. The share using social media every hour or more did not change between 2023 and 2025, at 36%.
Adolescent suicide rates have risen. US suicide rates among teens, particularly girls, rose meaningfully from around 2010 onward. The number remains low in absolute terms - suicide is rare even in stressed cohorts - but the trend was clearly upward and concentrated in younger groups. Data from the UK and other developed countries shows similar patterns. The rise has not continued uninterrupted: US rates for ages 10 to 24 peaked in 2021 and were lower in 2024, though the fall came among boys and young men while rates for girls have not changed significantly. That is too short a run to call a reversal, but it is worth holding against any account that treats the trend as still climbing.
Loneliness is widespread. The 2023 US Surgeon General's advisory on loneliness drew on multiple studies showing that about half of US adults reported measurable loneliness in recent surveys. A 2025 World Health Organization commission found loneliness most common among teenagers and linked it to some 871,000 deaths a year. The rise predates the pandemic but accelerated during it.
"Deaths of despair" rose sharply in working-age Americans. The combination of deaths from suicide, alcohol, and drug overdose - mostly fentanyl in recent years - rose dramatically in working-age Americans without college degrees from about 2000 through 2022. The pattern has been documented most carefully by Anne Case and Angus Deaton, whose framing has become widely used. The opioid crisis is the largest contributor; alcohol-related deaths have also risen. Overdose deaths have since fallen fast, to about 67,000 in the 12 months to April 2026 (CDC provisional counts), 40% below their 2023 peak and below the 2019 level.
Prescriptions have risen substantially. US antidepressant use rose nearly fivefold between 1988-1994 and 2005-2008, and in 2023 more than 1 in 10 US adults took a prescription medicine for depression. In England, 9.1 million people were prescribed an antidepressant in 2025/26, and the number of adults prescribed ADHD drugs rose 40% in a year. Some of this reflects expanded recognition and treatment of conditions that previously went undiagnosed; some reflects an increase in the underlying problem; some reflects a willingness to treat conditions pharmacologically that earlier generations would have treated through community support, work, religion, or simply tolerated.
Adolescents and the smartphone hypothesis
The single most-discussed explanation for the adolescent mental-health rise is the role of smartphones and social media. Jonathan Haidt's 2024 book "The Anxious Generation" laid out the case in detail and has been widely engaged with. The case has roughly four parts.
First, the timing. Smartphones reached majority adoption among adolescents around 2010-2012. Social media use spread to most adolescents around the same period. The mental-health declines began at roughly the same time, in the same demographic groups, in multiple countries. The time-correspondence is consistent enough across settings to be hard to attribute to coincidence.
Second, the dose-response pattern. Adolescents who use smartphones and social media more heavily have measurably worse mental-health outcomes on average. The pattern is stronger for girls than boys and stronger for some platforms (Instagram, TikTok, Snapchat) than others. In the 2026 World Happiness Report, Western European girls on social media seven or more hours a day rated their lives almost a point lower (out of 10) than those on it under an hour. Outside the English-speaking world and Western Europe, the links were more positive. Many studies have replicated the association; whether it reflects cause or effect is disputed.
Third, the mechanism. The smartphone-driven adolescence replaces in-person socializing with screen-mediated socializing, replaces play with passive consumption, replaces sleep with late-night scrolling, replaces real-world skill-building with comparison-driven status competition, and produces a dopamine cycle of variable-reward checking that resembles other addictive patterns. Each of these has measurable individual effects; together they describe a different developmental environment than previous generations had.
Fourth, the natural experiments. When Facebook reached US colleges one campus at a time in the mid-2000s, students' mental health worsened, a 2022 study found. Studies of school phone bans are mixed. In Norway, bans cut girls' visits for psychological care and reduced bullying, though with no average effect on mental health; in a large Florida district, test scores rose in the second year. But an English study of 30 schools found no difference in wellbeing between restrictive and permissive policies, and a national US study of lockable phone pouches found no average effect on test scores, with student wellbeing dipping in the first year and rising in the second.
The case is not closed. Andrew Przybylski (Oxford), Amy Orben (Cambridge), Candice Odgers (UC Irvine) and others argue that the effects are small: in Orben and Przybylski's 2019 analysis of 355,358 adolescents, technology use explained at most 0.4% of the variation in wellbeing. They also argue that distress may drive heavy use as much as the reverse, and that other shifts, such as academic pressure and post-pandemic disruption, deserve more weight. Haidt and colleagues replied in 2026 that such studies blend different groups and technologies, diluting harms that appear when, for example, girls' social media use is examined on its own. Both sides accept that heavy use goes with somewhat worse mental health on average; they disagree on how much of it social media causes.
Loneliness across age groups
Loneliness is not just a young-person problem or an old-person problem. The data shows elevated loneliness in multiple demographic groups, with different specific drivers in each.
Older adults living alone. The traditional loneliness population. Widows, people whose adult children live far away, people whose health limits social activity. The combination of declining family co-residence, falling religious participation, and declining community organizations has hit this group hard. The intervention research is reasonably strong: regular group activity, intergenerational programs, structured volunteering, and in some cases medication for treatable depression all help.
Working-age men, particularly those who are not partnered. A growing population in nearly every developed country. The combination of declining marriage rates (covered on the Family page), shrinking workplace social bonds, declining participation in religious and civic groups, and the specific cultural pattern in which men are less likely to maintain close friendships in adulthood produces measurable loneliness with serious consequences.
Mothers of young children. Despite being surrounded by people, parents of young children - particularly mothers, particularly first-time mothers - report some of the highest loneliness scores of any demographic group. The reasons are well documented: caring for young children is exhausting, isolating, and often poorly supported by families and workplaces. The pattern has worsened as nuclear-family households have replaced multi-generational ones in most developed countries.
Young adults. The 18-25 demographic shows higher loneliness scores than even older adults in many recent surveys. The specific drivers include the smartphone-and-social-media factor described above, the late timing of household formation, the geographic dispersion that follows university and early-career relocation, and the decline in young adults' civic and religious participation. The loneliness-among-young-adults pattern is relatively new.
Recent immigrants and people in life transitions. A more local pattern but a real one. The disconnection of being new in a country, a city, a job, or a life situation produces measurable loneliness even when the formal social environment includes many people.
How countries compare
Different countries report dramatically different mental-health and loneliness profiles, even with similar income levels.
The takeaway is less tidy than a family-versus-individualism story. Family and community ties matter, but the WHO's 2025 estimates find loneliness about twice as common in low-income countries (24%) as in high-income ones (11%), so poverty and insecurity matter too. Among rich countries, the high-competition societies of East Asia and the US show more strain than the Nordic countries, whose strong welfare institutions substitute for some of what family and community provide elsewhere.
The paths from here
Dated September 2026, with US teen distress easing from its 2021 peak, US overdose deaths down 40% from their 2023 peak, and social media age limits spreading from Australia. Mental-health crises are slow-moving social phenomena; reversal typically requires sustained change in multiple variables at once. Each path below is one realistic shape the next decade could take.
Continued worsening
Adolescent and young-adult mental-health rates continue rising. Loneliness spreads further. Mental-health treatment infrastructure expands but cannot keep up with demand. The pattern becomes the new normal rather than getting addressed.
Will it happen? Less likely than it looked in 2021. US teen measures have improved for four years, though they remain above 2015 levels, and the US suicide rate has been statistically flat since 2017. But the forces pessimists point to - smartphone culture, declining community, family fragmentation, economic stress on young adults - have not weakened, so a renewed rise cannot be ruled out.
A cultural correction on smartphones and social media
Schools, families, and governments make meaningful changes to how children and adolescents interact with smartphones and social media. Phone-free schools become the norm rather than the exception. Age verification is enforced for social-media access. Parents collectively delay smartphone provision until later in adolescence. The screen-driven adolescence of the 2010s and early 2020s becomes a temporary period rather than a permanent feature.
Will it happen? Under way. Australia has barred under-16s from major social media since December 2025; Indonesia and Malaysia followed in 2026, the UK plans a ban from spring 2027, and on September 17, 2026 the European Commission proposed a KIDS Act that would bar accounts under 13 and set 15 as the age for one's own. Several US states have laws of their own. Enforcement is the weak point: three months into Australia's ban, the eSafety regulator found 81.5% of 10-to-15-year-olds still on the restricted platforms (85.9% before). Whether any of this moves the mental-health numbers is the open question.
Mental-health system expansion
Insurance coverage of mental-health care expands. Telehealth makes therapy more accessible. The supply of mental-health clinicians grows. Cognitive-behavioral therapy and other evidence-based interventions become widely available. The gap between mental-health need and mental-health treatment narrows substantially.
Will it happen? Partly happening. Telehealth has widened access, but the supply of clinicians remains a binding constraint. AI chatbots are a new variable: in the first randomized trial of a generative-AI therapy bot (2025), 210 adults' depression and anxiety symptoms fell more over four weeks than a waitlist group's. But OpenAI said in October 2025 that 0.15% of ChatGPT's weekly users showed explicit signs of possible suicidal planning, and families have sued OpenAI and Character.AI over teenagers' suicides. Character.AI and Google agreed in January 2026 to settle the first cases; OpenAI denies ChatGPT was to blame.
Religious and community revival as a partial counter
The religious and community institutions that historically provided mental-health buffering recover ground in some cohorts. New community formations (running clubs, religious congregations, men's groups, neighborhood organizations, structured volunteering) absorb some of the loneliness and provide some of what previous generations got from churches, unions, and extended families. The institutional ecosystem of belonging partly rebuilds.
Will it happen? Possible but slow. The long decline in US religious participation has leveled off since about 2020, but without a revival (see the Religion page). Some new community forms have emerged, though most operate at small scale relative to the institutions they would need to replace. The longer-run question is whether the broader culture rediscovers the value of shared institutional belonging or continues toward more individualistic patterns.
Deaths of despair partially recede
The combination of fentanyl-control efforts, opioid-prescription reform, alcohol-policy improvements, and slowly improving labor-market conditions for non-college-educated workers reduces the mortality gap that has driven the deaths-of-despair pattern. The headline numbers improve even as the underlying conditions remain hard.
Will it happen? Partly happening: overdose deaths are now below their 2019 level. Alcohol deaths remain high; among adults aged 45 and over the rate was 25.2 per 100,000 in 2024, against 19.5 in 2014. The labor-market conditions that Case and Deaton blame have not fundamentally reversed.
The base case is uneven change: further improvement where it has begun (US teen measures, overdose deaths) and where policy is moving (age limits, phone-free schools, treatment capacity, perhaps AI tools), with slower movement on loneliness and the meaning-and-belonging deficit. The mental-health story is long-running; reversal will not be quick.
Where serious analysts disagree
Mental health is one of the topics where careful researchers disagree on both the severity of the crisis and the best response. Each reading below is held by named scholars worth engaging directly.
The smartphone-and-social-media story is real and underweighted
The timing, dose-response patterns, mechanism, and natural-experiment evidence all point toward smartphones and social media as a substantial cause of the adolescent mental-health crisis. Dismissing this evidence to avoid uncomfortable policy implications has cost a generation. Schools, governments, and parents should act on what is already known rather than waiting for definitive evidence that may not arrive in time.
Held by: Jonathan Haidt (NYU, "The Anxious Generation"), Jean Twenge (San Diego State, longtime generational researcher), and a substantial fraction of the developmental-psychology community. Beyond the evidence above, a 2025 court filing alleged that Meta halted an internal study after people who quit Facebook for a week reported less depression and anxiety; Meta says the study had methodological problems. In March 2026 a Los Angeles jury found Meta and YouTube negligent over addictive design; both vowed to appeal. The core observation that something changed in adolescent life around 2012 is widely accepted.
The smartphone story is overstated
The effect sizes in much of the smartphone-and-mental-health research are smaller than the headlines suggest. Other concurrent changes (academic pressure, family disruption, post-pandemic effects, climate anxiety, broader cultural shifts) deserve more weight. Targeting screens as the main intervention may be politically attractive but is empirically thin and could produce reform energy that misses the actual problem.
Held by: Andrew Przybylski (Oxford), Amy Orben (Cambridge), Candice Odgers (UC Irvine), and a body of methodological critique of the screen-time research. The evidence on their side includes small average associations, longitudinal studies with mixed results on which way causation runs, and school phone-ban studies that found no average effect on mental health.
Deaths of despair are about labor markets and economic shifts
The rise in mortality from suicide, alcohol, and drug overdose among working-age non-college-educated Americans tracks the decline in good-paying jobs without college credentials and the disruption of communities by deindustrialization. It also tracks the breakdown of the family and community structures that previously provided meaning. Addressing the medical surface (better treatment for opioid-use disorder, better mental-health access) helps but misses the deeper economic and social causes.
Held by: Anne Case and Angus Deaton (Princeton, "Deaths of Despair"), and a body of health-economics work. Their framing has been influential in policy. Critics such as the economist Christopher Ruhm find that local economic decline explains less than a tenth of the rise in drug deaths, and point instead to the drug supply, from prescription opioids to heroin and fentanyl.
The mental-health crisis is partly an over-medicalization story
Some of the apparent rise in anxiety and depression reflects expanded diagnostic categories, more aggressive screening, and a broader cultural willingness to label normal emotional struggles as treatable conditions. A self-help and therapy industry benefits from more diagnoses. The increase in suffering is real but smaller than the diagnostic figures suggest, and treating ordinary distress as mental illness produces its own harms.
Held by: Allen Frances (Duke, who led the DSM-IV task force and has written extensively on overdiagnosis), the psychologist Lucy Foulkes, and parts of the critical-psychiatry tradition. Foulkes's "prevalence inflation" hypothesis holds that awareness campaigns lead some people to read milder distress as a disorder. Critics reply that measures independent of labels rose too: the US suicide rate for girls aged 10 to 14 went from 0.5 per 100,000 in 1999 to 2.3 in 2024.
The deeper problem is the meaning-and-belonging deficit
The mental-health crisis is the surface symptom of a deeper structural shift: the decline of religious participation, the fragmentation of families, the erosion of community institutions, the rise of individualistic culture, and the disconnection of work from purpose. Treating the surface (more therapists, more medications, more apps) without addressing the underlying meaning-and-belonging deficit will continue to fall short. The real solutions involve rebuilding institutions of meaning on a large scale, which is hard and slow.
Held by: Robert Putnam (whose work on social capital is foundational here), Sebastian Junger ("Tribe"), Vivek Murthy (former US Surgeon General whose loneliness advisory framed the problem this way), and a tradition of community-focused public-health analysis. The argument is harder to translate into specific policy, and harder to test, than other framings.
Where the evidence is firmer: the post-2012 rise in adolescent distress was real and has partly reversed since 2021; heavy social media use goes with worse mental health, especially for girls; and overdoses, the largest part of the deaths-of-despair rise, have receded since 2023. Still disputed: how much social media causes, how much of the rise in diagnoses reflects changing labels, and how much weight the meaning-and-belonging deficit deserves. Proposals range from specific interventions (age limits, phone-free schools, treatment capacity, opioid policy) to the slower work of rebuilding institutions of belonging. None of this is fast.
What this means for you
Mental health and loneliness are unusually personal among the topics on this site. A few practical observations supported by the research:
If you are young or have school-age children
Researchers disagree about how much smartphones and social media harm adolescents, but the precautionary steps cost little. They include delaying smartphone access until later in adolescence, structuring phone-free time during the day, keeping phones out of bedrooms at night, and supporting in-person friendship and outdoor time. The collective version (parents agreeing together to delay phones) tends to work better than individual decisions, because no child is left as the only one without.
If you struggle with anxiety or depression
Most cases respond meaningfully to evidence-based treatment - cognitive-behavioral therapy, exercise, sleep regulation, social connection, and where appropriate medication. The cost of seeking treatment is much lower than it was a generation ago; the cost of not seeking it remains high. Reaching out to a primary-care physician or directly to a therapist is a small action that generates substantial value for most people who do it. None of this is medical advice; it is observing what the evidence consistently shows about treatment access.
If you feel lonely
Loneliness is more common than people who experience it usually realize. Many of the people around you are lonely too, even when their lives look full from outside. The evidence-based interventions are straightforward: regular structured activity with the same people over time (a sports league, a religious congregation, a class, a volunteer project), maintaining old friendships actively rather than letting them lapse, and being willing to be the person who initiates social contact rather than waiting for it. Loneliness is not a personal failure; it is a common feature of modern life that requires deliberate work to counter.
If you work in mental health or healthcare
The demand-supply gap in mental-health care is among the largest in any healthcare specialty. The supply of clinicians is tight in most rich countries, and the training pipeline produces fewer therapists than the population needs. The intersection of mental health and AI tools, of mental health and primary care integration, and of mental health with workplace wellness programs are all growing areas of need. Anyone with both clinical and technical or organizational skills is in unusually high demand.
If you are thinking about community
Most communities under-invest in the institutions of belonging - third places (cafes, parks, libraries, community centers), religious congregations whatever their tradition, civic organizations, sports leagues, neighborhood groups. The data on what produces resilient communities is reasonably clear: the people who join and sustain these institutions, even when it is inconvenient, are the backbone of community wellbeing. Showing up for things matters in ways the individual benefit rarely captures. Whatever policy does, much of the work of rebuilding belonging comes down to individual choices to participate.
The mechanics behind this
The mental-health story sits on top of three deeper mechanisms covered elsewhere on this site. If the analysis above depends on ideas you want to understand first, these fundamentals make the conversation more legible:
