select search filters
briefings
roundups & rapid reactions
Fiona fox's blog
Scientists comment on a series of papers, published in The Lancet, looking at health impacts of digital media use.
Dr Marco Bardus, Assistant Professor in Public Health, University of Birmingham, said:
“This Series makes an important contribution by moving beyond simplistic debates about screen time and focusing attention on the broader digital environments that shape young people’s health. The authors are appropriately cautious about the evidence presented and discussed (the umbrella review which will appear in eClinical Medicine). The findings of this large “review of reviews” are consistent with existing evidence, but are almost entirely based on observational, mostly cross-sectional studies rated low or very low certainty by the authors’ own assessment. The 2.8 million healthy years figure is a modelled scenario that assumes a causal link the paper elsewhere says has not been established, and should be reported as an illustration rather than an estimate.
“This is not a story about a uniformly harmful technology, as the same umbrella review finds educational content, active gaming and moderate social use associated with better development and wellbeing, and digital media remains a real channel for education, connection and health promotion. Harms are easier to measure than benefits, and the literature reflects that imbalance. The most valuable contribution here is diagnosing why we still don’t know the answer: the data that would settle it sit inside the companies.
“The Series therefore provides a useful framework for considering digital technologies as a public health issue, while also highlighting the urgent need for greater access to platform data (controlled by the industry), stronger longitudinal independent research, and more robust evaluations of regulatory and policy interventions.”
Dr Carolina Scarton, Senior Lecturer in Natural Language Processing, University of Sheffield, said:
What do we know about the governance, regulation and data collection of these platforms and whether what the researchers call for is possible to implement/ will achieve changes to behaviour / reductions in harm?
“The European Digital Service Act (DSA) is a first-of-its-kind regulation that, among other things, addresses the potential harms of very large online platforms (VLOPs) and very large online search engines (VLOSEs). It requires, for instance, that VLOPs and VLOSEs provide detailed reports regarding removed content, including reasons for removal [1]. For protecting minors, the European Commission has published specific guidelines under the DSA, including “Disabling by default features that contribute to excessive use, like communication “streaks,” ephemeral content, “read receipts,” autoplay, or push notifications, as well as removing persuasive design features aimed predominantly at engagement and putting safeguards around AI chatbots integrated into online platforms.” [2] In practice, enforcing the DSA can be challenging for the commission, since VLOPs and VLOSEs do not have their headquarters in Europe. In addition, the DSA still allows for cases where platforms can self-regulate. For instance, public data sharing with researchers has been a contentious topic, with platforms declining legitimate applications without providing reasoning or providing access only to limited data. Nevertheless, the DSA is a step in the right direction, it has already been enforced [3] with positive outcomes [4], and it can guide future policies worldwide. From the technical side, I do not see an implementation issue on the researchers call. However, the first item still expects that platforms will do a good job when performing harm assessments in new products or features before releasing them, and that they will share meaningful data on the effects on users’ health. Instead, I would recommend that the entire process (study, data, methods and evaluation) should be shared and/or conducted by vetted researchers. In other words, vetted researchers should be able to audit platforms’ decisions by conducting independent research and not only relying on the fidelity of the data provided by the platforms.”
[1] https://transparency.dsa.ec.europa.eu/
[2] http://digital-strategy.ec.europa.eu/en/library/commission-publishes-guidelines-protection-minors
[3] https://ec.europa.eu/commission/presscorner/detail/en/ip_25_2934
Prof Alan Woodward, Professor of Cybersecurity, University of Surrey, said:
“The press release is more cautious than most, and says clearly that the evidence is low-certainty and cannot establish cause and effect, but the phrase “emerging evidence of harm” still goes further than the underlying review, which states that its findings should not be interpreted as causal.
The umbrella review is a set of existing reviews, not new data. Seventy-one per cent of the 121 reviews it summarises were rated low or critically low quality, and most of the underlying studies are cross-sectional and rely on self-reported screen use. The odds ratios method they chose to use also make small effects look large: the figure linking problematic use to mental health problems corresponds to a correlation of about 0.2 which is much more modest.
The 2.8 million and 240,000 figures show what the burden would be if the association with screen time were entirely causal yet no experiment sits behind them, no uncertainty range is given, and the authors themselves say they should not be read as showing population-level harm. Depressed or less active young people may simply spend more time on screens.
“Problematic use” has no agreed definition, and the questionnaires that measure it overlap with the distress they are then correlated against. The strongest associations in the press release are therefore the ones most exposed to circularity.
On the technical proposals, platforms hold engagement logs and the results of their own experiments, but they do not generally hold health outcomes and often do not reliably know a user’s age. Data sharing alone will not settle the question. Independent access to randomised experiments on specific design features, linked to health outcomes measured separately, would. Risk assessments are already required of large platforms in the EU and UK and are largely self-assessed but pre-release checks will only work if regulators can audit code and data.
Calling for precautionary action is a legitimate position, but it is a value judgement, not a scientific finding. If governments do act, evaluation needs to be built in from the start, otherwise nobody will learn whether it worked.”
Dr Teodora Gliga, Professor of Developmental Cognitive Neuroscience, University of East Anglia, said:
“The conclusion that the ‘Scale of young people’s digital media use and emerging evidence of harm justify action now’ is both exaggerated and necessary.
“The umbrella review by Padmapriya and co-authors is carried to the best standards which means that we can confidently say that findings reflect the literature to date. This review finds consistent but small associations between digital media use and various mental and physical health outcomes, including poor sleep and neurodevelopmental conditions such as autism and AHDD. While the estimate of 2.8 million healthy years of life lost to depression is a large number, that corresponds to less than an hour per person, per year.
“This may well be un underestimate due to the very broad measures of digital media use – most studies included in the review only measure amount of time spent with screens or the content of what is watched might be more problematic.
“More importantly, the authors find that most of this evidence is observational rather than causal. This reviewed evidence does not tell us that screen time causes depression or autism/ADHD. The authors themselves caution against this interpretation since it remains possible that pre-existing characteristics like autistic traits or struggling to fall asleep, may lead to longer screen use.
“The problem is that obtaining causal evidence is a particularly challenging and slow process. Causal evidence requires intervention studies in which screen time is reduced (since the opposite would be ethically unacceptable) for a period of time that is long enough to see an impact. And this is why the leading conclusion, and the Lancet call for a public health approach to limiting exposure may be necessary. Digital media changes so fast that research, particularly the much needed but slower experimental research, becomes obsolete by the time a conclusion is reached. Yet, this research is critical, both for preventing harm caused by digital media use but also for preventing unnecessary anxiety or loss of potential benefits of such use.”
Dr Myles Jones, Senior Lecturer in the School of Psychology, University of Sheffield, said:
“The potential negative and positive outcomes for both physical and mental health resulting from digital media use in young people are of profound interest to the public, academics and practitioners. However, given that the term digital media encompasses an extremely wide range of online activities investigating its benefits and putative harm and can be difficult.
“Notwithstanding, Padmapriya et al., conduct a very thorough review of all the existing evidence (a review of reviews) and suggest that digital media use is generally associated with poorer physical and mental health although benefits are occasionally reported in certain situations, such as educational content with caregiver co-viewing.
“The fundamental problem is that most of the studies are cross-sectional (‘observational’) and thus it is unclear whether digital media use is the chicken or the egg. There are a good number of longitudinal reviews too, but some of these (e.g. Zhang et al., 2024) suggest a more complicated relationship with loneliness leading to more digital media use, but also vice-versa.
“The resulting three Lancet articles are essentially opinion pieces claiming that this umbrella review justifies ever greater regulation of digital media companies at the global level. Whether this should be the case, is a matter of public opinion.”
Dr Lizzy Winstone, Senior Research Associate in Applied Health Research, University of Bristol Centre for Public Health, said:
On the population harm estimates in paper 2
“The authors are right to urge caution in interpreting the large, illustrative population-level estimates. As noted by the authors, the underlying evidence is still largely observational and cannot establish that screen time itself causes depression or obesity.
“Although some types of digital media use may be harmful, others can support connection, wellbeing or physical activity. Consider, for example, the proliferation of online exercise classes, where screen time is actually facilitating physical activity. This also reinforces the recommendation from the authors (and others working in this field) to focus less on overall screen time, and more on understanding what young people are experiencing online, and how platform design and commercial practices shape those experiences.”
General comment on the Series:
“This Lancet Series promotes a welcome shift away from placing responsibility primarily on young people and support from their parents to manage their digital health, towards greater attention to the structural and commercial forces that shape what people encounter online.
“Digital literacy remains an essential tool in empowering individuals across age groups to flourish online, but its capacity to protect health is likely to be limited if we do not also address commercial practices such as targeted advertising and the algorithms that determine what content is repeatedly promoted to users. A public health approach therefore needs intervention at multiple levels, with particular attention to health inequalities.
“As a research community, we need to better understand whether exposure to the opportunities and risks of digital environments and people’s resources to navigate them are evenly distributed. The Series highlights an important research agenda. We need to study technology companies as commercial determinants of health, including how their business models and practices shape exposure to health relevant content, and how companies respond and adapt to emerging regulation.”
Dr Hisham Al-Assam, Associate Professor in Computing, The University of Buckingham, said:
“From a cybersecurity research perspective, I think the papers make a strong case for why age verification needs to be taken seriously, but they are more convincing on the governance case than on the technical evidence that these systems will actually work safely in practice.
“They rightly identify the weaknesses of the current 13+ threshold and the growing move towards more sophisticated age-assurance systems. They also recognise the central tension: a system designed to protect children must not create new privacy, security or exclusion risks. But the papers provide very limited empirical evidence on whether privacy-preserving age-verification technologies can reliably establish age, resist circumvention and abuse, and protect sensitive personal data at scale.
“That distinction is important. From a cybersecurity standpoint, it is not enough to say that a system is “privacy-preserving” in principle. We need evidence from real-world deployment, threat modelling, independent security testing and auditing to understand how it performs against determined attackers and how data could be exposed or misused.
“So I would see the proposals as robust at the level of governance and policy, but not yet backed by enough technical and empirical evidence to establish that the proposed approaches are secure and effective in practice. The papers themselves acknowledge this evidence gap, which makes independent access to platform data and rigorous security evaluation particularly important.”
Paper 1 ‘The digital determinants of health: a new era for global public health’ by James Shaw et al. , Paper 2 ‘The public health challenges of children’s and adolescents’ digital media use: applying the digital determinants of health as an interpretive framework’ by Louise Holly et al. and Paper 3 ‘Governing the digital determinants of health for children and young people: a global public health imperative’ by et al. were published in The Lancet at 23:30 UK time on Sunday 4th October.
Declared interests
Dr Marco Bardus: I am an Assistant Professor in Public Health at the University of Birmingham. I currently lead an NIHR School for Public Health Research-funded project examining industry responses, digital environments, and health inequalities under UK social media restrictions (due to start in January 2027). I have no financial relationships with technology companies relevant to this topic.
I learned about this Lancet commission series when I met DTH Lab’s Research and Policy Coordinator (Louise Holly) a couple of weeks ago at a workshop organised by the Technology, Internet & Policy Group. Louise sits on our project’s Expert Advisory Group and is the lead author of Paper 2 and co-author of Paper 1.
Dr Carolina Scarton: She has no conflicts of interest.
Prof Alan Woodward: I have no conflicts of interest in making these comments.
Dr Teodora Gliga: No conflicts of interest.
Dr Myles Jones: No interests to declare
Dr Lizzy Winstone: I have been funded by the NIHR School for Public Health Research to conduct research on social media, algorithmic literacy and young people’s mental health. I received a Churchill Fellowship, supported by the Mercers’ Company, to explore approaches to improving social media algorithmic literacy. I have collaborated with the Molly Rose Foundation on research into social media and algorithmic literacy in secondary schools and have provided advice on the development of their algorithmic literacy resources. I have no financial relationship with the Molly Rose Foundation and no funding from or financial relationships with social media or technology companies.
Dr Hisham Al-Assam: “I have no conflicts of interest to declare.”
For all other experts, no reply to our request for DOIs was received.