Should paediatric intensive care be centralised? Trent versus Victoria

Summary Background The mortality rate is lower among children admitted to specialist paediatric intensive care units (ICUs) than among those admitted to mixed adult and paediatric units in non-tertiary hospitals. In the UK, however, few children receive intensive care in specialist paediatric units. We compared the ICU mortality rate in children from the area of the Trent Health Authority, UK, with the rate in children from Victoria, Australia, where paediatric intensive care is highly centralised. Methods We studied all children under 16 years of age from Trent and Victoria who received intensive care between April 1, 1994, and March 31, 1995. Children younger than 1 month were excluded unless they had cardiac disorders. We developed a logistic regression model that used information gathered at the time of admission to ICU to adjust for risk of mortality. Findings The rates of admission of children to intensive care were similar for Trent and Victoria (1·22 and 1·18 per 1000 children per year), but the mean duration of an ICU stay was 3·93 days for Trent children compared with 2·14 days for children from Victoria. 74 (7·3%) of the 1014 children from Trent died, compared with 60 (5·0%) of the 1194 children from Victoria. With adjustment for severity of illness at the time of admission to ICU, the odds ratio for the risk of death for Trent versus Victoria was 2·09 (95% CI 1·37–3·19, pp Interpretation If Trent is representative of the whole country, there are 453 (200–720) excess deaths a year in the UK that are probably due to suboptimal results from paediatric intensive care. If the ratio of paediatric ICUs to children were the same in the UK as in Victoria, there would be only 12 paediatric ICUs in the country. Our findings suggest that substantial reductions in mortality could be achieved if every UK child who needed endotracheal intubation for more than 12–24 h were admitted to one of 12 large specialist paediatric ICUs.