Parent coaching cut young kids' screen time 28 minutes a day

It is late on an ordinary Tuesday in a house where the television went off an hour ago and a tablet did not. The parent standing in the doorway has already tried the thing every pamphlet recommends, which is a number. Two hours. One hour. Nothing after seven. The number is easy to say and, in most households, remarkably hard to make survive contact with a school night — and the guidance that hands it over rarely says what happens if the rule holds. It asserts a threshold and then goes quiet on the only question a tired parent actually has, which is whether any of this effort registers.

That gap is what a run of randomized trials has been working on. The newest of them, published in July 2026, is the one behind the week's headlines about screen limits in children. Two earlier trials, both from 2024, carry the claim for children generally. Between them they take up a narrower and more useful question than how much is too much: when a parent is coached, in a structured way, to cut screen time, does the screen time fall — and does anything else move with it.

The 2026 trial, by Kumari and colleagues in The Primary Care Companion for CNS Disorders, randomized 60 children aged 4 to 12 who were attending a child and adolescent psychiatry outpatient department at a tertiary hospital in North India, all of them assessed as having borderline intellectual functioning. Parents in the intervention arm received a three-session module delivered online, each session 50 to 60 minutes, covering the effects of heavy screen use, general guidelines for regulating it, the technical controls available on the devices themselves, and behavioural strategies for the moments when a screen goes off. Outcomes were measured six weeks later using a 24-hour screen diary, the Children's Sleep Habits Questionnaire, the Strengths and Difficulties Questionnaire, and a standardized physical-activity questionnaire. It is a small trial in a specific clinical population, and its numbers belong to that population rather than to children at large.

The two 2024 trials are the ones that speak to ordinary households. Poonia and colleagues, in Indian Pediatrics, enrolled 120 healthy infants aged 9 to 10 months at an immunization clinic in Delhi. Caregivers were randomized either to 30 minutes of in-person counselling built around reducing screen exposure, with a printed pamphlet to take home, or to routine counselling on general health, with monthly telephone reinforcement in both arms across six months. Kaur and colleagues, in Frontiers in Public Health, randomized 340 families in Chandigarh with children aged 2 to 5 to a two-month programme that reworked the home media environment alongside separate parent and child modules, then followed them for six months with light contact over WhatsApp; the control families received routine services.

The Chandigarh trial reports the cleanest arithmetic. Immediately after the programme, children in the intervention arm averaged 102.6 minutes of screen time on a typical day against 130.3 in the control arm — a difference of 27.7 minutes, with a confidence interval running from 5.1 to 50.3. Over the full baseline-to-follow-up window, a generalized estimating equation put the reduction at 35.81 minutes (CI 70.6 to 1.04 fewer). Physical activity rose alongside it: 48.4 additional minutes immediately after the programme (CI 6.6 to 90.3) and 73.4 at follow-up (CI 36.2 to 110.5). The standard deviations are worth noticing — roughly 100 minutes in both arms — which is a statistical way of saying households differ from one another far more than the arms differ from each other.

The Delhi infant trial produced the widest separation of the three, and the reason is probably its timing. At six months, 3% of the counselled group (2 of 61) were above an hour of screen time a day, against 53% of controls (32 of 61), and median daily screen time was 35 minutes against 75. Those children were 9 to 10 months old at enrolment — young enough that the counselling was preventing a habit rather than reversing one. The authors also report significant differences in behavioural scores and in fine motor and adaptive skills, measured with the Ages and Stages questionnaires.

The 2026 trial reports its results as effect sizes rather than minutes: significant reductions in screen time (partial eta-squared 0.4), improvements in sleep quality (0.3) and physical activity (0.2), and fewer behavioural difficulties (0.2), each at P below .05. A partial eta-squared of 0.4 is large by convention, but it is a proportion of variance, not a quantity of time, and the published abstract does not say whether it corresponds to fifteen minutes saved or ninety. That distinction matters more to a household than the significance test does.

All three trials were open-label, and in all three the outcome was reported by the same parents who received the intervention. Nobody wore a meter. A parent who has just spent three sessions learning why screen time matters has both a sharper eye for it and a human reason to record improvement, and no analysis can separate a household that changed from a diary that did. This is the single largest caveat on every number above, and it is a structural feature of the design rather than an oversight by the authors.

The geography is narrow too: Delhi, Chandigarh, and a tertiary hospital in North India. Device ownership, household composition and what a screen is used for all vary enough across settings that an effect size does not travel automatically. And the windows are short — six weeks in the 2026 trial, six months in the other two. The Chandigarh follow-up estimate is instructive in its own right: its interval reaches to within about one minute of no effect at all, which is a real result and a thin one at the same time.

Most screen-time guidance is a threshold in search of a trial. These three are the reverse: they leave the threshold alone and test the intervention, which is the part a household can actually be handed. And the intervention that worked was not resolve. It was structure — a few counselling sessions, a pamphlet, changes to where devices live in a home, a phone call once a month. Those are describable, repeatable things, and they moved roughly half an hour a day in populations that had not been selected for motivation.

The other quiet finding is about timing. The trial that started earliest, before any habit had formed, ended with the largest gap between its arms by a wide margin. Whether that reflects prevention being easier than reversal, or simply that infants have fewer competing routines, these trials cannot resolve. But it is the kind of result that reframes the question clinicians are asked. The evidence now says that coaching parents moves the number. It does not yet say what the moved number does to a child over the following decade, and the trials are honest about the difference.

The honest verdict is that this is a small, real effect measured well enough to believe and not nearly long enough to bank on. Three randomized trials agree that coaching parents in a structured way reduces young children's screen time, and the best-powered of them puts the size at about half an hour on a typical day, with more physical activity alongside it — which makes screen-time coaching one of the few pieces of parenting guidance that has been tested as an intervention rather than asserted as a threshold. Two things temper it. The first is that parents both delivered and measured the outcome in every trial, which is the weakest link in the chain and cannot be argued away. The second is that a reduction is not yet an outcome: no trial has followed these children long enough to say whether the recovered half-hour changes sleep, learning or mood in any way that persists. The most interesting result is the one the coverage skipped — the trial that began when its subjects were nine months old ended with 3% of the intervention group above an hour a day against 53% of controls, a gap far wider than anything achieved in older children. That asymmetry, if it holds up elsewhere, is a more useful finding than any hour limit.