A team led by the University of Calgary has published what may be the largest study ever conducted on children's screen time, synthesising data from more than 2.8 million children across 64 countries. The umbrella review — a study of studies — appeared in JAMA Pediatrics and delivers a headline finding that will rattle parents: the association between elevated screen exposure and poor developmental outcomes is "similar in magnitude to or larger than" the effects of family wealth, parenting behaviour, and premature birth. The domains of harm are broad. Higher screen use was linked to lower overall academic achievement, poorer language development, increased emotional and peer problems, and higher rates of both internalising behaviour (depression, anxiety symptoms) and externalising behaviour (aggression, ADHD symptoms). The researchers examined a range of devices — smartphones to televisions — and a range of activities including social media, gaming, and video calls. There is a conspicuous hole in the subject-level data. When researchers drilled into specific academic subjects, screen time showed no significant link to maths, reading, or science performance. The authors attributed this to a lack of evidence — those subjects were "among the least studied" — rather than to screen time being benign in those areas. That gap matters. An umbrella review is only as strong as the underlying evidence it synthesises. Critics moved quickly to frame the limits. Tatiana Íñiguez Berrozpe of the University of Zaragoza called the association real but small, and said the findings "in no way" support the claim that screens are the main driver of adolescent distress. Esther van Sluijs of Loughborough University praised the study's quality but flagged a critical weakness: the review could not differentiate between types of screen use. Lumping passive TikTok scrolling with an educational Zoom call produces a blunt instrument. This is the persistent problem in screen-time research. Association is not causation, and umbrella reviews amplify existing data — they do not generate new causal evidence. Children who spend more time on screens may already be experiencing the conditions (poverty, poor parenting, social isolation) that independently drive poor outcomes. Until longitudinal, causally designed studies control for these confounders at scale, the field will keep producing alarming correlations that resist actionable interpretation. The policy stakes are real regardless. The UK is actively considering time limits, curfews, or outright bans on social media for under-16s. This study will be cited in those debates — its sheer scale (2.8 million children, 64 countries) makes it politically useful even if methodologically limited. The question is whether regulators will act on the association or wait for causal proof that may never arrive in clean form. What the study does accomplish is normalising the comparison. Putting screen time alongside socioeconomic status and preterm birth as a predictor of developmental outcomes — even an associational one — reframes it from lifestyle choice to public health variable. Whether that framing survives scrutiny will depend on the next generation of studies that can actually isolate mechanism from correlation.