BackgroundShort-stay unplanned hospital admissions in children have much increased over the past decade everywhere. Timely primary care prevents unplanned admissions into hospital and short stays admissions can be regarded as a proxy indicator of conditions that could be dealt with in primary care. In April, 2004, major changes in primary care occurred across England, with the introduction of the new general practitioner (GP) contract. Out-of-hours health provision ceased to be the responsibility of GPs, and alternative services introduced to compensate were unpopular with patients. In-hours health-care provision also altered; the introduction of financial incentives for chronic disease management, having been aimed at adults, altered the availability of GP appointments for children. Reduced access to primary care in and out of hours could have delayed health seeking and potentially driven parents to use emergency care. We investigated the effect of these changes on children. MethodsWe applied a segmented population-based time trends study design, using Hospital Episode Statistics data, in children aged 0–14 years who were admitted in England between April 1, 2001, and March 31, 2004, and between April 1, 2004, and March 31, 2011. The outcome measures were age-specific unplanned admission rates for total and very short stays (same day discharge). The unit of analysis was middle super output area and we applied a generalised estimating equations model, adjusting for deprivation and autocorrelation. Indicator variables modelled a step change at and a trend change after service changes in April, 2004. Interaction coefficients estimated the effect modification of age on the baseline trend and point of intervention. FindingsAge-standardised rates increased by 15% between 2001–02 and 2010–11, from 64·7 (95% CI 64·6–64·9) to 74·3 (74·2–74·5) unplanned admissions per 1000 children. Segmented trends analysis showed a 1% step change (rate ratio [RR] 1·01, 95% CI 1·01–1·02) in the adjusted rate of all-cause unplanned admissions after the introduction of GP service changes in April, 2004, above the baseline trend of a 1% yearly increase in previous years (1·01, 1·01–1·01). The steepest age-specific baseline trends were in children aged less than 1 year, decreasing across older age bands (2% yearly increase in children under 1 year vs 1% in those aged 1–4 years). The largest step change in unplanned admission rates in 2004 was 5% in children aged 10–14 years (RR 1·05, 95% CI 1·04–1·07). Increases have been driven by very short stay rates, a yearly increase of 5% (1·05, 1·05–1·06) and a 4% step change (1·04, 1·03–1·05) in children aged under 15 years. The effect in 2004 ranged from no effect in children under 1 year (1·00, 0·99–1·01) to a 7% step change in children aged 5–9 years (1·07, 1·05–1·09). InterpretationWe noted a 5% increase in the rate of children admitted and discharged on the same day, coinciding with major reorganisation in primary care in 2004. The increase has not been uniform across age groups in children. Alternative explanations for the large increases in 2004 include the introduction of a 4-h waiting target in English emergency departments; changes in population demographics with European Union expansion; and the introduction of payment by results in English hospitals. There are limitations to routine hospital activity data and the observational study design allows only description of associations between exposure and outcome. However, Hospital Episode Statistics has been shown to be a useful method for exploring the effect of national policy change. FundingThis study is part of a National Institute for Health Research research project.