Children and young adults who use phones, social media and games compulsively are about twice as likely to be depressed or anxious, to harm themselves, to sleep badly and to feel lonely. That is the finding of the largest review of the evidence to date, which covers more than 9 million people under 25.
The analysis, published Monday in eClinicalMedicine, a Lancet journal, pools 121 earlier reviews. It anchors a three-paper series in The Lancet by 13 public health researchers. They concede the data can’t prove that screens cause the harm but argue the pattern is consistent enough, and the exposure common enough, for governments to act without waiting for proof.
“Governments have let technology companies write their own rules, leading to business models built to capture attention and collect data, with health treated as an afterthought,” said Ilona Kickbusch of the Digital Transformations for Health Lab in Geneva, who chairs the Series.
The harms, at a glance
Compulsive use, ages 10 to 24. The review calls it problematic use: lost control, preoccupation, and carrying on despite the damage. Compared with peers who use screens less or without that pattern, young people who fit it were:
-3.3 times as likely to have conduct problems and persistent negative emotions-2.3 times as likely to have depression, anxiety or other common mental health problems-2.3 times as likely to be lonely-2.2 times as likely to self-harm or have suicidal thoughts-2.2 times as likely to have sleep problems-2.2 times as likely to be physically inactive-1.75 times as likely to drink, smoke, use drugs or take other risks
{{/usCountry}}-3.3 times as likely to have conduct problems and persistent negative emotions-2.3 times as likely to have depression, anxiety or other common mental health problems-2.3 times as likely to be lonely-2.2 times as likely to self-harm or have suicidal thoughts-2.2 times as likely to have sleep problems-2.2 times as likely to be physically inactive-1.75 times as likely to drink, smoke, use drugs or take other risks
{{/usCountry}}more likely to report a poorer quality of life
Social media use, ages 10 to 24. Measured simply by how much or how often young people use it, social media was tied to:
-a 70% higher likelihood of common mental health problems,-a 50% higher likelihood of drinking,-a weaker link to eating disorders,-no statistically significant link to self-harm, sleep or overall well-being.
Who is most at risk
Girls. The link between social media and depression was much stronger for girls than boys, and passive scrolling was tied to lower mood and worse sleep among girls.
Young adults. The link between compulsive use and poor mental health was stronger among 18- to 24-year-olds than among younger teens.
Children up to 10
Under-fives who spent more time on screens had more emotional and behavioural problems.
Children aged 5 to 10 lost about 11 minutes of sleep for every extra hour on a phone or tablet.
Fantasy-heavy content was associated with weaker learning and attention.
Physical health, all ages
More screen time was linked to obesity, short-sightedness, metabolic risk, headaches and lower back pain.
Of the 316 links the researchers examined between a type of screen use and a health outcome, more than three in four pointed to harm.
The India undercount
Less than 5% of the evidence comes from low- and middle-income countries such as India, where young people’s screen use is climbing fast, the authors noted. The findings rest mostly on children in North America, western Europe, East Asia and Australia. Three in four Indian 14- to 16-year-olds use phones for social media, according to survey data. Young people’s fast-growing use of generative AI, for schoolwork and for company, barely registers in the research at all, the Series says.
India’s protections come from several laws: the IT Rules of 2021, the Protection of Children from Sexual Offences Act, and the Digital Personal Data Protection Act of 2023, and some of these statutes include companies to act more when it comes to children. But the most crucial of these, such as the provisions that requires verifiable parental consent before platforms process data from anyone under 18, are yet to come into effect since the data protection law as a whole is not fully operational.
The paper notes that enforcement “is uneven and child-specific design and systemic risk obligations are less developed than in some other jurisdictions,” the Series’ governance paper says.
“Governments shouldn’t wait for ‘perfect’ evidence before acting for our generation and the ones coming after us,” said Salman Fitrat Khan of OP Jindal Global University’s Jindal School of Public Health and Human Development in Sonipat, a co-author of the review and a member of the Series’ youth advisory group. “The health risks and harms associated with digital systems are preventable.”
Where screens helped
The review found benefits too:
-Under-fives who watched educational programmes, especially with a parent, had better language development.-Active video games, the kind that get players moving, were linked to higher well-being and more exercise.-Body-positive content was tied to better well-being; feeds of idealised bodies and “fitspiration” to worse.-Among 18- to 24-year-olds, moderate and social use went with better well-being and a stronger sense of support.
“The central question for public health is not only ‘how much’ but also ‘what, how, and with whom’,” the authors wrote.
How large the harm could be
A companion paper in the Series, led by Louise Holly of the Digital Transformations for Health Lab, put a global number on the risk. Using a common guideline of two hours of recreational screen time a day, it estimated that about 20% of depression cases and 15% of obesity cases among under-25s are associated with exceeding that limit. Only about a third of children aged 5 to 17, in a study of 57 countries, stay under it.
That works out to around 2.8 million years of healthy life lost to depression and 240,000 to obesity worldwide each year. The measure, disability-adjusted life years, counts time lost to illness as well as early death.
The authors treat the figure with caution. “They should not be taken as an accurate measurement of harm as, although they are based on the best available research, the certainty of the evidence is low,” said Falk Müller-Riemenschneider of the National University of Singapore, a senior author of the review. The point is scale. Small risks to each child add up to a large burden when nearly every child is exposed, and 93% of children in OECD countries are using digital media by age 10.
The case against platform design
Most studies count hours on screen because hours are easy to count. The Series argues that misses the point. The features that decide what young people see and how long they stay are harder to study, because the companies hold the data.
“Most digital platforms are designed by companies to capture attention and collect data, not to protect health,” said Jay Shaw of the University of Toronto, a Series author. Infinite scroll, autoplay and personalised feeds, he said, are “deliberate design choices, not simply a matter of individual self-control.”
Earlier this year, twenty-nine US states took that argument to court against Meta, in a product-liability case that treated the design of its platforms as defective. Meta settled for $18 billion in the third week of the trial. But 30% of the payout depends on TikTok and YouTube accepting matching terms, and the design changes Meta agreed to apply only to teen accounts in the US.
Six young adults aged 23 to 30, drawn from five continents, advised the authors of the new report, describing habits such as doomscrolling and constant comparison as responses to how platforms are built. The authors recorded that as lived experience and did not treat it as data.
The Series’ governance paper, led by Barbara Prainsack of the University of Vienna, accuses the industry of stalling regulation. “Technology companies are borrowing a playbook well known from tobacco and fossil fuels: manufacture doubt, discredit critics, frame regulation as a threat to freedom,” she said. “When regulation is missing, it doesn’t mean that a technology is ungoverned; it just means that companies set the rules instead of public policy.”
What the authors want
Age limits alone won’t do it, the Series says. It calls for:
-mandatory safety testing of new products and features before release,-company data on how features affect health, shared with vetted independent researchers,-a global treaty under the World Health Organization to set minimum standards, so weak rules in one country don’t undercut others,-an international body to monitor AI harms to children and young people.
It also points to the courts. “Legal action can do something regulation often cannot: compel companies to disclose internal documents and shift the public conversation about who is responsible for harm,” Kickbusch said. Litigation, she added, “is not a substitute for regulation”.
Until 2024, most platforms set their minimum age at 13, a threshold inherited from a 1998 US children’s privacy law that younger children easily get round. Australia raised the bar that year, barring under-16s under its Social Media Minimum Age Act. The Series also cites a preliminary finding by the European Commission that Meta may have breached the EU’s Digital Services Act by failing to enforce its own minimum age of 13.
The Union government and several state government are weighing limits on social media for under-16s – planning the study acknowledges.
How the study was done, and how sure it is
The paper is an umbrella review, a review of reviews. Researchers at the National University of Singapore’s Saw Swee Hock School of Public Health, led by Natarajan Padmapriya, ran no new experiments. They gathered every systematic review on young people’s digital media use and health published from January 2021 to December 2025: 121 in all, 52 with pooled statistics, drawing on about 2,790 studies. They removed overlap so no dataset was counted twice, then sorted results by age. The cutoff of 24 follows the Lancet Commission on Adolescent Health and Wellbeing, which notes that the brain’s systems for self-control and reward keep developing into the mid-twenties.
The authors are frank about the limits. Of the 121 reviews, 92% reached only low or very low certainty under GRADE, the standard scale for grading medical evidence.
“The associations between problematic digital media use and health harms identified in our review are modest in size and rest on low-certainty evidence that cannot establish cause and effect,” Holly said.
To be sure, the low certainty grades largely reflect the observational design of the research, and most reviews that graded their source studies rated them high or moderate in quality. Tests found no sign that studies showing harm were likelier to be published. And across hundreds of studies, outcomes and age groups, the findings leaned consistently toward harm. Together, the authors wrote, that is “a sufficient basis for precautionary public health action”.